URBAN TRAFFIC NETWORK FLOW MODELS.

This paper addresses the development and comparative assessment of macroscopic network-level traffic flow models, which describe the behavior and interrelation between traffic variables defined at the network level. These variables include average speed, concentration, flow, the fraction of vehicles stopped in the network, and the two-fluid running time variables. Three alternative sets of interrelated models, each with a different starting postulate, are presented and tested in terms of their performance against a series of microscopic simulation runs corresponding to different concentration levels. The analysis indicates that the network-level traffic variables are interrelated in a manner similar to that captured by the traffic models established for individual road sections. In particular, a well-known linear speed-concentration model as well as a nonlinear alternative is found to be generally applicable at the network level.