AI simulation corrections for lifetime extension of wind turbines – KIWi

Funding: BMBF
Total funding:  873,000 EUR
Partners: Fraunhofer IWES (applicant), Saarland University (network coordination), University of Bremen
Associated Partners: P.E. Concepts GmbH, marpitec GmbH
Term: 11/2022 – 10/2025


  • Wind turbines typically reach the end of their certified service lifetime after 20 years. However, many are still capable of producing green electricity. 
  • For extending the operating permit, an estimation of the remaining lifetime is required, which is based, among other things, on numerical simulation models. However, these models are inaccurate.
  • In the KIWi project, the project partners are incorporating a correction based on machine learning into the simulation procedure, thus enabling a more accurate determination of the remaining lifetime. 
  • In a subproject, Fraunhofer IWES is responsible for implementing the correction algorithm in the institute's own simulation environment.

Wind turbines usually have - based on the design lifetime - an operating permit for 20 years. However, many of them are fully capable of generating electricity beyond this period.

A separate operating permit is required for continued operation. Operators must provide evidence of the actual condition of the wind turbine, showing the predicted remaining lifetime. Among other things, comparison of the design site conditions with the real site conditions of the wind turbine is required. For this purpose, a numerical model and hundreds of simulations are used to estimate the fatigue loads. However, these models are generic and subject to errors, such as model simplifications or wind conditions that are difficult to reconstruct. The results of the simulations are therefore subject to large uncertainties in the load estimation, which means that the lifetime predictions do not reflect the actual potential of the plant.

This is where the KIWi research project comes in, drawing on new approaches such as artificial intelligence and machine learning. The aim is to integrate a machine learning-based correction and more accurate input parameter estimation directly into the simulation procedure for the model-based determination of the remaining service life. The simulation correction compensates for model errors and approximation errors of the simulation methods and thus enables a more precise determination of the possible lifetime extension of wind turbines.

In a subproject, Fraunhofer IWES is responsible for implementing the correction algorithm developed by the other project partners in the institute's own simulation environment and for validating it. In addition, the IWES research wind turbine Adwen AD-8 serves as a development platform and demonstrator. 


Press release: Für eine schnellere Energiewende: Mit KI werden Windenergieanlagen nachhaltiger (German only)

Logo Federal Ministry of Education and Research