Assessment of wind turbine structural integrity using response surface methodology

• A new approach to assessment of site specific wind turbine loads is proposed.
• The approach can be applied in both fatigue and ultimate limit state.
• Two different response surface methodologies have been investigated.
• The model uncertainty introduced by the response surfaces is determined to be small.
• The statistical uncertainty introduced by a limited number of seeds is determined.

In site suitability assessment of wind turbines, it is often the case that one or several wind climate parameters exceed the reference values for the wind turbine design class. In such cases, the IEC 61400-1 requires a load calculation based on the site specific wind climate conditions in order to document the structural integrity. This load calculation demands a significant number of aero-elastic simulations which are time consuming to perform and require expert knowledge. In this paper it is investigated, to which extent the site specific loads can be determined based on a response surface methodology (RSM).Two response surfaces are presented, formulated based on Taylor approximation and Central Composite Design. For each RSM, the model uncertainty is estimated for different combinations of the wind climate parameters, along with the statistical uncertainty introduced by a limited number of random seeds. The results show that fatigue loads during power production, in general, can be assessed accurately using both RSMs. However, central composite design introduces the smallest model uncertainty. For ultimate loads resulting from extreme turbulent inflow, a larger model uncertainty is introduced. Central composite design leads, again, to the lowest model uncertainty. The statistical uncertainty related to the number of aero-elastic simulations is modelled for each RSM using bootstrapping. In general, the statistical uncertainty related to the number of random seeds is larger than the model uncertainty related to the RSMs.