A heuristic algorithm for solving the discrete network design problem

[[abstract]]Transportation investment is a difficult decision, especially in developing countries where numerous improvements are needed but only limited resources are available. Although many studies have suggested that the network design approach be used to solve the investment decision problem, yet the problem solving algorithms previously developed are still not sufficient to provide optimal solutions with reasonable computing cost, especially when working with large networks. In this paper we suggest a new heuristic algorithm which can be used by general planners to deal with medium and large network design problems. This algorithm is based on the concept of expected generalized travel cost, and is developed by incorporating the Poorzahedy's modified model (a previous study) in an iterative optimization‐assignment procedure. A small network which was used in other studies is chosen to test the efficiency and effectiveness of the algorithm. In addition, a real highway network with 305 links and 135 candidate additions/alternative improvements has also been applied, and the computational efficiency is presented. All the computing work was executed by using simple commercial packages with a PC‐386 machine. The results conclude that their heuristic algorithm is very effective and is much more efficient than those developed in previous studies. An extended applicability of this algorithm for solving the network design problem with variable travel demand is also discussed.[[incitationindex]]SCI[[incitationindex]]EI[[booktype]]紙