This paper examines results from computational comparisons between different methods for choosing and scheduling transport investments. Using a road system which is widely used as a test network in the literature, two genetic algorithm specifications are compared and contrasted with existing procedures. The genetic algorithm approach presented in this paper differs from other methods, such as simulated annealing, in two aspects. First, the process of selecting and scheduling projects is done simultaneously, rather than in two stages, and second, some assumptions in modelling may be relaxed. The former is important as network design should consider at the outset when selected projects are to be implemented. In fact, it has been stated that it may be more important to find a good time schedule of investments than to determine the optimal investment program for a future time period. A dynamic planning concept is therefore required as a decision support tool that stresses the time scale of project realisation in the economic evaluation of projects. (A)
Samenvatting