Data assimilation of realistic boundary-layer flows for wind-turbine applications – An LES study

Providing observed date- and site-specific turbulent inflow fields for Large-eddy simulations (LES) of the flow through wind turbines becomes more and more important for realistic estimates of power production. In this study, data assimilation techniques are used to adapt the atmospheric inflow field towards measurement data. A Newtonian relaxation technique and a vibration assimilation method are implemented in the geophysical flow solver EULAG. Their capability of adapting mean wind profiles towards field measurements while maintaining the atmospheric turbulence of an idealized LES is investigated. The sensitivity of the methods to grid refinement and to parameter changes is analysed. The performance of the vibration assimilation technique is better suited for fine grids (dx=dy=dz=5 m) because of smaller damping effects on the atmospheric turbulence. Furthermore, the vibration method is used to nudge the inflow field of an idealized atmospheric simulation towards velocity profiles measured at the wind-farm site WiValdi at Krummendeich. A near neutral stratification is chosen from the measurements to test the assimilation technique. With the vibration assimilation method it is possible to adapt the zonal and meridional velocity components of an atmospheric flow. The LESs applying data assimilation are compared with the measurements and independent mesoscale simulations. A good accordance is found for the mean inflow velocity profiles and the turbulence intensities. In a final step, the assimilated flow field is taken as inflow for a wind-turbine simulation. The windturbine simulation shows characteristic structures of a wake in the atmospheric boundary layer. This study demonstrates that an efficient computing of different and realistic inflow fields for wind-turbine simulations is possible applying the vibration assimilation method.