The most critical aspect of signal design and timing at an intersection is the development of an appropriate phase plan, which is mainly driven by left-turn treatments.
Accommodating and addressing left-turning vehicles is challenging for traffic engineers as they seek balance between intersection capacity and safety; these are two conflicting goals in the operation of a signalized intersection that are mitigated through signal-phasing techniques. Exclusive left-turn lanes and protected left-turn phases are commonly used to minimize the impact of left-turning traffic. Cycle lengths typically have to increase, and the addition of extra time from through phases must be sacrificed. This may contribute to an increase in delay or decrease in operational performance at these intersections. Hence, to increase the left-turn capacity and reduce the delay at the intersections, researchers and traffic engineers found protected/permitted left-turn (PPLT) control to be the most effective thus far.
The traditional PPLT signal head has been a five-section configuration with a circular green (CG) indication for both permissive left turns and through traffic. While PPLT configurations allow most left-turning traffic to turn during the protected phase, the permitted phase exploits the efficiency of using available gaps in oncoming traffic to facilitate additional left turns. However, in many cases, concerns arise with left-turning traffic patterns during certain times of day, such as heavy pedestrian activity during school arrival and dismissal. When such concerns arise, engineers need to modify the signal from protected/permitted (PP) phasing to protected-only (PO) by changing the signal displays. Historically, this change required physical replacement of overhead signal displays and was essentially a permanent change, even though it may only be needed a few hours per day.
The use of a four-section head for the left-turn-only lane with a flashing yellow arrow (FYA) indication for permissive left turns overcomes this limitation. The FYA has been deemed to be the new standard for signalization as recommended in the 2009 Manual on Uniform Traffic Control Devices (MUTCD). FYA treatments at intersections are considered new and are evolving fast, especially in the central Florida area. With the advent of this new signal configuration, there was the opportunity to take the PPLT mode to a new level of operation. The new all-arrow configuration provides the opportunity to change the operation mode at any time during the day from fully protected to completely permissive or combinations of the PP signal phasing.
Though numerous studies have developed guidelines for selecting left-turn control types, to date, there are no clear or uniform standards for the selection of left-turn phasing mode changing by time of day, and there is no systematic approach that allows for scanning intersections and flagging ones that require attention to left-turn phasing mode.
Dr. Essam Radwan’s UCF research team, led by Dr. Hatem Abou-Senna, developed an interactive and efficient framework to serve as a decision support system (DSS) for the evaluation of left-turn phasing alternatives based on intersection conditions. This framework will allow (1) an interactive evaluation of left-turn phasing and ultimately recommend phasing mode by time of day, and (2) traffic-management center (TMC) data to be fed into the DSS so that intersections requiring attention/modification of left-turn mode can be flagged.
From the literature review, it was found that common guidelines for left-turn phasing did not apply to all intersections. A comprehensive approach was needed to cover all cases and to develop a deeper understanding of the range of parameters that affect left-turn phasing for efficient operation while maintaining safety.
Combining the two aspects is rarely achieved.
A list of candidate parameters related to traffic, safety, signal, geometry and land use was developed to determine the operational and safety impact measures of effectiveness (MOEs) for left turns. These parameters represented the basis of the interactive framework to evaluate the suitable left-turn mode under different time-of-day volume levels.
Five or flashing
The process for obtaining the left-turn parameters required several different methods of collection. There were several factors that required no field work and others that were obtained through databases or live data capture in the field. The research requirements demanded intersections having either a five-section signal head where the PP phase was used or a FYA signal already installed and operational.
Because of the wide spectrum of intersection types available in central Florida, the goal was for these intersections to be scattered around the area to obtain a fair sampling of sites. More than a dozen intersections were selected for data collection, ranging from small minor roads and ramp terminals to major arterials. Parameters that constitute the geometrics, safety and operational aspects of the intersection are important to classify the intersection. Additionally, specific categorical data parameters also were chosen that are considered significant enough to affect the characteristics of the traffic flow and behavior of the driver. This is a dramatic departure from the volume-based approach that has dominated in the past when determining the warrant for a protected left turn.
Specific categorical data that did not require field work included:
Time of day: to characterize the intersection throughout the day;
Peak hour: whether the analyzed hour is within the intersection’s peak time frame;
Geometry: special characteristics such as wide, skewed or dangerous by design;
Land use: to highlight traffic characterization and the environment surrounding drivers;
Criteria: to take the design environment into account (special features or situations);
Crossing lanes: to have a perspective on total lanes the driver is crossing;
Speed limit: to determine the need for larger gaps for the driver to accept a left turn; and
Left-turn crashes by time of day: protection in high-risk areas (special crash database).
Several other field parameters are included to snapshot the typical operation of the intersection especially during the permitted phase. These parameters included:
Permitted green times: total amount of permitted green time throughout the hour;
Permitted left-turn volume: number of left turns during permitted phase;
Total left-turn volume: total left-turn traffic for PP phases;
Permitted opposing volume: opposing volume during permitted phase;
Total opposing volume: total opposing traffic during PP phases; and
Left-turn truck percent (%): truck percentage in the subject left-turn lane.
The data extraction process began with identifying the left-turn approach that would be analyzed. The left-turn parameters related to the volume during the permitted green time and the extents of these periods were extracted in the laboratory by watching the videos second by second. Subject left turns also were timed from start to finish on the selected approaches to calculate the critical gap. Conversely, total turning movement counts were processed at the intersections using automated video detection.
The main objective was to study what happened during the permitted phases in a microscopic manner. Across all of the intersections, 23 left-turn approaches were analyzed totaling 229 hours of video data processed including both off-peak and peak conditions. Video data extraction was an essential process in constructing and analyzing the design of the experiment and eventually developing the new thresholds for the determination of left-turn modes by time of day.
Is it permitted?
Standard experimental designs either using full factorial or fractional factorial did not fit the research requirements, and therefore a novel optimal custom-design approach was selected for this research. Also, choosing an optimality criterion to select the design points was another requirement. JMP statistical software (created by SAS) was used to generate the custom design for this experiment. The custom-design approach in JMP generates designs using a mathematical optimality criterion. Optimal designs are computer-generated designs that aim at solving a specific research problem to optimize the respective criterion. It was imperative to investigate several model types, specifically generalized linear models.
The developed Poisson regression model provided better prediction profiles and showed the relationship between the significant parameters to a third-degree polynomial equation with an interactive capability of fitting a separate prediction equation for each dependent variable, such as volume or speed, to the observed response: number of predicted permitted left turns per cycle (PT LT Volume). This enables prediction of all combinations of parameters on the dependent variable at the same time.
The analysis of the experiment produced an interactive DSS for left-turn mode. Based on the predicted number of left turns during the permitted phase, the analyst can decide whether the permitted left-turn phase is warranted or not. When the characteristics of the intersection in a particular situation warrant that a left turn be permitted, the signal would be able to adapt and relay the results of the analysis via the controller through the DSS. The ultimate goal of the study is to eventually automate the process and have the controller make the determination.
After the model predicts the total number of left turns during the permitted phase, the analyst has to decide whether this left turn should operate as PO or PP. Three criteria were developed for this particular decision, two of which are related to operational aspects while the third one relates to safety. Specific thresholds also were determined for these criteria.
The first criterion included an indicator that takes into account three main factors:
1. The predicted number of permitted left turns during the hour (using the developed model) or if this value can be collected from the field (PT LT Vol);
2. The permitted opposing volume (PT Opp Vol) during the hour; and
3. The permitted green time (PT Grn Time) during the hour.
Using the three criteria along with their thresholds, the analyst can decide on accepting a permitted phase during a specified hour of the day or rejecting it. Rejecting a permitted phase means that it is not efficient.
A low number of permitted turns made during a specific time of the day represents the operational efficiency of the permitted phase along with other safety implications for drivers taking more risk and accepting smaller gaps. For example, comparing the amount of permitted green time given throughout the hour and the number of permitted left-turning traffic determines whether the opposing traffic flow was operating near or at saturation or whether other factors such as pedestrian activity are involved. This can be seen in the field when only one or two vehicles at the most can make the turn during the permitted phase every two to three cycles. Also, having an inefficient permitted phase along with some aggressive drivers can result in a crash. Therefore, eliminating the permitted phase during that time of the day improves the safety as well as the operation.
The critical gap is a judgment threshold—that is, different drivers or even the same driver at different times have different critical gaps. Acceptable gaps for drivers vary but usually have a trend of decreasing with the increase of flow rate.
When the opposing flow rate is low, left-turn drivers are inclined to reject some larger gaps since there are many available large gaps, so the critical gap increases. When the opposing traffic has a high flow rate, drivers are inclined to accept some smaller gaps due to the lack of available large gaps in the major stream, which is an important reason why drivers will take the risk to enter the intersection with the increase of waiting time. So the critical gap decreases and that’s why drivers tend to make more poor decisions as delay increases.
The developed guidelines would provide traffic engineers with the tools to utilize the efficiency of the permitted left-turn phase at both peak and off-peak times and reduce the delay at approaches with low volumes.
“Through this project, the UCF research team expanded the understanding of traffic flows involving left turns and put this new knowledge to practical use in the development of a software tool that traffic engineers can use to fine-tune time-of-day phasing at heavily traveled intersections. The result will be greater safety, higher throughput and fewer delays at these intersections, producing greater convenience and efficiency for Florida drivers,” said Rick Morrow, P.E., traffic operations engineer for District 5 at the Florida Department of Transportation, who sponsored this research and served as the project manager. It should be noted that the UCF research team is in the process of starting Phase II of the project for field testing and implementation of the DSS.