AI technology for optimal offshore wind farm performance

Data brings us knowledge and provides us with the opportunity to improve and develop. It is important, however, that we are able to gather and understand this data properly. What if the data stream is becoming too overwhelming? This is what is happening with the SCADA data coming from wind turbines in large-scale wind farms, says Pim Breukelman, Chief Commercial Officer of Jungle AI (Jungle).

Jungle is one of three nominees for the Offshore Wind Innovators Award 2021. As the company name suggests, the data from wind farms has become one big jungle, which makes it difficult to see what’s really going on. Nowadays, wind turbines are equipped with hundreds of sensors. For each sensor, static parameters are set which signal if there is a deviation from these parameters. However, this leads to a continuous flow of notification sounds via the SCADA-system. However, not all notifications are alarming. “This leads to so-called ‘alarm fatigue’ with operators, with the result that they don’t take action when in some cases they should,” explains Breukelman.

Dynamic alarming

Three young men, two Dutch and one Portuguese, who met at TU Delft, developed a software solution for this: Canopy. Using historical data from wind turbines, Jungle is able to identify patterns of normal operational behaviour under all conceivable circumstances. By applying Artificial Intelligence (AI), the normal behaviour of a wind farm can be learned within weeks. The actual behaviour is compared to the normal predictions, providing valuable insight into component health and turbine performance.

The result is that you now only receive a notification when a deviation from normal behaviour occurs,
so called dynamic alarm. Canopy clearly shows and ranks at which component the highest deviation or under performance was found, down to the level of the sensor(s) that detected the deviation. At that moment, it can be determined whether the deviation requires immediate intervention or whether it is something that could be taken into account during planned maintenance. By identifying the need for intervention early, you prevent unnecessary downtime at a later stage. Canopy can be used to detect component failure at a very early stage, but also, for example, undesired automatic power curtailment, icing or yaw misalignment. This leads to better performing wind farms that generally produce 1-2% more energy, thus bringing down the Levelized Cost of Energy.

Level playing field

Jungle mainly targets the wind farm owners and operators. They don’t always have access to the full picture or don’t have the tools to put it together, says Breukelman. Canopy enables them to keep a close eye on the performance of their wind farm and, where necessary, to exchange information with the party that performs the maintenance contract, usually the wind turbine supplier (OEM), on what steps to take. “This creates a level playing field between the wind farm operator and the OEM,” says Breukelman.

Asset agnostic

Jungle already has a number of customers in the wind industry using Canopy and is now looking to
scale up. However, Canopy is not only suitable for wind farms, says Breukelman. In principle, the software can be applied to any modern electromechanical equipment with sensors, as the software is asset
agnostic. Because the same algorithms can be used and no extra hardware is required, the software
can be applied quickly and is scalable as well.

Jungle is also planning to expand the deployment of Canopy to solar farms. “Solar farms use slightly different sensors, but the principles are the same. We are currently looking for solar farms to test Canopy”, says Breukelman.

Jungle and the other two nominees will pitch their innovation to the offshore wind sector during the hybrid TKI Wind op Zee LIVE event on 22 March, 2022.

Input your search keywords and press Enter.