Factors Influencing Operating Speeds and Safety on Rural and Suburban Roads

5. SPEED EFFECTS OF OPTICAL SPEED BARS ON RURAL AND SUBURBAN ROADS

This section describes the OSB site selection criteria, design, site characteristics, and data collection and analysis methods used for evaluation.

Overview

OSBs are 18-inch-long and 12-inch-wide white transverse markings placed on both sides of the lane perpendicular to the centerline, edge line, or lane line in a pattern of progressively reduced spacing. The pattern gives drivers the impression that their speed is increasing. The intended outcome of the treatment is to reduce vehicle-operating speeds.

The initial and final spacing between the OSBs depends on the initial speed (approach speed) and desired speed in the curve (advisory speed). The length of roadway treated with the OSBs depends on the speed difference between the initial and final speed. FHWA recommends that drivers be in the OSB segment for at least 4 s.

OSBs can be painted on the roadway or a thermoplastic material is used to improve durability and increase longevity of the markings. OSBs are typically installed in school zones, at horizontal curve locations, or on tangent roadway segments. This project analyzed the effects of OSBs on vehicle operating speeds on horizontal curves using an observational before-after study. Figure 58 shows an example of an OSB treatment.

Source: KLS Engineering, LLC

Figure 58. Photo. OSB example.

SITE SELECTION

The research team used extensive outreach with several State and local transportation agencies to identify potential locations for installing OSB treatment on two-lane rural/suburban horizontal curves. The research team contacted more than 40 agencies to determine their willingness to install the treatment, 3 of which came forward to install or assist in installing the treatment for the study purposes. The three agencies are Massachusetts Southeastern Regional Planning and Economic Development District (SRPEDD), Mohave County (Arizona), and the Alabama Department of Transportation (ALDOT).

SRPEDD conducted road safety audits (RSA) in 2009 on eight roadways in suburban/rural areas of southeastern Massachusetts. The roadways were selected based on the number of lane-departure crashes resulting in injury or death. OSBs were one of the treatments recommended in the RSA, especially on the approaches to the many dangerous curves. Initially, the research team selected 11 curved segments on 6 different two-lane roadways for OSB consideration. Based on further discussions with SRPEDD and the counties, the initial list of 11 was narrowed to 8locations on 4 roadways. During site visit, one location was found not to have an edge line and was thus eliminated, bringing the total number of sites in Massachusetts to seven. Table 40 shows the Massachusetts OSB treatment site(s) characteristics and location details.

Mohave County previously experimented with OSB at one location (Stockton Hill Road between MM 21 and 20) in July 2007, and was familiar with the OSB considerations. Mohave County examined speed and crash data maintained for its regional highway network and recommended four two-lane rural roadways as candidates. The locations had been the sites of a significant number of crashes, 85th percentile speeds significantly above posted, or both. Table 40 shows the Arizona OSB treatment site(s) characteristics and location details.

ALDOT provided 10 two-way rural roadways for OSB consideration. The roadways were selected based on crash data, at least three crashes on average per year over the last 10 years, on the approach, within, or immediately following a horizontal curve. Five roadways were deemed inappropriate for study purposes (residential streets, vicinity of major intersection, etc.). Eight curve segments on five different roadways were selected for OSB installation. Table 40 shows the Alabama OSB treatment site(s) characteristics. The identifying location information (e.g., route, milepost, county name) is not listed at the counties’ request.

To summarize, a total of 19 treatment sites (7 in Massachusetts, 4 in Arizona, and 8 in Alabama) were selected for field evaluations.

Treatment Design

The research team used two different types of OSB designs, as discussed in the following sections.

Design for Massachusetts and Arizona Sites

This layout was based on an experimental design by Mohave County Public Works (MCPW), which yielded positive results. The MCPW optical speed zone contained three speed-bar patterns―downstream, transition, and upstream―designed to convey to road users a sensory perception of increased speed while traveling through the zone. The MCPW design used a driver perception time of .75 s. The MCPW design layout involved computing the downstream bar-pattern spacing by multiplying the site-measured 85th percentile speed by 1 s and upstream bar spacing by 1.5 s (targeting a 50-percent increase in spacing from downstream to upstream). The transition pattern provided an incremental reduction in bar spacing, from upstream to downstream pattern spacing distance, between successive bars to disguise the physical change in bar spacing from the passing road user. The design used a fixed set of five bars in the upstream and downstream patterns while the transition set contained four bars.

To normalize the effects of differing PSLs, the research team slightly modified the MCPW design as follows:

• Used four bars each in the downstream pattern and upstream pattern and three bars in the transition zone.
  
  
• The upstream set bar spacing was established at 1-s headway.
  
  
• The downstream set bar spacing was established at 0.6-s headway.
  
  
• The transition set consisted of four sequential bars spaced at 0.9-, 0.8-, 0.7-, and 0.6-s headway.
  
  
• All spacing calculations were rounded to the next 5 ft.
  
  

This approach is demonstrated below using an example 40 mph, 85th percentile speed:

• Upstream set spacing at 1-s headway = (40*5280/3600)*1 ≈ 60 ft.

• Total upstream set spacing for four bars = 4*60 = 240 ft.
  
  
• Transition Set.
  
  
• First bar spaced at 0.9-s headway ≈ 54 ft.
  
  
• Second bar spaced at 0.8-s headway ≈ 48 ft.
  
  
• Third bar spaced at 0.7-s headway ≈ 42 ft.
  
  
• Total transition set spacing = 54 ft +48 ft +42 ft = 144 ft.
  
  
• Total downstream set spacing at 0.6-s headway = 4*36 = 144 ft.
  
  
• Total treatment length = 240 ft + 144 ft + 144 ft = 528 ft.
  
  

Design for Alabama Sites

The design principle adopted for the Alabama sites was the same design used in studies by Katz and Arnold et al.^(56,75) This design methodology considered an initial speed and a desired ending speed at each location. Based on these speeds, the length of OSB treatment is determined based on deceleration from the initial to the ending speed, and the bars are spaced such that a driver decelerating at a constant rate from the initial speed to the ending speed crosses four bars per second. The equation shown in figure 59 is used to determine the required length of the OSB treatment, and the equation shown in figure 60, developed by Katz, is used to find the spacing of the optical speed bar throughout the treatment.⁽⁷⁶⁾ A frequency of four bars per second was adopted for OSB design.

Figure 59. Equation. Length of OSB treatment.

Where:

D = distance traveled in slowing from v₀ to v₁.
α = deceleration rate.
v₀ = initial speed at the beginning of the treatment.
v₁ = final speed.

'

Figure 60. Equation. Individual placement of the OSBs.

Where:

x = placement of the optical speed bars.
x₀ = initial placement of the first bar. The value of 𝑥𝑥0 is set to zero when a first bar is placed at the beginning of the treatment.
n = number of the optical speed bar for which the placement is determined.
f = required frequency of the bars, which is the number of OSBs seen in a second by motorists travelling through the treatment.

The dimensions of the OSB installed were in accordance with the 2009 Manual on Uniform Traffic Control Devices (MUTCD) (Part 3B.22) recommended guidelines. The markings were 12inches wide by 18 inches long, installed on both sides of the lane perpendicular to the center line and edge line.

Site Characteristics

The range of features of the 19 study sites was quite broad. The characteristics of the study sites included the following:

• A minimum radius of 236 ft and maximum radius of 1,865 ft.
  
  
• Curves to the left and to the right.
  
  
• Speed limits of 25 to 55 mph.
  
  
• Advisory speeds of 20 to 35 mph.
  
  

Table 40 summarizes the site characteristics of the 19 treatment sites. The Before Period Data Summary section describes specific characteristics of each treatment site.

Table 40. OSB treatment location site characteristics.

	City/County	Route Name/ Number	Curve Direction/ Grade	AADT	PSL (mph)	Lane Width (ft)	Lc (ft)	Approx.Radius (ft)
1	Dolan Springs/ Mohave County, AZ	Pierce Ferry Road (northbound)	Left/Uphill	750	55/50a	12	993	497
2	Golden Valley/ Mohave County, AZ	Shinarump Road (southbound)	Left/Level	450	45	12	887	1,865
3	Meadview/Mohave County, AZ	Diamond Bar Road (southbound)	Right/Level	470	45	12	875	1,067
4	Golden Shores/ Mohave County, AZ	County Route 1 (southbound)	Right/Uphill	850	35	13	332	1,095
5	Alabama location #1	N/A	Right/Level	1,050	55	10	666	673
6	Alabama location #2	N/A	Right/Level	1,065	55	11	303	273
7	Alabama location #3	N/A	Right/Level	2,770	55	12	546	635
8	Alabama location #4	N/A	Right/Uphill	1,570	35	10	278	236
9	Alabama location #5	N/A	Right/ Downhill	380	40	10	1060	710
10	Alabama location #6	N/A	Right/Level	390	40	10	885	710
11	Alabama location #7	N/A	Right/Level	1,275	35	10	457	602
12	Alabama location #8	N/A	Right/ Downhill	1,370	35	11	223	486
13	Dartmouth/Bristol County, MA	Tucker Road (southbound)	Right/Level	2,900	30	13	817	1,099
14	Dartmouth/Bristol County, MA	Tucker Road (northbound)	Left/Level	3,950	35	12	293	569
15	Dartmouth/Bristol County, MA	Reed Road (southbound)	Left/Level	5,450	25	12	459	683
16	Fairhaven/Bristol County, MA	New Boston Road (southbound)	Left/Level	750	35	11	226	596
17	Fairhaven/Bristol County, MA	New Boston Road (northbound)	Right/Level	750	35	11	186	471
18	Rochester /Plymouth County, MA	Braley Hill Road (southbound)	Right/Level	1,150	30	11	673	695
19	Rochester/Plymouth County, MA	Braley Hill Road (northbound)	Right/Level	1,050	40	10	662	2,097

^a-50 mph is nighttime PSL
AADT = Annual Average Daily Traffic
PSL = Posted Speed Limit
L_c = Length of Curve
Approx. = Approximate

Installation

In all three States, thermoplastic tape, applied with heat, was used to install the OSBs. Glass beads were added while the tape was being placed to increase visibility. Spacing of the bars was measured prior to their installation.

Installation at the four sites in Mohave County, AZ, occurred between October 25 and 30, 2012. Installation at the seven Massachusetts sites occurred between November 19 and 23, 2012. Each county installed the OSBs in their areas. The eight Alabama locations were installed between September 25 and October 1, 2013, and contractors performed the installation for ALDOT.

Operating Speed Evaluation Methodology

Data-Collection Procedures

The OSB treatments started on the tangent approaching the horizontal curves and ended near the beginning of the curve (PC station). An observational before-after study was employed to evaluate the OSB treatment. Speed data were collected before and after OSBs were applied to four sites in Arizona, eight sites in Alabama, and seven sites in Massachusetts. The before period data were collected prior to applying the OSBs. Two after-period data collection efforts were undertaken. The first after-period data were collected approximately 1 month after the OSBs were installed in Alabama, Arizona, and Massachusetts. The second after-period data were collected approximately 6 months after the first after period in Arizona and 3 months after the first after period in Alabama. There was no second after-period data collected at the Massachusetts sites. The second after period was performed to assess long-term novelty effects of the OSBs and to determine whether they became less effective in reducing vehicle operating speeds over time.

The research team used a control point on the same roadway to determine whether vehicle operating speeds remained constant at a location that was not treated to ensure that another factor was not influencing vehicle speeds. The control point locations for the OSB evaluation were points located between 0.2 and 0.8 mi upstream of the treatment site location, on a tangent segment. The control point location was far enough upstream of the treatment so drivers were unable to see the treatment from that location.

Data were collected using on-pavement traffic sensors. A total of six sensors, four in the OSB travel direction and two in the opposing lane, were placed to allow vehicles to be tracked throughout the study site, thus enabling the determination of speed changes for individual vehicles. The first sensor in the travel direction was placed at the control point. The second sensor in the travel direction was placed at the first transverse marking that delineated the OSBs, which was located on the curve approach. The third sensor in the travel direction was placed at the last transverse marking that delineated the OSBs, which was near the PC. The fourth sensor was placed at the curve midpoint to determine whether any speed reduction was maintained throughout the curve. Two sensors were placed in the opposing lane to determine whether the presence of a vehicle travelling in the opposite direction influenced driver speed choice.
Figure 61 shows the layout of the speed data-collection equipment in relation to the OSB treatment.

Figure 61. Diagram. OSB data-collection setup (not to scale).

Speed data were collected during both daytime and nighttime periods at each site. At each study site, the research team collected data during favorable driving conditions: clear weather with normal visibility (no fog) and dry roadway conditions, without the presence of standing water from an earlier rain. The data were screened to include only free-flow vehicles, defined as those vehicles traveling with a minimum headway of 5 s.^(70,71) The analysis included only passenger vehicles, which were defined as vehicles having a length less than 24 ft.

The research team attempted to obtain the recommended 110 free-flow passenger car speeds for both the daytime and nighttime periods at each data-collection site in accordance with the sample size requirement described in table 7 and table 8. The 110 free-flow passenger car speeds recommended for each data-collection period based on the equation in figure 20 would yield 95‑percent statistical significance and more than 90-percent statistical power, if a 2.0-mph difference in means was computed among the before and after samples. At sites where data were collected over 2 days/nights, and 1 day/night did not provide the recommended sample of free-flow passenger cars, a two-sample t-test was performed to compare the mean speeds between the 2days/nights. If there was no statistically significant difference between the 2 days/nights, then they were combined to provide more free-flow speed observations for analysis.

An important data-collection condition to note relates to possible measurement error from the on-road sensors at several data-collection sites in the OSB evaluation. Outliers in the speed data were identified using the following process:

1. Speeds remained relatively consistent from the control point to approach at most sites, as would be expected, so it was assumed the speeds at these two points were precise.

2. At some sites, the speeds at the PC appeared to be unrealistically high, while at other sites, the speeds at the curve midpoint appeared to be unrealistically high. Many observations had acceleration rates exceeding 10 ft/s², which was nearly equivalent to deceleration rates assumed in stopping sight distance geometric design criteria. It was expected that vehicles either maintain their speed throughout the curve or decelerate, but not accelerate from the PC to curve midpoint. Previous research by Hu and Donnell found the maximum deceleration rate of vehicles entering a curve was 4.4ft/s².⁽⁷⁷⁾ Using this previous study, any observation with an acceleration rate less than -4.5 ft/s or greater than 1.0 ft/s (to be conservative) from the approach to PC or from the approach to curve midpoint, was eliminated from the database.

3. While each vehicle speed was tracked from the control point location through the midpoint of the horizontal curve, in accordance with the data-collection setup shown in figure 61, the acceleration/deceleration rates between the approach point, PC location, and midcurve location were compared. When a point speed at the locations produced rates of acceleration/deceleration outside the limits identified in item 2 above, the research team excluded this individual point speed from the analysis.

Performance Measures and Analysis Methods

The performance measures used to assess the speed effects of the OSB treatment included the mean speed, difference in speeds between the beginning and end OSB speed measurement locations (delta v as described in the HFST Statistical Analysis section), speed variance, and proportion of vehicles exceeding the PSL. Details of the statistical analysis methods are the same as for HFST. The mean speed will be evaluated using the equation in figure 21. The proportion of vehicles exceeding the PSL will be assessed using the equations shown in figure 22 through figure 25, and the speed variance will be appraised using the equation shown in figure 26.

Before Period Data Summary

This section of the report describes the before period speeds at each data-collection location.

Arizona

This section describes the characteristics of the four treatment sites in Arizona.

Northbound Pierce Ferry Road, AZ

The research team collected data on Pierce Ferry Road in Meadview, AZ. The direction of travel for the data collection was northbound, and the curve direction was to the left. The radius of curve and the curve length, calculated using Google Earth™, were 497.40 ft and 992.78 ft, respectively. There was a substantial cut-slope on the inside of the curve, which limits horizontal sight distance along the curve. The daytime PSL was 55 mph, and nighttime PSL was 50 mph. As figure 62 shows, there was a curve warning sign (W1-2) with an advisory speed plaque of 35mph (W13-1P) located prior to the PC. The travel lanes were 12 ft wide, and there were no paved shoulders on either side of the road. However, there was a gravel shoulder on both sides of the road approximately 5 ft wide.

The team collected speed data in the before period at the control point 0.4 mi before the curve approach, which was 800 ft before the PC. The research team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 660 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 62 shows the horizontal curve layout and the speed data-collection locations.

Figure 62. Diagram. Geometric layout of northbound Pierce Ferry Road (not to scale).

Table 41 shows the speed data collected in the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed (a vehicle present in the opposing lane) versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each of the curve locations. For this treatment site, there was a statistically significant difference in speeds at the control point, approach, PC, and curve midpoint for passenger cars versus heavy vehicles. There was also a statistically significant difference in speeds at the control point, approach, and PC for daytime passenger cars versus nighttime passenger cars. The nighttime speed was significantly lower at every data-collection point.

Table 41. Northbound Pierce Ferry Road before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	48.532	7.764	46.165	8.162	46.432	6.817	38.056	6.841	218	176	214
Passenger cars	48.867	7.565	46.793	7.855	46.789	6.628	38.513	6.739	203	166	199
Heavy trucks	44.000	9.220	37.667	7.697	40.500	7.546	32.000	5.237	15	10	15
Daytime passenger cars	49.577	7.116	47.521	7.429	47.669	6.277	39.271	6.744	142	118	140
Nighttime passenger cars	47.213	8.347	45.098	8.594	44.625	7.028	36.712	6.430	61	48	59
Opposed passenger cars	47.806	6.122	46.444	8.833	47.621	7.683	38.029	6.671	36	29	35
Unopposed passenger cars	49.096	7.838	46.868	7.655	46.613	6.400	38.616	6.769	167	137	164

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 63 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 63 also shows the mean speed for trucks within the study section. The figure shows that mean speeds for both heavy trucks and passenger cars remain relatively stable from the approach to the PC, but decrease substantially from the PC to the midpoint of the curve. The heavy-truck mean speeds align more with the 15th percentile passenger car speeds. Both the passenger car and truck mean speeds are consistent with the advisory speed of 35 mph at the midpoint of the curve. The mean acceleration rate from the PC to the midpoint of the curve was -3.534 ft/s for passenger cars and -3.803 ft/s for trucks. A negative value indicates deceleration.

Figure 63. Graph. Graphical representation of speeds on northbound Pierce Ferry Road.

Southbound Shinarump Road, AZ

The research team collected data on Shinarump Road in Golden Valley, AZ. The direction of travel for the data collection was southbound, and the curve direction was to the left. The radius of curve and the curve length, calculated using Google Earth™, were 1,865.05 ft and 887.35 ft, respectively. The PSL was 45 mph. The travel lanes were 12 ft wide, and there was no shoulder on either side of the road.

The team collected speed data at the control point, 0.6 mi before the curve approach, which was 800 ft before the PC. The research team also collected speed data at the PC and midpoint of the curve. One sensor was placed in the opposing direction of travel at the midpoint of the curve, and the other was placed 618 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 64 shows the horizontal curve layout, along with the speed collection locations.

Figure 64. Diagram. Geometric layout of southbound Shinarump Road (not to scale).

Table 42 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each curve location. For this treatment site, there was a statistically significant difference in speeds at the PC location for passenger cars versus heavy vehicles. There was also a statistically significant difference at the control point and PC for daytime versus nighttime passenger cars. There was a statistically significant difference at the approach and PC locations for opposed versus unopposed passenger cars.

Table 42. Southbound Shinarump Road before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	47.317	6.928	51.099	8.603	47.299	6.864	51.100	8.042	303	288	209
Passenger cars	47.566	6.768	51.241	8.023	48.017	6.341	51.537	7.773	249	238	175
Heavy trucks	46.167	7.583	50.444	10.956	43.880	8.188	48.853	9.099	54	50	34
Daytime passenger cars	46.686	7.001	50.777	8.431	47.009	6.033	50.894	7.375	121	115	94
Nighttime passenger cars	48.398	6.458	51.680	7.624	48.959	6.500	52.284	8.193	128	123	81
Opposed passenger cars	45.967	6.646	48.200	5.580	45.933	4.982	50.875	6.556	30	30	24
Unopposed passenger cars	47.785	6.770	51.658	8.224	48.317	6.468	51.642	7.964	219	208	151

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 65 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 65 also shows the mean speed for trucks within the study section. As the figure shows, the patterns of the speed changes were similar for the passenger car speeds and the truck mean speeds on the approach to curve. The speeds for passenger cars and the truck mean speeds increased from the control point to the approach of the curve, decreased from the approach to the PC, and then increased again from the PC to the midpoint of the curve. The 85th percentile speeds and the mean speeds for passenger cars along the curve were higher than the PSL of 45 mph. The mean acceleration rate from the PC to the midpoint of the curve was 4 ft/s for passenger cars and 6 ft/s for trucks.

Figure 65. Graph. Graphical representation of speeds on southbound Shinarump Road.

Southbound Diamond Bar Road, AZ

The research team collected data on Diamond Bar Road in Meadview, AZ. The direction of travel for the data collection was in the southbound direction, and the curve direction was to the right. The radius of curve and the curve length, calculated using Google Earth™, were 1,067.25 ft 875.32 ft, respectively. There were desert plants on the inside of the curve, which limited horizontal sight distance along the curve. The PSL was 45 mph. As figure 66 shows, there was a curve warning sign (W1-2) located after the PC. The travel lanes were 12 ft wide, and there were 5-ft paved shoulders on both sides of the road.

The team collected speed data at the control point, 0.4 mi before the curve approach, which was 800 ft before the PC. Speed data were also collected at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 640 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 66 shows the horizontal curve layout, along with the speed collection locations.

The OSBs were installed differently at this site compared with the other sites. The OSB treatment started 400 ft upstream of the curve start, compared with beginning at the curve start at the other Arizona, Massachusetts, and Alabama sites. Because the treatment was offset from the curve PC location, an additional sensor was placed at the end of the OSB treatment. Figure 67 shows the layout of the sensors in the after periods.

Figure 66. Diagram. Geometric layout of southbound Diamond Bar Road in the before period (not to scale).

Figure 67. Diagram. Geometric layout of southbound Diamond Bar Road in the after periods (not to scale).

Table 43 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team again performed a simple t-test for each of these comparisons at each of the curve locations. For this treatment site, there was a statistically significant difference in speeds at the control point, approach, PC, and curve midpoint for passenger cars versus heavy vehicles. The heavy vehicle speed was lower than the passenger car speed at all the data-collection points. There also was a significant difference at the PC for daytime versus nighttime passenger cars.

Table 43. Southbound Diamond Bar Road before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	53.421	8.312	49.171	6.730	50.299	7.116	50.435	7.621	292	214	209
Passenger cars	53.867	8.389	49.516	6.675	50.874	6.996	50.703	7.611	256	182	182
Heavy trucks	50.250	7.064	46.722	6.704	47.031	7.014	48.630	7.581	36	32	27
Daytime passenger cars	54.061	8.582	49.873	7.057	51.706	7.113	51.109	7.876	181	126	128
Nighttime passenger cars	53.400	7.939	48.653	5.598	49.000	6.399	49.741	6.918	75	56	54
Opposed passenger cars	53.472	9.376	50.057	6.320	51.000	8.218	50.057	8.359	53	35	35
Unopposed passenger cars	53.970	8.134	49.374	6.773	50.844	6.705	50.857	7.445	203	147	147

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 68 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 68 also shows the mean speed for trucks within the study section. As the figure shows, the passenger car speeds and the truck mean speeds were relatively stable along the curve. The speed changes between points were minor. The truck mean speeds were consistent from the control point to the PC; however, the speeds increased slightly from the PC to midpoint of the curve. The 85th percentile speeds and mean speeds for passenger cars are higher than the PSL of 45 mph. The mean acceleration rate from the PC to the midpoint of the curve was 1 ft/s for passenger cars and 2.835 ft/s for trucks.

Figure 68. Graph. Graphical representation of speeds on southbound Diamond Bar Road.

Southbound County Route 1, AZ

The research team collected data on County Route 1 in Golden Shores, AZ. The direction of travel for the data collection was southbound, and the curve direction was to the right. The curve was located on a moderate upgrade. The radius of curve and the curve length, calculated using Google Earth, were 1,095.39 ft and 332.27 ft, respectively. The PSL was 35 mph. The travel lanes were 13 ft wide, and there was a 4- to 5-ft paved shoulder on both sides of the road.

The team collected speed data at the control point, 0.8 mi before the curve approach, which was 800ft before the PC. The research team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 732 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 69 shows the horizontal curve layout, along with the speed collection locations.

Figure 69. Diagram. Geometric layout of southbound County Route 1 (not to scale).

Table 44 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each curve location. For this treatment site, there was a statistically significant difference in speeds at the approach, PC, and midpoint for daytime versus nighttime passenger cars. There was also a significant difference at the midpoint for opposed versus unopposed passenger cars.

Table 44. Southbound County Route 1 before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	53.471	8.882	47.731	5.559	44.021	5.211	43.919	7.259	342	329	307
Passenger cars	53.532	8.901	47.810	5.515	44.134	5.125	43.918	7.262	327	314	292
Heavy trucks	52.133	8.651	46.000	6.414	41.667	6.532	43.933	7.450	15	15	15
Daytime passenger cars	53.847	9.266	49.038	5.430	44.977	4.895	44.599	7.081	183	174	162
Nighttime passenger cars	53.132	8.428	46.250	5.238	43.086	5.228	43.069	7.421	144	140	130
Opposed passenger cars	53.479	9.299	47.125	5.354	44.111	4.900	41.091	7.844	48	45	44
Unopposed passenger cars	53.541	8.848	47.928	5.543	44.138	5.171	44.419	7.053	279	269	248

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 70 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 70 also shows the mean speed for trucks within the study section. The figure shows the passenger car speeds decelerated substantially from the control point to the PC and then stabilized from the PC to the midpoint of the curve. The truck mean speeds decreased from the control point to the midpoint of the curve. The passenger car speeds and the truck mean speeds along the curve were higher than the PSL of 35mph. The mean acceleration rate from the PC to the midpoint of the curve was 0.581 ft/s for passenger cars and 0.668 ft/s for trucks.

Figure 70. Graph. Graphical representation of speeds on southbound County Route 1.

Alabama

This section describes the characteristics of the eight treatment sites in Alabama. ALDOT had an agreement with the counties that route information would not be disclosed. Thus, in the discussion below all eight Alabama locations are referenced numerically (1 through 8).

Alabama Location #1

The direction of travel for the data collection was northbound, and the curve direction was to the right. The radius of curve and the curve length, calculated using Google Earth™, were 672.84 ft and 665.84 ft, respectively. The PSL was 55 mph. As figure 71 shows, there was a winding road sign (W1-5) 365 ft before the curve approach. The travel lanes were 11 ft wide, and there were 2‑ft paved shoulders on both sides of the road. There were trees offset approximately 10 ft from the inside edge of pavement, which limited horizontal sight distance along the curve.

The research team collected speed data at the control point, 0.4 mi before the curve approach, which was 500 ft before the PC. The team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 515 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 71 shows the horizontal curve layout, along with the speed data-collection locations.

Figure 71. Diagram. Geometric layout of Alabama Location #1 (not to scale).

Table 45 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each of the curve locations. For this treatment site, there was a statistically significant difference in speeds at the midpoint for passenger cars versus heavy trucks. The only other statistically significant difference in speeds was at the PC for opposed versus unopposed passenger cars.

Table 45. Alabama Location #1 before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	47.883	9.368	49.208	7.596	50.136	8.725	42.323	6.912	375	88	353
Passenger cars	47.917	9.380	49.218	7.540	50.429	8.700	42.516	6.892	362	84	341
Heavy trucks	46.923	9.332	48.923	9.394	44.000	7.789	36.833	5.132	13	4	12
Daytime passenger cars	47.638	9.418	48.805	7.641	50.795	8.551	42.288	6.764	174	39	160
Nighttime passenger cars	48.176	9.362	49.601	7.444	50.111	8.912	42.718	7.016	188	45	181
Opposed passenger cars	47.529	10.661	48.793	7.385	46.833	7.648	42.037	6.512	87	24	82
Unopposed passenger cars	48.040	8.955	49.353	7.596	51.867	8.736	42.668	7.014	275	60	259

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 72 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 72 also shows the mean speed for trucks within the study section. The figure shows the speeds for passenger cars increased slightly from the control point to the PC and decreased substantially from the PC to the midpoint of the curve. The truck mean speeds were relatively consistent from the control point the PC and dropped substantially from the PC to the midpoint of the curve. The 85th percentile speeds for passenger cars at the control point, the approach, and the PC were higher than the PSL of 55 mph. Only at the midpoint of the curve were the 85th percentile speeds for passenger cars lower than the PSL of 55 mph. The mean speeds and 15th percentile speeds for passenger cars and the truck mean speeds along the curve were all lower than the PSL of 55mph. The mean acceleration rate from the PC to the midpoint of the curve was -8.726 ft/s for passenger cars and -6.304 ft/s for trucks. A negative value indicates deceleration.

Figure 72. Graph. Graphical representation of speeds at Alabama Location #1.

Alabama Location #2

This curve was located approximately 0.4 mi upstream of the Alabama Location #1 curve. The direction of travel for the data collection was southbound, and the curve direction was to the right. The curve approach and curve were located on a downgrade. The radius of curve and the curve length, calculated using Google Earth™, were 272.92 ft and 303.38 ft. respectively. There was a substantial cut slope with trees on the inside of the curve, which limited horizontal sight distance along the curve. The PSL was 55 mph. As figure 73 shows, there was a winding road sign (W1-5) with an advisory speed plaque (W13-1P) of 30 mph 175 ft before the PC. The travel lanes were 11 ft wide, and there were 2-ft paved shoulders on both sides of the road. There was a drainage ditch located on the inside of the curve and guardrail on the outside of the curve with chevron markers.

The research team collected speed data at the control point, 800 ft before the curve approach, which was 500 ft before the PC. The team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 440 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 73 shows the horizontal curve layout, along with the speed data-collection locations.

Figure 73. Diagram. Geometric layout of Alabama Location #2 (not to scale).

Table 46 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each of the curve locations. For this treatment site, there was a statistically significant difference in speeds at the control point and approach for passenger cars versus heavy vehicles. There was also a statistically significant difference in speeds at the control point, approach, and curve midpoint for unopposed passenger cars versus opposed passenger cars.

Table 46. Alabama Location #2 before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	49.269	7.604	44.291	5.822	42.472	5.747	38.337	7.101	398	290	359
Passenger cars	49.569	7.238	44.479	5.623	42.630	5.554	38.418	7.137	376	273	337
Heavy trucks	44.136	11.319	41.091	8.053	39.941	8.050	37.091	6.560	22	17	22
Daytime passenger cars	49.884	8.070	45.072	5.722	43.263	5.737	37.821	7.802	138	99	123
Nighttime passenger cars	49.387	6.721	44.134	5.548	42.270	5.432	38.762	6.720	238	174	214
Opposed passenger cars	47.809	7.949	43.022	6.312	41.721	5.057	36.378	7.241	89	61	82
Unopposed passenger cars	50.115	6.928	44.930	5.323	42.892	5.674	39.075	6.991	287	212	255

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 74 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 74 also shows the mean speed for trucks within the study section. The figure shows the patterns of the speed changes were similar for the passenger car speeds and the truck mean speeds at this curve location. The speeds for passenger cars and the truck mean speeds decreased consistently from the control point to the midpoint of the curve. The speeds for passenger cars and the truck mean speeds along the curve were lower than the PSL of 55 mph but higher than the advisory speed of 30 mph. The mean acceleration rate from the PC to the midpoint of the curve was -2.128 ft/s for passenger cars and -2.211 ft/s for trucks. A negative value indicates deceleration.

Figure 74. Graph. Graphical representation of speeds at Alabama Location #2.

Alabama Location #3

The direction of travel for the data collection was southbound, and the curve direction was to the right. This curve to the right was part of a reverse curve, with a curve back to the left immediately after. The radius of curve and the curve length, calculated using Google Earth™, were 635.15 ft and 545.62 ft, respectively. The PSL was 55 mph. As figure 75 shows, there was a winding road sign (W1-5) with an advisory speed plaque (W13-1P) of 35 mph positioned 460ft before the PC; there were also chevrons on the outside of the curve. The travel lanes were 12 ft wide, and there were 2-ft paved shoulders on either side of the road.

The research team collected speed data at the control point, 0.2 mi before the curve approach, which was 500 ft before the PC. The team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 500 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 75 shows the horizontal curve layout, along with the speed data-collection locations.

Figure 75. Diagram. Geometric layout of Alabama Location #3 (not to scale).

Table 47 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each of the curve locations. For this treatment site, there was a statistically significant difference in speeds at the approach, PC, and the curve midpoint for passenger cars versus heavy vehicles. There was also a statistically significant difference in speeds at the approach for daytime passenger cars versus nighttime passenger cars.

Table 47. Alabama Location #3 before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	65.122	13.222	49.842	6.461	41.329	4.665	41.192	5.936	499	474	474
Passenger cars	65.372	13.359	50.011	6.519	41.652	4.653	41.691	5.714	441	417	417
Heavy trucks	63.224	12.070	48.552	5.891	38.965	4.066	37.544	6.299	58	57	57
Daytime passenger cars	66.174	13.109	50.566	6.965	41.960	4.681	41.609	6.025	265	248	253
Nighttime passenger cars	64.165	13.675	49.176	5.702	41.201	4.589	41.817	5.214	176	169	164
Opposed passenger cars	65.516	13.031	49.979	6.684	41.449	5.061	41.928	6.036	188	176	180
Unopposed passenger cars	65.265	13.622	50.036	6.408	41.801	4.336	41.511	5.463	253	241	237

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 76 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 76 also shows the mean speed for trucks within the study section. As the figure shows, the passenger car speeds and truck mean speeds decelerated substantially from the control point to the PC. Both passenger car speeds and the truck mean speeds then stabilized from the PC to midpoint of the curve. The passenger car speeds and the truck mean speeds along the curve were all higher than the advisory speed of 35mph. The mean acceleration rate from the PC to the midpoint of the curve was 0.321ft/s for passenger cars. The truck acceleration rate remained the same from the PC to the midpoint.

Figure 76. Graph. Graphical representation of speeds at Alabama Location #3.

Alabama Location #4

The direction of travel for the data collection was southbound, and the curve direction was to the right. The radius of curve and the curve length, calculated using Google Earth™, were 236.38 ft and 277.70 ft, respectively. There was a substantial cut slope on the inside of the curve, which limited horizontal sight distance along the curve. There was guardrail and also chevrons on the outside of the curve. The PSL was 35 mph. As Figure 77 shows, there was a turn warning sign (W1-1) with an advisory speed plaque (W13-1P) of 20 mph and flashing amber beacons above the sign located at the curve approach. The travel lanes were 10 ft wide, and there was no shoulder on either side of the road.

The research team collected speed data at the control point, 0.3 mi before the curve approach, which was 500ft before the PC. The team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve. The other one was placed 440 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 77 shows the horizontal curve layout, along with the speed data-collection locations.

\

Figure 77. Diagram. Geometric layout of Alabama Location #4 (not to scale).

Table 48 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each of the curve locations. For this treatment site, there was only a statistically significant difference in speeds at the curve midpoint for passenger cars versus heavy trucks and daytime versus nighttime passenger cars.

Table 48. Alabama Location #4 before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	50.785	9.441	46.478	7.630	39.042	4.961	34.213	9.051	502	478	461
Passenger cars	50.660	9.345	46.506	7.597	39.100	4.908	34.038	9.068	480	459	443
Heavy trucks	53.500	11.237	45.864	8.481	37.632	6.103	38.500	7.649	22	19	18
Daytime passenger cars	50.656	8.996	46.149	8.725	39.462	4.987	32.726	10.008	195	182	179
Nighttime passenger cars	50.663	9.592	46.751	6.723	38.863	4.849	34.928	8.273	285	277	264
Opposed passenger cars	51.007	9.307	46.188	9.237	39.053	5.305	33.523	9.955	144	132	128
Unopposed passenger cars	50.512	9.371	46.643	6.786	39.119	4.746	34.248	8.690	336	327	315

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 78 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 78 also shows the mean speed for passenger cars and trucks within the study section. The figure shows the passenger car speeds and truck speeds decelerated substantially from the approach to the midpoint of the curve. The passenger car speeds and the truck mean speeds along the curve were closer to the 35 mph PSL than the advisory speed of 20 mph. The mean acceleration rate from the PC to the midpoint of the curve was -0.931 ft/s for passenger cars and -0.183 ft/s for trucks. A negative value indicates deceleration.

Figure 78. Graph. Graphical representation of speeds at Alabama Location #4.

Alabama Location #5

The direction of travel for the data collection was southbound, and the curve direction was to the right. The curve approach and curve were located on a downgrade. This curve to the right was part of a reverse curve, with the site for Alabama Location #6 located immediately downstream. The radius of curve and the curve length, calculated using Google Earth™, were 709.81 ft and 1,059.71 ft, respectively. The PSL was 40 mph. The travel lanes were 9 ft wide, and there was no shoulder on either side of the road.

The research team collected speed data at the control point, 1,000 ft before the curve approach, which was 250 ft before the PC. The team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 455 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 79 shows the horizontal curve layout, along with the speed data-collection locations.

Figure 79. Diagram. Geometric layout of Alabama Location #5 (not to scale).

Table 49 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each of the curve locations. For this treatment site, there was a statistically significant difference in speeds at the control point, approach, and the curve midpoint for daytime versus nighttime passenger cars. There was also a statistically significant difference in speeds at the PC and midpoint of the curve for opposed versus unopposed passenger cars.

Table 49. Alabama Location #5 before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	41.944	7.212	43.742	8.327	45.781	7.524	39.810	6.842	233	64	221
Passenger cars	42.299	7.094	44.054	8.250	45.794	7.584	40.028	6.782	224	63	212
Heavy trucks	33.111	3.822	36.000	6.614	45.000	N/A	34.667	6.557	9	1	9
Daytime passenger cars	41.947	6.887	44.405	7.941	45.619	7.450	39.787	6.772	131	42	122
Nighttime passenger cars	42.796	7.383	43.559	8.686	46.143	8.021	40.356	6.821	93	21	90
Opposed passenger cars	40.235	7.242	41.353	11.045	38.833	7.333	34.933	7.045	17	6	15
Unopposed passenger cars	42.469	7.072	44.275	7.972	46.526	7.290	40.416	6.621	207	57	197

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint
N/A = Not Applicable

Figure 80 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 80 also shows the mean speed for trucks within the study section. As the figure shows, the speeds for passenger cars and the truck mean speeds gradually increased from the control point to the PC of the curve and decreased from the PC to the midpoint of the curve. The 85th percentile speeds of passenger cars along the curve were higher than the PSL of 40 mph. The mean acceleration rate from the PC to the midpoint of the curve was -7.817 ft/s for passenger cars and -0.369 ft/s for trucks. A negative value indicates deceleration.

Figure 80. Graph. Graphical representation of speeds at Alabama Location #5.

Alabama Location #6

This curve to the right was part of a reverse curve associated with the site for Alabama Location #5 located immediately upstream. The direction of travel for the data collection was northbound, and the curve direction was to the right. The radius of curve and the curve length, calculated using Google Earth™, were 710.84 ft and 885.08 ft, respectively. At the midpoint of the curve, the curve starts to go upgrade with a driveway located on the outside of the curve. The PSL was 40 mph. The travel lanes were approximately 9 ft wide, and there was no shoulder on either side of the road.

The research team collected speed data at the control point, 750 ft before the curve approach, which was 500ft before the PC. The team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 490 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 81 shows the horizontal curve layout, along with the speed data-collection locations.

Figure 81. Diagram. Geometric layout of Alabama Location #6 (not to scale).

Table 50 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each curve location. For this treatment site, there was a statistically significant difference in speeds at the approach for passenger cars versus heavy trucks. There was also a statistically significant difference at the PC for opposed versus unopposed passenger cars.

Table 50. Alabama Location #6 before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	43.386	9.196	45.081	7.189	46.442	6.431	45.031	8.436	197	86	98
Passenger cars	43.370	9.277	45.402	6.928	46.447	6.469	45.128	8.568	189	85	94
Heavy trucks	43.750	7.498	37.500	9.457	46.000	N/A	42.750	4.272	8	1	4
Daytime passenger cars	43.462	9.112	45.400	6.716	46.586	6.489	44.317	9.561	130	58	63
Nighttime passenger cars	43.169	9.708	45.407	7.433	46.148	6.538	46.774	5.869	59	27	31
Opposed passenger cars	41.056	7.565	42.944	10.067	43.400	4.881	46.750	4.070	18	10	12
Unopposed passenger cars	43.614	9.425	45.661	6.499	46.853	6.571	44.890	9.032	171	75	82

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint
N/A = Not Applicable

Figure 82 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 82 also shows the mean speed for trucks within the study section. The figure shows that the 85th percentile and the mean speeds for passenger cars remain relatively stable from the control point to the midpoint of the curve. The 15th percentile speeds for passenger cars increased substantially from the control point to the approach of the curve and stabilized from the approach to the midpoint of the curve. The truck mean speeds were not consistent; the speeds decreased from the control point to the approach of the curve and increased from the approach to the PC and then decreased from the PC to the midpoint of the curve. The 85th percentile speeds and the mean speeds for passenger and the truck mean speeds along the curve were all higher than the PSL of 40 mph. The mean acceleration rate from the PC to the midpoint of the curve was 0.627 ft/s for passenger cars and -1.038 ft/s for trucks. A negative value indicates deceleration.

Figure 82. Graph. Graphical representation of speeds at Alabama Location #6.

Alabama Location #7

This curve to the right was part of a reverse curve, with a curve back to the left immediately after. The direction of travel for the data collection was southbound, and the curve direction was to the right. The white edge lines and yellow centerline were severely faded on the tangent and throughout the curve. The radius of curve and the curve length, calculated using Google Earth™, were 601.93 ft and 456.89 ft, respectively. The PSL was 35 mph. The travel lanes were 10 ft wide, and there was no shoulder on either side of the road. There was a driveway located on the right side of the road approximately 100 ft before the PC. As figure 83 shows, there was a SCHOOL BUS STOP AHEAD sign (S3-1), with a REDUCED SPEED AHEAD sign (R2-5) above it, located on the curve approach. There was also a winding road sign (W1-5) with a speed limit sign (R2-1) of 35 mph 250 ft before the PC.

The research team collected speed data at the control point, 0.3 mi before the curve approach, which was 500 ft before the PC. The team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 450 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 83 shows the horizontal curve layout, along with the speed data-collection locations.

Figure 83. Diagram. Geometric layout of Alabama Location #7 (not to scale).

Table 51 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team also conducted simple t-tests for each of these comparisons at each curve location. For this treatment site, there was a statistically significant difference in speeds at the control point for passenger cars versus heavy vehicles. There was also a statistically significant difference in speeds at the approach and PC for daytime versus nighttime passenger cars.

Table 51. Alabama Location #7 before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	52.977	9.890	43.666	5.396	36.953	4.288	38.806	5.266	431	423	422
Passenger cars	52.850	9.896	43.653	5.388	36.973	4.305	38.755	5.239	421	413	412
Heavy trucks	58.300	8.433	44.200	6.033	36.100	3.604	40.900	6.208	10	10	10
Daytime passenger cars	53.000	9.668	44.420	5.224	37.647	4.536	39.188	5.310	193	190	191
Nighttime passenger cars	52.724	10.104	43.004	5.450	36.399	4.020	38.380	5.160	228	223	221
Opposed passenger cars	52.148	8.668	43.241	5.701	36.657	4.310	38.355	4.989	108	105	107
Unopposed passenger cars	53.093	10.287	43.796	5.277	37.081	4.306	38.895	5.325	313	308	305

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 84 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 84 also shows the mean speed for trucks within the study section. As the figure shows, the passenger car speeds and the truck mean speeds decelerated substantially from the control point to the PC of the curve. Passenger car speeds and the truck mean speeds remained relatively constant from the PC through the curve, averaging between 35 and 40 mph. The 85th percentile speeds and mean speeds of passenger cars and the truck mean speeds along the curve were higher than the PSL of 35 mph. The mean acceleration rate from the PC to the midpoint of the curve was 0.885 ft/s for passenger cars and 0.713 ft/s for trucks.

Figure 84. Graph. Graphical representation of speeds at Alabama Location #7.

Alabama Location #8

This curve was located approximately 1,000 ft downstream of Alabama Location #7 in the opposite direction. The direction of travel for the data collection was northbound, and the curve direction was to the right. There was a cut slope on the inside of the curve, which partially restricted horizontal sight distance entering the curve. The white edge lines and yellow centerline were severely faded on the tangent and throughout the curve. The radius of curve and the curve length, calculated using Google Earth, were 486.26 ft and 222.95 ft, respectively. The PSL was 35 mph. The travel lanes were 10 ft wide and there was no shoulder on either side of the road. There was a driveway on the outside of the curve at the midpoint. As figure 85 shows, there was a winding road sign (W1-5) with a speed limit sign (R2-1) of 35 mph located 700 ft before the curve approach. The end of a 280 ft concrete bridge crossing a ravine ended approximately 180ft before the PC.

The research team collected speed data at the control point, 1,500 ft before the curve approach, which was 500 ft before the PC. The team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, The other one was placed 415 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 85 shows the horizontal curve layout, along with the speed data-collection locations.

Figure 85. Diagram. Geometric layout of Alabama Location #8 (not to scale).

Table 52 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team again performed a simple t-test for each of these comparisons at each curve location. For this treatment site, there was a statistically significant difference in speeds at the PC and curve midpoint for passenger cars versus heavy vehicles. There was also a statistically significant difference in speeds at the control point for daytime versus nighttime passenger cars. There was also a statistically significant difference in speeds at the approach for opposed versus unopposed passenger cars.

Table 52. Alabama Location #8 before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	41.540	6.029	44.644	8.437	40.661	5.324	42.266	6.329	376	342	312
Passenger cars	41.566	6.053	44.672	8.317	40.754	5.326	42.375	6.337	366	334	304
Heavy trucks	40.600	5.254	43.600	12.624	36.750	3.655	38.125	4.612	10	8	8
Daytime passenger cars	43.145	5.833	44.612	8.251	40.842	4.928	42.401	5.908	165	146	137
Nighttime passenger cars	40.269	5.934	44.721	8.391	40.686	5.628	42.353	6.686	201	188	167
Opposed passenger cars	41.750	6.708	42.841	7.546	40.823	6.354	42.000	7.255	88	79	71
Unopposed passenger cars	41.507	5.842	45.252	8.477	40.733	4.979	42.489	6.042	278	255	233

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 86 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 86 also shows the mean speed for trucks within the study section. As the figure shows, the patterns of the speed changes were similar for the passenger car speeds and the truck mean speeds at this curve location. The speeds for passenger cars and the truck mean speeds increased from the control point to the approach of the curve, decreased from the approach to the PC of the curve, and then increased again from the PC to the midpoint of the curve. The speeds of passenger cars and the truck mean speeds along the curve were both higher than the PSL of 35 mph. The mean acceleration rate from the PC to the midpoint of the curve was 1.94 ft/s for passenger cars and -0.153 ft/s for trucks. A negative value indicates deceleration.

Figure 86. Graph. Graphical representation of speeds at Alabama Location #8.

Massachusetts

This section describes the characteristics of the seven treatment sites in Massachusetts.

Southbound Tucker Road, MA

The research team collected data on Tucker Road in Dartmouth, MA. The direction of travel for the data collection was southbound, and the curve direction was to the right. The radius of curve and the curve length, calculated using Google Earth™, were 1,098.66 ft and 816.67 ft, respectively. The PSL was 30 mph. The travel lanes were 13 ft wide, and there was no shoulder on either side of the road.

The team collected speed data at the control point, 0.2 mi before the curve approach, which was 319 ft before the PC. The research team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 347 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 87 shows the horizontal curve layout, along with the speed data-collection locations.

Figure 87. Diagram. Geometric layout of southbound Tucker Road (not to scale).

Table 53 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each of the curve locations. For this treatment site, there was a statistically significant difference in speeds at the control point and PC for passenger cars versus heavy trucks. There also was a statistically significant difference in speeds at the control point for daytime versus nighttime passenger cars, and also at the approach for opposed versus unopposed passenger cars.

Table 53. Southbound Tucker Road before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	44.458	7.485	36.492	5.228	35.223	4.467	35.329	5.014	577	539	514
Passenger cars	44.466	7.510	36.506	5.236	35.241	4.475	35.334	5.022	573	535	512
Heavy trucks	43.250	0.957	34.500	3.873	32.750	2.500	34.000	1.414	4	4	2
Daytime passenger cars	45.136	6.983	36.782	5.064	35.488	4.127	35.420	4.809	316	293	286
Nighttime passenger cars	43.642	8.048	36.167	5.431	34.942	4.855	35.226	5.289	257	242	226
Opposed passenger cars	44.306	7.210	36.972	5.543	35.272	3.792	35.036	5.203	252	235	224
Unopposed passenger cars	44.592	7.745	36.140	4.960	35.217	4.951	35.566	4.874	321	300	288

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 88 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 88 also shows the mean speed for trucks within the study section. As the figure shows, the passenger car speeds and truck speeds decelerated substantially from the control point to the approach of the curve. Both passenger car speeds and the truck mean speeds were relatively stable with minimal speed changes between the approach and the midpoint of the curve. The mean acceleration rate from the PC to the midpoint of the curve was 0.604 ft/s for passenger cars and 2.7 ft/s for trucks.

Figure 88. Graph. Graphical representation of speeds on southbound Tucker Road.

Northbound Tucker Road, MA

This curve was located approximately 0.9 mi upstream of the southbound Tucker Road curve. The direction of travel for the data collection was northbound, and the curve direction was to the left. This curve to the left was part of a reverse curve, with a curve back to the right. The radius of curve and the curve length, calculated using Google Earth™, were 568.76 ft and 293.24 ft, respectively. The PSL was 35 mph. The travel lanes were 12 ft wide, and there was no shoulder on either side of the road. There was a winding road sign (W1-5) located approximately 125 ft before the PC. An advisory speed plaque (W13-1P) of 25 mph was located approximately 100 ft upstream of the winding road sign. As figure 89 shows, there was one chevron on the outside of the curve at approximately the curve midpoint.

The research team collected speed data at the control point, 0.4 mi before the curve approach, which was 270 ft before the PC. The team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 262 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 89 shows the horizontal curve layout, along with the speed data-collection locations.

Figure 89. Diagram. Geometric layout of northbound Tucker Road (not to scale).

Table 54 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each of the curve locations. For this treatment site, there was a statistically significant difference in speeds at the control point, approach, and PC for passenger cars versus heavy trucks, with the heavy truck speed lower at these points. There also was a statistically significant difference in speeds at the PC and curve midpoint for daytime versus nighttime passenger cars. There also was a difference at the curve midpoint for opposed versus unopposed passenger cars.

Table 54. Northbound Tucker Road before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	33.895	4.361	37.609	4.247	33.381	3.820	35.507	4.852	573	565	458
Passenger cars	33.919	4.361	37.633	4.251	33.403	3.820	35.525	4.858	569	561	455
Heavy trucks	30.500	3.000	34.250	1.500	30.250	2.500	32.667	3.055	4	4	3
Daytime passenger cars	34.135	4.263	37.768	4.219	33.683	3.551	36.145	4.549	341	338	276
Nighttime passenger cars	33.596	4.495	37.430	4.300	32.978	4.167	34.570	5.168	228	223	179
Opposed passenger cars	33.681	4.172	37.453	4.207	33.369	3.635	35.839	4.758	351	347	280
Unopposed passenger cars	34.303	4.634	37.922	4.315	33.458	4.111	35.023	4.986	218	214	175

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 90 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 90 also shows the mean speed for trucks within the study section. As the figure shows, the patterns of the speed changes were similar for the passenger car speeds and the truck mean speeds at this curve location. The speeds for passenger cars and the truck mean speeds increased from the control point to the approach of the curve, decreased from the approach to the PC of the curve, and then increasing again from the PC to the midpoint of the curve. The speeds for passenger cars and the truck mean speeds along the curve were both higher than the PSL of 30 mph. The mean acceleration rate from the PC to the midpoint of the curve was 2.928 ft/s for passenger cars and 2.35 ft/s for trucks. A negative value indicates deceleration.

Figure 90. Graph. Graphical representation of speeds on northbound Tucker Road.

Southbound Reed Road, MA

The research team collected data on Reed Road in Dartmouth, MA. The direction of travel for the data collection was southbound and the curve direction was to the left. The radius of curve and the curve length, calculated using Google Earth™, were 683.31 ft and 458.56 ft, respectively. The PSL was 40 mph until 150 ft before the curve approach, where it changed to 25mph. The travel lanes were 13 ft wide, and there was a 4-ft shoulder on both sides of the road. As figure 91 shows, there was a curve warning sign (W1-2) located between the curve approach and the PC. There were also three chevrons on the outside of the curve.

The team collected speed data at the control point, 0.3 mi before the curve approach, which was 431ft before the PC. The research team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 311 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 91 shows the horizontal curve layout, along with the speed data-collection locations.

Figure 91. Diagram. Geometric layout of southbound Reed Road (not to scale).

Table 55 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each of the curve locations. For this treatment site, there was only a statistically significant difference in speeds at the control point and approach speed for passenger cars versus heavy trucks, with the heavy truck speed lower at these points.

Table 55. Southbound Reed Road before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All Observations	39.966	5.426	39.382	6.015	38.826	4.288	39.363	4.990	775	482	590
Passenger cars	40.036	5.395	39.450	5.985	38.811	4.227	39.416	4.945	756	471	575
Heavy trucks	37.211	6.070	36.684	6.758	39.455	6.654	37.333	6.355	19	11	15
Daytime passenger cars	39.754	5.282	39.249	6.262	38.558	3.773	39.692	4.767	337	224	247
Nighttime passenger cars	40.263	5.481	39.611	5.754	39.040	4.596	39.207	5.072	419	247	328
Opposed passenger cars	39.971	5.365	39.753	6.653	38.669	3.911	39.469	4.916	377	242	275
Unopposed passenger cars	40.100	5.432	39.148	5.227	38.961	4.542	39.367	4.979	379	229	300

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 92 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 92 also shows the mean speed for trucks within the study section. The figure shows that the speeds for passenger cars remained relatively stable along the curve, and the speed changes were minimal. The truck mean speeds increased slightly from the control point to the PC and decreased slightly from the PC to the midpoint of the curve. The passenger car speeds and the truck mean speeds along the curve were higher than the PSL of 25 mph, averaging between 37 to 39 mph. The mean acceleration rate from the PC to the midpoint of the curve was -0.092 ft/s for passenger cars and -1.863 ft/s for trucks. A negative value indicates deceleration.

Figure 92. Graph. Graphical representation of speeds on southbound Reed Road.

Southbound New Boston Road, MA

The research team collected data on New Boston Road in Fairhaven, MA. The direction of travel for the data collection was southbound, and the curve direction was to the left. The radius of curve and the curve length, calculated using Google Earth™, were 592.52 ft and 226.37 ft, respectively. The PSL was 35 mph. The travel lanes were 11 ft wide, and there was no shoulder on either side of the road.

The team collected speed data at the control point, 0.2 mi before the curve approach, which was 530 ft before the PC. The research team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 460 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 93 shows the horizontal curve layout, along with the speed data-collection locations.

Figure 93. Diagram. Geometric layout of southbound New Boston Road (not to scale).

Table 56 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each of the curve locations. For this treatment site, there was a statistically significant difference in speeds at curve midpoint for passenger cars versus heavy trucks. There also was a statistically significant difference in speeds at the control point and approach for opposed versus unopposed passenger cars.

Table 56. Southbound New Boston Road before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	37.004	6.532	37.704	5.967	37.489	5.607	36.624	6.206	257	229	229
Passenger cars	36.968	6.565	37.766	5.980	37.465	5.636	36.656	6.224	252	226	227
Heavy trucks	38.800	4.712	34.600	4.722	39.333	2.082	33.000	0.000	5	3	2
Daytime passenger cars	36.690	5.930	37.655	5.583	37.576	4.603	36.760	5.551	113	99	100
Nighttime passenger cars	37.194	7.052	37.856	6.302	37.378	6.343	36.575	6.728	139	127	127
Opposed passenger cars	35.638	5.307	36.259	5.872	36.788	5.410	35.962	6.145	58	52	52
Unopposed passenger cars	37.366	6.859	38.216	5.953	37.667	5.702	36.863	6.250	194	174	175

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 94 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 94 also shows the mean speed for trucks within the study section. The figure shows that the speeds for passenger cars remained relatively stable along the curve, and the speed changes were minimal. The truck mean speeds decreased slightly between the control point and the approach of the curve, increased significantly from the approach to the PC of the curve, and then decreased substantially from the PC to the midpoint of the curve. The 85th percentile speeds and the mean speeds for passenger cars along the curve were higher than the PSL of 35 mph. The mean acceleration rate from the PC to the midpoint of the curve was -0.490 ft/s for passenger cars and -0.960 ft/s for trucks. A negative value indicates deceleration.

Figure 94. Graph. Graphical representation of speeds on southbound New Boston Road.

Northbound New Boston Road, MA

This curve was located approximately 800 ft downstream of the southbound New Boston Road curve. The direction of travel for the data collection was northbound, and the curve direction was to the right. The radius of curve and the curve length, calculated using Google Earth™, were 470.83 ft and 186.21 ft, respectively. The PSL was 35 mph. The travel lanes were 11 ft wide, and there was no shoulder on either side of the road.

The research team collected speed data at the control point, 0.4 mi before the curve approach, which was 470 ft before the PC. The team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 464 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 95 shows the horizontal curve layout, along with the speed data-collection locations.

Figure 95. Diagram. Geometric layout of northbound New Boston Road (not to scale).

Table 57 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each of the curve locations. For this treatment site, there was a statistically significant difference in speeds at the control point and approach for passenger cars versus heavy trucks, with the truck speed lower at these points. There also was a statistically significant difference in speeds at the curve midpoint for daytime versus nighttime passenger cars.

Table 57. Northbound New Boston Road before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	39.796	5.704	36.876	5.691	39.500	4.955	33.742	6.553	225	114	209
Passenger cars	39.857	5.680	36.928	5.686	39.558	4.939	33.720	6.563	223	113	207
Heavy trucks	33.000	5.657	31.000	2.828	33.000	N/A	36.000	7.071	2	1	2
Daytime passenger cars	40.074	5.352	37.475	5.801	40.066	4.694	34.627	6.531	122	61	110
Nighttime passenger cars	39.594	6.070	36.267	5.499	38.962	5.194	32.691	6.478	101	52	97
Opposed passenger cars	40.119	4.835	36.810	4.830	39.714	3.989	32.789	4.911	42	21	38
Unopposed passenger cars	39.796	5.870	36.956	5.878	39.522	5.149	33.929	6.874	181	92	169

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint
N/A = Not Applicable

Figure 96 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 96 also shows the mean speed for trucks within the study section. The figure shows that the speeds for passenger cars remained relatively stable from the control point to the PC of the curve, but decreased substantially from the PC to the midpoint of the curve. The truck mean speeds decreased slightly from the control point to the approach of the curve and increased slightly from the approach through the curve. The 85th percentile speeds for passenger cars along the curve were higher than the PSL of 35 mph. The mean acceleration rate from the PC to the midpoint of the curve was -5.478 ft/s for passenger cars and -0.2 ft/s for trucks. A negative value indicates deceleration.

Figure 96. Graph. Graphical representation of speeds on northbound New Boston Road.

Southbound Braley Hill Road, MA

The research team collected data on Braley Hill Road in Rochester, MA. The direction of travel for the data collection was southbound, and the curve direction was to the right. The radius of curve and the curve length, calculated using Google Earth™, were 695.24 ft and 672.97 ft, respectively. There were trees and other growth on the inside of the curve, which partially restricted horizontal sight distance. The PSL was 30 mph. The travel lanes were 11 ft wide, and there was no shoulder on either side of the road. As figure 97 shows, there was a Dangerous Curves sign with a 30-mph speed limit sign (R2-1) located between the curve approach and the PC.

The research team collected speed data at the control point, 0.4 mi before the curve approach, which was 461 ft before the PC. The team also collected speed data at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 410 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 97 shows the horizontal curve layout, along with the speed data-collection locations.

Figure 97. Diagram. Geometric layout of southbound Braley Hill Road (not to scale).

Table 58 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each of the curve locations. For this treatment site, there was a statistically significant difference in speeds at the control point and approach for passenger cars versus heavy vehicles, with the truck speed lower at these points. There also was a statistically significant difference in speeds at the control point, approach, and curve midpoint for daytime versus nighttime passenger cars, with the nighttime speeds higher at these points. There also was a statistically significant difference in speeds at the curve midpoint for opposed versus unopposed passenger cars.

Table 58. Southbound Braley Hill Road before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	39.360	5.097	38.508	5.835	34.585	4.791	36.365	6.947	317	294	252
Passenger cars	39.472	5.136	38.600	5.872	34.665	4.693	36.502	6.905	305	284	243
Heavy trucks	36.500	2.844	36.167	4.345	32.300	6.977	32.667	7.483	12	10	9
Daytime passenger cars	38.640	4.423	37.953	4.997	34.433	4.591	35.697	7.398	150	141	122
Nighttime passenger cars	40.277	5.640	39.226	6.566	34.895	4.797	37.314	6.297	155	143	121
Opposed passenger cars	39.211	4.485	37.958	5.247	34.758	4.736	35.185	6.355	71	66	54
Unopposed passenger cars	39.551	5.324	38.795	6.046	34.638	4.691	36.878	7.025	234	218	189

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpointe

Figure 98 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 98 also shows the mean speed for trucks within the study section. The figure shows that the 85th percentile speeds and the mean speeds for passenger cars decreased slightly from the control point to the PC and then increased from the PC to the midpoint of the curve. The 15th percentile speeds of passenger car decreased slightly along the curve. The truck mean speeds increased slightly between the control point and the approach of the curve, decreased from the approach to the PC of the curve, and then increased significantly from the PC to the midpoint of the curve. The passenger car speeds and the truck mean speeds along the curve were higher than the PSL of 30 mph. The mean acceleration rate from the PC to the midpoint of the curve was 3.544 ft/s for passenger cars and 4.908 ft/s for trucks.

'

Figure 98. Graph. Graphical representation of speeds on southbound Braley Hill Road.

Northbound Braley Hill Road, MA

The northbound Braley Hill Road curve was located approximately 0.4 mi downstream of the southbound Braley Hill Road curve. The curve is located at the start of a broken back curve and compound curve. The direction of travel for the data collection was northbound, and the curve direction was to the right. The radius of curve and the curve length, calculated using Google Earth™, were 2,097.07 ft and 661.74 ft, respectively. There were trees and other growth approximately 5 to 10 ft off the inside of the curve, but horizontal sight distance was not restricted because of the large radius. The PSL was 40 mph. The travel lanes were 10 ft wide, and there was no shoulder on either side of the road. There were driveways spaced approximately 100 to 400 ft on both sides of the road on the curve approach and on the left side throughout the curve.

The research team collected speed data at the control point, 0.4 mi before the curve approach, which was 575 ft before the PC. Speed data were also collected at the PC and midpoint of the curve. One sensor in the opposing direction of travel was placed at the midpoint of the curve, and the other one was placed 415 ft downstream of the first sensor to determine whether vehicles were present in the opposing lane. Figure 99 shows the horizontal curve layout, along with the speed data-collection locations.

Figure 99. Diagram. Geometric layout of northbound Braley Hill Road (not to scale).

Table 59 shows the speed data for the before period. The data are shown as all observations, passenger car versus heavy trucks, daytime versus nighttime speeds for passenger cars, and opposed versus unopposed passenger cars. All of the comparisons exclude vehicles with high accelerations. The research team performed a simple t-test for each of these comparisons at each of the curve locations. For this treatment site, there was a statistically significant difference in speeds at the approach, PC, and curve midpoint for passenger cars versus heavy vehicles, with the truck speeds lower at these points.

Table 59. Northbound Braley Hill Road before data.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations	44.188	7.842	38.650	4.290	37.565	3.981	38.530	3.990	277	271	268
Passenger cars	44.169	7.738	38.692	4.294	37.603	3.993	38.572	3.992	273	267	264
Heavy trucks	45.500	15.022	35.750	3.096	35.000	1.826	35.750	3.096	4	4	4
Daytime passenger cars	44.238	6.538	38.524	4.084	37.431	3.703	38.565	3.823	147	144	147
Nighttime passenger cars	44.087	8.965	38.889	4.536	37.805	4.315	38.581	4.210	126	123	117
Opposed passenger cars	43.794	7.456	38.730	4.178	36.983	4.312	38.381	4.046	63	60	63
Unopposed passenger cars	44.281	7.835	38.681	4.338	37.783	3.889	38.632	3.983	210	207	201

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 100 shows the observed mean, 15th percentile, and 85th percentile operating speeds for passenger cars along the study section. Figure 100 also shows the mean speed for trucks within the study section. As the figure shows, the passenger car speeds and truck speeds decelerated from the control point to the approach of the curve and then stabilized from the approach to midpoint of the curve. The 85th percentile speeds for passenger cars along the curve were higher than the PSL of 40 mph. Both the passenger car and truck mean speeds along the curve were lower than the PSL of 40 mph. The mean acceleration rate from the PC to the midpoint of the curve was 0.634 ft/s for passenger cars and 0.425 ft/s for trucks.

Figure 100. Graph. Graphical representation of speeds on northbound Braley Hill Road.

Comparison of Before and After Period Speed Data

After applying the OSBs, the research team collected the first after period data in Arizona in November 2012 and the second after period data in April 2013. In Massachusetts, the research team collected after period data in December 2012. In Alabama, the research team collected the first after period data in October 2013 and the second after period data in December 2013. The following sections describe the results of statistical comparisons between the before and after conditions at these sites.

Before beginning the data analysis, the research team screened all raw data to exclude vehicles that could not be tracked completely from the control point to the curve midpoint. The analysis considered only free-flow vehicles. As previously defined, a vehicle was considered free-flow if it had headway of 5 s or greater. It was anticipated that vehicles present in the opposing travel lane would influence the speed of free-flow vehicles in the subject travel lane; therefore, the research team added an indicator variable to the data files to identify free-flow vehicles in the subject travel lane that were measured in the presence of an opposing vehicle. Missing data values were excluded from the analysis. Data points considered outliers were carefully evaluated to determine whether they should be eliminated or included in the analysis. Examples of outliers include vehicles traveling at very low speeds prior to entering or exiting nearby driveways.

Mean Operating Speeds

After data screening, the research team calculated mean operating speeds at each sensor location for all sites and all data-collection periods. The research team also performed an initial comparison between corresponding speed parameters at each site for each data-collection period by calculating the numerical differences in these speed parameters. The t-statistic for independent samples was then applied to determine if the differences in mean speeds were statistically significant. The mean operating speed comparisons were identified for each data-collection site, organized by the State.

Arizona

This section of the report presents the before-after comparisons for the four Arizona OSB sites.

Northbound Pierce Ferry Road, AZ

As table 60 shows, in the first after period at northbound Pierce Ferry Road, operating speeds remained the same for most metrics. The only metric that changed at the control point was nighttime passenger car speeds, which decreased by approximately 3.6 mph, while all other metrics did not change from the before to after period. The speeds at the other data-collection points remained the same for nighttime passenger cars. The only other significant differences between the before and first after period were daytime passenger cars and unopposed passenger cars speeds at the curve midpoint, which decreased by approximately 1.4 and 1.5 mph, respectively.

In the second after period, there were several additional significant changes between the before and second after period and between the two after periods. The only change at the control point was for opposed passenger cars, where speeds significantly increased by 3.1 mph in the second after period from the first after period. The speeds at the approach significantly increased for the following speed metrics: all observations, passenger cars, heavy trucks, nighttime passenger cars, and unopposed passenger cars, with speed increases of 1.7, 1.4, 6.8, 4.4, and 1.4 mph, respectively. Despite these significant increases at the approach, the speeds at the PC significantly decreased for all speed metrics, except for heavy trucks. These reductions ranged from approximately 4.6 to 5.6 mph. However, these reductions were not maintained throughout the curve, and speeds actually significantly increased at the curve midpoint for all speed metrics. These increases ranged from approximately 5.1 to 11.0 mph.

Overall, the OSBs significantly reduced speeds at the PC (end of the OSB), but these reductions were not observed throughout the curve, and speeds actually increased at the curve midpoint after the application of the OSBs. It is possible that drivers adjusted their speeds upward to compensate for the slower speeds on the approach to the curve.

Table 60. Before-after operating speeds at northbound Pierce Ferry Road.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	48.532	7.764	46.165	8.162	46.432	6.817	38.056	6.841	218	176	214
All observations 1st after	48.777	8.326	47.179	8.799	46.416	7.597	37.233	6.347	296	173	288
All observations 2nd after	49.318	7.503	47.860	6.765	41.763	4.879	44.271	6.737	258	253	240
Passenger cars before	48.867	7.565	46.793	7.855	46.789	6.628	38.513	6.739	203	166	199
Passenger cars 1st after	49.411	8.095	47.863	8.559	46.873	7.625	37.536	6.327	270	157	263
Passenger cars 2nd after	49.868	7.405	48.209	6.768	41.969	4.943	44.395	6.954	234	229	218
Heavy trucks before	44.000	9.220	37.667	7.697	40.500	7.546	32.000	5.237	15	10	15
Heavy trucks 1st after	42.192	7.965	40.077	8.236	41.938	5.802	34.040	5.755	26	16	25
Heavy trucks 2nd after	43.958	6.355	44.458	5.831	39.792	3.753	43.045	3.885	24	24	22
Daytime passenger cars before	49.577	7.116	47.521	7.429	47.669	6.277	39.271	6.744	142	118	140
Daytime passenger cars 1st after	50.353	7.825	48.263	8.303	47.519	7.133	37.836	6.204	232	129	225
Daytime passenger cars 2nd after	50.117	7.318	48.089	6.534	42.133	4.912	44.414	6.913	214	211	198
Nighttime passenger cars before	47.213	8.347	45.098	8.594	44.625	7.028	36.712	6.430	61	48	59
Nighttime passenger cars 1st after	43.658	7.390	45.421	9.747	43.893	9.138	35.763	6.832	38	28	38
Nighttime passenger cars 2nd after	47.200	7.997	49.500	9.012	40.056	5.047	44.200	7.537	20	18	20
Opposed passenger cars before	47.806	6.122	46.444	8.833	47.621	7.683	38.029	6.671	36	29	35
Opposed passenger cars 1st after	49.787	8.172	47.872	7.881	47.536	7.923	39.370	6.839	47	28	46
Opposed passenger cars 2nd after	50.870	7.589	48.130	7.055	41.981	5.443	43.731	6.754	54	52	52
Unopposed passenger cars before	49.096	7.838	46.868	7.655	46.613	6.400	38.616	6.769	167	137	164
Unopposed passenger cars 1st after	49.332	8.096	47.861	8.712	46.729	7.583	37.147	6.160	223	129	217
Unopposed passenger cars 2nd after	49.567	7.344	48.233	6.700	41.966	4.803	44.602	7.023	180	177	166

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 101 shows a graphical representation of the speed profiles for northbound Pierce Ferry Road. At the approach point and PC, the general shape of the speed profiles for the three collection periods remained relatively similar. At the midpoint, the mean and 85th percentile speeds showed an increase in the second after period.

Figure 101. Graph. Operating speeds comparison on northbound Pierce Ferry Road (PSL = 55 mph).

Southbound Shinarump Road, AZ

Table 61 shows speeds increased at the control point for the following speed metrics: all observations, passenger cars, nighttime passenger cars, and unopposed passenger cars, with the increases being 1.2, 1.4, 2.2, and 1.4 mph, respectively. All observations and heavy truck speeds decreased at the control point by 1.4 and 5.8 mph, respectively. The only other change in the first after period was heavy truck speeds, which decreased at the PC by 3.8 mph and at the curve midpoint by 4.7 mph. There was no strong trend initially in speed changes after the OSBs were installed, with only heavy truck speeds decreasing at the PC and curve midpoint.

There were more changes in speeds during the second after period. Speeds remained relatively similar at the control point, with only daytime passenger car speeds increasing by approximately 1.6 mph. There were no speed changes at the approach. Speeds increased at the PC for all observations, heavy trucks, and daytime passenger cars, with speed increases of approximately 1.9, 6.1, and 2.0 mph, respectively. However, the 2.0 mph increase for daytime passenger cars was not significant after accounting for the speed increase at the control point. Speeds at the curve midpoint significantly decreased for every speed metric. The speed reductions ranged from 6.0 to 8.2 mph. Based on the second after period, speeds at the curve midpoint significantly decreased after the application of the OSBs, with a practical decrease (greater than 5 mph).

Table 61. Before-after operating speeds at southbound Shinarump Road.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	47.317	6.928	51.099	8.603	47.299	6.864	51.100	8.042	303	288	209
All observations 1st after	48.542	6.689	49.716	7.630	46.393	6.947	49.838	8.216	236	224	154
All observations 2nd after	47.979	8.414	51.277	9.756	49.244	9.528	44.662	6.987	242	176	225
Passenger cars before	47.566	6.768	51.241	8.023	48.017	6.341	51.537	7.773	249	238	175
Passenger cars 1st after	48.930	6.887	50.592	7.471	47.409	6.490	50.780	8.036	201	193	132
Passenger cars 2nd after	48.375	8.105	51.491	9.502	49.158	9.245	45.099	6.954	216	158	203
Heavy trucks before	46.167	7.583	50.444	10.956	43.880	8.188	48.853	9.099	54	50	34
Heavy trucks 1st after	46.314	4.928	44.686	6.597	40.065	6.413	44.182	7.062	35	31	22
Heavy trucks 2nd after	44.692	10.248	49.500	11.724	50.000	12.020	40.636	6.067	26	18	22
Daytime passenger cars before	46.686	7.001	50.777	8.431	47.009	6.033	50.894	7.375	121	115	94
Daytime passenger cars 1st after	47.571	6.276	50.420	7.582	46.654	6.513	50.759	7.871	112	107	79
Daytime passenger cars 2nd after	48.316	8.146	51.181	9.422	49.017	9.247	44.805	6.618	171	119	159
Nighttime passenger cars before	48.398	6.458	51.680	7.624	48.959	6.500	52.284	8.193	128	123	81
Nighttime passenger cars 1st after	50.640	7.266	50.809	7.365	48.349	6.374	50.811	8.353	89	86	53
Nighttime passenger cars 2nd after	48.600	8.032	52.667	9.819	49.590	9.346	46.159	8.046	45	39	44
Opposed passenger cars before	45.967	6.646	48.200	5.580	45.933	4.982	50.875	6.556	30	30	24
Opposed passenger cars 1st after	46.263	5.086	49.211	5.127	46.105	5.259	52.636	5.608	19	19	11
Opposed passenger cars 2nd after	47.889	7.192	49.741	5.769	48.526	7.516	44.400	5.276	27	19	25
Unopposed passenger cars before	47.785	6.770	51.658	8.224	48.317	6.468	51.642	7.964	219	208	151
Unopposed passenger cars 1st after	49.209	7.000	50.736	7.670	47.552	6.608	50.612	8.218	182	174	121
Unopposed passenger cars 2nd after	48.444	8.242	51.741	9.907	49.245	9.477	45.197	7.165	189	139	178

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 102 shows a graphical representation of the speed profiles for southbound Shinarump Road. The general shape of the mean and 85th percentile speed profiles remained relatively similar in the before and first after period. In the second after period, the mean and 85th percentile speeds increased at the midpoint.

Figure 102. Graph. Operating speeds comparison on southbound Shinarump Road (PSL = 45 mph).

Southbound Diamond Bar Road, AZ

As table 62 shows, in the first after period at southbound Diamond Bar Road, operating speeds remained the same for most metrics. The only significant change at the control point was heavy truck speeds decreased by approximately 4.0 mph. Heavy truck speeds also decreased at the approach by approximately 2.7 mph, while all other speed metrics did not change significantly. The only speed metric at the PC that changed was nighttime passenger car speeds, which decreased by approximately 4.9 mph. The only speed metric at the curve midpoint that changed was heavy truck speeds, which decreased by approximately 4.7 mph. After taking into account the reduction in heavy truck speeds at the control point speed, truck speeds did not decrease in the first after period at the curve midpoint. The OSBs had virtually no effect on any of the other speed metrics in the first after period.

There were more speed changes in the second after period when compared with the before period, than there were between the first after period and before period. The only change at the control point was heavy truck speeds, which decreased from both the before period and first after period. These speed reductions were 4.0 and 6.9 mph, respectively. Speeds for all metrics increased at the approach, except heavy truck speeds did not change from the before period. Speed increases ranged between approximately 3.7 and 6.7 mph at this data-collection point. Speeds did not significantly increase at the PC, except heavy truck speeds decreased by approximately 5.2 mph. All measured speed metrics decreased at the curve midpoint, with speed decreases ranging between approximately 5.0 and 10.6 mph.

Based on the second after period, speeds at the curve midpoint significantly decreased after the application of the optical speed bars, with a practical decrease (greater than 5 mph).

Table 62. Before-after operating speeds at southbound Diamond Bar Road.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	53.421	8.312	49.171	6.730	50.299	7.116	50.435	7.621	292	214	209
All observations 1st after	52.111	9.266	48.144	6.865	49.969	8.313	49.091	8.654	153	64	110
All observations 2nd after	53.317	8.817	54.410	7.940	51.506	6.705	42.695	6.953	183	89	177
Passenger cars before	53.867	8.389	49.516	6.675	50.874	6.996	50.703	7.611	256	182	182
Passenger cars 1st after	53.476	9.418	49.097	7.084	50.925	8.655	50.375	8.743	124	53	88
Passenger cars 2nd after	54.071	8.616	55.035	7.831	52.083	6.446	43.061	7.017	170	84	164
Heavy trucks before	50.250	7.064	46.722	6.704	47.031	7.014	48.630	7.581	36	32	27
Heavy trucks 1st after	46.276	5.738	44.069	3.760	45.364	4.202	43.955	6.122	29	11	22
Heavy trucks 2nd after	43.462	4.502	46.231	3.811	41.800	1.483	38.077	3.904	13	5	13
Daytime passenger cars before	54.061	8.582	49.873	7.057	51.706	7.113	51.109	7.876	181	126	128
Daytime passenger cars 1st after	53.718	9.678	49.500	7.121	52.318	8.596	50.831	9.011	110	44	77
Daytime passenger cars 2nd after	53.832	8.857	54.904	8.035	52.576	6.361	42.475	7.134	125	66	122
Nighttime passenger cars before	53.400	7.939	48.653	5.598	49.000	6.399	49.741	6.918	75	56	54
Nighttime passenger cars 1st after	51.571	7.035	45.929	6.120	44.111	5.183	47.182	5.930	14	14	11
Nighttime passenger cars 2nd after	54.733	7.964	55.400	7.309	50.278	6.614	44.762	6.446	45	18	42
Opposed passenger cars before	53.472	9.376	50.057	6.320	51.000	8.218	50.057	8.359	53	35	35
Opposed passenger cars 1st after	52.500	10.988	48.611	7.047	52.714	9.050	52.077	9.996	18	7	13
Opposed passenger cars 2nd after	53.889	5.989	53.722	6.115	52.692	4.461	42.588	6.820	18	13	17
Unopposed passenger cars before	53.970	8.134	49.374	6.773	50.844	6.705	50.857	7.445	203	147	147
Unopposed passenger cars 1st after	53.642	9.175	49.179	7.120	50.652	8.665	50.080	8.548	106	46	75
Unopposed passenger cars 2nd after	54.092	8.890	55.191	8.012	51.972	6.765	43.116	7.060	152	71	147

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 103 shows a graphical representation of the speed profiles for southbound Diamond Bar Road. The general shape of the mean and 85th percentile speed profiles remained relatively similar in the before and first after period. In the second after period, the mean and 85th percentile speeds increased at the approach and midpoints.

Figure 103. Graph. Operating speeds comparison on southbound Diamond Bar Road (PSL = 45 mph).

Southbound County Route 1, AZ

As table 63 shows, in the first after period at southbound County Route 1, operating speeds remained the same for most metrics. There was a significant increase at the control point for the following speed metrics: all observations, passenger cars, daytime passenger cars, nighttime passenger cars, and unopposed passenger cars, with speed increases of approximately 2.0, 2.2, 2.8, and 2.1 mph, respectively. There were no significant speed changes at the approach. Speeds at the PC remained the same in the first after period, except daytime passenger car speeds significantly decreased by approximately 1.1 mph. Speeds also remained constant at the curve midpoint, with the exception that opposed passenger car speeds significantly increased by approximately 3.4 mph. The optical speed bars had virtually no effect on speeds in the first after period.

There were more speed changes in the second after period when compared with the before period than there were between the first after period and before period. There were no significant speed differences at the control point between the second after period and before period. Speeds generally decreased from the first to second after period, with speeds returning closer to the before period. Speeds remained the same at the approach, except speeds significantly increased for all observations, passenger cars, and unopposed passenger cars, increasing by approximately 1.1, 1.0, and 1.1 mph, respectively. Speeds significantly increased at the PC for all measured speed metrics, expect opposed passenger car speeds remained the same. Speed increases ranged between approximately 1.8 and 6.3 mph. These speed increases were also observed at the curve midpoint, with all measured speed metrics, except heavy trucks, increasing significantly by approximately 3.5 to 7.1 mph.

The OSBs did not have a strong effect on operating speeds in the first after period, but after several months, the speeds at the PC and curve midpoint significantly increased between 2.1 and 7.1 mph. It is possible that drivers decelerated on the curve approach but increased speeds through the curve influenced by the large curve radii.

Table 63. Before-after operating speeds at southbound County Route 1.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	53.471	8.882	47.731	5.559	44.021	5.211	43.919	7.259	342	329	307
All observations 1st after	55.448	9.376	47.924	5.972	43.565	5.031	43.677	7.210	382	363	341
All observations 2nd after	53.622	9.856	48.820	5.650	46.521	5.742	48.014	7.115	233	219	145
Passenger cars before	53.532	8.901	47.810	5.515	44.134	5.125	43.918	7.262	327	314	292
Passenger cars 1st after	55.746	9.419	48.045	5.896	43.635	5.013	43.877	7.147	355	337	316
Passenger cars 2nd after	53.532	9.792	48.792	5.656	46.417	5.719	47.986	7.166	216	204	138
Heavy trucks before	52.133	8.651	46.000	6.414	41.667	6.532	43.933	7.450	15	15	15
Heavy trucks 1st after	51.519	7.939	46.333	6.811	42.654	5.284	41.160	7.679	27	26	25
Heavy trucks 2nd after	54.765	10.889	49.176	5.736	47.933	6.076	48.571	6.477	17	15	7
Daytime passenger cars before	53.847	9.266	49.038	5.430	44.977	4.895	44.599	7.081	183	174	162
Daytime passenger cars 1st after	55.689	9.357	48.222	5.420	43.895	4.746	43.969	7.107	257	248	229
Daytime passenger cars 2nd after	54.046	9.582	49.408	5.225	47.034	5.326	48.316	7.149	152	145	98
Nighttime passenger cars before	53.132	8.428	46.250	5.238	43.086	5.228	43.069	7.421	144	140	130
Nighttime passenger cars 1st after	55.898	9.628	47.582	7.004	42.910	5.656	43.632	7.287	98	89	87
Nighttime passenger cars 2nd after	52.313	10.249	47.328	6.375	44.898	6.383	47.175	7.236	64	59	40
Opposed passenger cars before	53.479	9.299	47.125	5.354	44.111	4.900	41.091	7.844	48	45	44
Opposed passenger cars 1st after	56.000	9.372	48.301	5.721	43.843	5.029	44.506	6.613	93	89	85
Opposed passenger cars 2nd after	53.407	9.616	47.963	4.642	45.373	5.564	48.171	6.667	54	51	41
Unopposed passenger cars before	53.541	8.848	47.928	5.543	44.138	5.171	44.419	7.053	279	269	248
Unopposed passenger cars 1st after	55.656	9.452	47.954	5.965	43.560	5.015	43.645	7.333	262	248	231
Unopposed passenger cars 2nd after	53.574	9.879	49.068	5.943	46.765	5.745	47.907	7.399	162	153	97

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 104 shows a graphical representation of the speed profiles for southbound County Route1. The general shape of the mean and 85th percentile speed profiles, at the curve approach point and PC, for the three collection periods remained relatively similar in the before and first after period. However, the midpoint speeds increased in the second after period.

Figure 104. Graph. Operating speeds comparison on southbound County Route 1 (PSL = 35 mph).

Arizona Summary of Results

Overall, the OSBs did not have an effect on operating speeds at the Arizona sites. There were no changes in speeds that were consistent across all of the sites.

Alabama

This section of the report presents the before-after comparisons for the eight Alabama OSB sites.

Alabama Location #1

As table 64 shows, in the first after period at Alabama Location #1, operating speeds decreased for several speed metrics. There was a significant decrease at the control point for all speed metrics. These speed decreases ranged between approximately 2.1 and 6.7 mph. Speeds at the approach significantly decreased for all speed metrics. However, after accounting for the speed decrease at the control point, none of these reductions were significant, and speeds for passenger cars, daytime passenger cars, and unopposed passenger cars actually increased. Speeds at the PC significantly decreased for the following speed metrics: all observations, passenger cars, daytime passenger cars, and unopposed passenger cars. However, after accounting for the speed decrease at the control point, none of these reductions were significant, and speeds for passenger cars and opposed passenger cars actually increased. Speeds at the curve midpoint did not change significantly from the before period. If the speed reduction at the control point was taken into account, speeds increased slightly at the curve midpoint.

There were more speed changes in the second after period when compared with the before period, than there were between the first after period and before period. Speeds at the control point increased for all speed metrics, except for heavy trucks. The range of speed increases was between approximately 1.7 and 2.7 mph. Not taking the speed increase at the control point into account, speeds at the approach increased from the first after period, but they remained the same when compared with the before period, except all observations significantly decreased by approximately 0.8 mph. Not taking into account the speed increase at the control point, speeds at the PC increased from the first after period, but remained the same when compared with the before period, except opposed passenger cars significantly increased by approximately 3.3 mph. Speeds significantly decreased at the curve midpoint for all speed metrics, except for heavy trucks. The speed decreases were between approximately 1.4 and 2.6 mph. If the speed increases at the control point were taken into account, these decreases became more significant.

The OSBs did not have a strong effect on operating speeds in the first after period, but after several months, the speeds at the curve midpoint significantly decreased between 1.4 and 2.6mph.

Table 64. Before-after operating speeds at Alabama Location #1.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	47.883	9.368	49.208	7.596	50.103	8.770	42.323	6.912	375	87	353
All observations 1st after	43.306	6.928	45.497	6.752	47.431	5.886	41.691	6.951	340	204	333
All observations 2nd after	50.190	7.863	48.380	6.788	49.776	6.830	40.364	9.437	822	223	733
Passenger cars before	47.917	9.380	49.218	7.540	50.429	8.700	42.516	6.892	362	84	341
Passenger cars 1st after	43.575	6.880	46.134	6.490	48.148	5.530	42.523	6.342	313	183	306
Passenger cars 2nd after	50.226	7.931	48.522	6.792	49.906	6.914	40.475	9.413	793	213	710
Heavy trucks before	46.923	9.332	48.923	9.394	41.000	6.083	36.833	5.132	13	3	12
Heavy trucks 1st after	40.185	6.850	38.111	5.243	41.190	5.269	32.259	6.671	27	21	27
Heavy trucks 2nd after	49.207	5.722	44.483	5.468	47.000	3.944	36.957	9.740	29	10	23
Daytime passenger cars before	47.638	9.418	48.805	7.641	50.795	8.551	42.288	6.764	174	39	160
Daytime passenger cars 1st after	42.964	6.865	46.016	6.568	48.020	5.060	42.470	6.112	252	151	247
Daytime passenger cars 2nd after	49.417	7.985	48.296	6.904	49.578	7.058	40.840	9.469	348	102	306
Nighttime passenger cars before	48.176	9.362	49.601	7.444	50.111	8.912	42.718	7.016	188	45	181
Nighttime passenger cars 1st after	46.098	6.395	46.623	6.186	48.750	7.427	42.746	7.282	61	32	59
Nighttime passenger cars 2nd after	50.858	7.838	48.699	6.705	50.207	6.797	40.198	9.373	445	111	404
Opposed passenger cars before	47.529	10.661	48.793	7.385	46.833	7.648	42.037	6.512	87	24	82
Opposed passenger cars 1st after	43.250	5.583	44.775	4.699	46.667	4.151	41.725	6.345	40	21	40
Opposed passenger cars 2nd after	49.228	7.986	47.772	6.816	50.130	5.361	39.479	10.074	158	46	142
Unopposed passenger cars before	48.040	8.955	49.353	7.596	51.867	8.736	42.668	7.014	275	60	259
Unopposed passenger cars 1st after	43.623	7.057	46.333	6.696	48.340	5.666	42.643	6.345	273	162	266
Unopposed passenger cars 2nd after	50.474	7.904	48.709	6.778	49.844	7.297	40.724	9.233	635	167	568

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 105 shows a graphical representation of the speed profiles for Alabama Location #1. The mean and 85th percentile operating speeds increased in the first after period but remained relatively similar during the before and second after data-collection periods.

Figure 105. Graph. Operating speeds comparison at Alabama Location #1 (PSL = 55 mph).

Alabama Location #2

As table 65 shows, in the first after period at Alabama Location #2, operating speeds remained the same for most metrics. The only significant change at the control point was unopposed passenger car speeds decreased by approximately 1.1 mph. Speeds at the approach increased for the following speed metrics: all observations, passenger cars, and opposed passenger cars, with speed increases of 1.0, 1.1, and 2.4 mph, respectively. There were no significant speed changes at the PC. Speeds at the curve midpoint significantly increased for the following speed metrics: all observations, passenger cars, daytime passenger cars, and opposed passenger cars, with speed increases of approximately 1.0, 1.1, 1.9, and 3.4 mph, respectively. After installing the OSBs, speeds significantly increased at the curve midpoint during the first after period, but most of the speed increases were not practically significant.

The speed changes that occurred in the first after period continued through the second after period, and speeds in the second after period were similar to the first after period. The only significant change at the control point was opposed passenger car speeds increased by approximately 2.1 mph. Speeds at the approach increased for the same speed metrics as in the first after period: all observations, passenger cars, and opposed passenger cars, with speed increases of approximately 1.0, 0.9, and 3.5 mph, respectively. There were no significant speed changes at the PC. Speeds increased at the curve midpoint for the following speed metrics: all observations, daytime passenger cars, and opposed passenger cars, with speed increases of approximately 5.6, 1.5, and 3.0 mph, respectively. However, if the speed increase at the control point was taken into account, the speed increase at the curve midpoint for opposed passenger cars was no longer significant. Speeds significantly increased for several speed metrics in both after periods at the curve midpoint, but many speed metrics did not increase.

Table 65. Before-after operating speeds at Alabama Location #2.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	49.269	7.604	44.291	5.822	42.472	5.747	38.363	7.094	398	290	358
All observations 1st after	48.843	6.991	45.328	5.034	42.245	4.882	39.310	5.429	470	413	462
All observations 2nd after	49.374	6.823	45.335	5.407	43.163	5.419	43.949	6.719	495	307	316
Passenger cars before	49.569	7.238	44.479	5.623	42.630	5.554	38.418	7.129	376	273	336
Passenger cars 1st after	49.039	6.984	45.549	4.977	42.552	4.742	39.556	5.325	439	386	432
Passenger cars 2nd after	49.472	6.787	45.413	5.181	43.192	5.263	39.003	5.624	632	499	585
Heavy trucks before	44.136	11.319	41.091	8.053	39.941	8.050	37.091	6.560	22	17	22
Heavy trucks 1st after	46.065	6.582	42.194	4.861	37.852	4.825	35.767	5.764	31	27	30
Heavy trucks 2nd after	46.429	7.386	43.000	10.015	42.200	9.466	37.111	8.345	21	15	18
Daytime passenger cars before	49.884	8.070	45.072	5.722	43.263	5.737	37.821	7.802	138	99	122
Daytime passenger cars 1st after	49.302	6.916	45.764	4.782	42.782	4.626	39.770	5.158	318	280	313
Daytime passenger cars 2nd after	49.584	6.915	45.776	5.031	43.692	5.102	39.381	5.647	375	295	344
Nighttime passenger cars before	49.387	6.721	44.134	5.548	42.270	5.432	38.762	6.720	238	174	214
Nighttime passenger cars 1st after	48.347	7.143	44.983	5.437	41.943	5.008	38.992	5.725	121	106	119
Nighttime passenger cars 2nd after	49.307	6.607	44.883	5.359	42.471	5.419	38.465	5.557	257	204	241
Opposed passenger cars before	47.809	7.949	43.022	6.312	41.721	5.057	36.378	7.241	89	61	82
Opposed passenger cars 1st after	49.026	6.713	45.421	4.674	42.338	5.027	39.787	5.602	76	68	75
Opposed passenger cars 2nd after	49.918	6.386	46.541	4.829	44.012	4.633	39.337	5.721	98	81	92
Unopposed passenger cars before	50.115	6.928	44.930	5.323	42.892	5.674	39.075	6.976	287	212	254
Unopposed passenger cars 1st after	49.041	7.048	45.576	5.044	42.597	4.686	39.507	5.272	363	318	357
Unopposed passenger cars 2nd after	49.390	6.861	45.206	5.221	43.033	5.367	38.941	5.609	534	418	493

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 106 shows a graphical representation of the speed profiles for Alabama Location #2. The general shape of the speed profiles for the three collection periods remained relatively similar.

Figure 106. Graph. Operating speeds comparison at Alabama Location #2 (PSL = 55 mph).

Alabama Location #3

As table 66 shows, in the first after period at Alabama Location #3, operating speeds changed for many speed metrics. As in the before period, operating speeds at the control point were high. It was originally thought that a sensor malfunction caused these high speed measurements, but these high speeds persisted through all three time periods. Speeds at the control point increased for all speed metrics, except nighttime passenger cars. These speed increases were between 2.5and 3.4 mph. Speeds at the approach decreased for all speed metrics, except nighttime passenger cars and opposed passenger cars. These speed decreases were between approximately 1.0 and 3.0 mph. The only change at the PC was passenger car speeds, which increased by approximately 0.6 mph. Speeds at the curve midpoint decreased for all speed metrics, except heavy trucks. These speed reductions were between approximately 1.9 and 2.8 mph. Speeds at the curve midpoint were slightly lower immediately after the OSBs were installed.

Most of the changes in speeds from the first after period did not persist through the second after period. Speeds at the control point increased for all speed metrics, with speed increases between approximately 2.0 and 3.5 mph. Speeds at the approach remained the same, except all observations decreased by approximately 0.6 mph. There were no significant speed changes at the PC. Speeds at the curve midpoint increased from the first after period and returned closer to the before period. There were no significant speed changes at the curve midpoint.

Speeds slightly decreased at the curve midpoint immediately after installing the OSBs, but the decrease was not maintained through the second after period.

Table 66. Before-after operating speeds at Alabama Location #3.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	65.253	12.908	49.831	6.463	41.307	4.663	41.233	5.914	498	472	472
All observations 1st after	67.813	11.357	48.246	5.062	41.365	4.737	38.677	6.919	513	493	511
All observations 2nd after	68.027	11.822	49.208	5.335	41.622	4.770	41.085	6.888	960	891	879
Passenger cars before	65.520	13.004	50.000	6.522	41.629	4.652	41.740	5.683	440	415	415
Passenger cars 1st after	68.215	11.509	49.018	4.728	42.239	4.371	39.085	7.219	400	381	398
Passenger cars 2nd after	68.302	11.785	49.606	5.297	42.065	4.712	41.571	6.823	819	752	748
Heavy trucks before	63.224	12.070	48.552	5.891	38.965	4.066	37.544	6.299	58	57	57
Heavy trucks 1st after	66.389	10.732	45.513	5.276	38.393	4.748	37.239	5.534	113	112	113
Heavy trucks 2nd after	66.433	11.955	46.901	4.969	39.223	4.366	38.313	6.618	141	139	131
Daytime passenger cars before	66.424	12.482	50.549	6.973	41.923	4.682	41.689	5.979	264	246	251
Daytime passenger cars 1st after	68.880	11.622	49.098	4.733	42.278	4.364	38.861	7.470	317	302	316
Daytime passenger cars 2nd after	69.931	11.578	50.006	5.282	42.450	4.634	41.924	6.831	466	436	432
Nighttime passenger cars before	64.165	13.675	49.176	5.702	41.201	4.589	41.817	5.214	176	169	164
Nighttime passenger cars 1st after	65.675	10.759	48.711	4.725	42.089	4.421	39.951	6.118	83	79	82
Nighttime passenger cars 2nd after	66.150	11.726	49.076	5.278	41.535	4.774	41.089	6.793	353	316	316
Opposed passenger cars before	65.516	13.031	49.979	6.684	41.423	5.063	41.928	6.036	188	175	180
Opposed passenger cars 1st after	68.953	11.438	48.929	5.002	41.900	4.482	39.318	6.634	85	80	85
Opposed passenger cars 2nd after	68.935	12.198	49.245	5.109	41.894	4.961	42.335	7.003	200	180	185
Unopposed passenger cars before	65.524	13.010	50.016	6.413	41.779	4.332	41.596	5.405	252	240	235
Unopposed passenger cars 1st after	68.016	11.538	49.041	4.659	42.329	4.344	39.022	7.378	315	301	313
Unopposed passenger cars 2nd after	68.097	11.651	49.722	5.355	42.119	4.634	41.320	6.751	619	572	563

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 107 shows a graphical representation of the speed profiles for Alabama Location #3. The general shape of the speed profiles for the three collection periods remained relatively similar.

Figure 107. Graph. Operating speeds comparison at Alabama Location #3 (PSL = 55 mph).

Alabama Location #4

As Table 67 shows, in the first after period at Alabama Location #4, operating speeds decreased for many speed metrics. There was a significant decrease at the control point for all measured speed metrics. Speed decreases were between approximately 1.6 and 3.3 mph for passenger cars and 5.3 mph for heavy trucks. There was a significant decrease at the approach for all speed metrics, except heavy trucks and opposed passenger cars. These speed decreases were all approximately 1.5 mph. However, if the decrease in speed at the control point was taken into account, the speed decreases at the approach were no longer significant. There was a significant decrease at the PC for all measured speed metrics, except heavy trucks. Speed decreases were between 2.0and 2.5 mph. However, if the speed decreases at the control point were taken into account, the only speed decrease at the PC that was significant was for daytime passenger cars, and nighttime speeds actually increase slightly. Speeds were also significantly lower at the curve midpoint for all measured speed metrics. However, if the speed decreases at the control point were taken into account, none of the speed decreases at the curve midpoint were significant. Overall, installing the OSBs did not have an effect on operating speeds in the first after period.

Most of the speed changes in the first after period were no longer present in the second after period. Speeds at the control point were closer to speeds in the before period, and there were no significant speed changes. Speeds at the approach were also closer to speeds in the before period, and there were no significant speed changes, except daytime passenger speeds increased by approximately 1.1 mph. Speeds at the PC significantly decreased for the following speed metrics: all observations, passenger cars, and opposed passenger cars, with speed decreases of approximately 0.6, 0.7, and 1.2 mph, respectively. Speeds at the curve midpoint did not significantly change, except nighttime speeds were approximately 1.5 mph lower.

Speeds slightly decreased at the curve midpoint after installing the OSBs, but the decrease was not maintained through the second after period.

Table 67. Before-after operating speeds at Alabama Location #4.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	50.761	9.466	46.666	7.082	39.091	4.939	34.664	8.205	494	471	455
All observations 1st after	48.515	8.022	45.152	5.409	36.891	4.357	31.938	6.086	513	512	513
All observations 2nd after	50.322	7.655	47.172	6.128	38.469	4.754	34.433	8.094	615	603	564
Passenger cars before	50.633	9.369	46.703	7.019	39.153	4.883	34.506	8.197	472	452	437
Passenger cars 1st after	48.541	7.986	45.242	5.431	37.000	4.330	31.819	6.162	475	474	475
Passenger cars 2nd after	50.381	7.664	47.155	6.038	38.455	4.709	34.434	8.125	582	571	535
Heavy trucks before	53.500	11.237	45.864	8.481	37.632	6.103	38.500	7.649	22	19	18
Heavy trucks 1st after	48.184	8.561	44.026	5.059	35.526	4.519	33.421	4.864	38	38	38
Heavy trucks 2nd after	49.273	7.543	47.485	7.649	38.719	5.566	34.414	7.623	33	32	29
Daytime passenger cars before	50.588	9.045	46.631	7.464	39.528	5.023	33.861	8.062	187	176	173
Daytime passenger cars 1st after	48.966	8.241	45.564	5.389	37.020	4.325	31.969	5.975	351	350	351
Daytime passenger cars 2nd after	50.324	7.331	47.770	6.113	38.953	4.793	35.084	8.230	343	338	320
Nighttime passenger cars before	50.663	9.592	46.751	6.723	38.913	4.786	34.928	8.273	285	276	264
Nighttime passenger cars 1st after	47.339	7.111	44.331	5.466	36.944	4.362	31.395	6.668	124	124	124
Nighttime passenger cars 2nd after	50.464	8.132	46.272	5.829	37.734	4.499	33.465	7.887	239	233	215
Opposed passenger cars before	50.886	9.349	46.429	8.484	39.141	5.234	34.328	8.582	140	128	125
Opposed passenger cars 1st after	49.156	8.039	45.227	5.317	36.836	4.814	31.953	6.551	128	128	128
Opposed passenger cars 2nd after	51.089	7.337	47.274	6.375	37.910	4.802	34.535	8.608	124	122	114
Unopposed passenger cars before	50.527	9.390	46.819	6.311	39.157	4.746	34.577	8.051	332	324	312
Unopposed passenger cars 1st after	48.314	7.966	45.248	5.480	37.061	4.143	31.769	6.021	347	346	347
Unopposed passenger cars 2nd after	50.190	7.746	47.122	5.950	38.604	4.678	34.406	8.000	458	449	421

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 108 shows a graphical representation of the speed profiles for Alabama Location #4. The general shape of the speed profiles for the three collection periods remained relatively similar.

Figure 108. Graph. Operating speeds comparison at Alabama Location #4 (PSL = 35 mph).

Alabama Location #5

As table 68 shows, in the first after period at Alabama Location #5, operating speeds changed for several speed metrics. There were no significant speed changes at the control point or approach. Speeds at the PC significantly decreased for the following speed metrics: all observations, passenger cars, and unopposed passenger cars, with speed decreases of approximately
2.2, 2.1, and 2.9 mph, respectively. Speeds for opposed passenger cars significantly increased by approximately 5.4 mph. Speeds at the curve midpoint significantly increased for all speed metrics, except heavy trucks. These speed increases were between approximately 4.2 and 7.8mph. Speeds at the PC significantly decreased after the installation of the OSBs, but vehicles maintained a more constant speed throughout the curve, and speeds increased at the curve midpoint.

The speed changes that occurred in the first after period continued through the second after period, and speeds in the second after period were similar to the first after period. There were no significant speed changes at the control point or approach. Speeds at the PC significantly decreased for the following speed metrics: all observations, passenger cars, heavy trucks, and unopposed passenger cars, with speed decreases of approximately 1.9, 1.8, 5.2, 1.7, and 2.6 mph. Speeds for opposed passenger cars significantly increased by approximately 6.5 mph. Speeds at the curve midpoint significantly increased for all speed metrics. Speed increases were between approximately 3.6 and 8.5 mph for passenger cars and 7.8 mph for heavy trucks.

Installing the OSBs decreased operating speeds at the PC during both after periods between approximately 1.7 and 2.9 mph for passenger cars, while speeds at the curve midpoint increased during both after periods between approximately 3.6 and 8.5 mph.

Table 68. Before-after operating speeds at Alabama Location #5.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	41.944	7.212	43.742	8.327	45.781	7.524	39.810	6.842	233	64	221
All observations 1st after	41.467	7.184	43.170	7.194	43.636	6.437	44.182	7.264	353	236	242
All observations 2nd after	41.889	7.277	43.069	7.426	43.847	6.047	43.949	6.719	495	307	316
Passenger cars before	42.299	7.094	44.054	8.250	45.794	7.584	40.028	6.782	224	63	212
Passenger cars 1st after	41.851	7.086	43.460	7.037	43.689	6.435	44.384	7.164	335	222	232
Passenger cars 2nd after	42.185	7.110	43.344	7.168	44.024	5.960	44.020	6.672	471	294	302
Heavy trucks before	33.111	3.822	36.000	6.614	45.000	0.000	34.667	6.557	9	1	9
Heavy trucks 1st after	34.333	5.018	37.778	8.128	42.786	6.647	39.500	8.383	18	14	10
Heavy trucks 2nd after	36.083	8.214	37.667	10.120	39.846	6.842	42.429	7.773	24	13	14
Daytime passenger cars before	41.947	6.887	44.405	7.941	45.619	7.450	39.787	6.772	131	42	122
Daytime passenger cars 1st after	41.644	6.846	43.222	7.071	43.668	6.212	44.273	7.151	284	193	198
Daytime passenger cars 2nd after	42.272	6.988	43.276	7.003	43.921	6.131	43.858	7.030	301	190	204
Nighttime passenger cars before	42.796	7.383	43.559	8.686	46.143	8.021	40.356	6.821	93	21	90
Nighttime passenger cars 1st after	43.000	8.280	44.784	6.763	43.828	7.888	45.029	7.309	51	29	34
Nighttime passenger cars 2nd after	42.029	7.340	43.465	7.471	44.212	5.660	44.357	5.877	170	104	98
Opposed passenger cars before	40.235	7.242	41.353	11.045	38.833	7.333	34.933	7.045	17	6	15
Opposed passenger cars 1st after	41.905	9.286	42.762	8.630	44.200	6.609	42.692	6.851	42	30	26
Opposed passenger cars 2nd after	41.867	10.281	44.567	8.063	45.294	5.828	43.389	5.782	30	17	18
Unopposed passenger cars before	42.469	7.072	44.275	7.972	46.526	7.290	40.416	6.621	207	57	197
Unopposed passenger cars 1st after	41.843	6.733	43.560	6.790	43.609	6.421	44.597	7.190	293	192	206
Unopposed passenger cars 2nd after	42.206	6.858	43.261	7.106	43.946	5.970	44.060	6.732	441	277	284

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 109 shows a graphical representation of the speed profiles for Alabama Location #5. The PC mean operating speeds decreased in the two after periods while the midpoint speeds increased. The midpoint 85th percentile speeds increased in the two after periods.

Figure 109. Graph. Operating speeds comparison at Alabama Location #5 (PSL = 40 mph).

Alabama Location #6

As table 69 shows, in the first after period at Alabama Location #6, operating speeds decreased for many speed metrics. There was a significant decrease at the control point for the following speed metrics: all observations, passenger cars, nighttime passenger cars, and unopposed passenger cars, with speed decreases of approximately 2.0, 2.0, 4.4, and 2.2 mph, respectively. There was a significant increase at the approach for all speed metrics. These increases ranged from approximately 8.3 to 9.5 mph for passenger cars. However, these increased speeds were not maintained, and speeds at the PC significantly decreased for the following speed metrics: all observations, passenger cars, heavy trucks, and unopposed passenger cars, with speed decreases of approximately 1.6, 1.4, 4.4, and 1.8 mph, respectively. Speeds were also significantly lower at the curve midpoint for all passenger car speed metrics. These speed reductions were approximately 3.9 to 8.8 mph. Speeds throughout the curve decreased between 1.4 and 8.8 mph at this site after installing the OSBs.

In the second after period, there were several additional significant changes between the before and second after period and between the two after periods. Speeds generally remained the same at the control point. Only all observations and heavy trucks had a significant reduction in speeds, with speed reductions of approximately 1.3 and 7.0 mph, respectively. Speeds at the approach were lower than in the first after period, but they were still significantly higher in the second after period for the following speed metrics: all observations, passenger cars, nighttime passenger cars, and opposed passenger cars, with speed increases of approximately 1.3, 1.2, 1.9, and 4.3 mph, respectively. Despite these significant increases at the approach, the speeds at the PC remained constant from the before to second after period with the exception of nighttime passenger cars. There was a significant increase of approximately 2.9 mph; however, after accounting for the speed increase at the approach, this increase was no longer significant. Speeds increased slightly at the curve midpoint in the second after period when compared with the first after period for several speed metrics, but speeds were lower than the before period for all speed metrics. These speed reductions were approximately 3.2 to 4.75 mph.

Overall, the OSBs significantly reduced speeds at the curve midpoint in both the first and second after periods.

Table 69. Before-after operating speeds at Alabama Location #6.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	43.413	9.211	45.102	7.201	46.442	6.431	45.495	7.111	196	86	97
All observations 1st after	41.382	9.064	54.627	8.674	44.879	6.628	40.780	7.109	220	190	205
All observations 2nd after	42.082	10.108	46.412	7.360	47.644	7.706	41.762	7.679	478	104	433
Passenger cars before	43.399	9.294	45.426	6.939	46.447	6.469	45.613	7.199	188	85	93
Passenger cars 1st after	41.371	9.028	54.709	8.760	45.005	6.642	40.848	7.159	213	183	198
Passenger cars 2nd after	42.326	10.069	46.652	7.339	47.859	7.775	41.925	7.674	457	99	415
Heavy trucks before	43.750	7.498	37.500	9.457	46.000	0.000	42.750	4.272	8	1	4
Heavy trucks 1st after	41.714	10.889	52.143	5.242	41.571	5.682	38.857	5.581	7	7	7
Heavy trucks 2nd after	36.762	9.705	41.190	5.828	43.400	4.930	38.000	6.979	21	5	18
Daytime passenger cars before	43.504	9.135	45.434	6.731	46.586	6.489	45.032	7.759	129	58	62
Daytime passenger cars 1st after	41.937	8.633	54.931	8.493	45.219	6.581	40.810	7.160	175	151	163
Daytime passenger cars 2nd after	42.604	9.832	46.247	7.189	46.725	7.582	41.664	7.769	283	51	253
Nighttime passenger cars before	43.169	9.708	45.407	7.433	46.148	6.538	46.774	5.869	59	27	31
Nighttime passenger cars 1st after	38.763	10.393	53.684	9.957	44.000	6.942	41.029	7.258	38	32	35
Nighttime passenger cars 2nd after	41.874	10.455	47.310	7.552	49.063	7.875	42.333	7.530	174	48	162
Opposed passenger cars before	41.056	7.565	42.944	10.067	43.400	4.881	46.750	4.070	18	10	12
Opposed passenger cars 1st after	40.273	10.612	52.545	6.743	43.545	6.362	37.909	7.368	11	11	11
Opposed passenger cars 2nd after	42.824	11.301	47.235	6.602	44.800	6.301	43.605	6.670	51	5	43
Unopposed passenger cars before	43.647	9.443	45.688	6.508	46.853	6.571	45.444	7.556	170	75	81
Unopposed passenger cars 1st after	41.431	8.961	54.827	8.855	45.099	6.666	41.021	7.129	202	172	187
Unopposed passenger cars 2nd after	42.264	9.917	46.579	7.431	48.021	7.840	41.731	7.767	406	94	372

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 110 shows a graphical representation of the speed profiles for Alabama Location #6. The general shape of the speed profiles for the three collection periods remained relatively similar.

Figure 110. Graph. Operating speeds comparison at Alabama Location #6 (PSL = 40 mph).

Alabama Location #7

As table 70 shows, in the first after period at Alabama Location #7, operating speeds changed for many speed metrics. The only significant change at the control point was heavy truck speeds, which decreased by approximately 6.0 mph. Speeds at the approach decreased for all measured speed metrics. These speed decreases were between approximately 2.2 and 2.9 mph for passenger cars and 5.6 mph for heavy trucks. However, if the speed decrease at the control point was taken into account, this speed decrease was no longer significant. Speeds at the PC significantly decreased for the following speed metrics: all observations, passenger cars, daytime passenger cars, opposed passenger cars, and unopposed passenger cars. Speed decreases were practically insignificant, with speed decreases between approximately 0.7 and 1.3 mph. Speeds at the curve midpoint significantly decreased for all measured speed metrics. These speed decreases were between approximately 2.8 and 3.6 mph for passenger cars. Speeds for heavy trucks decreased by approximately 7.3 mph. However, if the speed decrease at the control point was taken into account, this speed decrease was no longer significant. After installing the OSBs, speeds initially decreased through the curve.

Many of the changes in the first after period were also present in the second after period. There were no significant speed changes at the control point. Speeds at the approach increased from the first after period, but speeds were still significantly lower than the before period for all speed metrics, except heavy trucks. Speed decreases were between approximately 1.4 and 1.7 mph. Speeds at the PC were similar to speeds in the first after period and were significantly lower for all passenger car speed metrics, except opposed passenger cars. These speed decreases were small, with decreases between approximately 0.8 and 1.2 mph. Passenger car speeds at the curve midpoint increased from the first after period, but they were all still lower than the before period, except opposed passenger car speeds were not significantly lower. These speed decreases were between approximately 1.2 and 1.9 mph.

Overall, speeds decreased at the approach, PC, and curve midpoint after the installation of the OSBs. However, these speed decreases were not practically significant.

Table 70. Before-after operating speeds at Alabama Location #7.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All Observations before	53.159	9.499	43.648	5.400	36.950	4.293	38.819	5.257	429	422	419
All Observations 1st after	52.538	9.914	41.247	5.013	36.176	4.156	35.527	4.771	580	574	579
All Observations 2nd after	53.113	9.203	42.260	5.270	36.148	4.034	37.234	5.337	795	764	745
Passenger cars before	53.036	9.497	43.635	5.391	36.971	4.310	38.768	5.230	419	412	409
Passenger cars 1st after	52.546	9.918	41.341	4.992	36.249	4.150	35.594	4.777	560	554	559
Passenger cars 2nd after	53.127	9.141	42.285	5.234	36.164	4.048	37.225	5.259	780	751	730
Heavy trucks before	58.300	8.433	44.200	6.033	36.100	3.604	40.900	6.208	10	10	10
Heavy trucks 1st after	52.300	10.069	38.600	5.020	34.150	3.884	33.650	4.320	20	20	20
Heavy trucks 2nd after	52.400	12.362	41.000	7.000	35.231	3.059	37.667	8.566	15	13	15
Daytime passenger cars before	53.408	8.731	44.387	5.235	37.646	4.548	39.175	5.316	191	189	189
Daytime passenger cars 1st after	52.361	9.748	41.537	5.062	36.360	4.189	35.566	4.962	402	397	401
Daytime passenger cars 2nd after	53.165	9.232	42.644	5.289	36.448	4.154	37.260	5.336	514	500	481
Nighttime passenger cars before	52.724	10.104	43.004	5.450	36.399	4.020	38.418	5.141	228	223	220
Nighttime passenger cars 1st after	53.019	10.354	40.842	4.788	35.968	4.048	35.665	4.284	158	157	158
Nighttime passenger cars 2nd after	53.053	8.980	41.590	5.065	35.598	3.773	37.157	5.118	266	251	249
Opposed passenger cars before	52.148	8.668	43.241	5.701	36.657	4.310	38.355	4.989	108	105	107
Opposed passenger cars 1st after	51.773	10.502	41.000	4.558	35.625	3.671	35.042	4.632	97	96	96
Opposed passenger cars 2nd after	53.025	8.605	41.893	5.107	36.182	3.870	37.514	4.969	159	154	148
Unopposed passenger cars before	53.344	9.763	43.772	5.282	37.078	4.312	38.914	5.313	311	307	302
Unopposed passenger cars 1st after	52.708	9.795	41.413	5.079	36.380	4.235	35.708	4.803	463	458	463
Unopposed passenger cars 2nd after	53.153	9.280	42.385	5.266	36.159	4.096	37.151	5.332	621	597	582

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 111 shows a graphical representation of the speed profiles for Alabama Location #7. The general shape of the speed profiles for the three collection periods remained relatively similar.

Figure 111. Graph. Operating speeds comparison at Alabama Location #7 (PSL = 35 mph).

Alabama Location #8

As table 71 shows, in the first after period at Alabama Location #8, operating speeds remained the same for most metrics. The only significant change at the control point was daytime passenger car speeds, which decreased by approximately 1.9 mph. There were no significant speed changes at the approach or PC for any speed metric. Speeds decreased at the curve midpoint for all speed metrics, except heavy trucks. The speed decreases were between approximately 2.7 and 3.5 mph. After the OSBs were installed, speeds significantly decreased at the curve midpoint during the first after period.

There were more speed changes in the second after period, but most these changes were from the first after period and not the before period. The only significant changes at the control point compared with the before period were speeds for all observations, passenger cars, and nighttime passenger cars increased, with increases of approximately 0.9, 0.9, and 1.2 mph, respectively. There were no significant speed changes at the approach when compared with the before period. Speeds did not significantly change at the PC, except heavy truck speeds increased by approximately 4.0 mph. Speeds at the curve midpoint decreased for the following speed metrics: all observations, passenger cars, daytime passenger cars, and unopposed passenger cars, with speed decreases of approximately 1.0, 1.0, 1.3, and 1.5 mph. Based on the second after period, speeds at the curve midpoint slightly decreased after applying the OSBs, but these speed reductions were not practical, with less than a 2.0-mph reduction.

Table 71. Before-after operating speeds at Alabama Location #8.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	41.540	6.029	44.644	8.437	40.657	5.331	42.239	6.340	376	341	310
All observations 1st after	41.095	6.122	43.958	6.022	40.503	4.697	39.032	5.118	379	370	375
All observations 2nd after	42.467	5.824	45.129	6.511	41.091	4.821	41.275	6.115	614	571	538
Passenger cars before	41.566	6.053	44.672	8.317	40.751	5.334	42.348	6.349	366	333	302
Passenger cars 1st after	41.108	6.173	43.919	5.967	40.560	4.701	39.058	5.138	369	361	365
Passenger cars 2nd after	42.416	5.734	45.213	6.434	41.106	4.798	41.320	6.144	587	548	515
Heavy trucks before	40.600	5.254	43.600	12.624	36.750	3.655	38.125	4.612	10	8	8
Heavy trucks 1st after	40.600	3.978	45.400	8.072	38.222	4.116	38.100	4.458	10	9	10
Heavy trucks 2nd after	43.593	7.582	43.296	7.927	40.739	5.429	40.261	5.446	27	23	23
Daytime passenger cars before	43.145	5.833	44.612	8.251	40.842	4.928	42.375	5.922	165	146	136
Daytime passenger cars 1st after	41.245	6.185	43.931	5.902	40.428	4.732	38.951	5.148	306	299	303
Daytime passenger cars 2nd after	42.984	5.802	45.665	6.153	40.865	4.563	41.108	6.125	367	347	334
Nighttime passenger cars before	40.269	5.934	44.721	8.391	40.679	5.642	42.325	6.696	201	187	166
Nighttime passenger cars 1st after	40.444	6.117	43.857	6.319	41.194	4.533	39.581	5.101	63	63	62
Nighttime passenger cars 2nd after	41.468	5.502	44.459	6.825	41.522	5.164	41.713	6.177	220	201	181
Opposed passenger cars before	41.750	6.708	42.841	7.546	40.823	6.354	42.000	7.255	88	79	71
Opposed passenger cars 1st after	40.991	6.438	42.616	5.091	39.773	4.692	38.545	4.781	112	110	110
Opposed passenger cars 2nd after	42.582	5.029	45.277	6.065	40.983	4.497	42.173	5.403	184	177	168
Unopposed passenger cars before	41.507	5.842	45.252	8.477	40.728	4.988	42.455	6.057	278	254	231
Unopposed passenger cars 1st after	41.160	6.066	44.486	6.235	40.904	4.673	39.278	5.279	257	251	255
Unopposed passenger cars 2nd after	42.340	6.032	45.184	6.603	41.164	4.940	40.908	6.439	403	371	347

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 112 shows a graphical representation of the speed profiles for Alabama Location #8. The general shape of the speed profiles for the three collection periods remained relatively similar.

Figure 112. Graph. Operating speeds comparison at Alabama Location #8 (PSL = 35 mph).

Alabama Summary of Results

Overall, the optical speed bars had no effect on operating speeds at the Alabama sites. There were no changes in speeds that were consistent across all sites.

Massachusetts

This section of the report presents the before-after comparisons for the seven Massachusetts OSB sites.

Southbound Tucker Road, MA

As table 72 shows, operating speeds changed significantly at the southbound Tucker Road site in the after period. Oddly, speeds at the control point decreased between approximately 5.6 and 7.3mph for all passenger car speed metrics, and approximately 9.4 mph for heavy trucks. This difference in operating speeds at the control point cannot be explained, except it may be a result of a measurement error caused by the on-pavement sensor. Speeds in the after period remained relatively constant from the control point to the approach. Speeds increased slightly at the approach for the following speed metrics: all observations, passenger cars, daytime passenger cars, and unopposed passenger cars, with speed increases of approximately 0.8, 0.9, 1.2, and 1.1mph, respectively. These speed increases were minimal and were not practically significant. Speeds also increased at the PC for all measured speed metrics, with speeds increasing between approximately 3.0 and 3.7 mph for all passenger car speed metrics, and approximately 4.3 mph for heavy trucks. Speeds in the first after period remained relatively constant from the control point to the approach. There was a sensor malfunction at the curve midpoint, and no speeds were collected and no t-test was performed. Speeds at the PC increased between 3.0 and 3.7 mph for passenger cars after the OSBs were installed at this site.

Table 72. Before-after operating speeds at southbound Tucker Road.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	44.458	7.485	36.492	5.228	35.223	4.467	35.329	5.014	577	539	514
All observations after	37.651	5.164	37.337	5.164	38.547	6.650	0.000	0.000	341	139	0
Passenger cars before	44.466	7.510	36.506	5.236	35.241	4.475	35.334	5.022	573	535	512
Passenger cars after	37.753	5.147	37.428	5.169	38.569	6.692	0.000	0.000	332	137	0
Heavy trucks before	43.250	0.957	34.500	3.873	32.750	2.500	34.000	1.414	4	4	2
Heavy trucks after	33.889	4.567	34.000	3.873	37.000	2.828	0.000	0.000	9	2	0
Daytime passenger cars before	45.136	6.983	36.782	5.064	35.488	4.127	35.420	4.809	316	293	286
Daytime passenger cars after	38.616	5.033	37.966	5.145	38.493	7.690	0.000	0.000	203	71	0
Nighttime passenger cars before	43.642	8.048	36.167	5.431	34.942	4.855	35.226	5.289	257	242	226
Nighttime passenger cars after	36.395	5.049	36.581	5.114	38.652	5.476	0.000	0.000	129	66	0
Opposed passenger cars before	44.306	7.210	36.972	5.543	35.272	3.792	35.036	5.203	252	235	224
Opposed passenger cars after	38.686	5.503	37.822	5.583	38.390	8.812	0.000	0.000	118	41	0
Unopposed passenger cars before	44.592	7.745	36.140	4.960	35.217	4.951	35.566	4.874	321	300	288
Unopposed passenger cars after	37.238	4.876	37.210	4.926	38.646	5.603	0.000	0.000	214	96	0

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 113 shows a graphical representation of the speed profiles for southbound Tucker Road. In the after period, mean and 85th percentile speeds decreased significantly at the control point while they increased at the PC. Midpoint speeds were not available because of a sensor malfunction.

Figure 113. Graph. Operating speeds comparison at southbound Tucker Road
(PSL = 30 mph).

Northbound Tucker Road, MA

As table 73 shows, operating speeds significantly changed at northbound Tucker Road, but the changes did not follow a clear trend. Speeds at the control point did not change in the first after period. Speeds decreased slightly at the approach for the following speed metrics: all observations, heavy trucks, daytime passenger cars, and opposed passenger cars, with speed decreases of approximately 0.8, 0.7, 0.7, and 0.9 mph, respectively. These changes in speeds were minimal and not practically significant. Speeds at the PC increased between approximately 1.8 and 3.1 mph for all speed metrics. However, speeds at the curve then decreased for all speed metrics, except heavy trucks. Speeds at the curve midpoint decreased between approximately 1.1and 1.7 mph. There was no strong trend in the effects of the OSBs at this site, with speeds significantly increasing at the PC but then decreasing at the curve midpoint.

Table 73. Before-after operating speeds at northbound Tucker Road.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	33.895	4.361	37.609	4.247	33.381	3.820	35.507	4.852	573	565	458
All observations after	33.761	4.403	36.850	5.004	35.768	4.249	33.904	5.674	394	367	355
Passenger cars before	33.919	4.361	37.633	4.251	33.403	3.820	35.525	4.858	569	561	455
Passenger cars after	33.903	4.347	36.911	5.037	35.821	4.230	33.980	5.645	383	358	348
Heavy trucks before	30.500	3.000	34.250	1.500	30.250	2.500	32.667	3.055	4	4	3
Heavy trucks after	28.818	3.516	34.727	3.133	33.667	4.743	30.143	6.283	11	9	7
Daytime passenger cars before	34.135	4.263	37.768	4.219	33.683	3.551	36.145	4.549	341	338	276
Daytime passenger cars after	34.416	3.787	37.024	4.764	36.771	3.545	34.455	5.391	209	192	187
Nighttime passenger cars before	33.596	4.495	37.430	4.300	32.978	4.167	34.570	5.168	228	223	179
Nighttime passenger cars after	33.287	4.877	36.776	5.358	34.723	4.680	33.429	5.896	174	166	161
Opposed passenger cars before	33.681	4.172	37.453	4.207	33.369	3.635	35.839	4.758	351	347	280
Opposed passenger cars after	34.126	4.032	36.581	4.621	36.017	3.937	34.320	4.864	191	179	172
Unopposed passenger cars before	34.303	4.634	37.922	4.315	33.458	4.111	35.023	4.986	218	214	175
Unopposed passenger cars after	33.682	4.640	37.240	5.412	35.626	4.506	33.648	6.313	192	179	176

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 114 shows a graphical representation of the speed profiles for northbound Tucker Road. In the after period, the mean and 85th percentile speeds increased slightly at the PC and midpoints. The control point and approach speeds remained relatively similar during both the before and after data-collection periods.

Figure 114. Graph. Operating speeds comparison at northbound Tucker Road
(PSL = 35 mph).

Southbound Reed Road, Massachusetts

As table 74 shows, operating speeds changed significantly at the southbound Reed Road site in the first after period. Operating speeds decreased slightly at the control point for the following speed metrics: all observations, passenger cars, nighttime passenger cars, opposed passenger cars, and unopposed passenger cars, with speed decreases of approximately 1.2, 1.2, 2.7, 1.2, and 1.1 mph, respectively. These speed decreases were not practically significant, except for the nighttime speed decrease of 2.7 mph. There was a sensor malfunction at the approach, so no speeds were collected, but the research team performed a t-test. Speeds increased significantly for all passenger car speed metrics at the PC, with speed increases between approximately 1.4and 3.8 mph. These speed increases become more significant and larger in magnitude if the speed decrease at the control point was taken into account. However, speeds were then significantly lower at the curve midpoint for all passenger car speed metrics. These speed decreases were approximately 2.7 to 4 mph. There was no clear trend in operating speeds at this site after the OSBs were installed. Operating speeds increased at the PC in the after period, but then operating speeds decreased at the curve midpoint.

Table 74. Before-after operating speeds at southbound Reed Road.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	39.966	5.426	39.382	6.015	38.826	4.288	39.363	4.990	775	482	590
All observations after	38.723	4.995	0.000	0.000	41.525	6.728	36.190	5.133	328	324	327
Passenger cars before	40.036	5.395	39.450	5.985	38.811	4.227	39.416	4.945	756	471	575
Passenger cars after	38.831	4.957	0.000	0.000	41.541	6.732	36.232	5.111	320	316	319
Heavy trucks before	37.211	6.070	36.684	6.758	39.455	6.654	37.333	6.355	19	11	15
Heavy trucks after	34.375	4.868	0.000	0.000	40.875	6.978	34.500	6.094	8	8	8
Daytime passenger cars before	39.754	5.282	39.249	6.262	38.558	3.773	39.692	4.767	337	224	247
Daytime passenger cars after	39.691	4.417	0.000	0.000	42.326	7.233	36.963	4.597	188	184	187
Nighttime passenger cars before	40.263	5.481	39.611	5.754	39.040	4.596	39.207	5.072	419	247	328
Nighttime passenger cars after	37.606	5.422	0.000	0.000	40.447	5.816	35.197	5.618	132	132	132
Opposed passenger cars before	39.971	5.365	39.753	6.653	38.669	3.911	39.469	4.916	377	242	275
Opposed passenger cars after	38.781	4.926	0.000	0.000	41.372	6.445	36.445	4.910	228	226	227
Unopposed passenger cars before	40.100	5.432	39.148	5.227	38.961	4.542	39.367	4.979	379	229	300
Unopposed passenger cars after	38.957	5.056	0.000	0.000	41.967	7.427	35.707	5.570	92	90	92

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 115 shows a graphical representation of the speed profiles for southbound Reed Road. In the after period, mean speeds increased at the midpoint while 85th percentile speeds increased at the PC point. Approach speeds were not available because of a malfunctioning sensor.

Figure 115. Graph. Operating speeds comparison at southbound Reed Road
(PSL = 25 mph).

Southbound New Boston Road, MA

Overall, operating speeds were higher in the after period at southbound New Boston Road. As table 75 shows, operating speeds increased for all passenger car speed metrics, with speed increases between approximately 1.3 to 4.2 mph. Speeds increased at the approach for the following speed metrics: all observations, passenger cars, heavy trucks, and opposed passenger cars, with speed increases of approximately 1.0, 0.9, 6.1, and 3.2 mph, respectively. However, if the speed increase at the control point was taken into account, speeds at the approach for all observations and passenger cars actually decreased slightly. Speeds increased at the PC for all passenger car speed metrics, with speed increases between approximately 1.6 and 2.7 mph. However, if the speed increase at the control point was taken into accounted, the speeds at the PC did not change from the before period for any passenger car speed metric. Speeds at the curve midpoint increased for all speed metrics, with speed increases between approximately 0.9 and 2.6 mph for passenger cars and 5.6 mph for heavy trucks. Once again, none of these speed increases for passenger cars were significant if the speed increase at the control point was taken into accounted. Operating speeds were higher at southbound New Boston Road in the after period, but the OSBs had no effect on speeds after accounting for higher speeds at the control point.

Table 75. Before-after operating speeds at southbound New Boston Road.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	37.004	6.532	37.704	5.967	37.489	5.607	36.624	6.206	257	229	229
All observations after	39.003	6.359	38.675	5.154	39.345	5.446	38.025	6.192	381	293	325
Passenger cars before	36.968	6.565	37.766	5.980	37.465	5.636	36.656	6.224	252	226	227
Passenger cars after	38.962	6.366	38.630	5.160	39.323	5.470	38.013	6.211	373	288	318
Heavy trucks before	38.800	4.712	34.600	4.722	39.333	2.082	33.000	0.000	5	3	2
Heavy trucks after	40.875	6.081	40.750	4.713	40.600	4.037	38.571	5.653	8	5	7
Daytime passenger cars before	36.690	5.930	37.655	5.583	37.576	4.603	36.760	5.551	113	99	100
Daytime passenger cars after	38.567	6.228	38.424	5.389	39.403	5.390	38.073	6.152	210	154	178
Nighttime passenger cars before	37.194	7.052	37.856	6.302	37.378	6.343	36.575	6.728	139	127	127
Nighttime passenger cars after	39.472	6.525	38.896	4.852	39.231	5.579	37.936	6.307	163	134	140
Opposed passenger cars before	35.638	5.307	36.259	5.872	36.788	5.410	35.962	6.145	58	52	52
Opposed passenger cars after	39.796	6.541	39.500	5.061	39.500	5.780	38.525	6.375	108	86	99
Unopposed passenger cars before	37.366	6.859	38.216	5.953	37.667	5.702	36.863	6.250	194	174	175
Unopposed passenger cars after	38.623	6.275	38.275	5.167	39.248	5.346	37.781	6.136	265	202	219

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 116 shows a graphical representation of the speed profiles for southbound New Boston Road. In the after period, 85th percentile speeds increased at the PC and midpoint. All other speed profiles remained relatively similar in the before and after data-collection periods.

Figure 116. Graph. Operating speeds comparison on southbound New Boston Road (PSL = 35 mph).

Northbound New Boston Road, MA

As table 76 shows, operating speeds changed significantly at the northbound New Boston Road site in the after period. All observations and passenger car speeds increased at the control point by approximately 1.0 and 0.9 mph, respectively. These slight speed increases do not have any practical significance. Approach speeds increased for the following speed metrics: all observations, passenger cars, heavy trucks, nighttime passenger cars, and unopposed passenger cars, with speed increases of approximately 1.5, 1.4, 6.8, 2.3, and 1.5 mph, respectively. Speeds at the PC did not change in the after period. Speeds at the curve midpoint increased between approximately 2.8 and 4.5 mph for all speed metrics, except heavy trucks. Speeds significantly increased at the curve midpoint after the application of the OSBs at this site. Overall, the installation of the OSBs did not result in reduced operating speed at this site.

Table 76. Before-after operating speeds at northbound New Boston Road.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	39.796	5.704	36.876	5.691	39.500	4.955	33.742	6.553	225	114	209
All observations after	40.769	7.248	38.356	6.573	39.559	6.720	37.568	7.593	320	118	229
Passenger cars before	39.857	5.680	36.928	5.686	39.558	4.939	33.720	6.563	223	113	207
Passenger cars after	40.780	7.267	38.366	6.555	39.535	6.781	37.546	7.608	314	114	227
Heavy trucks before	33.000	5.657	31.000	2.828	33.000	0.000	36.000	7.071	2	1	2
Heavy trucks after	40.167	6.706	37.833	8.159	40.250	5.315	40.000	7.071	6	4	2
Daytime passenger cars before	40.074	5.352	37.475	5.801	40.066	4.694	34.627	6.531	122	61	110
Daytime passenger cars after	40.719	7.186	38.270	6.268	39.134	6.080	37.443	7.915	196	67	131
Nighttime passenger cars before	39.594	6.070	36.267	5.499	38.962	5.194	32.691	6.478	101	52	97
Nighttime passenger cars after	40.881	7.430	38.525	7.031	40.106	7.704	37.688	7.207	118	47	96
Opposed passenger cars before	40.119	4.835	36.810	4.830	39.714	3.989	32.789	4.911	42	21	38
Opposed passenger cars after	40.662	8.230	37.954	5.792	38.048	6.561	37.289	7.884	65	21	38
Unopposed passenger cars before	39.796	5.870	36.956	5.878	39.522	5.149	33.929	6.874	181	92	169
Unopposed passenger cars after	40.811	7.012	38.474	6.747	39.871	6.819	37.598	7.572	249	93	189

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 117 shows a graphical representation of the speed profiles for northbound New Boston Road. Mean and 85th percentile speeds increased at the midpoint in the after period. The general shape of the speed profiles at all other locations remained relatively similar.

Figure 117. Graph. Operating speeds comparison on northbound New Boston Road
(PSL = 35 mph).

Southbound Braley Hill Road, MA

As table 77 shows, operating speeds changed significantly at the southbound Braley Hill Road site in the after period. At the control point, daytime passenger car speeds increased by approximately 1.2 mph, and nighttime passenger car speeds decreased by approximately 1.7mph. Speeds at the approach were significantly higher for all speed metrics, with increases between 5.6 and 7.3 mph for the different passenger car speed metrics and 11.6 mph for heavy trucks. The speeds at the approach were approximately 5 to 6 mph higher than speeds at the control point in the after period. This does not reflect how drivers typically drive, so this significant increase at the approach may have resulted from a measurement error from the on-pavement sensor. Speeds then decreased at the PC for the following speed metrics: all observations, passenger cards, nighttime passenger cars, opposed passenger cars, and unopposed passenger cars, with speed decreases of approximately 1.2, 1.2, 1.6, 1.9, and 1.0 mph, respectively. The speed decreases at the PC were small in magnitude and not practically significant. Speeds at the curve midpoint also decreased in the after period for the following speed metrics: all observations, passenger cars, daytime passenger cars, nighttime passenger cars, and unopposed passenger cars, with speed decreases of approximately 2.2, 2.3. 1.7, 2.8, and 2.5 mph, respectively. Speeds throughout the curve decreased between 1 and 3 mph at this site after the application of the OSBs; this did not consider the significant increase at the approach speed because that appears to be the result of a measurement error.

Table 77. Before-after operating speeds at southbound Braley Hill Road.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	39.360	5.097	38.508	5.835	34.585	4.791	36.365	6.947	317	294	252
All observations after	39.040	5.667	45.191	7.031	33.415	4.160	34.176	7.407	277	260	256
Passenger cars before	39.472	5.136	38.600	5.872	34.665	4.693	36.502	6.905	305	284	243
Passenger cars after	39.208	5.645	45.075	6.760	33.520	4.174	34.244	7.497	265	250	246
Heavy trucks before	36.500	2.844	36.167	4.345	32.300	6.977	32.667	7.483	12	10	9
Heavy trucks after	35.333	5.033	47.750	11.671	30.800	2.821	32.500	4.577	12	10	10
Daytime passenger cars before	38.640	4.423	37.953	4.997	34.433	4.591	35.697	7.398	150	141	122
Daytime passenger cars after	39.801	4.950	45.298	6.401	33.701	4.205	34.038	7.345	141	134	132
Nighttime passenger cars before	40.277	5.640	39.226	6.566	34.895	4.797	37.314	6.297	155	143	121
Nighttime passenger cars after	38.532	6.297	44.823	7.164	33.310	4.146	34.482	7.696	124	116	114
Opposed passenger cars before	39.211	4.485	37.958	5.247	34.758	4.736	35.185	6.355	71	66	54
Opposed passenger cars after	38.981	5.634	44.096	7.604	32.837	4.520	33.820	7.176	52	49	50
Unopposed passenger cars before	39.551	5.324	38.795	6.046	34.638	4.691	36.878	7.025	234	218	189
Unopposed passenger cars after	39.263	5.660	45.315	6.534	33.687	4.080	34.352	7.591	213	201	196

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 118 shows a graphical representation of the speed profiles for southbound Braley Hill Road. In the after period, the mean and 85th percentile speeds increased significantly at the PC point. Speed profiles at all other points remained relatively similar during both the before and after data-collection periods.

Figure 118. Graph. Operating speeds comparison at southbound Braley Hill Road
(PSL = 30 mph).

Northbound Braley Hill Road, MA

As table 78 shows, operating speeds changed significantly at the northbound Braley Hill Road site in the after period. Oddly, speeds at the control point decreased between approximately 6.1and 7.8 mph for all speed metrics, except heavy trucks. Heavy truck speeds decreased by approximately 9.2 mph, but this was not significant because the SD was so high and there were few observations in both periods. This difference in operating speeds at the control point cannot be explained, except that it may be a result of a measurement error caused by the on-pavement sensor. Speeds at the approach did not change for any speed metric if the significant reductions in speeds at the control point were not taken into account. Speeds at the PC increased between approximately 2.8 and 4 mph for all speed metrics. These speed increases at the PC increase in magnitude and become more significant if the speed decreases at the control were taken into accounted. Speeds at the curve midpoint also increased between approximately 1.0 and 1.4 mph for all speed metrics, except heavy trucks. Heavy truck speeds decreased by approximately 2.7mph, but this was not significant because there were few observations in both periods. These speed increases at the curve midpoint increased in magnitude and become more significant if the speed decreases at the control were taken into account. Based on the first after period, speeds at the PC and curve midpoint significantly increased after the application of the OSBs. The speed increase at the PC was higher than at the curve midpoint, with the speed increase at the curve midpoint not being practically significant at a maximum speed increase of 1.4 mph for all speed metrics.

Table 78. Before-after operating speeds at northbound Braley Hill Road.

Speed Metric	Control Point Speed	Control Point SD	Approach Speed	Approach SD	PC Speed	PC SD	Mid Speed	Mid SD	N1	N2	N3
All observations before	44.188	7.842	38.650	4.290	37.565	3.981	38.530	3.990	277	271	268
All observations after	37.166	6.209	38.456	4.774	40.500	4.758	39.719	4.198	331	224	320
Passenger cars before	44.169	7.738	38.692	4.294	37.603	3.993	38.572	3.992	273	267	264
Passenger cars after	37.185	6.201	38.469	4.759	40.541	4.795	39.748	4.168	324	218	313
Heavy trucks before	45.500	15.022	35.750	3.096	35.000	1.826	35.750	3.096	4	4	4
Heavy trucks after	36.286	7.017	37.857	5.815	39.000	2.966	38.429	5.623	7	6	7
Daytime passenger cars before	44.238	6.538	38.524	4.084	37.431	3.703	38.565	3.823	147	144	147
Daytime passenger cars after	37.994	5.366	38.893	4.567	40.518	4.922	39.921	4.115	169	137	164
Nighttime passenger cars before	44.087	8.965	38.889	4.536	37.805	4.315	38.581	4.210	126	123	117
Nighttime passenger cars after	36.303	6.910	38.006	4.934	40.580	4.604	39.557	4.230	155	81	149
Opposed passenger cars before	43.794	7.456	38.730	4.178	36.983	4.312	38.381	4.046	63	60	63
Opposed passenger cars after	37.703	5.891	38.099	3.981	40.457	5.413	39.644	4.193	91	70	90
Unopposed passenger cars before	44.281	7.835	38.681	4.338	37.783	3.889	38.632	3.983	210	207	201
Unopposed passenger cars after	36.983	6.319	38.614	5.031	40.581	4.493	39.789	4.166	233	148	223

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Figure 119 shows a graphical representation of the speed profiles for northbound Braley Hill Road. The mean and 85th percentile control point speeds decreased in the after period while speed profiles at all points remained relatively similar in both the before and after period.

Figure 119. Graph. Operating speeds comparison at northbound Braley Hill Road
(PSL = 40 mph).

Massachusetts Results Summary

Overall, the OSBs did not have an effect on operating speeds at the Massachusetts sites. There were no changes in speeds that were consistent across all of the sites.

Speed Difference Analysis

The next measures computed using the speed data were the change in speed from the approach to PC and the change in speed from the control point to the midpoint of the horizontal curve. These measures were taken in the before and both after periods at each of the treatment sites. As discussed previously, the research team used a t-test for independent samples to compare the two after periods with the before period and to each other. A positive value of delta speed indicates that a mean decrease in speed occurred from the approach to PC or from the control point to the curve midpoint. A negative value represents a mean increase in the speed difference between the two locations.

Arizona

Table 79 shows the change in speed (delta) from the approach to the PC and from the control point to the curve midpoint.

Table 79. Change in speeds for Arizona sites.

Speed Metric	Delta Speed (Approach to PC)	Standard Deviation	Delta Speed (Control Point to Midpoint)	Standard Deviation	N1	N2
NB Pierce Ferry Road before	0.267	4.527	10.514	7.776	90	107
NB Pierce Ferry Road 1st after	1.059	5.513	12.806	8.689	102	180
NB Pierce Ferry Road 2nd after	5.733	4.654	5.392	7.379	161	148
SB County Route 1 before	4.142	3.903	8.907	9.947	141	129
SB County Route 1 1st after	4.243	4.167	12.123	10.490	169	154
SB County Route 1 2nd after	2.216	3.438	5.183	10.640	97	60
SB Diamond Bar Road before	-1.535	4.277	3.152	7.979	99	99
SB Diamond Bar Road 1st after	-1.676	5.662	2.031	9.751	37	64
SB Diamond Bar Road 2nd after	1.547	3.856	11.600	8.375	53	105
SB Shinarump Road before	4.656	6.367	-3.973	7.219	90	73
SB Shinarump Road 1st after	4.043	4.798	-3.056	6.111	93	72
SB Shinarump Road 2nd after	3.149	6.500	3.566	9.846	101	136

Bold indicates significance at the 95-percent confidence level (α = 0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α = 0.05) between the 1st and 2nd after periods
NB = Northbound
SB = Southbound
PC = Point of Curvature
N₁ = Number of observations approach to PC
N₂ = Number of observations control point to midpoint

As table 79 shows, there was no difference in delta speed from the approach to the PC for any sites in the first after period. There was a significant increase in delta speed (control point to midpoint) of approximately 2.3 mph on northbound Pierce Ferry Road in the first after period. This was a result of vehicles having a slower speed at the curve midpoint in the first after period. There was also a significant increase in delta speed (control point to midpoint) of approximately 3.2 mph on southbound County Route 1. This was a result of vehicles travelling faster at the control point in the first after period.

There were more significant changes in delta speed in the second after period than there were in the first after period, with delta speed changing at every site. Delta speed (approach to PC) from the approach to the PC increased by approximately 5.5 mph on northbound Pierce Ferry Road, which was a result of higher speeds at the PC in the second after period. Conversely, delta speed (control point to midpoint) decreased by approximately 5.1 mph on northbound Pierce Ferry Road, which was a result of higher speeds at the curve midpoint in the second after period. Speeds significantly decreased at this site at the PC on northbound Pierce Ferry Road, but then they significantly increased at the curve midpoint in the second after period. Delta speed (approach to PC) decreased by approximately 1.9 mph on southbound County Route 1, which was a result of higher speeds at the curve approach, but even higher speeds at the PC. Delta speed (control point to midpoint) also decreased by approximately 3.7 mph on southbound County Route 1, which resulted from vehicles accelerating out of the curve. Delta speed (approach to PC) was actually negative in the before period on southbound Diamond Bar Road, representing speeds increased from the approach to PC. However, delta speed (approach to PC) was positive in the second after period at this site. There was a significant increase of approximately 3.1 mph in delta speed, which was a result of vehicles accelerating slightly from the control point to the approach in the second after period. Delta speed (control point to midpoint) changed drastically in the second after period on southbound Diamond Bar Road. Delta speed increased by approximately 8.4 mph, which was a result of vehicles decelerating at a much greater rate from the PC to the curve midpoint in the second after period. Finally, delta speed (control point to midpoint) was actually negative in the before period, but delta speed was positive in the second after period. Delta speed (control point to midpoint) increased by approximately 7.5 mph on southbound Shinarump Road, which resulted from vehicles decelerating from the PC to the curve midpoint in the second after period, where vehicles accelerated from the PC to the curve midpoint in the before period.

Overall, delta speed did not change in the first after period after the OSBs were installed at the Arizona sites, but delta speeds were significantly different at all sites in the second after period. However, there was no clear trend in the change in delta speed as a result of installing the OSBs. Delta speed from the PC to the approach increased by approximately 5.5 and 3.1 mph at two sites, while delta speed decreased by approximately 1.9 mph at another site. Delta speed from the control point to the curve midpoint decreased by approximately 5.1 and 3.7 mph at two sites, while delta speed increased by approximately 8.4 and 7.5 mph at two other sites. The change in delta speed varied both in direction and magnitude across the different Arizona sites, with no strong trend present. Overall, the OSBs did not have an effect on operating speeds at the Arizona sites. There were no changes in speeds that were consistent across all of the sites.

Alabama

Table 80 shows the change in speed (delta) from the approach to the PC and from the control point to the curve midpoint.

Table 80. Change in speeds for Alabama sites.

Speed Metric	Delta Speed (Approach to PC)	Standard Deviation	Delta Speed (Control Point to Midpoint)	Standard Deviation	N1	N2
Alabama Location #1 before	0.037	3.653	5.252	8.817	27	115
Alabama Location #1 1st after	-0.933	4.173	0.549	7.154	134	213
Alabama Location #1 2nd after	-0.685	3.947	8.695	10.378	73	226
Alabama Location #2 before	3.068	5.175	11.612	9.533	74	85
Alabama Location #2 1st after	3.127	3.909	9.533	7.555	220	246
Alabama Location #2 2nd after	2.526	3.984	10.210	7.864	234	276
Alabama Location #3 before	8.796	4.115	24.444	13.239	142	144
Alabama Location #3 1st after	6.418	3.244	30.227	13.700	225	234
Alabama Location #3 2nd after	7.160	3.253	27.872	12.219	306	298
Alabama Location #4 before	7.253	4.303	16.891	11.210	99	101
Alabama Location #4 1st after	8.534	3.981	16.823	9.590	236	237
Alabama Location #4 2nd after	8.560	4.400	14.489	10.179	241	229
Alabama Location #5 before	0.541	5.450	1.820	8.132	37	111
Alabama Location #5 1st after	0.097	4.304	-3.264	6.864	165	174
Alabama Location #5 2nd after	-0.210	3.920	-1.738	6.337	176	191
Alabama Location #6 before	-0.611	3.568	-0.964	12.250	54	56
Alabama Location #6 1st after	7.915	4.975	0.732	9.227	141	153
Alabama Location #6 2nd after	0.327	4.516	1.197	10.233	49	228
Alabama Location #7 before	6.827	3.501	14.381	8.213	127	126
Alabama Location #7 1st after	4.925	2.926	16.709	9.636	319	323
Alabama Location #7 2nd after	6.089	2.951	15.883	8.963	382	369
Alabama Location #8 before	3.838	4.295	0.290	6.497	105	100
Alabama Location #8 1st after	3.607	3.797	2.405	6.296	196	200
Alabama Location #8 2nd after	4.479	3.945	2.308	7.735	217	211

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
PC = Point of Curvature
N₁ = Number of observations at the PC
N₂ = Number of observations at the control point and approach

As table 80 shows, there were many changes in the delta speed between the approach and the PC and the delta speed between the control point and the curve midpoint in Alabama. As table 80 shows, the speed difference from the approach to the PC increased for two sites in the first after period, while the difference also decreased for two sites. The difference increased by approximately 8.5 mph at Alabama Location #6. This was a result of vehicles having a higher speed at the approach in the first after period. There was also a significant increase in delta speed of approximately 1.3 mph at Alabama Location #6. This was a result of vehicles having a slower speed at the PC in the first after period. The difference decreased by approximately 2.4 mph at Alabama Location #3. This was a result of vehicles having a slower speed at the approach in the first after period. There was also a significant decrease of approximately 1.9 mph at Alabama Location #7. This was a result of vehicles having a slower speed at the approach in the first after period.

Most of the significant changes in delta speed (approach to PC) from the first after period were also significant in the second after period. The delta speed at Alabama Location #6 returned closer to the before period, and the difference was no longer significant in the second after period. The delta speed at Alabama Location #3 returned closer to the before period, but the difference was still significant, with a decrease of approximately 1.6 mph. The delta speed at Alabama Location #4 was similar to the first after period, with a delta increase of approximately 1.3 mph from the before period. The delta speed at Alabama Location #7 returned closer to the before period, but the difference was still significant, with an increase of approximately 0.7 mph.

As table 80 shows, the speed difference from the control point to the curve midpoint increased for three sites in the first after period, while the difference also decreased for three sites. The delta speed at Alabama Location #1 significantly decreased by approximately 4.7 mph. This was a result of vehicles having a slower speed at the control point. There was also a decrease in delta speed at Alabama Location #5 of approximately 5.1 mph. This was a result of speeds actually being higher at the curve midpoint than at the control point. Delta speed also decreased at Alabama Location #2 by approximately 2.1 mph. This was a result speeds at the curve midpoint increasing in the first after period.

In addition to the sites where there was a delta speed (control point to midpoint) difference from the before to first after period, there were also other sites where there was a change in delta speed in the second after period. Delta speed actually increased at Alabama Location #1 by approximately 3.4 mph, while there was a decrease in the first after period. This was a result of vehicles having a higher speed at the control point. The delta speed at Alabama Location #8 of was similar to the first after period, with a delta increase of approximately 2.0 mph from the before period. This was a result of vehicles having a higher speed at the control point and a lower speed at the curve midpoint. There was an increase in delta speed at Alabama Location #3 of approximately 3.4 mph. This was a result of vehicles having a higher speed at the control point. There was a decrease in delta speed at Alabama Location #4 of approximately 2.4 mph. This was a result of control point speeds being slightly lower. There was a decrease in delta speed at Alabama Location #5 of approximately 3.6 mph. Speeds actually increased from the control point to the curve midpoint. There was an increase in delta speed at Alabama Location #7 of approximately 1.5 mph. This was a result of vehicles having a slower speed at the curve midpoint.

There were many changes in delta speed, but there was no clear trend in the change. Delta speed increased at several sites, while it also decreased at other sites. Many of the speed changes were a result of control point speeds changing from the before to after periods. These changes at the control point make it difficult to determine whether the OSBs had an effect on delta speed. The change in delta speed varied both in direction and magnitude across the different Alabama sites, with no strong trend present.

Massachusetts

Table 81 shows the change in speed (delta) from the approach to the PC and from the control point to the curve midpoint.

Table 81. Change in speeds for Massachusetts sites.

Speed Metric	Delta Speed (Approach to PC)	Standard Deviation	Delta Speed (Control Point to Midpoint)	Standard Deviation	N1	N2
NB Braley Hill Road before	0.714	2.685	5.769	5.783	91	91
NB Braley Hill Road after	-1.313	3.009	-2.042	5.485	80	96
NB New Boston Road before	-1.409	3.472	4.805	6.582	44	77
NB New Boston Road after	-0.551	3.731	3.373	9.249	49	102
NB Reed Road before	0.380	2.432	11.155	7.481	50	110
NB Tucker Road before	3.889	2.603	-1.931	5.164	72	58
NB Tucker Road after	1.027	3.820	0.293	6.415	75	75
SB Braley Hill Road before	4.021	4.786	2.402	7.960	94	82
SB Braley Hill Road after	11.126	5.021	5.765	7.726	103	102
SB Tucker Road before	0.957	3.440	9.717	7.580	117	120
SB Tucker Road after	0.976	5.015	-	-	42	-
SB Reed Road before	1.563	3.956	0.343	5.573	64	73
SB Reed Road after	-	-	3.566	6.116	-	53
SB New Boston Road before	0.125	3.485	-0.167	7.053	64	66
SB New Boston Road after	-0.809	3.465	0.1	6.208	110	120

Bold indicates significance at the 95-percent confidence level (α = 0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α = 0.05) between the 1st and 2nd after periods
NB = Northbound
SB = Southbound
PC = Point of Curvature
N₁ = Number of observations approach to PC
N₂ = Number of observations control point to midpoint

As table 81 shows, there were many changes in the delta speed between the approach and the PC and the delta speed between the control point and the curve midpoint in Massachusetts. Delta speed (approach to PC) on northbound Braley Hill Road decreased by approximately 2.0 mph. In the before period delta speed was positive, which represents a decrease in speed from the approach to the PC. However, in the after period delta speed (approach to PC) was negative. Delta speed (control point to midpoint) also decreased in the after period on northbound Braley Hill Road by approximately 7.8 mph. Delta speed (control point to midpoint) was also negative in the second after period, which was a result from the control point speed being significantly lower and vehicles maintaining a more constant speed throughout the curve. As previously discussed, this was most likely a result of measurement error with the on-pavement sensor at the control point. There was also a decrease in delta speed (approach to PC) on northbound Tucker Road of approximately 2.9 mph. However, delta speed (control point to midpoint) increased in the first after period by approximately 2.2 mph. Delta speed was negative in the before period, but was positive in the first after period. Next, delta speed (approach to PC) and delta speed (control point to midpoint) increased by approximately 7.1 and 3.4 mph, respectively, on southbound Braley Hill Road. The large increase in delta speed (approach to PC) was the result of a significant increase in speeds at the approach, which was most likely the result of a measurement error with the on-pavement sensor, as previously discussed. Delta speed (control point to midpoint) increased as a result of speeds being lower at the curve midpoint in the first after period. Next, there was an increase in delta speed (control point to midpoint) on southbound Reed Road of approximately 3.2 mph. This was a result of lower speeds at the curve midpoint in the after period. Because approach speeds were not collected at this site in the first after period, a delta speed between the approach and PC was not calculated. Finally, delta speed (approach to PC) on southbound New Boston Road decreased by approximately 0.9 mph. This was a minimal decrease, and vehicles essentially maintained their speed from the approach to PC in both the before and first after period.

Unlike the Arizona sites, there was a weak trend in the change of delta speed at the Massachusetts sites. If the change in delta speed between the approach and PC on southbound Braley Hill Road was ignored because there most likely was a problem with the on-pavement sensor, the delta speed between the approach and PC decreased at three sites and remained the same at two others. If the change in delta speed between the control point and curve midpoint on southbound Braley Hill Road was ignored because there most likely was a problem with the on-pavement sensor, the delta speed between the control point and curve midpoint decreased at two sites, increased at one site, and remained the same at two sites. The OSBs may have a small effect on the delta speed between the approach and the PC and the delta speed between the control point and curve midpoint by decreasing the difference slightly.

Analysis of Variance for Speed Differences

An analysis of variance was used to compare speed differences in the before and after periods. The results are organized and grouped by State.

Arizona

As table 82 shows, the speed difference between the control point and the curve midpoint in the first after period significantly increased by 2.82 mph, which was a result of speeds at the curve midpoint decreasing and speeds at the control point increasing. The speed difference also increased in the second after period; however, this increase was not statistically significant. After installing the additional OSBs, the speed difference did not decrease from the control point to curve midpoint after the novelty effect period.

Table 82. Speed difference between the control point and the curve midpoint.

Time Period	F Statistic	Significance	Magnitude of Difference (mph)
First after period	35.62	< 0.001	2.82
Second after period	1.35	0.246	0.57
Second after period (additional OSBs installed)	2.71	0.100	-0.92

Bold indicates significance at the 95-percent confidence level (α=0.05)
OSB = Optical Speed Bar

As table 83 shows, there was an increase in the speed difference from the approach to the PC in the first after period; however, this increase was not significant. There was a significant increase in the speed difference in the second after period and after installing the additional OSBs of 1.33and 1.41 mph, respectively. The increase in the difference from the approach to the PC in the second after period may show that the OSBs were effective in reducing drivers’ speeds while in the OSBs.

Table 83. Speed difference between the approach and the PC.

Time Period	F Statistic	Significance	Magnitude of Difference (mph)
First after period	8.39	0.004	0.73
Second after period	26.11	<0.001	1.33
Second after period (additional OSBs installed)	23.08	<0.001	1.41

Bold indicates significance at the 95-percent confidence level (α=0.05)
OSB = Optical Speed Bar

As table 84 shows, the speed difference from the control point to the PC in the first before period significantly increased by 2.20 mph. The speed difference also increased in the second after period and after installing the additional OSBs; however, these increases were not significant.

Table 84. Speed difference between the control point and the PC.

Time Period	F Statistic	Significance	Magnitude of Difference (mph)
First after period	26.00	<0.001	2.20
Second after period	1.11	0.292	0.47
Second after period (additional OSBs installed)	0.35	0.552	0.30

Bold indicates significance at the 95-percent confidence level (α=0.05)

Alabama

As table 85 shows, the speed difference from the control point to the curve midpoint in the first after period did not significantly change. The speed difference significantly increased in the second after period by approximately 0.68 mph. After installing the additional OSBs, the speed difference decreased slightly from the control point to the curve midpoint. This decrease was not practically significant.

Table 85. Speed difference between the control point and the curve midpoint.

Time Period	F Statistic	Significance	Magnitude of Difference (mph)
First after period	2.14	0.143	0.51
Second after period	4.38	0.036	0.68

Bold indicates significance at the 95-percent confidence level (α=0.05)

As table 86 shows, the speed difference from the approach to the PC significantly decreased in the after periods, which means that vehicles maintained a more constant speed as they approached the curve.

Table 86. Speed difference between the approach and the PC.

Time Period	F Statistic	Significance	Magnitude of Difference (mph)
First after period	34.65	<0.001	-0.81
Second after period	28.68	<0.001	-0.7

Bold indicates significance at the 95-percent confidence level (α=0.05)

As table 87 shows, the speed difference from the control point to the PC significantly decreased in the first after period, but there was no significant change in the second after period. The decrease in the first after period was approximately 1.68 mph. This corresponds with the decrease in the difference from the approach to the PC, and vehicles maintained a more constant speed from the control point to the PC.

Table 87. Speed difference control Point to PC.

Time Period	F Statistic	Significance	Magnitude of Difference (mph)
First after period	20.93	<0.001	-1.68
Second after period	0.05	0.829	0.08

Bold indicates significance at the 95-percent confidence level (α=0.05)

Massachusetts

As table 88, table 89, and table 90 show, all three speed differences decreased in the after period at the Massachusetts sites, with speed becoming more uniform throughout the horizontal curves. There was much variability in the speed changes at the different sites in Massachusetts, with the speed differences increasing at some sites, while decreasing at others. Overall, the analysis of variance shows the speed difference decreased from the control point to the curve midpoint, the approach to the PC, and the control point to the PC, with speed difference decreases of 1.80,1.88, and 4.87 mph, respectively.

Table 88. Difference between the control point and the curve midpoint.

Time Period	F Statistic	Significance	Magnitude of Difference (mph)
After period	48.78	<0.001	-1.80

Bold indicates significance at the 95-percent confidence level (α=0.05)

Table 89. Difference between the approach and the PC.

Time Period	F Statistic	Significance	Magnitude of Difference (mph)
After period	71.87	<0.001	-1.88

Bold indicates significance at the 95-percent confidence level (α=0.05)

Table 90. Difference between the control point and the PC.

Time Period	F Statistic	Significance	Magnitude of Difference (mph)
After period	508.45	<0.001	-4.87

Bold indicates significance at the 95-percent confidence level (α=0.05)

Vehicles Exceeding the Posted Speed Limit

Finally, in addition to the t-test, the percentage of vehicles exceeding the PSL and the advisory speed at the treatment and control sites was calculated and compared between data-collection periods. The following sections present the results for each State.

Arizona

Table 91 shows the number of passenger cars observed as exceeding the speed limit at each location at every site in Arizona in the before and both after periods. A test of proportions was used to determine whether the proportion of speeding vehicles changed at each location of every site from the before to after treatment application. From table 91, it was clear there were no visible trends from the before period to after treatment application. There was no change in the proportion of passenger cars exceeding the speed limit at the control point at any site, in any analysis period. The only change in proportion of passenger cars exceeding the speed limit at the approach was in the second after period on southbound Diamond Bar Road, where slightly fewer vehicles exceeded the speed limit. These results were expected because these data collection points were before or at the beginning of the OSBs, so no change would be expected. The only change at the PC between the before and after periods was the proportion of vehicles exceeding the PSL on northbound Pierce Ferry Road decreased in the second after period. The effect of the OSBs on the proportion of vehicles exceeding the speed limit varies from site to site and from the first to second after period. On northbound Pierce Ferry Road, the proportion decreased in the first after period, but then it increased in the second after period. On southbound County Route 1, the proportion decreased in the first after period, but then returned to the proportion in the before period. On southbound Diamond Bar Road, the proportion increased in the first after period, but then decreased in the second after period. On southbound Shinarump Road, the proportion decreased in the second after period. The fact that there was no trend in the change of the proportion of vehicles exceeding the speed limit and those with decreased speeds, while others increased, probably means the OSBs had no effect.

Table 91. Number of vehicles exceeding the PSL in Arizona.

Location and Data Collection Period		Speed Observation Site
		NB Pierce Ferry Rd	SB County Route 1	SB Diamond Bar Rd	SB Shinarump Rd
Control Point PCs	Before	83	145	115	55
	1st after	138	173	73	62
	2nd after	127	100	89	90
Approach PCs	Before	69	147	95	71
	1st after	106	174	63	70
	2nd after	101	102	96	101
PC PCs	Before	57	139	80	59
	1st after	60	166	25	52
	2nd after	39	97	47	61
Midpoint PCs	Before	21	122	80	55
	1st after	15	135	43	56
	2nd after	77	58	30	62
Number of PCs at Control Point and Approach	Before	108	147	136	96
	1st after	186	175	92	98
	2nd after	162	102	108	146
Number of PCs at PC	Before	90	141	99	90
	1st after	102	169	37	93
	2nd after	161	97	53	101
Number of PCs at Midpoint	Before	107	129	99	73
	1st after	180	154	64	72
	2nd after	148	60	105	136

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
NB = Northbound
SB = Southbound
PC = Point of Curvature

Alabama

Table 92 shows the number of passenger cars observed to exceed the speed limit at each location at every site in Alabama in the before and both after periods. A test of proportions was used to determine whether the proportion of speeding vehicles changed at each location of every site from the before to after treatment application. From table 92, it was clear there were no long-term effects on the number of vehicles speeding after the application of the OSBs.

Alabama locations #1, #2, and #3 were the only sites where there was a change in the proportion of vehicles exceeding the speed limit at the control point. The proportion decreased at Alabama Location #1 in the first after period, but returned to the before period proportion in the second after period. The proportion decreased at Alabama Location #2 in the second after period, but increased at Alabama Location #3. The proportion of vehicles exceeding the speed limit at the approach decreased in the first after period at Alabama locations #1, #3, and #7, but the decreased proportions were only maintained through the second after period at Alabama Location #3. The proportion of vehicles exceeding the speed limit at the approach increased for Alabama Location #6 in the first after period and Alabama Location #8 in the second after period. The only change in the proportion of vehicles exceeding the speed limit at the PC was a decrease as in the first after period. The proportion of vehicles exceeding the speed limit decreased during the first after period at the following locations: #6, #8, #4, #2, and #7. However, the reductions were only maintained through the second after period at locations #6 and #7. The proportion of vehicles exceeding the speed limit increased at Location #5 during both after periods. There was no visible long-term trend in the effects of the OSBs on the proportion of vehicles exceeding the PSL. There was a decrease at the curve midpoint for only two sites in the second after period, but there was also an increase at another site. The fact that there was no consistent trend in the change of the proportion of vehicles exceeding the speed limit, but no change at most sites, probably means the OSBs had no effect.

Table 92. Number of vehicles exceeding the PSL in Alabama.

Location and Data Collection Period		Speed Observation Site
		Alabama Location #6	Alabama Location #1	Alabama Location #8	Alabama Location #3	Alabama Location #4	Alabama Location #5	Alabama Location #2	Alabama Location #7
Control Point PCs	Before	85	28	107	131	100	75	22	126
	1st after	109	5	171	214	226	135	40	314
	2nd after	180	50	216	302	239	177	43	383
Approach PCs	Before	96	24	109	31	99	90	2	122
	1st after	156	14	193	19	231	165	7	284
	2nd after	196	26	229	42	238	192	7	364
PC PCs	Before	47	9	92	0	86	30	0	89
	1st after	110	8	174	1	157	117	2	197
	2nd after	38	13	193	4	199	125	3	236
Midpoint PCs	Before	42	5	92	1	44	50	2	105
	1st after	79	3	152	0	62	128	0	168
	2nd after	119	7	178	7	108	143	1	230
Number of PCs at Control Point & Approach	Before	118	125	117	152	103	118	97	128
	1st after	165	218	201	235	237	244	250	323
	2nd after	253	256	234	323	244	277	303	393
Number of PCs at PC	Before	54	27	105	142	99	37	74	127
	1st after	141	134	196	225	236	165	220	319
	2nd after	49	73	217	306	241	176	303	382
Number of PCs at Midpoint	Before	56	115	100	144	101	111	85	126
	1st after	153	213	200	234	237	174	246	323
	2nd after	228	226	211	298	229	191	234	369

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
PC = Point of Curvature

Massachusetts

Table 93 shows the number of passenger cars observed that exceeded the speed limit at each location at every site in Massachusetts in the before and both after periods. A test of proportions was used to determine whether the proportion of speeding vehicles changed at each location of every site from the before to after treatment application. From table 93, it was clear there were no visible trends from the before period to after treatment application. Northbound Braley Hill Road and northbound Tucker Road were the only sites where there was a change in the proportion of vehicles exceeding the speed limit at the control point. The proportion decreased on northbound Braley Hill Road, but increased on northbound Tucker Road. Northbound Tucker Road was the only site where there was a change in the proportion of vehicles exceeding the speed limit at the approach. The proportion increased, which corresponds with the increase in the proportion at the control point. Northbound Braley Hill Road and northbound Tucker Road were the only sites where there was a change in the proportion of vehicles exceeding the speed limit at the PC. The proportion increased at both sites, which corresponds with the increase in the proportion at the control point of northbound Tucker Road. The increase on northbound Braley Hill Road does not correspond with the change at the control point because the proportion decreased there. Finally, these two sites were the only sites where there was a change in the proportion of vehicles exceeding the speed limit at the curve midpoint. The proportion increased at both sites, which corresponds with the increase in the proportion at the control point of northbound Tucker Road. Again, the increase does not correspond with the change at the control point, because the proportion decreased there. There was no visible trend in the effects of the OSBs on the proportion of vehicles exceeding the PSL. There was a change at two sites only, and at one site, there was also an increase in the proportion at the control point. The fact that there was no trend in the change of the proportion of vehicles exceeding the speed limit, with an increase at several sites but no change at most sites, probably means the OSBs had no effect.

Table 93. Number of vehicles exceeding the PSL in Massachusetts.

Location and Data Collection Period		Speed Observation Site
		NB Braley Hill Rd	NB New Boston Rd	NB Reed Rd	NB Tucker Rd	SB Braley Hill Rd	SB Tucker Rd	SB Reed Rd	SB New Boston Rd
Control Point PCs	Before	69	69	116	27	95	124	102	49
	1st after	33	116	-	68	106	110	53	110
	2nd after	0	0	0	0	0	0	0	0
Approach PCs	Before	28	54	114	51	95	116	101	56
	1st after	39	97	-	77	109	108	-	104
	2nd after	0	0	0	0	0	0	0	0
PC PCs	Before	18	38	50	27	79	103	64	43
	1st after	40	37	-	72	90	38	49	85
	2nd after	0	0	0	0	0	0	0	0
Midpoint PCs	Before	28	38	72	32	69	106	72	42
	1st after	45	64	-	58	77	-	51	85
	2nd after	0	0	0	0	0	0	0	0
Number of PCs at Control Point & Approach	Before	91	85	116	73	98	127	103	75
	1st after	100	143	-	81	109	114	53	148
	2nd after	0	0	0	0	0	0	0	0
Number of PCs at PC	Before	91	44	50	72	94	117	64	64
	1st after	80	49	-	75	103	42	51	110
	2nd after	0	0	0	0	0	0	0	0
Number of PCs at Midpoint	Before	91	77	110	58	82	120	73	66
	1st after	96	102	-	75	102	-	53	120
	2nd after	0	0	0	0	0	0	0	0

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
NB = Northbound
SB = Southbound
PC = Point of Curvature

Speed Variance Analysis

The final speed performance metric considered in the current study was speed variance. A two-sided F-test was used to compare the variances of vehicle operating speeds in the before and after periods for passenger cars and heavy trucks. The results are organized by State.

Arizona

Table 94 shows the SD of speeds at each location at every site in Arizona for the before and both after periods. From table 93, it was clear there were no visible trends in a change in SD from the before period to first after period. The SD of control point speeds did not change at any site in the first after period. Northbound Pierce Ferry Road was the only site where there was a change in the SD of approach speeds, which was an increase of approximately 1.3 mph. There was also an increase of approximately 1.2 mph in the SD of speeds at the PC of northbound Pierce Ferry Road. There was no change in the SD of speeds at the PC at any other site. Finally, there was no change in the SD of speeds at the curve midpoint of any site.

As in the first after period, there were no visible trends in the change in SD from the before to second after period. The only change at the control point was on southbound Shinarump Road, where the SD increased by approximately 1.3 mph. There was no change in the SD of the speeds at the approach. There was a change in the SD of speeds at the PC for both northbound Pierce Ferry Road and southbound Shinarump Road. The SD decreased by approximately 1.0 mph on northbound Pierce Ferry Road and increased by approximately 3.3 mph on southbound Shinarump Road. The increase on southbound Shinarump Road was still significant even after accounting for the increase at the control point. Finally, there was no change in the SD of speeds at the curve midpoint between the second after period and the before period. If the OSBs had any effect on the speed deviation, it would have been most prevalent at the PC and curve midpoint, because these were after the OSBs. The fact that there was no consistent trend in the change of the SD of speeds at most sites probably means the OSBs had no effect.

Table 94. Change in standard deviation of speed in Arizona.

Site	Control Point SD	Approach SD	PC SD	Midpoint SD	N1	N2	N3
NB Pierce Ferry Road before	7.308	7.121	5.795	6.695	108	90	107
NB Pierce Ferry Road 1st after	7.836	8.415	6.970	5.989	186	102	180
NB Pierce Ferry Road 2nd after	7.377	6.373	4.753	7.012	162	161	148
SB County Route 1 before	9.132	5.325	4.846	6.655	147	141	129
SB County Route 1st after	9.497	5.280	4.544	7.241	175	169	154
SB County Route 2nd after	9.920	5.562	5.347	7.531	102	97	60
SB Diamond Bar Road before	8.543	7.230	6.758	7.590	136	99	99
SB Diamond Bar 1st after	9.450	7.161	8.636	8.862	92	37	64
SB Diamond Bar 2nd after	9.244	8.364	6.781	7.215	108	53	105
SB Shinarump Road before	7.047	8.893	6.248	7.692	96	90	73
SB Shinarump Road 1st after	6.374	7.863	6.713	8.109	98	93	72
SB Shinarump Road 2nd after	8.309	9.902	9.530	6.813	146	101	136

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
NB = Northbound
SB = Southbound
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Alabama

Table 95 shows the SD of speeds at each location at every site in Alabama for the before and both after periods. From table 95, it was clear there were no visible trends in the change in SD from the before period to first after period. The only change in the SD at the control point was at Location #1, which was a decrease of approximately 1.8 mph. The OSBs were not visible at this location, which corresponds to the SDs remaining the same as the before period. The SD decreased at the approach at locations #8 and #4 by approximately 2.1 and 2.2 mph, respectively. The SD also decreased at the PC at Location #1 and Location #2 by approximately 3.4 and 1.5mph, respectively. The SD decreased at the curve midpoint at Location #4 and Location #2 by approximately 2.0 and 2.4 mph, respectively. However, the SD increased at the curve midpoint at Location #3 by approximately 2.4 mph.

As in the first after period, there were no visible trends in the change in the SD from the before to second after period. The SD decreased at the control point at Location #4 by approximately 1.7 mph. As in the first after period, the SD decreased at the approach at Location #8 and Location #3 by approximately 1.9 and 1.5 mph, respectively. The SD also increased at the approach at Location #6 by approximately 1.3 mph. The only change at the PC was the SD increased at Location #3 by approximately 0.6 mph. The SD increased at the curve midpoint at Location #1 by approximately 2.4 mph. As in the first after period, the SD increased at the curve midpoint at Location #3 by approximately 1.3 mph. The SD decreased at the curve midpoint at Location #2 by approximately 1.8 mph. If the OSBs had any effect on the speed deviation, it would have been most prevalent at the PC and curve midpoint, because these were after the OSBs. The SD decreased at the PC for two sites in the first after period, but these decreases did not persist throughout the second after period. The fact that there was no consistent trend in the change of the SD of speeds at most sites, with an increase at some sites, probably means the OSBs had no effect.

Table 95. Change in standard deviation of speed in Alabama.

Site	Control Point SD	Approach SD	PC SD	Midpoint SD	N1	N2	N3
Alabama Location #1 before	8.953	7.773	8.541	6.979	125	27	115
Alabama Location #11st after	7.113	6.823	5.181	6.102	218	134	213
Alabama Location #1 2nd after	7.959	6.901	7.593	9.344	256	73	226
Alabama Location #2 before	7.218	5.239	5.980	7.485	97	74	85
Alabama Location #2 1st after	6.986	4.882	4.532	5.038	250	220	246
Alabama Location #2 2nd after	7.132	5.105	5.221	5.651	303	234	276
Alabama Location #3 before	12.267	6.872	4.058	5.495	152	142	144
Alabama Location #3 1st after	11.686	4.676	4.349	7.731	235	225	234
Alabama Location #3 2nd after	11.413	5.385	4.655	6.809	323	306	298
Alabama Location #4 before	9.000	5.784	4.627	7.770	103	99	101
Alabama Location #4 1st after	8.434	5.462	4.126	5.790	237	236	237
Alabama Location #4 2nd after	7.286	5.883	4.677	8.131	244	241	229
Alabama Location #5 before	6.783	7.346	6.959	6.603	118	37	111
Alabama Location #5 1st after	6.687	6.808	6.126	7.155	244	165	174
Alabama Location #5 2nd after	6.704	7.059	6.159	7.098	277	176	191
Alabama Location #6 before	9.141	5.970	6.466	8.027	118	54	56
Alabama Location #6 1st after	8.521	8.573	6.578	7.108	165	141	153
Alabama Location #6 2nd after	9.615	7.279	7.619	7.857	253	49	228
Alabama Location #7 before	9.446	5.276	4.730	5.497	128	127	126
Alabama Location #7 1st after	9.866	5.229	4.307	5.052	323	319	323
Alabama Location #7 2nd after	9.392	5.449	4.246	5.431	393	382	369
Alabama Location #8 before	5.742	8.262	4.442	5.804	117	105	100
Alabama Location #8 1st after	6.192	6.198	4.723	5.335	201	196	200
Alabama Location #8 2nd after	6.249	6.372	4.895	6.486	234	217	211

Bold indicates significance at the 95-percent confidence level (α=0.05) between the 1st/2nd after period and before period
Italics indicates significance at the 95-percent confidence level (α=0.05) between the 1st and 2nd after periods
SD = Standard Deviation
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Massachusetts

Table 96 shows the SD of speeds at each location at every site in Massachusetts for the before and both after periods. Unlike at the Arizona sites, there may have been a change in the SD of speeds after installing the OSBs. The SD of speeds at the control point increased by approximately 1.6 mph on northbound New Boston Road, decreased by approximately 0.6 mph on northbound Tucker Road, and decreased by approximately 2.4 mph on southbound Tucker Road. These changes were not the result of applying the OSBs because they were not visible from the control point. The SD of speeds at the approach increased by approximately 1.0 mph on northbound Braley Hill Road, 1.3 mph on northbound Tucker Road, and 1.5 mph on southbound Braley Hill Road. The optical speeds bars may have affected these SDs as they would be visible prior to reaching the approach. The SD of speeds at the PC increased by approximately 1.1 mph on northbound Braley Hill Road, 4.1 mph at southbound Reed Road, and 0.8 mph on southbound New Boston Road. Finally, the SD of speeds at the curve midpoint increased by approximately 2.2 mph on northbound Tucker Road. All of these changes were still significant after accounting for the changes at the control point, because there was no change at the control point for any of these sites, except northbound Tucker Road, but the SD decreased at that site. The SD at the approach increased between 1.0 and 1.5 mph at three sites and remained the same at the other three in the after period. The SD at the PC increased between 0.8 and 4.1 mph at three sites and remained the same at the other four sites in the after period. The SD at the curve midpoint increased by 2.2 mph at one site and remained the same at the five other sites. It appears the SD may increase slightly at the approach and PC, while remaining the same at the curve midpoint after installing the OSBs.

Table 96. Change in standard deviation of speed in Massachusetts.

Site	Control Point SD	Approach SD	PC SD	Midpoint SD	N1	N2	N3
NB Braley Hill Road before	5.855	3.912	3.202	3.698	91	91	91
NB Braley Hill Road 1st after	5.390	4.908	4.269	3.891	100	80	96
NB New Boston Road before	5.523	6.106	4.877	7.002	85	44	77
NB New Boston Road 1st after	6.820	6.336	5.826	8.009	143	49	102
NB Tucker Road before	4.471	3.708	3.388	4.059	73	72	58
NB Tucker Road 1st after	3.653	5.005	3.625	6.242	81	75	75
SB Braley Hill Road before	4.519	4.577	4.536	7.847	98	94	82
SB Braley Hill Road 1st after	4.856	6.085	3.957	7.277	109	103	102
SB Tucker Road before	7.357	4.281	4.870	4.579	127	117	120
SB Tucker Road 1st after	4.964	4.673	5.861	0.000	114	42	0
SB Reed Road before	5.203	5.096	4.094	5.093	103	64	73
SB Reed Road 1st after	4.587	0.000	8.267	4.757	53	51	53
SB New Boston Road before	6.132	5.332	4.496	5.412	75	64	66
SB New Boston Road 1st after	5.952	5.376	5.277	6.162	148	110	120

Bold indicates significance at the 95-percent confidence level (α=0.05)
SD = Standard Deviation
NB = Northbound
SB = Southbound
PC = Point of Curvature
N₁ = Number of observations at the control point and approach
N₂ = Number of observations at the PC
N₃ = Number of observations at the midpoint

Summary

Results of this study showed that OSBs may have some minor effects on vehicle speeds. However, the magnitude of speed reductions was generally small, and because of the inconsistent magnitude of speed reductions at all the test sites, no conclusion can be drawn regarding the effect of OSB treatment to reduce speeds. Even though minor speed reductions occurred at some locations, average and 85th percentile speeds observed at all the sites were still higher than the PSLs, indicating the OSBs were not effective enough in providing the desired speed limit compliance. Future analyses may consider featuring an enforcement component, which was out of the scope of this study.

It is clear from the data that driver operating speeds are influenced by the curve radius. Further, the benefits gained by a reduction in driver operating speeds before the PC may be lost as drivers compensate and accelerate through the curve (for treatments that end prior to the PC). While this study did not show any consistent benefits as a result of installing the OSBs, consideration should be given to extending the treatment through the curve—keeping in mind that the entrance and midpoint are the most severe areas.