Safety Performance of Part-Time Shoulder Use on Freeways, Volume 2: Conduct of Research Report (2021)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

218 C H A P T E R 9 - C O N C L U S I O N S Conclusions NCHRP Project 17-89 developed a safety prediction method for freeways that enables the evaluation of part-time shoulder use (PTSU). The method describes a process for evaluating the safety of a freeway facility for a specified time period. The method includes crash prediction models that are used to evaluate each of the segments and speed-change lanes that comprise the freeway facility. The database assembled for the project includes data for 728 study sites, collectively representing 14 freeway facilities in five states, with a total length of 164.8 miles. About 25 percent of the total mileage consists of highway facilities with PTSU operation during 1 or more hours of the day. The assembled data were used to develop six crash prediction models that collectively address three freeway site types and two severity categories. The three site types addressed include: freeway segments, ramp entrance speed-change lanes, and ramp exit speed-change lanes. The two severity categories include: fatal-and-injury (FI) crashes and property-damage-only (PDO) crashes. Severity distribution functions were also developed. They can be used to predict the distribution of FI crashes by the K, A, B, and C severity levels. The crash prediction models include adjustment factors (AFs) for PTSU, so sites can be analyzed with and without PTSU and the results can be compared. The models also include AFs for a number of other freeway geometric design elements and traffic control features. The variables in these AFs that are related to PTSU design and operation are identified in the following list:  Proportion of time that PTSU operates.  PTSU lane width.  Proportion of segment length with PTSU transition zone present between, upstream of, or downstream of a PTSU lane.  Number of through lanes on segment, including managed lanes but not including auxiliary lanes or PTSU.  Proportion of segment length with turnout present.  Turnout spacing. Transition zones are locations upstream, downstream, or between portions of freeway with a PTSU typical section. Turnouts are paved areas adjacent to a shoulder used for PTSU that function as refuge areas for disabled vehicles. An examination of the model predictions indicates that sites with PTSU are often associated with larger FI and PDO crash frequencies than sites without PTSU, all other factors (e.g., traffic volume, lane width, etc.) being the same. However, the larger crash frequency associated with PTSU operation is partially offset by a shift in the severity distribution away from the most severe crashes. In fact, when a turnout is provided every 0.5 miles and the PTSU operates 2 hours per day or less during weekdays, there is a lower average cost per crash such that there is a safety benefit associated with PTSU operation. The Federal Highway Administration recommends turnout spacing of 0.5 miles (Jenior et al. 2016). Among PTSU sites, the crash frequency was found to be larger at sites where the proportion of time that PTSU operated was larger. Also, among PTSU sites, crash frequency was found to be lower at sites where turnouts were present.

219 In addition to this final report, NCHRP Project 17-89 produced the following two documents  PTSU Informational Guide  PTSU Safety Evaluation Guidelines The PTSU Informational Guide presents an overview of PTSU in the United States, the results of PTSU operational studies, and the results of PTSU safety research. The PTSU Safety Evaluation Guidelines describe a safety prediction method for evaluating freeways with PTSU (or freeways where PTSU is being considered for implementation). This method was prepared by the researchers as a proposed chapter for the next edition of the Highway Safety Manual (AASHTO 2010). Additionally, this project developed a spreadsheet tool that implements the crash prediction models and is described in the PTSU Safety Evaluation Guidelines. Additional analysis of the assembled data was conducted to examine differences in safety performance of various PTSU features and operating strategies. The specific features and strategies examined are identified in the following list.  Shoulder Open versus Shoulder Closed. The safety performance of sites with PTSU operation was evaluated during times when the shoulder was closed to traffic and again during times when the shoulder was open to traffic. This analysis was conducted with a dataset consisting of 48 observations per site—one observation for each weekday hour and one for each weekend hour of the study period. The analysis results indicate that sites with the shoulder open during a given hour are associated with 137 percent higher total crash frequencies (all types and severities combined) than sites at which the shoulder is closed, all other factors being the same (e.g., volume, lane width, etc.).  Shoulder Closed versus no PTSU. The safety performance of sites with PTSU (during hours for which the shoulder was closed to traffic) was compared to “comparison” sites without PTSU operation. This analysis was conducted with a dataset consisting of 48 observations per site—one observation for each weekday hour and one for each weekend hour of the study period. The analysis results indicate that there is no practical difference between the crash frequency of sites with PTSU (during hours the shoulder was closed to traffic) and the crash frequency of sites without PTSU. Thus, the increase in predicted total crash frequency on PTSU facilities discussed earlier in this chapter is largely associated with PTSU being operational rather than simply present but not in operation.  Left-side versus Right-side PTSU. The ability to assess the safety performance of left-side versus right-side PTSU was limited by the relatively small number of left-side PTSU sites in the dataset. When this project began, there were only a few miles of left-side PTSU in existence in states with crash data of sufficient quality for modeling purposes. Within this limited dataset, the sites with left- side PTSU tended to have fewer crashes than sites with right-side PTSU. However, the difference between the crash frequency of sites with right-side PTSU and the crash frequency of sites with left- side PTSU is not statistically significant. This finding is qualified as being applicable to PTSU facilities with ramps providing right-side ramp access to the freeway because this project’s dataset only included right-side ramp access. In recent years, several states have opened left-side PTSU facilities and will soon have several years of crash data available from these facilities.  Dynamic Signs versus Static Signs. Dynamic signs electronically change their display when the shoulder is open or closed. Typically, a green arrow is used to indicate the shoulder is open and a red X is used to indicate the shoulder is closed. Static signs use reflective sheeting to list the days and hours in which the shoulder is open. The analysis results indicate that there is no practical difference between the crash frequency of sites with dynamic signs and the crash frequency of sites with static signs.  Dynamic Operation versus Static Operation. Dynamic PTSU (D-PTSU) is characterized by variable hours of operation. By definition, D-PTSU facilities must have dynamic signs located along the freeway and displaying the status of the shoulder lane (open or closed). Static PTSU (S-PTSU) has

220 fixed hours of operation. Some S-PTSU facilities have dynamic signs, and others have static signs. Two PTSU facilities were converted from S-PTSU to D-PTSU during the study period. An empirical Bayes before-after study found a 7.3 percent decrease in total crashes (all types and severities combined) following conversion from S-PTSU to D-PTSU. NCHRP Project 17-89 also assessed the safety performance of bus-on-shoulder (BOS) operation. BOS facilities allow buses driven by trained drivers—but no other vehicles—to travel on the shoulder when the travel speed in the general-purpose lanes drops below a certain threshold. The assessment was based on the analysis of road inventory and crash data for several hundred miles of freeways with BOS operation. The analysis indicated (1) the presence of BOS operation is associated with a small difference in predicted crash frequency when compared to sites without BOS operation, and (2) this difference is not statistically significant. Additionally, the difference is considered too small to be of practical significance. Finally, the NCHRP Project 17-89 researchers recommend that future research investigate the following topics:  Differences in safety performance between right-side and left-side PTSU. This project included left-side and right-side PTSU sites, but the number of left-side sites was small. Several left-side PTSU facilities have opened in recent years, and additional data could be collected from these facilities. These data should provide sufficient sample size to more definitively address the question of whether left-side PTSU operation is associated with a larger or smaller crash frequency than right-side PTSU operation.  Observational before-after study of freeways where PTSU was implemented. Several PTSU facilities have opened in recent years, and sufficient data from the “after” period should be available to conduct a before-after study. This type of study was not available to Project 17-89 due to a lack of historical information about the “before” period at the sites studied. If a before-after study is conducted, it should disaggregate the freeway facility into sites and compute the change in crash frequency at each site. This approach is intended to isolate the effect of individual PTSU design and operational features on crash frequency and, thereby, produce crash modification factors that are transferrable to other states and regions. Such a study could also include areas beyond the limits of PTSU to understand the safety effects of shifting and/or removing a bottleneck through the implementation of PTSU.  Safety effects of freeway active traffic management (ATM) treatments. The safety effects of ATM treatments, especially if used in combination with PTSU, should be quantified in the form of crash modification factors. Many newer PTSU facilities have ATM treatments such as lane control signals over all lanes, queue warning systems, and variable speed limits. Many facilities have more than one of the treatments in place.  Calibration of all HSM freeway models to a common state or region. It is likely that the next edition of the HSM will include three freeway safety prediction methods: (1) the method in Chapter 18 of the Highway Safety Manual Supplement (AASHTO 2014), (2) the method developed in this project, and (3) the method developed by NCHRP Project 17-89A for freeways with HOV/HOT lanes. The models produced by these projects have been developed with data from different states and for different time periods. As a consequence, the results from the uncalibrated versions of these models will not be comparable to one another. In preparation for the second edition of the HSM, NCHRP Project 17-72 is calibrating to a common state some of the crash prediction models developed after the first edition of the HSM was published. However, NCHRP Project 17-72 will not be calibrating the models for the three aforementioned freeway methods.  Extend models to rural areas. The method developed for this project is applicable to urban freeways. The applicability of this method to rural areas is unknown. Additional data could be collected in rural

221 areas and used to estimate models for PTSU operation in rural areas, such as areas with seasonal or recreational-related congestion.  Replace bi-directional freeway model in HSM with single-direction method. Using data from this project, NCHRP 17-89A, possibly NCHRP 17-45, and possibly new sites, develop a comprehensive single-direction model to replace the current bi-directional model in Chapter 18 of the HSM. This model would be useful for projects in which only one direction of a freeway is being studied.  Use of random parameters and latent class models for HSM applications. Random parameters models and latent class modeling techniques are receiving increased interest by safety analysts. These techniques appear to be well suited to some safety-management applications (e.g., network screening) where the sites of interest are in the data used for model estimation. However, the models in Part C of the HSM are generally used by practitioners for design applications (e.g., to analyze existing or planned facilities) where the sites of interest are not included in the data used for model estimation. Unfortunately, for design applications, there is no established procedure for using the model parameters (whether it be a random parameter or a parameter from one of two or more classes) to obtain reliable estimates of predicted average crash frequency. Future research should identify the means by which practitioners could most effectively operationalize the random parameters and latent class regression modeling techniques for safety-management applications and for design applications, recognizing that such use would likely require software implementation.