Traffic incident modelling in mixed urban networks
MONGEOT
Type de document
RAPPORT DE STAGE
Langue
anglais
Auteur
MONGEOT
Résumé / Abstract
Traffic incidents causing a reduction in capacity can result in significant disruptions to traffic flows in road networks. Incident management strategies are then required, using traffic control and information systems, to minimise the disruption and congestion in the network. This thesis has been concerned with the specification, development and application of a new traffic modelling approach for incident conditions. This provides new fundamental understandings of incident impacts as well as offering a basis to support incident management strategies. Reviews of traffic incident characteristics and of urban network simulation models are presented and underline the dual incident impacts on the traffic flow and on the traffic assignment. These reviews form the basis for the traffic flow model developed. This model is based on the first-order macroscopic theory, to represent the dual incident severity on the carriageway width and on the free-flow speed, at any position of a network link and along any extent. The perturbation consequences of the proposed model on the flow dynamics were studied by extending the Lighthill and Whitham (LW) theory to incident discontinuities. This enabled traffic characteristics and shock-wave trajectories to be developed both graphically in a Time-Space plane and analytically; in addition, analytical expressions were derived for queue lengths and perturbation c1earing times in relation with the incident severity and duration and the demand level. The incident model was also implemented into the STRADA simulation tool as the 'STRADA-I' flow model, whose consistency with the extended LW theory was checked on a number of simulation runs. The STRADA-I flow model consistency was also verified by simulation comparisons carried out against the well-established CONTRAMI simulation tool on a small network, by studying queue build-up, queue decay and spill-back processes through network links and nodes. One result of the comparison was the accuracy of the dynamics representation in the STRADA-I flow model. The issue of traffic assignment in incident conditions is discussed, and initial requirements for developing an assignment module for incident conditions are specified. This incident assignment module is intended to take account of real-time traffic information and DRG, in addition to other more traditional traffic control strategies. Its application, together with the incident flow model, can be found in the development of simulation of integrated real-time traffic control and management in a mixed network in incident conditions. Recommendations are included for further research and development necessary. Thèse de doctorat de philosophie