We buy far more goods from China than we sell to the Chinese. When a container ship has sailed goods from Shanghai to Aarhus in Denmark, and the containers have been unloaded and emptied, it can be difficult to refill all the containers—there simply are not enough goods to be shipped in the opposite direction.
The picture is the same throughout the EU and the USA, which has a massive trade deficit with the Far East. For shipping companies such as Maersk, the imbalance in world trade is a major challenge because it means that vessels have to sail with empty containers. This costs billions of dollars and puts a strain on the environment.
But the problem can be minimized if Maersk becomes better at predicting where in the world the empty containers will be needed. It is a question of getting them to wherever a new demand is emerging and moving them around as little as possible.
This is where Pierre Pinson is lending a helping hand. He is professor of operations research at DTU Management, and in collaboration with Maersk, he is in the process of developing mathematical models for the relocation of empty containers.
“When containers are used to carry goods from China, you could just sail the empty ones back, but it’s not quite that simple because the need for containers varies dynamically throughout the year and is also influenced by geopolitical decisions and numerous other factors,” says Pierre Pinson, who continues:
“Maersk collects large amounts of data, which can be used to predict where and when containers will be needed. If the company can become better at these predictions, huge sums can be saved.”
According to Maersk, in 2020 the company moved 12.6 million 40 foot containers at sea, but the company cannot say how many of the containers were empty.
Solving a global puzzle
"We’re expanding our competences and upgrading our academic understanding of applied mathematics because there’s a large talent pool at DTU—we get fantastic feedback from very talented people."
Klaus Holst, Senior Researcher in Maersk’s Research and Development Department
Pierre Pinson is developing new mathematical methods to turn the large amounts of data into valuable knowledge that can be used as a basis for decision-making.
This is done through operations research, which is a mathematical approach to optimizing large and complex systems, and one that is often used in production and route planning, in finance, and in decision-making. Operations research combines mathematical modelling, computer science, stochastic methods, and numerical analysis.
Developing, testing, and refining models that are at the forefront of development and which can predict the need for empty containers in different places on the planet takes a lot of time because it is a giant puzzle that has to be solved. Demand has to be predicted for many hundreds of locations several weeks into the future—Maersk typically looks between one and 13 weeks ahead. Moreover, not all containers are created equal, and it is no use sending an empty 40 foot dry cargo container to a destination where the future need is a 20 foot refrigerated container that can be used for shipping food.
The forecast must be based on historical data and the expert knowledge available to employees around the world. The puzzle is dynamic because the world economy is constantly changing, and the models must also be able to be fed with the changes Maersk is making internally. For example, if the company increases or reduces capacity on certain routes, the models must take this into account.
In addition to being able to continuously adjust for sudden changes in global trade, models must be able to quickly find a new equilibrium when unpredictable events such as a sudden pandemic put pressure on the system. So there is plenty to address, and Pierre Pinson is delighted to have access to the many data from Maersk:
“Data are essential for a university researcher like me who wants to conduct best science. Data are like raw diamonds fresh from the mine; they need to be sanded and polished to be turned into something beautiful.”
Pinson elaborates:
“In the logistics field, Maersk is one of the major players with access to the best data. Maersk has the raw material I need as a data science researcher.”
Industrial PhD provides link
Even the world’s biggest container shipping company needs the expertise available at the universities, and Klaus Holst, a senior researcher in Maersk’s research and development department, is enthusiastic about the collaboration:
“We’re expanding our competences and upgrading our academic understanding of applied mathematics because there’s a large talent pool at DTU—we get fantastic feedback from very talented people—also from among the students, some of whom can become future employees. So we try to find exciting problems we can solve together.”
In this case, the link between Maersk and DTU consists of industrial PhD Benedikt Sommer, who has been working on the project with the empty containers since January 2020. As an industrial PhD, Benedikt Sommer is employed by Maersk and at the same time enrolled at DTU with Pierre Pinson as supervisor. This is an arrangement that everyone involved benefits from—not least Klaus Holst:
“It’s a fantastic arrangement for us and the industrial PhD student, who can go on to pursue an academic career or continue in business with a head start over ordinary PhD students,” he says.
“Now we’re anchored to a three-year project, which gives a sense of stability. It gives us a longer focus than we would typically have with such projects at Maersk, and we work in a different way—we conduct research with a different cadence, which offers new opportunities,” says Klaus Holst.