Abstract
Multi-Modal freight models are traditionally built following the well known “four steps model” in which generation, distribution, modal-split and assignment are seen as separated modules. An alternative approach is to represent the multi-modal network by means of a “mono-modal” one, in which each particular transport operation (loading or unloading operation, transhipment, ...) is represented by a dedicated “virtual link”.
This approach, promoted by several authors, often referenced to as “super networks” or “virtual networks” is proven to give interesting results, but has some kind of “hidden trap”, linked to transport distances, that will be presented in this paper and that can only be solved using appropriate assignment techniques.
In a previous paper by Jourquin and Limbourg [15], it appeared that the equilibrium assignment techniques only partially solve the “distance trap”, because the obtained solutions are very close to those obtained by a simple “all-or-nothing” assignment. This can mainly be explained by the fact that, in inter-urban, inter-regional or international traffic, only a small part of the journey is done during the peak-hours in congested areas, and that the demand data that can be collected on large areas is often available only on a yearly basis, making it practically impossible to have a good estimate of the demand at a given period of the day.
However, one can not consider that all the traffic between a given pair of origin and destination will be charged on the same transport mode and/or route. This paper outlines a new usable deterministic multi-flow/multi-modal assignment technique that can be applied on large (virtual) freight transportation networks and that offers a satisfactory solution to the “distance trap”.
