Disruption in Congested Networks

Paper in proceedings (conference paper)

en

The purpose of the paper is to introduce and discuss a special problem related to congested traffic networks. In 1968, Braess presented an example of careless network changes, which led in congested networks to higher density of traffic. Since then, various versions of network design problems have been studied and clarified. We focus on phenomena accompanying many network design applications, specifically, the reconstruction of a network and its partial closure. A bilevel structure of the studied problem is tackled by the use of Karush-Kuhn-Tucker conditions. The obtained nonlinear integer program can be solved by the use of modelling optimization software as, e.g., GAMS. However, only small instances can be successfully solved within the acceptable time limits. For the large model instances, as the considered modification of the known Sioux Falls test case, the hybrid algorithmic framework combining the use of deterministic solvers encapsulated in GAMS with the genetic algorithm has been designed.

multicommodity network flow, traffic assignment problem, network design, congested network, nonlinnear integer programming, noncooperative equilibrium, Karush-Kuhn-Tucker condditions, genetic algorithms.

2013-06-26

VUT v Brně

Brno

978-80-214-4755-4

1803-3814

Proceedings of 19th International Conference on Soft Computing MENDEL 2013

191–196

6