A Critical Evaluation of Structural Analysis Tools Used for the Design of Large Composite Wind Turbine Rotor Blades Under Ultimate and Cycle Loading

Rotor blades for 10-20MW wind turbines may exceed 120m. To meet the demanding requirements of the blade design, structural analysis tools have been developed individually and combined with commercial available ones by blade designers. Due to the various available codes, understanding and estimating the uncertainty introduced in the design calculations by using these tools is needed to allow assessment of the effectiveness of any future design modification. For quantifying the introduced uncertainty a reference base was established within INNWIND.EU in which the several structural analysis concepts are evaluated. This paper shows the major findings of the comparative work performed by six organizations (universities and research institutes) participating in the benchmark exercise. The case concerns a 90m Glass/Epoxy blade of a horizontal axis 10MW wind turbine. The detailed blade geometry, the material properties of the constitutive layers and the aero-elastic loads formed the base by which global and local blade stiffness and strength are evaluated and compared. Static, modal, buckling and fatigue analysis of the blade were performed by each partner using their own tools; fully in-house developed or combined with commercially available ones, with its specific structural analysis approach (thin wall theory and finite element models using beam, shell or solid elements) and their preferable analysis type (linear or geometrical non-linear). Along with sectional mass and stiffness properties, the outcome is compared in terms of displacements, stresses, strains and failure indices at the ply level of the blade structure, eigen-frequencies and eigen-modes, critical buckling loads and Palmgren-Miner damage indices due to cycle loading. Results indicate that differences between estimations range from 0.5% to even 40%, depending on the property compared. Modelling details, e.g. load application on the numerical models and assumptions, e.g. type of analysis, lead to these differences. The paper covers these subjects, presenting the modelling uncertainty derived