Global sensitivity analysis of a semi-submersible floating wind turbine using a neural network fitting method

The stochastic platform motions of a floating offshore wind turbine (FOWT) significantly affect its load characteristics and power production. It is vital to understand the influence of various input parameters on the global performance of an FOWT to ensure structural safety and power efficiency. This study investigates the influence of key design parameters related to metocean conditions and natural periods of platform motions on wind turbine performance. A global sensitivity analysis (GSA) framework is proposed to map the input-output relation of a 4-MW semi-submersible FOWT. This framework uses fully-coupled numerical analysis and neural network fitting (NNF). Several important parameters are selected, including natural periods of the platform motions, mean wind speed, turbulence intensity, wave spectral peak period and significant wave height. Results demonstrate that natural periods of platform motions and wave conditions have a significant influence on the dynamic responses of the FOWT. Furthermore, statistical analysis reveals that severe sea states and low natural periods of platform motions can increase the operation cycles of the blade pitch system by 20%. The use of NNF method effectively increases the efficiency of the GSA by at least forty times. The proposed framework can be applied to other areas of FOWTs.