A stochastic transit assignment model considering differences in passengers utility functions

Otto Anker Nielsen


The paper presents a framework for public traffic assignment that builds on the probit-based model of Sheffi and Powell [Sheffi, Y., Powell, W.B., 1981. A comparison of stochastic and deterministic traffic assignment over congested networks. Transportation Res. B. 15 (1), 53–64; Sheffi, Y., Powell, W.B., 1982. An algorithm, for the equilibrium assignment problem with random link Times. Networks 12 (2), 191–207.]. Hereby, the problems with overlapping routes that occur in many public transport models can be avoided. The probit-based model with modifications similar to the principles in Nielsen [Nielsen, O.A., 1996. Do stochastic traffic assignment models consider differences in road users utility functions? Twentyfourth European Transport Forum (PTRC Annual Meeting). London, UK, Seminar M.] is used as a starting point. This makes it possible to describe passengers’ different preferences towards different sub-modes and against transfers. This also considers dependencies of choices through chains of sub-modes. The simulation of perceived travel times is extended to describe differences in the distribution of travel- and waiting times for different sub-modes. Parallel lines are frequency aggregated in order to handle waiting times appropriately. Initial tests on a full-scale case show that the methodology can describe route choices in public transport very well. This is both due to the model’s ability to describe overlapping routes and due to the many different coefficients, error components and distributions that make it possible to calibrate the model. In practice, the many parameters might also be the main weakness, since this complicates the calibration. At the end of the paper, proposals to coefficients are presented based on a Danish SP-analysis. This demonstrated the applicability of the method.