Citation

Park BB, Kamarajugadda A. Int. J. Sustain. Transp. 2007; 1(3): 193-207.

Copyright

(Copyright © 2007, Informa - Taylor and Francis Group)

DOI

10.1080/15568310600737568

PMID

unavailable

Abstract

Signalized intersections are one of the most common ways of sharing right of ways among vehicular and pedestrian traffic. The design of isolated signalized intersections involves determination of signal timing plan (i.e., the cycle length, phase sequence, and green time allocation). These selections are, in general, conducted through computerized software such as Synchro and TRANSYT-7F. These programs generate optimal signal timing plans by minimizing delay obtained from either analytical models such as the Highway Capacity Manual (HCM) delay equation or traffic simulation models. One of the shortcomings in the state-of-the-practice in signal optimization is that signal timing plans do not necessarily account for day-to-day variability in traffic demand as they are developed on the basis of the average demand volumes from a day or two worth of traffic counts. This paper presents a genetic algorithm–based stochastic signal optimization method that explicitly considers variability in the HCM delay equation due to day-to-day traffic demand fluctuations. The variability in the delay arising from varying demand conditions is evaluated using an integration technique. The proposed approach was compared with Synchro, a traffic signal timing optimization software, through a stochastic microscopic simulation program, CORSIM. The simulation results indicate that signal timing plans generated from genetic algorithm (GA)-based stochastic optimization generally outperformed those from Synchro.