DO STOCHASTIC TRAFFIC ASSIGNMENT MODELS CONSIDER DIFFERENCES IN ROAD USERS' UTILITY FUNCTIONS?

This paper discusses whether the approach of existing stochastic traffic assignment models, such as stochastic user equilibrium (SUE) models, is sufficient to describe road users' behaviour. Up to a certain point, perceived travel resistances can compensate for road users' ignorance of actual travel resistances. However, this approach does not fully allow for variations in road users' utility functions, for example the weighting of travel length against time. A heuristic modification of SUE is presented, in which two types of stochastic component occur, which consider: (1) road users' perception of the road network at the link level; and (2) differences between road users' utility functions. The original SUE model is the special case where component (2) is zero. The convergence problems of the algorithms are discussed, and a modification to accelerate convergence is proposed. To illustrate the paper's theoretical discussions, bundles of roads between two urban zones in Copenhagen, Denmark, are presented according to the different principles considered, and compared with the results of a comprehensive traffic survey. In particular, the main route choice of the ring motorway versus the bundle of routes through the city was modelled. Several topics for further research are proposed. For the covering abstract, see IRRD 889965.