Insights in wind turbine drive train dynamics gathered by validating advanced models on a newly developed 13.2 MW dynamically controlled test-rig

► Flexible multibody models can accurately capture gearbox dynamics. ► We show the importance bearing off diagonal stiffness terms. ► The importance of flexibility is highlighted. ► A dynamic test facility is developed. ► Experimental measurements update and verify the model. Guaranteeing reliable and cost-effective wind turbine drive trains requires expert insights in dynamics during operation. A combination of advanced modeling techniques and detailed measurements are suggested to realize this goal. The flexible multibody modeling technique enables the simulation of dynamic loads on all drive train components. Moreover it facilitates estimation of structural component deformation caused by dynamic loading. This paper gives a detailed overview of the assumptions made in this modeling approach. Furthermore the influence of the different structural component flexibilities is investigated in detail. To gain confidence in the models created, model validation by means of a comparison with measurements is necessary. To overcome issues concerning test repeatability experienced in field testing, test-rig testing is suggested as a valid alternative. In order to be representative, dedicated dynamic load cases, which represent specific dynamic behavior of the gearbox in a wind turbine need to be realized on the test-rig. However a highly dynamic test-rig complying with the specifications was not commercially available. Therefore Hansen developed a high dynamic test-rig with a nominal power of 13.2 MW and a peak power capacity of 16.8 MW. A back-to-back gearbox configuration was used. The complexity of controlling dynamics of the test-rig was solved by identifying dedicated load cases which represent specific wind turbine behavior. This paper describes the development process of the project consisting of four phases. During two phases a scaled set-up was used, which enabled iterative optimization of the complex interaction between the mechanical dynamics and the electrical controller of the test-rig. In the final part of the paper the two previously discussed approaches are combined, as it discusses results from the validation of simulation models using measurements performed on the 13.2 MW test-rig.