A strategy has been defined and is under evaluation, where it is assumed that each intersection of a "guidance network" is equipped with local controllers responsible for computing and distributing route suggestions to individual vehicles. The strategy is characterised by "short term" and "local" adaptations due to both local disturbances and accidents. The assumption is made that a nominal "assignment" of traffic volumes is known to all these controllers in term of flows and percentage of vehicles having destination D to be routed from the ingoing link i to the outgoing link j. Based on the observation of downstream links state variables and their deviation from nominal values as coming from nominal assignment (e.g. system equilibrium assignment) the strategy has been defined so as to adapt the splitting rates at each node to these variations in real-time context. The second part of the paper deals with a first analysis of the results: two different scenarios have been considered with different penetration rates. In the first scenario a medium sized grid network of 117 nodes with a high junction per square km density is modelled. The beacon density is assumed to be high (one per intersection) and the traffic light control is assumed to be fixed time. The average network speed was between 20 and 30 km/hr. In the second scenario the effects of the strategy are assessed in those cases where accidents induce a severe loss of capacity on some links. The accuracy of the results and the indications coming from simulations (by using NEMIS network microsimulator) indicate that the implemented control strategies are really effective. (A) For parts A and C see IRRD 848020 and 848022
Samenvatting