EVALUATION OF HEURISTIC TRANSIT NETWORK OPTIMIZATION ALGORITHMS

Changes in urban land use and travel demand have created the need to restructure many existing mass transit networks. Heuristic network optimization as one of the available methodologies to improve transit networks is described. The characteristics and results of the algorithms developed in Europe are summarized and a short description of the American algorithms is given. The potential for applying network optimization methodologies in the context of small to medium-sized American cities is evaluated. The review and evaluation of 13 heuristic methodologies revealed a wide range of approaches that are generally theoretically sound, have reasonable potential for generating improved networks, and are computationally and otherwise feasible. Application of an unproven new algorithm by Mandl to the bus network for Madison, Wisconsin, and the light rail network for Duesseldorf, West Germany, showed that a fairly complex heuristic algorithm can be implemented quickly and easily. Mandl's algorithm, however, did not generate an improved network, primarily because the initial computer-generated network does not follow demand. Better results were obtained with two other heuristic methodologies that have been applied to the Duesseldorf network. The Madison and Duesseldorf applications form the basis for recommendations for further improvement of heuristic methodologies.