Code-to-code comparison of realistic wind turbine instability phenomena

Kurzfassung

Preceding studies (Hach et al, 2020) show significant differences in the prediction of dynamic aeroelastic instabilities with state-of-the-art simulation tools. Instabilities were provoked by an overspeed scenario where the wind speed increases, but no counteracting generator moment was imposed. Significant differences between the tools were found in the critical speed and instability behavior. A negative side effect of the runaway procedure is that the operating conditions at which the instability occurred could be vastly different. Differences in the instability mechanisms could therefore be in part allocated to these differing operating conditions. The aim of this study is to find a critical configuration model which becomes unstable under nominal, controlled operating conditions. The expectation was that the establishing instability mechanism would be more representative for potential aeroelastic phenomena which could be experienced on current or future turbines (Volk et al, 2020). State-of-the-art simulation models are used to give insight in the instability phenomena and to compare their respective modeling capabilities. These included two general purpose multi-body simulation tools (alaska/Wind and Simpack) and three industry relevant turbine simulation tools (Bladed, HAWC2, OpenFAST). Simulations were executed both in the time domain (all tools) and in the frequency domain (Bladed and HAWCStab2). The publicly available reference wind turbine model IWT-7.5-164 served as reference for the comparison. The global flapwise, edgewise and torsional stiffnesses were reduced over the full blade to enforce instabilities under nominal operating conditions. A recomputation of the stiffness matrices was performed with the adjusted input in order to assure equivalent stiffness reductions across all tools. Consistency across the models was verified by a blade and full model modal analysis, static structural deformation tests and a steady aeroelastic deformation test. The final stability analysis was performed for multiple points along the nominal control curve. Unstable time domain simulations were analyzed by a frequency analysis to determine the instability mechanism. This also allowed a comparison between the time domain and the frequency domain simulation tools. A comparison is shown for the aeroelastic modes which play a role in the instability mechanisms. The agreement between the linearization results of Bladed and HAWCStab2 is satisfying. All simulation tools exhibit comparable unstable behavior with a significant participation of the 1st and/or 2nd edge bending modes of the blades. Large differences are observed between critical wind speeds in the time domain simulation results. Preliminary analyses reveal discrepancies in the damping associated with the vibrations. The presented comparison demonstrates that the observed instability mechanism matches across the aeroelastic simulation tools if the operating conditions are close enough - which is not always the case in a runaway analysis. To provoke an instability in the normal operating range the stiffness of the rotor blades was decreased significantly. It was found that the stiffness matrices needed to be re-computed since independently reducing the principle stiffnesses in flap/edge/torsion leads to inconsistent models.