In progress Analysis of the e ect of the Ultra Large Container Ship on container terminals



In this PhD project, we aim to analyze the e ects of the Ultra Large Container
Ship (ULCS) on ports. Although there are a handful of publications
that have already attempted this, they are outdated and use oversimpli ed
statistical models. We believe that operations research models are more capable
of tackling this problem, as the relationships between the input are
explicitly modeled.
We propose to analyze the e ects of ULCSs by considering both land
and sea perspectives. For the sea perspective we investigate how liner shipping
companies could eciently utilize ULCSs in their service network. For
the land perspective we investigate how the port call size is a ected, and
also what the consequences are for terminal operations and the hinterland
transportation network.
An important aspect of this project is to apply research in practice. Not
only does this mean that we work closely together with practitioners, but
also that this research is able to provide insight to the actors in the maritime
supply chain.


Container terminals, containers, shipping, large ships, logistics, optimization, hinterland transportation

Time frame

2016 - 2020


Container liner companies have their ships sail fixed routes with a predefined time scheme. Furthermore, to facilitate the shippers they often apply a weekly call in each port, which means that several ships are needed on a route.

Connecting the routes with transhipment yields a network which connects many ports with each other. Finally, adding feeder services from main ports to smaller ports completes the network.

Designing such a network is a strategic problem, in which trade lanes and ship sizes have to be decided upon. To date, almost all liner companies operate long routes with many calls, as transhipment costs are typically high. In the last decade network optimisation models have been defined by researchers, allowing liner companies to determine an optimal network in terms of maximizing a profit function of shipping containers given a demand between a number of ports and costs due to the ship, fuel and handling.

In the last decade liner companies have also been competing vigorously, by driving down the cost of shipping one container by increasing the sizes of their ships. This strategy requires huge investments and smaller companies have had problems in keeping up with this race. At the moment ships have been ordered with a size larger than 20.000 TEU which represents a doubling of size in 10 years. The increase of ship size has been in all dimensions, length, breadth and draft. This means that terminals need to have cranes long enough to cover the full breadth of a ship and ports need to ensure that their entrance has enough depth. The latter is difficult for ports like Hamburg as it has a long and not that deep entrance canal. Rotterdam was receiving already larger ore ships, so it did not have a problem in this respect.

The increase in ship size went hand in hand with a substantial increase in the number of containers shipped, both from Asia to Europe as well as the Americas. This increase in international trade was created by a high growth of Asian and especially Chinese exports to the continents.

Although ship sizes increase, it is the question whether the call size (the number of containers loaded and unloaded in one port) will increase as well. It is likely that larger ships will call less ports, as the costs in the port are higher for large ships than for small ones.

We therefore pose our first two research questions:

  • Will the increase in ship size continue, stabilize or be reversed?
  • Will the call size increase in certain ports?

We will study these questions by inserting these large ships as options in the network optimisation models (available from Mulder and Dekker (2014)) and investigate whether and when liner companies will employ these ships and also whether the call sizes will increase. We will use standard data collected in the LinerLIb created by Brouer et al. (2015), together with own assessments as well as assessments of the companies involved in the research.

These questions will require extensive experimentation and a scaling up of the models as the whole world needs to be covered, implying large models. Moreover, our own results indicate that results obtained so far by researchers for large scale networks are not optimal (thesis Mulder (2015)). Hence the optimisation methods need to be improved.

The next two steps will consider the effect on the terminal and hinterland.

In the terminal larger call sizes may or may not increase productivity because cranes can work longer on each bay. On the other hand, cranes may need to travel longer and deeper in the ship. Moreover, if liner companies would use split calls as standard for the Port of Rotterdam (i.e. first unloading in Rotterdam, then sailing to Hamburg, next returning to Rotterdam for loading), the terminals may have no longer synergy from combined loading and unloading operations. Furthermore, a larger call would also mean a larger peak load for the stack. We will analyse these aspects with analytical and simulation models for the terminal operations. We have such models ourselves, but they are also available with the terminals. Yet it is envisaged that the models need to be adapted and more data needs to be gathered to obtain the required results.

The larger call sizes would imply larger peak loads for the terminal and hence also larger peaks for the hinterland transport. The latter would like to see a more steady operation (e.g. it is much more difficult to have 6 trains each on two days, than 2 trains each during 6 days). Using hinterland network design and operation models we will study this effect, by defining and analysing several scenarios. We will also investigate whether mitigating measures can be taken.

The research will be carried out within the LIS program of ERIM. One of the themes is terminal and transport optimisation. The problem has been brought forward by container terminals and the port of Rotterdam. As there are four main problem topics we will also employ econometric MSc students writing their thesis on one problem topic in order to speed up the research. This will allow the PhD student to make a head start on these topics.

Supervisory Team

Rommert Dekker
Professor of Econometrics (Operations Research and Informatics)
  • Promotor
René de Koster
Professor of Logistics and Operations Management
  • Promotor