The conventional traffic assignment is generally based on static assumptions of travel demand and route choice behaviour. Thus the prediction of the rush-hour traffic conditions using the conventional methods probably provides unrealistic results. In order to improve the prediction accuracy or to apply the traffic assignment methods to the appraisal of various traffic control policies, recent research in this field tends to focus on the development of the dynamic traffic assignment model. This paper presents a semi-dynamic traffic assignment model, where a day is divided into 24 time-segments and for each time-segment the user equilibrium assignment model with elastic demand is applied; a static assignment model is applied to each time-segment. Travel demand functions are not used in the usual sense, instead, these are derived from the functions which describe the inter-relationships of flows between successive time-segments. The model presented in this paper enables us to predict flow patterns of each time- segment without the usage of a special type of traffic assignment method and may contribute to resolve the inherent computational difficulty involved in the other types of dynamic traffic assignment method proposed so far. Model formulation by the non-linear mathematical optimisation programme and experimental results applied the real transportation networks are reported.
Samenvatting