This paper proposes a highway traffic assignment technique basedon an improved behavioral model of drivers' route choice as developed in a recently completed study in the netherlands. Route choice models are developed from data collected in three corridors in the netherlands. The models presented here, which are based on evidence of drivers' varying valuations of a number of road characteristics, are(a) probabilistic and (b) based on more variables than were used inprevious models. They contrast with the underlying route choice models of conventional traffic assignment procedures, which are typically based on a single measure of travel impedance (e.G., Travel time, generalized travel cost). A key feature of the models developed in the present study is that they are based on data describing the actual routes chosen by individual drivers. The paper describes how these models are used to generalize assignment methods through the exploitation of a multiclass-user technique. In an uncongested network, several routes typically will be predicted to be used between a givenorigin and destination; as congestion increases, so will the diversity of routes used. Several models appropriate for use in varying circumstances of data availability are presented and compared. Model inputs (e.G., Road attribute data) are described, and practical implications of the underlying structural assumptions are discussed. Spatial transferability of the models is appraised on the basis of the differing results obtained for the three corridors studied. Finally, advantages and limitations of application of the proposed assignment method compared with conventional procedures are discussed. This paper appears in transportation research record no. 1220, Forecasting.
Samenvatting