A systematic investigation of common gradient based model updating approaches applied to high-fidelity test-data of a wind turbine rotor blade

Abstract

Within the context of the "SmartBlades2"-project, a wind turbine rotor blade was designed and extensively tested. The rotor-blade uses a lightweight composite structure and bend-twist coupling. The bend-twist coupling facilitates a passive load-reduction by changing the angle of attack under load. Due to a high sensor-density of 265 accelerometers in the experimental modal test of the blade, the sophisticated structural dynamics of the model are captured. Apart from the commonly measured first flapwise-, edgewise-bending and torsion mode, 35 modes up to a frequency of 60 Hz are identified. Unlike in many other wind turbine rotor blade investigations, the Finite Element (FE) model uses shell elements instead of beam elements and is directly based on production drawings. This experimental and simulative setup is particularly relevant, since a significant number of mode shapes exhibit a distinct local behavior which was in previous studies not accounted for. The differences between experimental and simulated results are minimized using computational model updating procedures. In this case-study, two formerly underrepresented aspects of the updating of large-scale FE models are examined. One is the use of different parameterizations and the other is the possibility of insufficient experimental data. The parametrizations are based on well-established criteria like error-localization and sensitivity. Moreover, the updating is performed with different (i.e. reduced) subsets of the modal data and the results are then compared to the model updating results achieved with the entire dataset. This in-depth investigation of the model updating of a composite structure allows the deduction of general guidelines in the model updating of industrial-sized FE models.