Posters

Abstract

Conventional independent wind resource estimates rely on using measured data to inform energy yield. Where these measurements have been conducted in the vicinity of operating wind farms. Neglecting these effects will lead to double accounting when conducing a wind farm simulation of wakes and blockage. Accounting for these effects will be referred to as “dewaking” in this study. As more wind farms are constructed in the North Sea the need to dewake measurements is increasing and understanding of the dewaking process should be improved. The aim of the study is to compare two dewaking methodologies and provide recommendations where the methodology can be improved. Two models will be used; firstly, using the TurbOPark deficit model coupled with the Forsting induction model as implemented in conventional software and another using the NWP weather prediction model (as detailed in “A simple axial induction modification to WRF’s Fitch wind farm parameterisation”, Vollmer et al., Wind Energy Sci., 9, 1689–1693, 2024.) For each model, two time-based simulations were executed with and without turbine interaction effects. From these simulations at each time step a dewaking factor was derived. It is this dewaking factor that is then applied to measured data to dewake before calculating a wind farm gross power. The study will focus on a two-year period (2023-2024) that coincides with public measurement data in the German Bight from the N9 and N10 lidars. The results highlight some important findings, an upstream slowdown, a speed up tangentially to the wind farms modelled and a strong seasonality to when dewaking should be applied. Although the Forsting induction model is contributing to a slowdown before the wind farms, at a distance of ~10km these effects are negligible. In comparison the NWP is estimating a significant global blockage effect that is larger than the combined individual turbine induction effects modelled by Forsting. For wind directions (330-030) where the measurements were fully upstream of the wind farms at distances between 10 and 30km, the engineering models were not predicting any deficit whereas the NWP was predicting an average correction of 1% - 2%. When the measurements were fully shaded behind the operating wind farms, the NWP was predicting a larger dewaking correction factor, which was expected. However, for wind directions where the measurements were between a gap in the wind farm cluster, this was reversed with the NWP predicting lower correction factors. This was due to lateral speed ups being observed in the NWP model that are not accounted for in the engineering models. The impact of this study is that when conducting dewaking assessments the final wind data will be materially different using the NWP method compared to the engineering model due to more physical phenomenon being resolved. This can results in mean wind speed 1%-2% higher using NWP. Using the NWP method provides a better way to capture the complex turbine interaction effects and results in a higher predicted wind resource which is corrected for global blockage effects and complex speed ups.