Network Transmission Model

Abstract

Traffic flow can be modelled at different levels, of which the vehicle level and the link level are mostly used. Recently, traffic relationships at a higher, network level are found, and a dynamic traffic flow model has been introduced. This paper tests such a model, the Network Transmission Model (NTM), for the city of The Hague. In this model, for each zone (i.e., parts of a network, neighbourhoods in this case) the number of vehicles in the network as well as the outflow is determined in each time step. The basic information from the city planners and the network is taken (OD, road lengths and capacities), and used as input to the NTM. Afterwards, the model parameters are adapted with input from practitioners of municipality. The parameters in the model had an understandable effect on the traffic congestion patters. After adaptation, the model predicted a traffic patterns in line with the expectations of experts of the The Hague municipality. Generally the model predicts the congestion in the right zones and the propagation is predicted correctly. Moreover, the parameters have an interpretation so tuning is done. Finally, the model is quick enough to serve as on-line support for traffic management decisions. As face validation two different cases (adapted OD matrix and a modelled incident) were assessed with the model without adapting model parameters. The predicted traffic patterns were assessed as reasonable. The test also revealed further improvements in including traffic waiting for parking spaces and the inabilities to include turning restrictions.