The Transport Research Centre of Rijkswaterstaat has started work on a new version of the Dutch National Model for Traffic and Transport (LMS - Landelijk Model Systeem verkeer en vervoer). The new version of the model (LMS v 8.0) will be dynamic: it will account for 11 time periods of varying levels of demand. In the preparatory study for LMS v 8.0, RAND Europe carried out a study of possible dynamic traffic assignment algorithms for the new LMS. As a consequence of the study by RAND Europe, three organisations were asked to examine their dynamic traffic assignment models for the possibility of implementation in the new version of the LMS. This report contains the results of the study of the INDY model. INDY (Interactive Dynamic traffic assignment model) is a dynamic traffic assignment model that combines analytical simulation with a route choice component. INDY also works with different user-classes and vehicle types. The model can account for different types of infrastructure, and variable travel costs and travel times. On the basis of vehicle and users attributes, INDY accounts for the route choice behaviour and the traffic flows on the network. The model can account for the dynamic effects of peak and off-peak traffic, and takes as input dynamic origin destination (O/D) matrices. The major results of the study are that INDY can take as input the modified LMS network, dynamic O/D matrices and different user-classes and produce results. The calculations were performed on a network with over 25,434 links and 109,292 non-zero O/D pairs. The dynamic O/D matrices were delivered to AVV Transport Research Centre by RAND Europe, and used by all three organisations in the model tests. In total, 181,567 routes were identified in the network. The number of routes per O/D pair varied from a minimum of 1 to a maximum of 14. With a simulation time step of 10 seconds, the run time for a single iteration was 10 hours for the single user-class case. INDY also produced the necessary data on total kilometres travelled and cumulative delay for different user-classes and for different types of roads, including the "top 5 most congested traffic links" in the morning and afternoon peaks. The results show consistency both between the different runs and with the outputs of the LMS QBlok assignment and with trip statistics for the Netherlands. Finally, graphics indicating the percentage of the free-flow speed achieved and the intensity/capacity ratio were generated. The scope of the study excluded verification of the results, although comparisons with statistics are provided when feasible. Finally, this report contains an assessment of INDY's current and potantial performance based on a table of hard, soft and optional requirements provided by AVV Transport Research Centre. INDY 1 1.0 meets all but a few of the requirements. Furthermore, this report describes what must be carried out to adjust INDY to meet the (few) unmet requirements. The conclusion is that there are few requirements that are hard to meet. Finally, the report contains suggestions for how INDY can be made more efficient, both in terms of memory use and run time. (Author/publisher)
Samenvatting