Posters

Abstract

Accurate modeling of wake losses is a key element of wind resource assessment and wind farm performance prediction. Beyond the formulation of the velocity deficit itself, the choice of wake superposition model plays a critical role, particularly for downstream turbine rows where multiple wakes interact, and recovery processes strongly influence energy production.  This study investigates the influence of wake overlap regimes on the performance of different wake superposition models. We propose a new sector-dependent  approach  rather than selecting a single superposition configuration for all conditions. This approach acknowledges that turbine rows may experience distinct wind regimes and wake overlap conditions on wind direction, leading to different optimal modeling choices.  Two widely used analytical wake models – the Jensen-Park model and the TurboGauss model, which corrects the top-hat wake distribution but relies heavily on empirical parameters – are considered to account for the impact of wake deficit formulation and wake expansion behavior. For each wake model, three superposition methods (linear sum, quadratic sum, and max-sum) are evaluated under a range of wind conditions, including fully aligned flow resulting in full wake overlap and oblique inflow cases leading to partial wake interactions.  The proposed hybrid sector-based approach is applied to three offshore wind farms to address wake-related challenges, where wakes tend to persist over long distances. In contrast, terrain complexity on onshore environments promote faster wake recovery. The selected wind farms exhibit contrasting layout characteristics: large-scale farms with regular geometries and wide turbine spacing, such as Horns Rev and Nysted, and a denser wind farm characterized by reduced inter-turbine spacing and a less regular layout caused by the presence of a central void. For each wind farm, the analysis focuses on the normalized power production of turbines within the same row, compared against SCADA data. For Horns Rev and Nysted, additional validations are performed using 5° wind direction sectors, based on the mean normalized production of the eighth turbine column.  The relative performance of the superposition models under each wind direction is assessed via statistical error metrics such as RMSE and MSE. The results demonstrate that this hybrid sector-based modeling framework is a practical and effective approach for improving wind farm energy yield predictions, leading to a considerable reduction in prediction error,  and a clear potential enhancement to  industrial wind farm design and performance assessment.