One possible approach to dynamic assignment is the extension of microscopic traffic flow models by route choice models. Some of these models are already available for practical applications, but inherently require long computing times. Approaching the problem from the other side, the paper presents a dynamic traffic assignment method based on the extension of an existing static assignment method developed by Lohse. The static version attempts to model the drivers' increasing knowledge of a road network within the course of time using an iterative approach. Starting with a shortest-path search in the unloaded network the trips of one o-d pair are equally distributed onto the known routes of this o-d pair after each iteration step. Each iteration step adds one route per o-d pair which can be either a new route or a route which has already been identified during an earlier step. The link impedance for the next route search is deduced from the impedance of the current traffic volume and the previously estimated impedance. The iteration process terminates when the link traffic volumes of two iteration steps differ only within a user-defined limit.
Abstract