Evaluation of traffic signal timing optimization methods using a stochastic and microscopic simulation program.

This study evaluated existing traffic signal optimization programs including Synchro,TRANSYT-7F, and genetic algorithm optimization using real-world data collected in Virginia. As a first step, a microscopic simulation model, VISSIM, was extensively calibrated and validated using field data. Multiple simulation runs were then made for signal timing plans such that drivers' behavior, day-to-day traffic variation, etc. were considered in the evaluation. Finally, long-term demand growth or changes were statistically modeled and evaluated, again using multiple simulation runs. Five timing plans were evaluated using the simulation test bed. The timing plans under evaluation included (1) the former timing plan of the Virginia Department of Transportation(VDOT), (2) VDOT's current timing plan, (3) the genetic algorithm optimized timing plan, (4)the Synchro optimized timing plan, and (5) the TRANSYT-7F optimized timing plan. The simulation study results indicated that the current practice of VDOT's current signal optimization procedure significantly improved upon its former one by reducing travel times by 17% and total system delay by 36%. The three "optimized" timing plans did not provide significant improvements. Evaluation of the Lee-Jackson Memorial Highway network showed that the current VDOT signal optimization procedure significantly improved the performance of network operations. Thus, the study recommended that VDOT continue using its procedure for developing new timing plans but that it evaluate its signal timing plan regularly so that it does not become outdated.