Complex wake merging phenomena in large offshore wind farms

Despite recent advances in both measurement and modeling of wind and turbulence in wind farms, a gap in knowledge regarding wind power production in large offshore wind farms still remains. Data from Horns Rev wind farm shows that wind farm models used by industry to predict power production in large wind farms tend to underestimate losses due to wind turbine wake losses. Fully eddy resolving simulations are herein used as guidance for improving analytical wake models as applied to the Lillgrund wind farm. Comparisons between the eddy resolving model and the analytical model with SCADA observations from the wind farm for exact row wind direction and offset wind direction provide insight into mechanisms of wake merging. For wind aligned with a turbine row, estimates of power production from the analytical model, LES, and WAsP compare well with observations taken from the farm, but overall the LES predicts a deeper wake with slightly faster recovery than the analytical model. A second experiment with wind direction skewed relative to exact rows reveals the limitations of the analytical model. The LES performs better than the analytical model in handling complex wake situations such as unsteady partial waking and complex lateral wake merging. Physically motivated corrections based on internal boundary layer growth and merged-wake reduced velocity deficit replenishment are applied to the simplified wake model to attempt to account for the ‘deep array' power loss effect and partial waking. Additional refinements to the simplified parameterizations of growth of wake depth and wake width with downstream distance are introduced based on observations from a scanning lidar.