DNVGL.no

How Jotun uses artificial intelligence to deliver paint to ships around the world

SHARE:
PRINT:
Jorun Drydock
Using machine learning and huge amounts of data, it is now easier for Jotun to plan how much paint they will deliver to which destinations.

When a ship arrives to port, it often brings goods and services. Everything from refuelling, food for the crew, garbage disposal – and regular painting of the hull.

"It's not just the paint itself that is our product, but also how the paint arrives in the right place at the right time," says Geir Axel Oftedahl, Director of Business Development at the Norwegian paint company Jotun.

For a long time, ships have reported their real-time positions using so-called transponders. Then, those who want to know can find out where the ships are at any time, and when they are expected at their destined port.

“Everyone can anticipate that. A more demanding question is which port the ship will be in if you see four or five voyages ahead," says Oftedahl.

In collaboration with DNV GL, the Norwegian paint group has conducted a pilot project to predict the destinations of the various ships – not the day before, but weeks ahead.

Doctoral degree in spatial statistics: “We see that the algorithm works”

On the team that developed the solution was DNV GL's David Volent Lindberg (32), a civil engineer in Industrial Mathematics from NTNU and with a doctorate in spatial statistics from the same university.

"The paint job is the same, but the distribution of paint is improved," he explains.

By processing huge amounts of historical data from previous voyages, one of Lindberg's colleagues, Thomas Mestl, developed an algorithm that estimates when the ships Jotun will deliver paint to, arrive at the various ports around the world.

“Based on historical data, the algorithm predicts where the ships will dock at what time. It took a long time to develop the algorithm, but we see that it works,” he says.

“How well do you succeed with the predictions?”

“We hit some ships, but not all. The solution is not supposed to give a solid prediction," says Lindberg.

According to the statistician, it must first and foremost indicate probabilities.

“For example, it’s an 85 percent probability that a particular ship will arrive at a specific port in Singapore on Tuesday in three weeks. If you know that, you know a lot. And as the day approaches, the more accurate will be the prediction.”

“Proves that even paint can be digitized”

Through machine learning, the solution makes it possible for Jotun to optimize its stock and delivery plans – in practice making it easier to determine how much paint should be at which warehouse, and when the paint should be distributed to the vessels arriving at a port.

"The solution proves that even paint can be digitized," says Jotun’s Oftedahl.

"We must exploit the opportunities provided by the technology. At the same time, this solution will benefit many providers of goods and services to the shipping industry.”

Assisting customers in the digital race: “Rewarding assignments”

The solution is part of DNV GL's digital industrial platform “Veracity”. Lindberg believes that many shipping companies' service providers will benefit from the solution, including DNV GL itself.

"The one who manages to handle data most efficiently is the one who wins. As a technology company, it is important for us to create tools that can help customers utilize new technology," he says.

Lindberg has worked in the international quality assurance and risk management company for over three years, and thinks the task of developing the solution is rewarding.

“Absolutely. This is pioneering work. By utilizing existing technology, we provide a whole industry with increased sustainability through increased productivity. That is meaningful,” Lindberg concludes.

David Volent Lindberg
DNV GL's David Volent Lindberg
Jotun veracity video thumb
Watch the Jotun case video