OPTIMAL ALLOCATION OF SERVICE TIMES FOR VEHICULAR QUEUES AT SIGNALIZED TRAFFIC INTERSECTIONS

The use of traffic signals has been widely accepted as an effective means of controlling traffic at intersections. However, in some situations, traffic signals actually increase the magnitude of the problems they are intended to solve. This is because most widely used traffic signal optimization models are based on simplified assumptions such as constant vehicular arrivals and service, and therefore, give inaccurate results in some traffic conditions.;The suitability of the above optimization methods for oversaturated flow conditions, and steady state and transient demand conditions are discussed. The advantages and disadvantages of each of the two optimization methods, and the application of the simulation optimization method in state dependent server-vacation signal timing are also addressed.;A computer program which uses the Akaike's Information Criteria (AIC) in selecting the best fitting model among several competing probability distribution models is also developed and discussed. The reliability of the AIC model-fitting computer program is also verified by performing Monte Carlo experiments.;The accuracy of the traffic signal optimization models is validated by comparing the results obtained to observed data. The estimated results are not significantly different from those observed in the field at 5% significant level. Also, when applied to two signalized intersections we obtained more efficient results.;Two innovative stochastic traffic signal optimization methods for signalized traffic intersections are developed and discussed. Determination of optimal cycle and service times is based on minimization of the expected delay for a given period of observation. Traffic signal timing is formulated as a stochastic inventory problem, which is then solved using simulation and dynamic programming.