Advances in Freight Transport Demand Modelling: An Assessment with Research Perspectives.
F Combes, F Leurent, LVMT, FR
The paper has a twofold objective: first, to review the models and approaches; second, to discuss which specificities of the freight transportation system are correctly accounted for
For a long time, freight transport demand models have been inspired by, and transferred from, the models for passenger demand. However, in recent years specific methods have been developped to address the modelling of the freight transportation system along various strategies which led to a handful of models, each of which being targeted at a specific aspect of freight demand.
The paper has a twofold objective: first, to review the models and approaches; second, to discuss which specificities of the freight transportation system are correctly accounted for.
The models for freight demand have been classified into three categories, depending on whether they are focused on representation of the supply side of the freight transport system, or the demand one, or the relationship between them. Such categories do not intend to be well-delineated, but rather to highlight the objective chosen by the modellers; moreover, there are ?composite? models which link together components taken from two or more of the categories.
The three categories are as follows:
1. models of the supply-demand relationship focus on the matching between supply and demand i.e. the transactions whereby the quantities and prices and qualitative conditions are set up, which lead to the shipments of goods. These models are oriented to the description of the transactions and the shipments; their primary goal is to describe in a statistical manner the freight transport market, and its long-term trends. The objective may be addressed in two alternative ways, either top-down from aggregate to disaggregate or bottom-up. A top-down approach may be adopted, as in the REDEFINE project, by working directly with indicators aggregated up to a high level. Conversely, a bottom-up approach may be taken starting from the disaggregate level of the shipments, as in the French survey ?ECHO?. In some ?composite? models the dependency between freight transportation as an economic sector and the rest of the economy is emphasized in order to analyse explicitly the linkage between the transportation conditions and the demand. This applies to Spatial Input-Output models such as the SCENES model (ME&P, 2002), the SMILE model (Tavasszy, 2003), and the demand generation component of EUNET (Jin & Williams, 2005).
2. supply-oriented models of freight demand focus on how the supply side behaves and is developed in a demand-responsive way, without neglecting the demand side nor reducing it to a mere matrix of fixed origin-destination flows. Models that address only the design and management of a supply firm, for a fixed, given demand, fall outside the scope of this paper. Groothedde (2003) proposed a model to optimize the development of waterway transport and of logistic platforms in the Netherlands, in relation with servicing logistic chains for a set of identified, regular customers, across whom the goods can be allocated dynamically. In the MOSES model of Kettner et al (2003), the focus is on the efficient bulk and break of a set of shipments by a rail services operator: the engines, wagons and tracks are allocated to the services so as to meet the requirements of a quality-responsive demand.
3. Demand oriented models: priority is granted to the representation of the demand side i.e. the customers? behaviours. Some models follow a macroscopic approach and establish relations of a statistical nature between the demand behaviour and an aggregate representation of the supply side: e.g. the EUNET model of Jin & Williams (2005), in which a statistical linkage is modelled at the zone level between the amount of logistic facilities and services, and the logistic decisions taken by the shippers. In other models the customers? costs and decisions are described in a disaggregate way: the Nodus model of Jourquin & Beuthe (2005) enables one to represent the logistic options available to the shippers in a very detailed way in terms of both costs and operation time, leading to a per-ton cost based on the logistic process. Eventually, a highly disaggregated representation of the demand is combined to an aggregate representation of the supply: e.g. the Freturb model of Routhier et al (2002).
After introducing the models that are currently available (at least for research purposes), an assessment is performed along the following set of issues:
1. The representation and its level of disaggregation
2. The segmentation of the demand.
3. The modelling of modal choices.
4. The representation of times and costs throughout the logistic processing of a shipment.
5. The representation of the relationships between the economic agents.
6. The appraisal of the social costs, the measurement of the social welfare.
7. The ability to test policies.
Association for European Transport