Validation of FAST.Farm considering structural loads at alpha ventus
| Main Authors: | Kretschmer, Matthias, Jonkman, Jason, Cheng, Po Wen |
|---|---|
| Format: | info Proceeding |
| Terbitan: |
, 2019
|
| Online Access: |
https://zenodo.org/record/3373361 |
Daftar Isi:
- In past years, wind farm control has gained interests significantly by both academia and industry to reduce levelized cost of electricity of wind energy. The primary goal of wind farm control is nowadays power output maximization but the importance of other control objectives such as grid stability and reduction of structural loads will presumably rise in the future. Since wind farms interact with the atmospheric boundary layer, designing new wind farm controllers is a challenging task and is highly dependent on adequate simulation tools. On one side, it is important to have a computationally cheap simulation environment to be able to perform many runs for controller design and testing; on the other side, good accuracy in predicting the system dynamics is required especially when considering structural loads of wind turbines. The recently developed tool FAST.Farm [1] fulfils these criteria in general and is therefore in scope of this study for validation against measurement data. FAST.Farm is an extension of the widely used tool FAST (now OpenFAST) aero-hydro-servo-elastic tool with the added capability of simulating wind farm effects. The wind farm modelling is based on the Dynamic Wake Meandering (DWM) model and incorporates further developments, which are for example: Allowing wakes to merge, meandering of wakes both laterally and axially, and including a wind farm controller. In previous work, FAST.Farm was calibrated and verified against high fidelity simulations (Large Eddy Simulations using SOWFA and OpenFOAM) [2]. For this validation study, measurement data acquired at the alpha ventus wind farm in the North Sea was used. Hereby, the focus is on turbines AV4 and AV5 which are Senvion 5M turbines having a rated power of 5 MW. The turbines are equipped with strain gauges at multiple tower positions as well as the blade root. Meteorological conditions were derived from the Fino 1 met mast. In a first step, environmental conditions of the measurements were filtered by using the procedure presented in [3]. Based on that, wind fields were created using an LES precursor in SOWFA, as well as with TurbSim to increase the amount of random seeds, thus enabling a better statistical convergence of the simulations. In a next step, the simulation models of turbines AV4 and AV5 were adjusted individually to account for asymmetries of the real wind turbine (e.g. pitch errors, mass imbalances). The validation was focused on single wake effects where the wake of turbine AV4 hits the turbine AV5. Because FAST.Farm runs in real time, many simulations are performed with varying wind directions, wind speeds, atmospheric stabilities and different yaw misalignments. This was necessary to match possible scenarios in the measurement campaign and thus increase the statistical variability. The comparison of FAST.Farm with measurement data of modern utility-scale wind turbines shows promising results for calculating power losses and structural loads of turbines in wake. The computational efficiency of FAST.Farm allows running many cases, which is required for wind farm controller design and tests. [1] Jonkman et al., 35th Wind Energy Symposium, p. 454 (2017) [2] Jonkman et al., Journal of Physics: Conference Series 1037, 062005 (2018) [3] Kretschmer et al., Journal of Physics: Conference Series 1037, 052009 (2018)