Network Aggregation Effects upon Equilibrium Assignment Outcomes: An Empirical Investigation

Abstract

The level of spatial detail (i.e. zone size and network detail) used in transportation analyses is commonly regarded as an important factor affecting the accuracy of the resulting estimates of the impacts of a transportation plan. The precise effects of the level of detail are, however, largely unknown. To investigate these effects empirically for the car traffic assignment module, an experiment was designed to allow especially the study of the individual as well as combined effects of the level of detail and the type of assignment model. It involves the application of various assignment models at different levels of detail. Three network models were developed for the road network of Eindhoven (population: 200,000): a fine, a medium and a coarse network model. In this article the results of the equilibrium assignments are presented, which are occasionally compared with all-or-nothing outcomes. Mainly load figures are dealt with here. The experiment indeed showed a significant effect of the level of detail on most assignment outcomes. This effect proved to be consistent but diminishing: an increase in the level of detail always yielded better results but only marginal improvement could be obtained beyond a certain level. Compared with all-or-nothing assignment results, equilibrium loads agree much better with the counts at all levels of spatial detail.