C4 - Robust model updating of digital twins considering uncertain input variables

The dynamic behaviour of a wind turbine (WT) in the range > 20 MW is characterised by various non-linear effects. In order to predict the dynamic behaviour already during the design phase, meaningful simulation models are required.

However, it can be assumed that simulation results regarding the turbine dynamics, which were calculated during the design, and measurements of the dynamics of the WT, which are carried out during later operation, differ from each other. Reasons for this discrepancy can be, for example, inaccuracies in the model (e.g. neglect of manufacturing tolerances), changes in the real structure over time (e.g. corrosion) or changes in the environmental conditions (e.g. scour).

In order to minimise these deviations, the numerical structure model can be subsequently adapted to the measured data with the help of model updating procedures. The goal of model updating can thus be, first, to improve the quality of the initial simulation model. Second, an adaptation to long-term changes of the structure and the boundary conditions over the lifetime is possible. And third, model updating can be used for damage localisation and quantification. Especially the last two points are needed to turn a classical simulation model into a digital twin.

In this subproject, model updating procedures are being researched with which these three goals can be realised. The methods will be validated using two different test structures: first, on a beam structure which is tested under laboratory conditions (Figure 1) and second, on a lattice tower structure placed in the open field and exposed to real environmental conditions (Figure 2).

Both structures have reversible damage mechanism. Thus, the structure can be damaged and repaired as desired and the model updating procedures can be tested and validated. In the final step, the procedures are applied to the digital twin of an offshore megastructure, where damage is introduced into the simulation model.

The scientific novelties of this subproject are that model updating procedures are investigated, which take into account measurement and model uncertainty. In addition, the procedures are optimised in terms of robustness and accuracy of the model updating. The research objective of the sub-project is thus to investigate robust model updating procedures for the digital twin of a future wind turbine based on global multi-objective optimisation.