Towards a sustainable multimodal freight transport system for the Randstad

Prof. dr. ir. L.A. Tavasszy, TUD en VU
2009-2013

Duurzamer goederenvervoer

'Towards a sustainable multimodal freight transport system for the Randstad' richt zich vooral op het sterk groeiende goederenvervoer zoals zich dat in en rond de drie Randstedelijke mainports (de havens en Schiphol) afspeelt. De wetenschappers onderzoeken productie-, consumptie-, handels- en toevoerketens en buigen zich onder meer over het concept van overslagterminals en het vraagstuk van de stedelijke distributie.

Lees hier een interview met Lóri Tavasszy  uit april 2009 

Download hier een presentatie uit oktober 2009 over het programma.

General aim

Freight transport in the Randstad is characterized by an intensive internal transport flow and worldwide connections through three mainports. The expected growth poses serious challenges to the functioning of transport networks and the sustainability of local living environments. Surprisingly however, research and models on freight transport until now have had a too narrow scope to treat the problems relevant to the Randstad, omitting important information about changes in trade patterns and logistics chains. In this programme we develop an integrative model for freight transport to, from, and within the Randstad area with a multi-modal perspective. The modelling problem requires a consistent and interlinked development in 3 areas:

1. The major trade and logistics linkages of the Randstad. The main focus is on modelling production, consumption, trade and supply chains. Existing models like SMILE need to be re-framed and refined to fit the specific data availability and complexities of the region which do not appear at the national level.
2. Multimodal routing of freight flows, including transshipment of freight at terminals within and around the Randstad, are a new focus area for reliability and cost improvements. Different organizational and technical designs will be evaluated, taking into account environmental constraints and objectives.
3. City distribution: current issues such as cooperation, consolidation, transport reorganisation and routing improvement require more detail in modelling than available at present. We develop an agent based model which translates the movement data produced by the first and second models into detailed city level freight deliveries.

Expected scientific results

The main results include:

• A detailed dataset on freight transport flows according to source, destination, type of goods, and employed mode of transport;
• A method for the estimation of interregional freight matrices using partial data from multimodal transport chains;
• A model that describes inventory location choices of shippers in international logistics chains;
• A method for system optimization of services and facilities in multi-product, multimodal networks;
• A situated multi-agent system model for optimization of city logistics in poly-nuclear urbanized regions.

All models will be theoretically specified building on existing research and empirically validated with the data available for the Randstad. Also applications will be made showing the real world value of the models. Note that the database created is an attractive base for further research. The project connects to a stream of work concerned with the problem of data fusion (e.g. the OPUS project), where regional data management approaches are developed for complex systems working e.g. through forms of Bayesian analysis.

Expected results relevant for policy makers

The project will provide models, data that produce quantitative insights in the possible savings associated with the improvement of the freight networks in the Randstad and the improvement in international competitive position that can be gained. The project will indicate what is the importance of extending certain infrastructure or improving services. Thus a freight master plan can be developed based on a comprehensive system of models and data which is now lacking. The project will provide answers to the question which sectors need to be stimulated to use the Randstad as a key region in their logistics operations. Also, opportunities will be identified to extend logistics operations from servicing transit of freight to extending services towards logistics operations and advanced production systems, thus increasing the potential attractiveness of the region for international investments. Finally, the application of the models to develop a consistent strategic design of the freight system across various levels of governance and involving many different stakeholders, will be a unique policy experiment for the region. Reporting on the process and the results followed in the project will be of great importance.

Dissemination plan

The project results (models, databases) will be made available to the public to apply in practical policy development. The dissemination of findings will take place during the project in various ways.

• Project Steering group and user group
• Final conference for a broad audience
• Case study where practitioners will be involved

Project 1: Modelling complex freight demand structures (pt 1: trade
and transport data, pt 2: logistic chains)

Researcher project 1: Maureen Lankhuizen, VU
Supervisor: Henri de Groot, VU
Researcher project 2: Igor Davydenko, TUD
Supervisor: Lori Tavasszy, TUD

This project proceeds along two lines, to be carried out by two researchers who will work together closely. The first line, project 1a, supervised by Dr. de Groot, will expand on an existing cooperation with Statistics Netherlands. It aims to develop a detailed description of freight transport flows according to source, destination, type of goods, and employed mode of transport. As such, it also provides detailed information on the extent and characteristics of re-export flows. The research will largely focus on exploiting information from both the trade and transport statistics which are currently being linked and where the project aims to offer a substantial contribution (see Linders et al., 2007), mainly by refining the statistical models to derive consistent estimates for missing data at a detailed levels of aggregation. Further development of the integration methodology will focus on: 

• deriving more detailed integrated estimates for trade and transit flows in terms of value and weight, by country of origin and destination;
• refining the statistical model using minimum information theory and econometric methods to estimate with limited data in order to derive consistent estimates for missing data at a more detailed level of sector aggregation;
• projection methods for predicting future trade and transit flows by transport mode;
• improving data-quality assessment in order to determine uncertainty margins on source data;
• sensitivity analysis of the integration methodology and - outcomes to the prior assumptions with respect to the distribution of underlying, sometimes missing, data;
• estimation of missing data using structural models of trade and information on total logistics costs from the second line of the project.

In the second line of this project, project 1b, we make the link between trade and transport flows structurally dependent on firm behaviour, by introducing a of a model of decision making within logistics chains. The roots of this model lie in pioneering work on logistics modelling with the SMILE model (see Tavasszy et al, 1998). This model extends the conventional 4 step freight modelling approach by an additional step representing the logistics chains in the system. We extend the existing model by including chains ending at the county (gemeente) level, carrying out a more detailed calibration of the model using the new freight flow databases. Where needed the databases will be further detailed using spatial economic data (LISA). A survey among shippers is held to obtain shipment data needed to develop a representative picture of preference structures for various commodity types and logistics chains. New forecasts will be prepared for various segments using the projections available from the WLO scenarios, now adding new detail which is relevant for the Randstad including varying assumptions concerning policy relevant variables such as oil prices levels and road pricing. The results may provide different forecasts for freight transport than available at the moment, as changes in distribution chains are now taken into account. In addition, policy relevant information will be obtained by means of forecasts of re-export flows and associated geographical shifts in logistics functions.

Project 2: System optimization of multimodal freight networks

Researcher: Mo Zhang, TUD
Supervisor: Bart Wiegmans, TUD

A modelling approach will be developed in order to combine economics of transport network design and environmental economics into one integrated optimization model. This integrated approach will enable an investment and exploitation analysis combined with a valuation of external effects such as CO2 emissions, noise nuisance and particulate emissions. Extensions in port and terminal capacity in the mainports need to take place within the boundaries of sustainable accommodation of the expected freight flows. Besides capacity enlargements, also organisational and technical improvements could be implemented in ports and terminals. All improvements, either capacity enlargements or more innovative solutions must lead to better sustainability, better reliability and lower costs. In the analysis, investments and exploitation for different scenarios as well as organizational and technical designs will be evaluated, taking into account environmental constraints and objectives. The conclusions will help the Randstad area and its mainports to become more sustainable while at the same accommodating more freight flows.

Economic analysis will provide a first basis for the investment and exploitation analysis of the different proposed strategies and their respective investments (Newman et al., 2004). Different calculation methods will be applied in order to analyse the associated costs whit the respective alternatives. Furthermore, besides the financial analysis, also sustainability criteria (such as CO2 emissions, noise nuisance and particulates emission levels will be included the models that provide the input for the evaluation of alternatives. External costs will be included in the optimization procedures.

The organisational component focuses on the different actors involved in the needed investments. Especially, public private partnerships are needed in order to be able to fully accommodate the expected freight flow increases. Earlier research by Wiegmans et al. (2002) proved that it is quite a challenge to make the public and private sector work together in freight related infrastructure investments. However, current practices in investments in container terminals prove that in these cases a modus has been found where the private sector takes a considerable part of the risks and investments. This might serve as the example for the other needed investments in road, rail, barge and terminal capacity.

The technical part will focus on the technical possibilities to realise efficiencies, cost savings and better quality in the intermodal transport chains. Several researches have been performed into the possibilities of innovations to help improve the competitive position of intermodal freight transport (Wiegmans, 2005; Wiegmans et al., 2007). Many innovations are being developed but most of the innovations fail to enter the market successfully. We will analyse these innovations in sustainable, financial and organisational terms in order to see what they might contribute to a more sustainable and attainable Randstad area. The results of the research will be both methodological (approaches to optimize multimodal networks) and application oriented (insight into promising network configurations for the Randstad).

Project 3: A situated MAS for urban logistics in the Randstad

Researcher: Nilesh Anand, TUD
Supervisor: Bert van Wee, TUD 

Tools that support ex ante evaluation of spatial urban freight planning and spatial plans are a basic requirement to construct sustainable urban freight policy. There is a need for an artificial environment in which policy can be developed and tested in order to cope with the increased complexity of reality. The main characteristics of the urban logistic concepts are the high number of stakeholders and the heterogeneity of their needs, which force a very thin segmentation of proposed services, often going below the minimum scale for economic feasibility. Key research issues such as cooperation, consolidation, transport reorganisation and routing improvement require additional features and detail in modelling than available at present.

We develop an operational agent based model which makes use of the movement data produced by the first and second project to model detailed city level freight movements. Multi-agent systems provide a paradigm for modelling and developing self-adaptive distributed applications. A multi-agent system (MAS) models a distributed application as a set of autonomous, cooperating entities who reside in a common (distributed) environment. The use of agents for the representation of organizations or interest groups provides a realistic modelling of the process. It allows to address desires, beliefs, and preferences of actors in the planning process itself and translate them into visions of actors.

One interesting family of MAS, called Situated MAS, employs agents who themselves do not exhibit advanced intelligent behaviour. Instead, Situated MAS are designed to achieve system-wide behaviour through agents with purely localized (situated) perceptions and actions in the environment (i.e. a selected group of municipalities, products or services). The first step of the methodology compromises the ontology of logistics profiles or entities. The concept definition profile or entities is defined as the logistics characteristics of homogenised groups, defined by the features of territory, needs of the transported goods and the agents implied in each logistic chain. The second step encompasses the evaluation of different logistic solutions and measures based on different criteria (e.g., freight platforms, distribution centres, auctions, bus-lanes, time-windows etc.) in terms of their suitability for serving the different profiles by doing some (agent-based) simulation experiments. The prime advantage of Situated MAS is flexibility, i.e. their ability to adapt to change in the system (agents join or leave the system, the environment changes, the problem changes dynamically). Another advantage is scalability of the system. We will start with a hypothetical city and extend it easily to real-life experiments with more cities covering the Randstad. The result will be a model, which can be applied to poly-nuclear region systems, in contrast to the conventional single-city model systems.

Ultimately we will evaluate the economical and business feasibility of the adaptation of more sustainable urban logistic solutions for the Randstad. The outcomes will not be the result of comparing a relative static set of criteria (like MCA), but will be a result of a dynamic exchange of information between actors with different objectives and sets of criteria and therefore can support the definition of a sustainable Randstad future master plan.