Book Description

The efficiency of congested urban transportation networks can be improved by implementing appropriate traffic control strategies, such as signal control timing, turning movement control; implementation of one-way traffic policies, lane distribution controls etc. In this dissertation, the following strategies are addressed: 1) Intersection left turn addition/deletion, 2) Lane designation, . and 3) Signal optimization. The analogy between the network design problem (NDP) and the optimization of traffic control strategies motivated the formulation of an urban transportation network design problem (UTNDP) to optimize traffic control strategies. An UTNDP is a typical bi-level programming program, where the lower level problem is a User Equilibrium (UE) traffic assignment problem, while the upper level problem is a 0-1 integer programming problem. The upper level of an UTNDP model is used to represent the choices of the transportation authority. The lower level problem captures the travelers' behavior. The objective function of the UTNDP is to minimize the total UE travel time. In this dissertation, a realistic travel time estimation procedure based on the 1997 HCM which takes account the effects of the above factors is proposed. The UTNDP is solved through a hybrid simulated annealing-TABU heuristic search strategy that was developed specifically for this problem. TABU lists are used to avoid cycling, and the Simulated Annealing step is used to select moves such that an annealing equilibrium state is achieved so that a reasonably good solution is guaranteed. The computational experiments are conducted on four test networks to demonstrate the feasibility and effectiveness of the UTNDP search strategy. Sensitivity analyses are also conducted on TABU list length, Markov chain increasing rate and control parameter dropping rate, and the weight coefficients of the HEF, which is composed of the current link v/c ratio, the historical contribution factor, and the random factor.