Some New Approaches to the Microscopic Modelling of Traffic Flow and the Dynamic Route Assignment Problem

Abstract

In order to simulate the transportation system of a large region dynamically, three things have to be known and modelled: who wants to go where at what departure time (destination choice), which route to the destination is selected (route choice), and finally how the locomotion along this route is performed in time (travelling). This article deals with the second and third question. Firstly, it is shown how the dynamic route choice problem could be solved by a simulation—based approach. Surprisingly, when comparing this approach to the classic static one, it turns out to be simpler and applicable to larger networks than its classical counterpart. Secondly, by using a simplified model of car—following, most of the phenomena observed in real traffic can be reproduced by the simulation. The realism achieved with this model allows a reliable estimation of the emissions caused by traffic (car traffic amounts to roughly 80% of the total travel volume). The model is numerically very efficient, enabling the simulation of large road networks with more than 1 million individual cars in real time Finally, preliminary results are mentioned concerning the possibility of expanding this simplified microscopic approach to the destination choice modelling. Following this route of thinking further finally leads to an integrated, microscopic, and dynamic approach to transportation planning and control. The first steps toward this aim has been done within the FVU.