Transferability of Meta-Model Configurations for Different Wind Turbine Types

Abstract

Abstract The use of meta-models of aero-elastic simulation models is increasing in wind energy. Here, the question arises whether knowledge about meta-modelling, i.e. meta-model configurations, that has been gained for one wind turbine type can also be transferred to other wind turbine types. Here, knowledge of meta-modelling or meta-model configurations means the choice of meta-model settings, e.g. the underlying functions, the choice of the sampling method as well as the required amount of training data. In this work, the transferability of the Kriging meta-model configurations is analysed for two different offshore wind turbine types, an offshore wind turbine with a monopile foundation and an offshore wind turbine with a jacket foundation. First, the Kriging settings are analysed regarding their transferability. After this, a convergence study is conducted to answer the question whether knowledge regarding the required amount of training data and the selected sampling method are transferable. It turns out that it is possible to use the findings regarding the meta-model settings from the monopile for the jacket as well. The results of both offshore wind turbines are also comparable with regard to the chosen sampling method and the convergence behaviour of the meta-models regarding the required amount of training data. Thus, a transferability of meta-model configurations is generally feasible for the investigated offshore wind turbines and the choice for a specific meta-model is quite robust.