Skip to main content

Digital Twin – the road to autonomy

Most people encounter - perhaps without being aware of it - a digital twin when they see Magnus Carlsen play chess on Norwegian TV. But digital twins are also making their way into the industries, and their use is far more advanced than chess games.

On Norwegian TV, next to the livestream of the chess players, we see a computer that analyzes the game and gives the viewer an assessment of current positions and suggests the best possible next move. But what is a digital twin? 

(This article was first published in Analyse by Dagens Næringsliv)

- I want to start with our definition, says Chief Technology Officer Christian Møller in Kongsberg Digital. – One of the things we do is to collect real-time data from sensors on an asset. That gives us information about the asset’s condition, which would be the same as collecting information about the position of the pieces in a chess game. But that doesn’t really tell us anything about what should happen next. We need to know the rules to act. In chess, those are defined and quite simple. In the real world, the rules are substantially more complex but can, at least in part, be described by physical laws or data-driven models - or ideally, both physical models and data-driven models that describe behavioral rules.

If you have real-time data (where the chess pieces are positioned) and the rules (for chess), you can ask your computer what to do, or more specifically search, optimize and predict.

- We don't play chess. We work on a completely different scale, for instance with oil and gas operations, Møller points out.

Before starting to develop an oil field, the whole process is tested in a simulator to see if the plant will work as intended. That means we understand how the plant behaves.

- We have a simulator that can describe the future if we know what the conditions are. If you have 40 switches you can turn, the simulator can assess what the operator should do to control the production. The computer can explore combinations and respond with the optimal solution for the challenge presented by the operator.

In the long run, this process gradually becomes more autonomous. The digital twins can make more and more decisions on their own and may just send an SMS to the operator about the changes.

- You can't do this just by retrieving data or just by using a 3D model. Intelligence is required to understand the outcome of what the machine is doing, Møller explains. - The digital twins can use artificial intelligence alongside other conventional methods and understand the outcome of a set of actions.

The digital twins can use artificial intelligence alongside other conventional methods and understand the outcome of a set of actions.
Christian Møller, Chief Technology Officer, Kongsberg Digital

Kongsberg Digital also works with digital twins in the maritime sector, wind power and the utility industry in general.

- We are engaged in Yara Birkeland, the autonomous ship that will be operated from shore. An operator in the onshore control room needs to know ​​what the ship is going to do when unplanned events occur, and that is where a digital twin comes in. Let's say Yara Birkeland meets two ships. Input tells the digital twin the directions and speed of the ships. The digital twin analyzes the situation and can tell the operator what the ships will do in the next five minutes. Then the operator can rest assured that the systems in the ship make the right decision. In the future this technology can be scaled so that an operator can operate a fleet of autonomous ships. A digital twin will then become an integral part of the entire system.

Today, Kongsberg Digital also provides digitalization solutions to wind farms to make renewable energy production more cost efficient.  Renewable energy sources are growing in importance, but the inflexible and decentralized nature of renewable energy production, combined with increasing use, can put a new demand on power grids.

Kongsberg Digital elaborates: Moving from today’s situation where power generation happens at a few locations to a situation where power will be consumed and produced (the so called prosumer) scattered across nations, the demand on existing grid infrastructure will be challenged. The supply and demand situation will be more unpredictable and coupled with advancement in batteries for storage and electric cars, we will need a more intelligent systems to assist in operating the grid. This is the purpose of the digital twin. It can use real time data, historical data, physics-based models, machine learning models etc to optimize operations of the grid. We can even imagine twins in different nations connected and collaborating to the optimize power distribution on a continental scale.

- We want to be able to look at the total netload in connection with production and control the electricity grid efficiently. That requires flexibility for the times electric cars are being charged or Christmas dinner is cooked, but also batteries for storage of renewable energy. A digital twin will help organize and coordinate the power flow. In the future, a digital twin can analyse the wind conditions in Germany and suggest how the power producers in Telemark should optimize their production. One day digital twins might optimize and coordinate power production in Europe via the power markets.