Modelling of traffic flow dynamics in incident conditions using the first order macroscopic approach
MONGEOT
Type de document
ARTICLE DE PERIODIQUE
Langue
anglais
Auteur
MONGEOT
Résumé / Abstract
Traffic incidents contribute to urban network congestion by creating an impact on traffic which can be surprisingly fast and may involve rapid spilling-back congestion. In traffic science, traffic incidents are seldom considered from the point of view of the local perturbation of the traffic dynamics which they induce, but rather as a source of non-recurrent congestion disrupting the usual vehicle routes in the network. In consequence, urban network modelling tools, when used to simulate incidents, put emphasis on the impact of incidents on the traffic assignment and simplistically model their impact on the flow dynamics. However, the traffic flow and the assignment sub-models are inter-related in the sense that the link travel times output from the flow sub-model are a basic criterion for the assignment process, which in turn outputs traffic volumes to be propagated along links and may thus modify the previously-established link travel times. It therefore appears that an accurate modelling of the flow dynamics perturbation induced by an incident at the local link level is crucial for the modelling of the impact of the incident at the network level. In particular, the existing variety of incidents implies more than a representation involving a temporary reduction in capacity. Also, it should be possible to model the incident extent and position in a link, rather than assimilating the incident location to either a network point or a whole link length. A discretized version of a first-order macroscopic model allowing for an accurate representation of incidents precisely enables a flow dynamics perturbation evolution through space and time to be simulated. The paper presents the theory of the model and the first results from simulation using the newly-developed STRADA-I program for incident scenario modelling in mixed urban networks.
