A behavioral shipment-based model of freight tour formation

Abstract

Tour formation is a distinguishing feature of freight transportation. An increasing amount of research is dedicated to the consideration of tour formation in freight simulation models, but these tour formation models lack statistical calibration on empirical data, do not represent shipments explicitly, or focus on a narrow segment of freight transportation. This paper presents a novel shipment‐based approach to model tour formation behavior. We developed an algorithm that constructs tours through iterative allocation of an additional shipment. Two choice models provide an empirical foundation to this algorithm. Parameters of these choice models are estimated on a dataset with information regarding over two million shipments. This information is gathered automatically from the planning systems of carriers transporting goods in the Netherlands. With this model we are able to reproduce observed tour patterns excellently for a given set of shipments. In addition, the model considers many objectives and constraints that determine the tour formation process and acknowledges differences between goods, vehicle, and location types. For example, shipments at ports tend to be transported with direct tours, while tours starting at a distribution center have more stops. This model can be applied in a shipment‐based freight simulation framework to construct tours for large third‐party carriers in the Netherlands.