Building latent segments of goods to improve shipment size modelling: Confirmatory evidence from France

Freight transport demand is a demand derived from all the activities needed to move goods between locations of production to locations of consumption, including trade, logistics and transportation. A good representation of logistics in freight transport demand models allows us to predict the effects of changes in logistics systems on future transport flows. As such it provides better estimations of the costs of interaction and allows to predict changes in spatial patterns of freight transport flows more accurately. In recent years, the attention for freight modelling has been growing and new research work has appeared aimed at incorporating logistics in freight models. In this paper we review the state of the art in the representation of logistics considerations in freight transport demand models. Our focus is on the service and cost drivers of changes in logistics networks and how these affect freight transport. Our review proceeds along a conceptual framework for modelling that goes beyond the conventional 4-step modelling approach. We identify promising areas for freight modelling that have an integrative function within this framework, such as spatial computable general equilibrium models, supply chain choice models and hypernetwork models.

The application of discrete choice models (e.g., logit, probit) to study modal choice in passenger transportation has had a wide acceptance in the literature. However, little success had been reported on the application of these models to study the demand for freight transportation. This is mainly because in freight transportation a model that merely attempts to explain the choice of mode without taking into consideration other related factors, such as shipment size, is only one part of a complete model. Another type of models known as inventory-based models, which takes these factors into consideration, has been developed and applied with a greater success. However, the data requirement of these inventory models has hampered their applicability, especially in situations with limited data on goods movement. This paper presents a new approach to study the demand for intercity freight transportation. The model proposed in this paper utilizes the strength of discrete choice models (e.g., probit) in explaining the process of mode choice as one part of a complete model. The complete model is presented as a joint discrete/continuous choice model for the choices of mode and shipment size. The model is practical in that it requires the same amount and quality of data that would be required to develop a standard disaggregate mode choice model, and it can be estimated using simple two-stage estimation methods which utilizes standard probit maximum likelihood and ordinary least squares estimation techniques. Key words: disaggregate, freight transportation, maximum likelihood, mode, model, probit, shipment.

The freight transport system is a major determinant for the competitiveness of logistics and production activities. On the other hand, logistics and production shape freight transport demand. Therefore, freight transport demand models are needed that can capture the influence and requirements of ongoing trends in production and logistics. In this chapter, the German automotive industry is used as an example of how certain trends might influence the development of freight transport demand. Here, the impact of changes in the supply chain structure on freight transport demand is emphasised. Furthermore, data sources available on the sectoral level are discussed. Based on the insights into ongoing trends, available data and sectoral characteristics, a concept for a sectoral freight transport model for the German automotive industry is developed.

The increasing importance of logistics and its effects on transportation processes have been considered in freight transport demand modelling by new model developments incorporating several logistics aspects. However, transport logistics hubs – hubs without storage function that are primarily used for transhipment – seem to be not considered adequately in most freight models although they are crucial for freight transport. Currently it is not apparent which models do consider transport logistics hubs and how they do. The aim of this article is, therefore, to review how transport logistics hubs are considered in freight transport demand models and to discuss remaining gaps and challenges. Based on an extensive literature review a broad and internationally oriented analysis of transport models has been conducted. More than 100 models were examined regarding their consideration of logistics in general and transport logistics hubs in particular. The article points out the major importance of transport logistics hubs within freight transport and, thus, underpins the fact that they should also be considered sufficiently in freight demand models. It reveals that there are numerous models in application worldwide and that most of these models do not consider logistics or hubs. Only few models integrate logistics hubs, in general, and transport logistics hubs in particular. Furthermore, the consideration of transport logistics hubs within these models differs quantitatively (number and types of hubs) but also qualitatively (detail of hub integration) – also depending on model properties (e.g., scale of analysis, resolution or level of aggregation). Within freight transport processes, transport logistics hubs gained growing empirical significance in the recent past. This implies a major relevance to consider them in demand modelling, too. However, the article reveals that the integration of transport logistics hubs lags behind their empirical significance and that there are certain restrictions which hamper an adequate integration. The potential of many models is limited by reason of data availability. The lack of detailed data remains a major challenge that is, frankly, not new to the freight modelling community. Closing this data gap will help to consider transport hubs adequately and, thus, increase accuracy of models which would enable policy decision-makers to come to more effective decisions.

Analysing freight shippers’ mode choice preference heterogeneity using latent class modelling

5 - Inventory Theory and Freight Transport Modelling

A logit model for shipment size choice with latent classes – Empirical findings for Germany

Due to the structural heterogeneity of freight transport and its long-term subordinate consideration, there have been low research activities in this field as well as low motivations to integrate this segment into transport demand models for decades. There is also no generally valid framework for freight transport modelling in comparison to passenger transport modelling. Without a common framework, it is difficult to exchange ideas between scientists and to enable an efficient communication and solid agreements between a client and a contractor of a transport modelling service. This contribution presents a typological order of characteristics of freight transport demand models which covers their internal structures and employed methods. In addition, a systematic overview of selected international freight transport demand models is given.

Freight transport demand modelling (FTDM) from dispatch logs and partial observations play a key role in providing realistic expectations of freight transport needs and in assessing the environmental impacts of new policies. Shaping freight transport to answer the current challenges of green logistics – decarbonization, energy consumption optimization, intelligent and sustainable management – requires a continuous assessment and adaptation of the whole logistic system. Demand-centric models may support these ends. Understanding advances in freight modelling is thus crucial to overcoming FTDM challenges, including data scarcity. This paper provides two major contributions: i) a review of the state-of-the-art models on freight transport demand modelling, and ii) an integrated overview of how these models can support decision-making aligned with sustainability goals towards green (zero emissions) and digital logistics.

This paper addresses two important issues of econometric freight transportation demand analysis: the simultaneity between quantity shipped and mode/destination choice, and the effect of spatial price competition on the demand. The use of switching regression techniques allows estimation of the model because shipment size and mode/destination choice are derived from the same optimization problem. In the theoretical model, spatial price competition determines the company's market area and therefore determines its sales and shipment sizes. The empirical model provides consistent estimates of unconditional freight demand. The estimates are necessary to forecast transportation flows and derive elasticities for policy analysis.

Freight transportation activities in the present circumstances are closely related to a larger context of logistics decision-making, but logistic requirement is not relevant to all types of goods. Most existing freight transport demand models used for mode and route selection still focus only on the direct factors. This paper discusses the taxonomy of logistic and transport models, and the factors can be incorporated into freight transport demand models in order to improve the forecasting capability of the model.

The approach utilises a simultaneous equations system to derive a discrete/continuous choice model for the choices of mode and shipment size in the intercity freight market. The method is applied to the demand for intercity freight transport using data from the US Census of transportation.

Competition between freight and passenger transport for the use of the road infrastructure is an increasingly important problem. This research is carried out in the framework of the DIDAM (Disaggregated demand and assignment models for combined passengers and freight transport) research project which aims at proposing some advances in fundamental research in transportation modelling and analysis, and this paper presents some preliminary results obtained on the Belgian network. The DIDAM project's methodology is organized around two axes: on one hand, a disaggregated dynamic demand model for freight transport is proposed. In this model, the freight transport actors are represented by agents. These agents are extracted from existing databases and their behavior is then generated by means of a simulation that tries to represent the interactions between shippers and carriers who both try to minimize their costs. The carriers try to fill their trucks and to combine several trips to maximize their benefits (by minimizing the empty-trips for example). This step of the model faces a well-known optimization problem called Traveling Salesman Problem (TSP). In the meanwhile the shippers try to find the best transport opportunity by putting the carriers into competition. On the other hand, the Belgian road network is represented in a GIS based transport model software developed at FUCaM and called NODUS. Each arc of the network is associated with a dynamic cost function built using standard OD matrices for passenger transport combined with global traffic density data. By doing this, one takes into account the passenger flows everywhere and at every moment of the day. That way, the rush-hours are also taken into account. These dynamic cost functions are then used by the transporter-agents of the simulation that tries to minimize their costs. A time-dependent OD matrix for freight transport is built as result from this process. Finally, a prototype of a joint traffic assignment model is being developed. This new assignment procedure assigns the flow for each time-slice sequentially, keeping the still running flows from previous time slices into account. Still under development, this algorithm should provide the agents a feedback on their traveling time, which could be retained to allow them to change their habits if needed during a next trip. This paper will first present the time-dependent freight origin-destination matrices generation process, discussing the agent's generation, their characteristics and the implementation of their behavior in a simulation. Finally, the first time-dependent freight transport OD matrices built on the basis of this methodology will be assigned on the Belgian network using the experimental assignment model.

The paper discusses the theoretical and empirical evidence on the subject and concludes that freight mode choice can be best understood as the outcome of interactions between shippers and carriers, and that mode choice depends to a large extent on the shipment size that results from shipper-carrier interactions. These conclusions are supported by economic experiments designed to test the hypothesis of cooperative behavior. This was accomplished by conducting two sets of experiments (ones with the shipper playing the lead role in selecting the shipment size; and others in which the shipment size decision was left to the carriers), and by comparing their results to the ones obtained numerically under the assumption of perfect cooperation. The comparison of results indicated that the experiments converged to the perfect cooperation case. This is in line with the conclusion from game theory that indicates that under typical market conditions the shipper and carrier would cooperate. These results also imply that it really does not matter who “makes” the decision about the shipment size and mode to be used at a given time period, as over time the shipper—that is the customer—ends up selecting the bids more consistent with its own interest. In other words, these results do not support the assumption that freight mode choice is solely made by the carriers.

The choice of shipment size in freight transport depends on the logistical imperatives of shippers and the technical possibilities of carriers. The choice is closely related to the mode of transportation and therefore important for public policy. The theory of optimal shipment size and mode choice is robust. Numerous inventory-theoretical models of optimal shipment size are applied by shippers in operational contexts. However, none has been validated empirically over a large and heterogeneous population of shipments, and thus none are useful for modeling freight transportation demand, particularly because of the lack of adequate data. The simple economic order quantity (EOQ) model was assessed empirically on a national scale over a heterogeneous population of shipments. The French ECHO database, which observed commodity flow rates between shippers and receivers, was used to estimate the EOQ shipment size specification, and the validity of the EOQ model was confirmed. The study revealed the dominant role of the commodity flow rate between shipper and receiver and of commodity value density. The relationship between mode choice and shipment size is highlighted.