Factors Influencing Operating Speeds and Safety on Rural and Suburban Roads

CHAPTER 7. SUMMARY, CONCLUSIONS, AND CONSIDERATIONS

This section summarizes project findings, conclusions, and general considerations for future research.

High Friction Surface Treatment Evaluation Findings

The research team used observational before–after studies to evaluate HFST at four treatment sites and three control sites in West Virginia. The HFST was applied in June 2012 with the first after period data collection in August 2012 and the second after period data collection in June 2013. The research team performed operational, driver behavior, and friction evaluations. The following is a summary of the evaluation findings:

• Operating Speed Evaluation Results
  
  
  • Mean Operating Speed—The research team compared the mean operating speed for passenger cars and trucks before and at two different time periods after installing the HFST. The mean speeds were compared at a location on the approach tangent, curve PC location, and the midpoint of the horizontal curve. The analysis included only daytime and unopposed free-flow vehicle mean speeds. The results of the mean speed analysis indicate that the HFST did not significantly affect vehicle mean operating speeds in a consistent manner across all four treatment sites. Few of the independent samples t-tests showed that the mean passenger car and heavy truck operating speeds differed between the before and first or second after periods at the treatment sites. A multifactor ANOVA found that the data collection time period–treatment site interaction was not statistically significant (F_{2, 4160} = 0.49, p > 0.05) at the approach speed location, when including observations of free-flow vehicles only and controlling for the presence of vehicles in the opposing lane, time of day, and vehicle type. Similar results were found at the PC (F_2,4160 = 1.51, p > 0.05) and midcurve locations (F_2,4160 = 0.56, p > 0.05).
    
    
  • Change in Mean Operating Speed—The change in speed from the entry of the curve (PC) to the midpoint of the horizontal curve (referred to as delta speed hereafter) was also considered in the analysis. There is no consistent trend in the delta speed for passenger cars from the before to the second after period. There was a significant increase at two treatment sites, but there was also a very significant decrease at another treatment site. A multifactor ANOVA found that the treatment site–time period interaction was not statistically significant (F_2,4160 = 0.51, p > 0.05) when including observations of free-flow vehicles only and controlling for the presence of vehicles in the opposing lane, time of day, and vehicle type. This suggests that the HFST did not affect the change in vehicle speeds between the PC and midcurve locations in the current study.
    
    
  • Vehicles Exceeding PSL—The research team calculated and compared the percentage of vehicles exceeding the PSL and the advisory speed at the treatment and control sites. There were no visible trends from the before period to after treatment application in the number of passenger cars and heavy trucks observed to exceed the PSL. Alternatively, there appeared to be a slight reduction in the number of trucks exceeding the advisory speed, but there does not appear to be a strong effect on the number of passenger vehicles exceeding the advisory speed.
    
    
  • Speed Variance Analysis—A two-sided F-test was used to compare the variance of vehicle operating speed in the before and after periods for passenger cars and heavy trucks. The effects of the HFST on the speed deviation was considered for the PC and curve midpoint locations because the surface was applied starting at the PC, or just before it, and it was assumed that the surface would have no effect on the speed deviation at the approach data-collection location. There was no consistent trend in the effects of the HFST on the speed deviation of passenger cars. In the first after period, the speed deviation increased at the PC for two sites, but also decreased at two other sites. In the second after period, the speed deviation decreased at the PC for two sites. In the first after period, the speed deviation increased at the curve midpoint at one site, but the decreased at another site. In the second after period, the speed deviation increased at the curve midpoint for two sites, but also decreased at two sites. For trucks, the speed deviation decreased at the PC at one site in both after periods, and also decreased at one site in both after periods. There was no change in speed deviation at the curve midpoint for trucks at any site.
    
    
• Encroachment Evaluation Results
  
  
  • The purpose of the encroachment evaluation was to determine whether the HFST changed the proportion of vehicles that crossed either the edge line or centerline. There was no significant change in the braking and encroachment data that lasted throughout the treatment period except for an increase in the proportion of vehicles exiting the curve that encroached onto the shoulder at the U.S. 33 treatment site and also a decrease in the proportion of vehicles entering the curve that encroached onto the shoulder at the WV Route 32 treatment site. These results show the HFST does not have a consistent effect on the braking and encroachment behavior of vehicles
    
    .
• Friction Evaluation Results
  
  
  • The research team used the DF tester and CT meter to evaluate skid resistance of the HFST and comparison site locations. The friction level increased significantly from the before to the after period at the treatment sites, ranging from an increase of 0.21 to 0.30 in the SN65 value. At all four treatment sites, the second after period produced friction levels that were statistically higher than the before period, indicating the higher friction levels produced by the treatment were maintained for at least 1 year. For the margin of safety analysis, the research team estimated the available side (lateral) friction. While the available side friction is inconsistent from before to after at some comparison locations, the change in available side friction at 40 mph is quite substantial at the treatment sites.
    
    
  • In addition to the friction curves, speed data collected were also used to compute the difference in friction supply and the friction demanded by vehicles traversing horizontal curves at the treatment and control locations. The mean friction demand at all sites, for all time periods, is lower than the mean side friction supply. The 85th percentile friction demand at all treatment sites, for all time periods, except the before time period of the U.S. 33 treatment site, is lower than the mean side friction supply. The 85th percentile friction demand for all time periods of the U.S. 33 comparison site is greater than the side friction supply. The 95th percentile friction demand at all treatment sites, for all time periods, except the before time period of the U.S. 33 treatment site, is lower than the mean side friction supply. The 95th percentile friction demand for all time periods of the U.S. 33 comparison site is greater than the side friction supply. The only additional site where the friction demand at the 95th percentile exceeded the friction supply was at the U.S. Route 219 treatment MM 5.81 site in the before period.
    
    

HFST Conclusions

In summary, the operational and driver behavior analyses generally found no consistent differences at the treatment sites between the before and after time periods based on the data collected in the current study. The friction analysis, however, clearly demonstrated that the friction supply increased considerably at the four horizontal curve treatment locations in West Virginia. The friction levels generally remained high for a period of 1 year after the treatment was applied. A safety analysis, currently being completed under a separate FHWA contract, will reveal further information about the safety effects of the HFST.

HFST Considerations for Future Research

The current study collected speed, driver behavior, and friction data at several HFST countermeasure and control sites in West Virginia. An observational before–after safety evaluation is being completed under a separate FHWA project, “Evaluations of Low-Cost Safety Improvements Pooled Fund Study (ELCSI–PFS).” Phase VI of this effort, Safety Performance Evaluation, includes the HFST safety evaluation. Future research is recommended to determine the long-term durability of the HFST by continually assessing the friction supply at the pavement–tire interface. The friction supply should be considered in future safety performance evaluations of this treatment, by considering the time-varying nature of the HFST countermeasure.

Optical Speed Bar Evaluation Findings

OSBs were implemented and evaluated at four sites in Arizona, eight sites in Alabama, and seven sites in Massachusetts. Two different designs were tested as part of this research, and a robust data collection and analysis was undertaken to individually track each vehicle thus enabling the determination of speed changes for individual vehicles. 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. There was no second after period in Massachusetts. In Alabama, the research team collected the first after period data in October 2013 and the second after period data in December 2013. The findings of the evaluation study are as follows:

• Mean Operating Speeds—The mean operating speeds at each sensor location for all sites and all data-collection periods was calculated. An initial comparison was made between corresponding speed parameters at each site for each data-collection period by calculating the numerical differences in these speed parameters. The OSBs did not have an effect on the mean operating speeds at any of sites. There were no changes in speeds that were consistent across all of the sites.
  
  
• Speed Difference Analysis—This measure computed the change in speed (delta speed) from the approach to PC and the change in speed from the control point to the midpoint of the horizontal curve.
  
  
  • Arizona Sites—Overall, delta speed did not change in the first after period. 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. The change in delta speed varied both in direction and magnitude across the different Arizona sites, with no strong trend present.
    
    
  • Massachusetts Sites—Unlike the Arizona sites, there was a weak trend in the change of delta speed at the Massachusetts sites. The OSBs may have contributed to a small decrease on both the delta speeds between the approach and PC and between the control point and curve midpoint.
    
    
  • Alabama Sites—There were many changes in delta speed, but there was no clear trend in the change. Delta speed increased at several sites but 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.
    
    
• Analysis of Variance for Speed Differences—The research team used an ANOVA to compare speed differences in the before and after periods. 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 other sites.
  
  
• Vehicles Exceeding the PSL—The research team calculated the percentage of vehicles exceeding the PSL and the advisory speed at the treatment and control sites and compared between data-collection periods. There was no consistent trend in the change of the proportion of vehicles exceeding the speed limit, meaning the OSBs had no effect.
  
  
• Speed Variance Analysis—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. There was no consistent trend in the change of the SD of speeds at most sites, meaning the OSBs had no effect.
  
  
• Speed Threshold Analysis—The research team calculated the change in percentage of vehicles traveling a certain threshold over the PSL and compared between data-collection periods. The number of drivers traveling 10 and 15 mph or more over the PSL were compared before and after installation of the treatment. The data did not indicate a consistent reduction in high-end speeding for vehicles traveling over the PSL.

OSB Conclusions

The results of this evaluation yielded inconsistent speed reductions at all the test sites. Based on the results, the following conclusions can be drawn:

• The experimental MCPW design used at seven sites in Massachusetts and four sites in Arizona did not show a consistent trend in speed reductions. Previously, MCPW used a similar design at one site and reported mean speed reductions of 2 mph during daytime and almost 4 mph during nighttime. However, the MCPW study did not track individual vehicles, and only one data collection point was used.
  
  
• The design used at eight sites in Alabama did not show reductions in speed. Previous studies that used this design reported minor reductions (1 to 2 mph). However, it is not known whether previous studies employed the same rigorous data collection and analysis as was undertaken by this study (i.e., tracking free-flow vehicles).

Based on the results, it can be concluded that the OSB designs used in this research were unsuccessful in reducing vehicle speeds. The effectiveness of OSBs depends on the bar spacing and marking type, and for this reason, the results of this study cannot be generalized to confirm the ineffectiveness of OSBs.

OSB Considerations for Future Research

From the results of the field trials and previous OSB evaluations, it can be concluded that OSB treatments have little or no influence driver speeds. However, if further studies are undertaken, the following modifications should be considered:

• Marking Type
  
  
  • Consider evaluating non-MUTCD (2009 MUTCD, Part 3B.22, recommended markings to be 12 inches in width by 18 inches in length) experimental marking types. For instance, MCPW bar design contained two transverse markings spaced 8 inches apart, and each marking had a 24-inch length by 8-inch width to establish an overall speed bar dimension of 2 ft by 2 ft. Similarly, Iowa State University modified the MUTCD design to include a third bar (in the middle) for more visual effect. The middle bar provides additional visual contrast for the driver and also encourages drivers to place their vehicle between the bars, which is expected to cause drivers to slow as they concentrate on the driving task.
    
    
• Evaluate the influence of the OSBs on drivers of different vehicle classes (e.g., trucks).
  
  
• Because the beginning of the curve is the most crucial location for influencing driver speeds, consider extending the treatment length such that the OSBs end at the midpoint of the curve.
  
  
• Conduct a review of the crash history every 2 to 5 years after installing the markings. This would provide sufficient time to assess the true effects of the OSB on crash performance. In such cases the OSBs must be refreshed to maintain visibility (thermoplastics is preferred over paint).

Lane-Width–Shoulder-Width Combination Evaluation Findings

The study also estimated the safety effects of lane-width–shoulder-width combinations on rural two-lane, two-way road segments. Information was collected on 886 segments (minimum length of 0.5 mi) in Illinois (541) and Minnesota (345) with varying lane-width–shoulder-width combinations. Parameters for lane-width indicators showed that, with shoulder width ignored, the expected number of total (i.e., all types and severities) crashes increases as lane width decreases, but it is difficult to distinguish the performance of an 11-ft-lane width from a 12-ft lane width. The main effect of shoulder width was a decrease in the expected number of crashes as shoulder width increased. In addition, the interaction of the lane width indicator and shoulder width showed that shoulder width has the greatest effect on safety when the lane width equals 10ft. Shoulder width also has a greater effect on safety when the lane width is 11 ft than when the lane width is 12 ft.

Lane-Width–Shoulder-Width Combination Conclusions

The results of the lane-width–shoulder-width safety evaluations show more complex (but intuitive) interactions between expected crash frequency, lane width, and shoulder width than what is currently reflected in the Highway Safety Manual CMFs for rural, two-lane roads. For any given pavement width, there are combinations of lane width and shoulder width that result in the lowest expected crash frequency (for all crash types and severities as well as for fatal-plus-injury crashes). For narrower total paved widths, the optimal lane width appears to be 12 ft. This general conclusion is consistent with conclusions from Gross et al.⁽⁴⁸⁾ As total paved widths become larger, there is not necessarily a safety benefit from using a wider lane, and in some cases, using a narrower lane appears to result in lower expected crash frequencies.

Lane-Width–Shoulder-Width Combination Considerations for Future Research

FHWA’s A Guide to Developing Quality Crash Modification Factors indicates that similar conclusions from different cross-sectional studies may be one way to increase confidence in CMFs derived from regression models.⁽⁸¹⁾ Two separate studies, this effort together with Gross et al., have come to similar conclusions regarding lane-width–shoulder-width interactions on rural, two-lane roads using data from different States and different statistical analysis methods.⁽⁴⁸⁾ The findings for a safety optimal lane width for a given total paved width imply an underlying interaction between roadway cross-section dimensions, speed, and safety. No study has been able to quantify this interaction to date, although the need to consider users’ speed adaptation in safety evaluations is recognized as a high-priority research topic.⁽⁹⁸⁾ A before–after safety evaluation, with participating agencies willing to reallocate pavement width to different lane-width–shoulder-width combinations, may offer additional insight. Ideally, this study would also track driver adaptation over time to the lane width and shoulder width changes, specifically with respect to driver speed. This additional piece of information would enhance the interpretation of the safety findings. Increasingly robust cross-sectional studies (e.g., larger samples, additional States) are recommended in the absence of such a before–after study. Supplementing the cross-sectional safety data with a speed data-collection effort across a variety of lane-width–shoulder-width combinations would again strengthen the interpretation of the safety findings.