This study gives both a theoretical and an empirical evaluation of the QNet traffic assignment model. This model is available as a part of the Dutch National Model System. QNet has been developed in response to the major problem that occurs in traditional assignment models, when applying them in heavily congested networks. In traditional models, the trips are assigned simultaneously to all links along a given route. This implies that the trips might cause congestion on every link resulting in a prolongation in travel time on every link. However, this does not occur in reality. Congested links act as "faucets". They hold up travel. A car in a congested link cannot contribute to congestion on other links at the same time. QNet corrects this problem by upholding traffic on congested links. In this way, not all trips in a given period are assigned to every link along their routes. It is attempted to assess to what extent QNet is able to correctly predict the actual loads on links. The results are compared with both traditional 'all-or-nothing' assignments and with equilibrium assignment models.
Samenvatting