Abstract
Selection of the left-turn phasing mode is a crucial decision for the safe and efficient movement of left-turning traffic at signalized intersections. Because of different safety and operational effects caused by the signal mode, the two must be evaluated concurrently and be balanced based on capacity and crash potential when comparing protected-only, permissive-only, and protected-permissive (PPLT) modes. This left-turn phasing mode choice can be made on a time-of-day basis, so that changing traffic conditions are accommodated appropriately. This thesis seeks to define guidance that field traffic engineers can use to select the appropriate left-turn mode based on prevailing traffic conditions by time-of-day. In particular, guidance on the use of PPLT with flashing yellow arrows was of interest to the Virginia Department of Transportation (VDOT).
Prior to developing time-of-day guidance, the overall safety effects of converting left-turn phase modes and indications needed to be explored. This thesis examined the impact of converting from a circular green “ball” display for the permissive portion of PPLT phasing to the flashing yellow arrow (FYA) signal indication, as well as protected-only phasing to PPLT with FYA. To quantify these conversions, a before and after evaluation of signal conversions was performed using the Empirical Bayes (EB) approach to develop crash modification factors (CMFs) from data for 37 intersections in Virginia. Statistically significant CMFs were found for the PPLT with green ball to PPLT-FYA conversion for all severities and angle crash types as 0.71, 0.62, and 0.77 for all, 3-, and 4-leg intersections. For conversions from protected-only to PPLT-FYA, CMFs of 1.47 and 2.56 for all and angle crash types for all severities, as well as CMFs of 1.57 and 3.30 for all and angle types for fatal and injury severities were determined.
In evaluating different left-turn phasing modes on a time-of-day basis, crash risk, left-turn conflict, and capacity prediction models for permissive-only and PPLT modes were developed using simulation data. A total of 750 unique scenarios based on different combinations of intersection characteristics, traffic signal parameters, and traffic volumes were simulated in VISSIM and trajectory files were processed in SSAM for the number of conflicts per scenario. From the outputs from the simulation models, prediction models for determining left-turn capacities and the number of left-turn conflicts per 100 left-turning vehicles were created using multiple regression. Additionally, a crash risk model was created. A final model predicting the average crash risk per hour based on the predicted number of conflicts was developed to be used as a crash risk assessment tool. The three models created were incorporated into a new tool for the evaluation of signal phasing based on time-of-day. This spreadsheet incorporated all of the resulting models into one tool, and should be used by VDOT engineers in determining phasing mode on a time-of-day basis.
