This paper presents some of the main issues arising from stochastic user equilibrium (SUE) assignment, and compares the implementation of SUE by different commercial transport planning packages in the UK. SUE models the effects of congestion, and the impact of daily variability and routing errors on the spread of traffic through a road network. It embeds a cost randomisation procedure within its iterative solution method for capacity restraint or user equilibrium assignment. The SUE equilibrium model must be calibrated with respect to shape and the range of possible error values. SUE's assignment model requires convergence for the following elements: (1) the capacity restraint, requiring convergence in requiring convergence in the interdependence between link flows and costs; and (2) the stochastic element, to minimise flow changes resulting from new randomisations. SUE assesses these elements separately. When SUE is applied, the following issues concern benefit calculations: (1) a possible bias against certain road types; and (2) the impact of imperfect randomisations on benefit calculations. Of the five main UK assignment packages, SUE is not available in CONTRAM and EMME/2, but it is implemented in QVIEW, SATURN and TRIPS.
Samenvatting