Application of a Single-Wake Model Accounting for Wind-Direction Variability Based on Wind-Field Measurements Around Full-Scale Coastal Wind Turbines

Accurate wake prediction under realistic atmospheric conditions is essential for wind-farm energy assessment. This study investigated the influence of wind-direction variability on single-turbine wake prediction using SCADA and vertical-LiDAR measurements from two full-scale coastal wind turbines. First, the effects of inflow conditions on wake-induced velocity deficits were examined for sea-wind and land-wind cases. Predictions obtained by applying the NOJ model to 10 min mean data were then compared with those based on 0.25 Hz line-of-sight data. Accounting for instantaneous wind-direction variations improved agreement with the measured wind-speed ratios, reducing the mean bias error from −0.075 to 0.011 and the root-mean-square error from 0.156 to 0.062. Based on these findings, a rotor-averaging method was developed using the expected overlap between the wake region and the downstream rotor, calculated from the probability distributions of wind direction and its standard deviation. The proposed method reduced the peak overlap near the wake center and extended the wake influence over a wider directional range. Its influence was limited in the near-wake region but increased with downstream distance. Comparison with full-scale measurements showed that the proposed approach improved power-ratio predictions, achieving the lowest mean bias error and root-mean-square error of 0.026 and 0.047, respectively, under land-wind conditions. These results demonstrate that probabilistic rotor overlap provides a simple method for incorporating wind-direction variability into engineering wake models.