The stability of assignment results

Abstract

Paper presented at the 22nd Annual Southern African Transport Conference 14 - 16 July 2003 "National issues affecting the movement of people and goods - strategic approaches", CSIR International Convention Centre, Pretoria, South Africa. ABSTRACT: The most common technique used in the traffic assignment phase of transportation modelling is the user equilibrium method. This method takes the effect of congestion on travel times into account by implementing Wardrop’s first principle. This principle states that if two or more routes between an origin and destination pair are used then the travel times on these routes will be equal, i.e. in equilibrium. The equilibrium assignment technique is an iterative one, and one of the problems for transportation modellers is to decide on the number of iterations that should be performed. The majority of programs that implement the equilibrium assignment technique include default values for stopping the iterative process. These stopping conditions are usually the number of iterations or one or more types of stopping criteria. This paper presents the results of a study where a number of different values for stopping the iterative process were tested. It was found that the degree of convergence required for stable results is dependent on the reason for doing the assignment. More iterations are required if the results are to be used in an economic evaluation than if volumes are needed for road design purposes. The paper provides recommended values that should be used as stopping criteria when doing equilibrium assignments.