Wind turbine blade large deflections: A non-intrusive method for blade nonlinear reduced order models

The increasing length of the wind turbine blades has brought new challenges for the wind turbine load analysis, such as lack of an effective method to model and simulate large blade deflections. The effectiveness of a model can be measured by its accuracy to capture the blade deflections and the computational resources it requires. Although the existing geometrically nonlinear solvers are very accurate, their computational costs are still beyond the desired level for the wind turbine load analysis

This thesis aims to develop an effective method for generating accurate geometrically nonlinear blade models, which are computationally cheaper than the existing models used in the turbine load analysis. The research focuses on the blade kinematics, nonlinear blade modelling effects on the turbine load simulations, and nonlinear model reduction methods to obtain accurate and computationally cheap blade models.

The presented turbine models show that the linear models cannot accurately estimate the large blade deflections occurred in normal operating conditions. The linear models cause artificial elongation of the blade length, inaccurate deflection and load estimations, especially in torsion direction when compared to the nonlinear blade models. The linear relation between deflections and force response results in a constant coupling between different blade directions, which are a function of deflections in nonlinear models. Some couplings causes significant torsional deflection for very long and flexible blades, and thus it is essential to capture coupling effects accurately in wind turbine load simulations. Therefore in this thesis, a non-intrusive reduction method is proposed to generate reduced order blade models, which use nonlinear relations between force response and deflections. Non-intrusive methods need a geometrically nonlinear structural solver to compute deflections for the selected static load cases. The relations between force response and deflections are calculated by using the static load results. The selected load cases should result in realistic deflections that the structure goes through in its operational life. The proposed method uses a reduction basis, including bending mode shape and modal derivative vectors, which allow capturing nonlinear effects such as torsional deflection due to the coupling between flapwise and edgewise motions.

The effectiveness of the reduced order models generated by the proposed method is evaluated for some test cases, including three turbine blades. The results show that the reduced order models can estimate the deflections accurately compared to the original high fidelity nonlinear models, and they require much less computational time than the original models require.