Twin Hub Network: Identification of Promising Networks. Comparison of Approaches and Results

Twin hub network is about bundling rail container flows from and to the seaports Rotterdam and Antwerp and
smaller seaports in the range Duinkerke – Amsterdam. The bundling is to lead to larger trainloads, higher service
frequencies, a higher network connectivity (serving more inland terminals) and a better utilization of rail infrastructure,
all leading to lower (generalized) transport costs and – derived from the modal shift – lower external transport costs.
The concept contributes to making freight transport more sustainable, increasing the territorial / economic cohesion
of Europe, and to a decline of regional disparities, as also smaller seaports and inland terminals can be accessed by
rail.
The bundling is to take place by hub-and-spoke networks. The spokes begin in different seaports or different rail
terminals of a seaport or end at different rail inland terminals v.v. The hubs lie in the gravity points of the envisaged flows, namely Antwerp and Rotterdam. Twin hub network actually is a title for a larger number of hub-and-spoke
networks, each consisting of 2-8 train connections with mutual rail-rail exchange at the hub. Some hub-and-spoke
networks run via Antwerp, others via Rotterdam, dependent on the geographical orientation of a network and the
availability of hub nodes.
The question now is which rail terminals should be connected by Twin hub network and which hub is to be used by a
certain hub-and-spoke network. This service network design problem focuses on the minimisation of system costs by
maximising the size of trainloads, hence minimising fixed train costs per load unit. The challenge is a short term (pilot)
and a long term one (imaging the general potential). The paper gives an overview of the different modelling and other
approaches applied.
The identification of a promising pilot network took place by 1) identifying promising Twin hub regions, 2) selecting
connections between promising regions and the hub to use, 3) selecting the best inland terminal to use in a promising
Twin hub region, 4) analyzing the feasibility of the connections and network, 5) based on analysing the potential
modal shift . To image the long term potential, a large scale optimization challenge, 6) the project developed a the
so-called Twin hub bundling tool. Step 1 took place by spreadsheet modelling of flows in search for regions that can
be accessed by train as the bundled flow sizes exceed certain pre-defined thresholds. The results were compared
with the marketing results of the commercial departments of the rail operators. The selection of connections and hubs
(step 2) took place by discussion, the rail operators proposing options and all partners arguing the consistency of the
proposals. Step 3 was the result of a mixed approach of applying the Europe terminal model, an extended version
of the LAMBIT model of VUB, and parallel spreadsheet calculations. Step 4 consisted of a comparison of intermodal
door-to-door costs with the costs of the reference mode (unimodal road or shortsea chains), on the basis of market
prices (train, terminals, pre- and post-haulage, reference modes) delivered by the rail operators. All costs were
recalculated to achieve a good understanding of market prices (train costs by using the actualised Rail Cost Model
of TU Delft). The comparison was carried out for different flow levels (sensitivity analysis) anticipating on the results
of the modal shift analysis (step 5). The bundling tool transforms a complete flow matrix into intermodal train and
unimodal truck services (hub-and-spoke, direct, otherwise), minimising system costs, and without introducing service
and network designs first.