The tools that allow simulation and evaluation of the effects of different control strategies and informative systems need the explicit modellisation of time-dependent costs flows and queues due to time-dependent demand and/or changes in supply that is a dynamic modellisation of traffic assignment. Most of the existing methods do not clearly define link flows and/or can hardly be extended to include en-route diversions from the initially chosen path because of the computational burden of keeping the identity of diverted flows. Moreover some of them rely on assumptions which do not necessarily rule out overtaking between users. A new dynamic network loading model, MICE, is presented which is able to represent flow propagation without FIFO rule violations or other inconsistencies even in a multi destination framework. The travel time function imposed gives automatic respect of a capacity value so that network bottlenecks can be adequately reproduced. The presented model is part of a new framework on travel time function models, this new approach overcomes flow propagation inconsistencies that are characteristic of exit link function formulations. Queue spill-holes link to link can be reproduced and examples are given on test networks to show the ability of the model to explicitly deal with bottlenecks and to represent while-trip rerouting.
Abstract