Global sensitivity analysis of model uncertainty in aeroelastic wind turbine models

A framework is presented for performing global sensitivity analysis of model parameters associated with the Blade Element Momentum (BEM) models. Sobol indices based on adaptive sparse polynomial expansions are used as a measure of global sensitivities. The sensitivity analysis workflow is developed using the uncertainty quantification toolbox UQLab that is integrated with TNO's Aero-Module aeroelastic code. Uncertainties in chord, twist, and lift- and drag-coefficients have been parametrized through the use of NURBS curves. Sensitivity studies are performed on the NM80 wind turbine model from the DanAero project, for a case with 19 uncertainties in both model and geometry. The combination of parametrization and sparse adaptive polynomial chaos yields a new efficient framework for global sensitivity analysis of aeroelastic wind turbine models, paving the way to effective model calibration