2012 Volume 42 Issue 1 Pages 77-91
The standard method for predicting traffic flows on urban road networks, static traffic assignment, is based on the principle that drivers seek their own least cost routes from their origins to their destinations. This principle corresponds to a network user-equilibrium state in which for every origin-destination pair, all used routes have equal costs and no unused route has a lower cost. Under somewhat mild assumptions, the network equilibrium problem can be formulated as a convex optimization problem with linear constraints, and solved with an iterative algorithm. The precision and speed of such solutions has increased remarkably during the past ten years. Although the total flows on links of the urban road network are uniquely determined in this formulation, route flows and multi-class link flows are not. An additional assumption, called the condition of proportionality, may be imposed to determine these flows uniquely. This condition is the basis for the design of a new algorithm, Traffic Assignment by Paired Alternative Segments (TAPAS). Results of computational experiments for two network representations are presented. The difference in the network representations pertains to restrictions on the use of certain links of the network by trucks in a two-class assignment. The findings illustrate the importance of class-specific network representations and the condition of proportionality in road traffic assignments at the route and link level.
JEL Classification: R41
