Despite the significant amount of research on alternative designs ([1-18]), there are still existing gaps in the literature, for example, with regards to RCUT intersections in rural areas with challenging topography. Improved access to highways has been identified as a critical equity issue for rural areas [19]. Several rural communities rely on 2-way STOP-controlled intersections with high-speed expressways for access to markets or tourism traffic. These intersections do present operational and safety challenges [20]. If the economic cost of converting Stop-controlled approaches into a grade-separated interchange cannot be justified, often the concerned agencies shut down the intersections severely impacting businesses such as local farms and wineries [21, 22]. RCUT intersections are an innovative low-cost intersection design where all traffic on the minor road must turn right at the main intersection, thereby eliminating the conflicts leading to severe far-side angle crashes (Fig. 1). Despite the lower cost and potential to address operational and safety concerns through the elimination of severe conflicts [1-18], their application is not uniformly distributed throughout the US. For example, they are widely used in North Carolina and Michigan, but to the best of our knowledge, there is no known application of these designs on rural expressways in CA.

This research presents a framework to estimate the expected safety and operational performance of RCUT intersection designs as possible substitutes for a conventional Two-way Stopped Controlled (TWSC) intersection. The framework is demonstrated with an application case study at a rural California intersection (Figs. 2 and 3) with challenging geometric features and limited right-of-way (ROW). The analysis framework for considering and evaluating such treatments could provide new insights for transportation designers and engineers. Moreover, published research that matches their local or regional context may push practitioners in large states like California to at least evaluate such alternatives where they may be appropriate. The framework used in this study will allow agencies to:

Fig. (1). A typical RCUT design on rural intersections [14].

Fig. (2). The existing intersection on US 101 @ Tassajara Creek Road.

Fig. (3). The significant elevation difference between the two directions.

1. Define possible RCUT alternatives and identify geometric constraints that may affect the range of the alternatives.

2. Estimate the expected future operational conditions, including the vehicle travel times for the RCUT designs in comparison to existing TWSC.

3. Establish a procedure to estimate the safety performance of the RCUT designs in comparison to the existing intersection by using the following methods:

a. Empirical Bayes approach prescribed by the Highway Safety Manual [23, 24].

b. A surrogate measure-based approach;

4. Discuss the relative merits and select from the two approaches to estimating safety

The study utilized the microsimulation approach to estimate operational performance. Simulation output from PTV-VISSIM [25] is also used in conjunction with SSAM (Surrogate Safety Assessment Model) [26] for the surrogate measure-based safety estimation.

The first traces of alternative to the traditional 4-legged intersections go back about 70 years ago when jug-handle intersections were constructed in New Jersey [27]. Median U-turn (MUT) intersections were introduced in Michigan in the 1960s, and there are many of them all around the US that still perform well. Based on past studies, MUT intersections could improve traffic operation and safety in locations with lower left-turn traffic by reducing the number of signal phases and conflict points [28]. Alternative intersections gained great attention again during the 1990s with growing traffic volume and tighter budgets for funding new infrastructure in the US. In the last three decades, hundreds of alternatives to traditional 4-legged intersections designs, including roundabouts, quadrants, RCUTs, and Contraflow intersections (CFIs), have been built in the US.

RCUT intersection (it is also called a superstreet, synchronized j-turn, or reduced conflict intersection) is a variation of the MUT design. However, in contrast to the MUT, traffic coming from the minor roads is not allowed to proceed straight across the major roads. Also, left-turn traffic on the major roads would access the minor roads using exclusive left-turn lanes at the intersection (instead of using U-turns like MUTs). RCUT intersections could be stop-controlled, yield controlled or signalized. Fig. (1) shows a typical rural RCUT intersection [14].

According to past studies [4, 29, 30], RCUT intersections illustrated excellent performance in terms of traffic operation as well as safety when there is a low through and left-turn demand on the minor road. A study on a rural four-lane high-speed major road in Missouri [12] showed high potential for RCUT in terms of traffic safety performance; however, the vehicle travel times at RCUT intersections were slightly higher than the existing TWSC intersection. A before-after analysis of fatal and injury, angle, and left-turn crashes showed that these crashes reduced by more than half after upgrading a conventional stop-controlled intersection to an RCUT intersection. Sideswipe, rear-end, and other types of crashes decreased by a lesser degree or slightly increased [14]. Another study examined the safety performance of five RCUT (with stop signs) installations on a rural, four-lane highway in comparison to a conventional stop-controlled intersection on minor roads [11]. Data was gathered three years before and after the installations. It was found that the RCUT intersections reduced the total number of reported crashes by an average of 35%. Additionally, no fatal crashes happened during the study period after installation. While the literature provides significant evidence of the efficacy of these intersections based on post hoc analysis, the literature is lacking in framework to evaluate the future expected performance of RCUT intersections for jurisdictions (e.g., Caltrans) where they have not yet been implemented. This study provides such a framework for future safety and operational performance evaluation framework using case study from a site in rural California.

The research team installed a video camera at the location to record the PM peak hour traffic. The video was reviewed to estimate average speed, headway, and allowable gap for the turning traffic. Due to the negligible turning traffic demand at the study location, another video camera was installed at a similar intersection (State Route 65 and Avenue 182) in Tulare County, California, to include a larger number of observations to estimate the distribution of the allowable gap. Among the measures from the recorded videos, vehicle speed and the acceptable gap for turning movements were used to calibrate the simulation model, while headways were used for the validation.

Table 2 shows a summary of data collected from videos. Based on the data in Table 2, speed distributions with a mean speed of 72 mph were added into VISSIM on all lanes of the major road, while vehicles had an average speed of 25 mph (same as the speed posted) on the minor road approach. Also, traffic coming from the minor road, the left-turn demand from the major road, and the U-turn vehicles (in RCUT designs) could start their movements when there was a minimum gap equal to 4.5 seconds between the conflicting vehicles. By default, VISSIM considers a 3-sec minimum gap for conflicts between through traffic and turning traffic movements. However, the 3-sec gap would not be acceptable in this study because of the high speed on the major road.

After changing the simulation model setting to desirable levels, we estimated traditional validation measures, including travel time and delays at the Stop sign. While these measures signified that the model was well-calibrated, such measures aren’t most meaningful in the context of this location. For appropriate SSAM-based safety evaluation, the distribution of simulated headways vs. field headways was the most critical validation measure. Based on simulation runs of the TWSC model, the mean headway was identified to be equal to 2.90 s and 3.01s on the left and right lanes of the major road, respectively. By conducting ANOVA, differences between real data and simulation outputs were found to be statistically insignificant at a 95% Confidence Interval.

Traffic volume was collected during AM peak (7:00 AM to 8:00 AM), mid-day (Noon to 1:00 PM), and 4:30 PM to 5:30 PM) on a weekday in June 2020. (Fig. 5) shows existing traffic volumes during PM peak hour between 4:30 PM and 5:30 PM. As shown in Fig. (5), the intersection operated almost as a T-intersection due to the very low demand from the WB direction. It should be noted that there are only a few residential houses on this side of the intersection.

For this study, in addition to exiting traffic demand conditions for three different times of day, a wide range of scenarios were defined to cover various states of current and future traffic demand, traffic distribution (on different approaches), and turning traffic ratios. Overall, 54 simulation scenarios were tested in this study. Table 3 shows the factors considered to create the 54 (3*2*3*3) different traffic condition/intersection design scenarios. These factors include:

Fig. (5). Traffic volume (veh/hr) diagram during PM peak hour at the study location.

• 3: Time of day scenarios
• 2: Existing and 2030 conditions based on a 2% annual growth rate
• 3: Variations of EB and NB-Left traffic demand (existing demand; 4 times the existing demand; and 8 times the existing demand)
• 3: Intersection designs (TWSC, RCUT#1, RCUT #2)

The variations in the only EB and NB-Left demand were considered based on discussions with stakeholders. The stakeholders don’t expect the WB demand to increase due to the topography of the area. Overall, there were 18 traffic volume scenarios ranging between 3,150 and 5,100 vehicles/hr for each of the three intersection designs. The next section elaborates on the key results of the evaluation.

Intersection treatments, even with proven effectiveness, often have a wide discrepancy in their applications from one jurisdiction to the next. Agencies applying a certain new treatment often need a way to estimate future operational and safety benefits for their sites. Such estimations are not trivial to obtain merely based on post hoc studies from other jurisdictions. A robust framework to estimate the future operational and safety benefits of intersection treatments is proposed here. The framework may be summarized as follows:

• Develop microsimulation models for existing and potential treatment designs
• Estimate operational benefits using a well-calibrated simulation model using Measures of Performance such as travel time and delays
• Review CMF clearinghouse and other literature for assessing the safety benefits of the treatment. If the intersection treatment is typical, use the HSM-prescribed Empirical Bayes method to estimate the safety performance of the proposed design. Otherwise, use a surrogate safety-measure based approach.

This paper demonstrated this framework in the context of potential RCUT designs for a rural California intersection. Overall, 54 simulation scenarios were models using VISSIM for estimating operational performance at varying levels of travel demand for the intersection. The scenario included the base condition with TWSC intersection and two RCUT designs. Note that given the design constraints of the location, the U-turn length for the RCUT alternative was restricted in one direction to only 400 ft. This atypical design meant that available CMFs in the literature were not applicable to this location. Hence a surrogate measure-based safety analysis was conducted. The following bullet points have summarized some of the important findings:

• The proposed RCUT designs reduced or eliminated the more severe crossing conflicts.
• For both RCUT designs, there was an increase in rear-end and lane-change conflicts compared to the TWSC design. It may indicate a future trade-off between (reduced) severe vs. (increased) non-severe crashes following the installation of RCUT.
• Depending on the site context, increasing the U-turn distance may eliminate some of the weaving maneuvers that lead to increased conflicts. Of course, given the geometric constraints, it was not a viable option for this work.