Anders is teaching wind turbines to speak up before something goes wrong
Anders Malund Dammark Jensen has helped develop a new method for building digital twins of wind turbines, bridges and skyscrapers. The method can detect signs of wear and tear, even when they cannot be measured directly.
We go to the doctor when we get sick. We can do that because we’re human. A wind turbine can’t tell us when something is wrong, and a bridge can’t complain about fatigued steel members.
But Anders Malund Dammark Jensen and his colleagues at CP-SENS may have found a way to make large structures speak up when something is wrong.
“By creating a digital twin of a wind turbine, we can use data from sensors on the physical turbine to monitor its condition. Combined with computer models, this allows us to detect or estimate when the steel is experiencing fatigue or when something is at risk of failing.”
That’s according to Anders Malund Dammark Jensen, a PhD student at the Department of Civil and Architectural Engineering at Aarhus University.
Sensors allow him to monitor a wind turbine in real time and collect data on its condition. Combined with simulation models, the data can also be used to estimate things that cannot be measured directly.
The method can be applied to bridges, buildings and offshore wind turbines. And that could make a real difference.
Maintaining offshore wind turbines, for example, is both difficult and expensive.
You can’t simply send a technician out to inspect one. The job may require specialised vessels, suitable weather conditions and careful planning.
“If the digital twin detects that a particular component is developing problems, maintenance can be scheduled before something goes wrong,” he says.
“And even more interestingly, the model may be able to detect problems that cannot yet be seen or measured directly.”
Even onshore wind turbines, people are regularly sent out to inspect their condition. Manual inspections take considerable time and labour, whereas automated digital monitoring could make the process much more efficient.
“Clearly, we can save time and resources if we can make the process not just simpler, but also more accurate,” says Anders Malund Dammark Jensen.
Could extend the life of structures
Greater accuracy is useful not only for identifying worn or “sick” components. It could also help extend the service life of a structure.
A wind turbine’s expected service life is always based on conservative estimates, but better monitoring could potentially add valuable years.
“It may be possible for a turbine to remain in service for several more years if we have better, more accurate data on its condition,” says Anders Malund Dammark Jensen.
“And that would benefit both the climate and our use of resources if we could keep more wind turbines, bridges and other structures in service instead of demolishing them and replacing them with new ones.”
One method for different types of structures
The idea of wind turbines telling us when they are “sick” may sound a little like science fiction.
But several of Anders Malund Dammark Jensen’s experiments show that it is possible.
“I wanted to teach them to put their condition into words, so to speak. To do that, we had to combine algorithms and models capable of processing the data we collect from them,” he says.
The approach has worked. Digital twins are not a new technology in themselves, but what is new here is the way the algorithms are combined, allowing the same method to be applied to different types of structures.
“You could say we’re not inventing new LEGO bricks. We’re finding a new way to put them together,” he says, explaining further:
“We’ve developed a system of algorithms that can be used to build digital twins of wind turbines, bridges and buildings. In other words, it’s not just about creating a digital twin of one specific structure, but developing a general method.”
The key is that the algorithms are modular and can be adapted to different types of structures. This makes it possible to tailor the system to each case rather than developing a completely new solution from scratch every time.
Sensors can be used to monitor a wind turbine in real time and collect data on, for example, its movements and vibrations. Combined with simulation models, this data can help estimate conditions that cannot be measured directly.
“I use sensor measurements as input to an algorithm that estimates external loads, such as wind and waves. These estimates can then be used to assess things like fatigue in the steel,” explains Anders Malund Dammark Jensen.
Wind is one of several external loads acting on a wind turbine. For an offshore wind turbine, these include wind and waves, while a bridge is exposed to loads such as wind and traffic.
By combining sensor measurements with models, it is possible to build a more detailed picture of the loads a structure is exposed to and estimate conditions that cannot be measured directly.
Making knowledge freely available
When Anders Malund Dammark Jensen submits his PhD thesis on 31 October, his results and algorithms will not be limited to him and the university.
“We share our knowledge through software that is freely available as open source, so anyone can use it,” he says.
He has developed most of the software infrastructure needed to build a digital twin. Companies can then use it to create their own models based on data from specific structures.
“That could be data from a particular skyscraper. You can then create a digital twin of that building, giving you a wealth of useful information in real time,” he says.
The software will be freely available to everyone.
“Anders has spent three years developing tools that can speed up the implementation of digital twins using open-source software,” says Giuseppe Abbiati.
He is an associate professor at the Department of Civil and Architectural Engineering and supervises the PhD project.
“Anders’ work helps bring this emerging technology closer to practical application in the construction industry,” he says.
(copy 8)
Algorithms for digital twins
With growing global interest in technologies such as floating offshore wind turbines, the industry faces increasingly expensive and complex maintenance challenges.
One valuable tool for addressing these challenges is a digital twin that monitors a wind turbine in real time and assesses when maintenance may be needed.
A digital twin combines a physical system — such as a wind turbine — with one or more computer simulation models that exchange data with the physical system. Sensors installed on the turbine collect data which, together with the simulation models, can be used to estimate quantities that cannot be measured directly.
These may include fatigue in the steel or changes in the stiffness of the soil surrounding the foundation, caused by the varying loads acting on the turbine.
Already in use on the Great Belt Bridge
The Great Belt Bridge is one example of a structure for which digital twin technology is being used. The aim is to extend the bridge’s service life from 100 to 200 years. Drones, cameras and 15,000 sensors collect large amounts of data for the digital twin, providing an up-to-date picture of the bridge’s condition.
Algorithms can then identify areas of concrete damage that need repair, helping maintenance teams keep the bridge in service for many more years than originally anticipated.