Presentations

Abstract

Wind farm wakes have been shown to propagate and persist offshore over distances of tens of kilometres, when the atmospheric conditions are stable (e.g. [1]). With the planning and steady build-up of wind farms in the North Sea and US Eastern Seaboard, the impact of cluster wakes on wind farm annual energy production (AEP) increases over time. Therefore, cluster wakes are an increasingly emergent risk to LCOE. To better understand the potential biases and uncertainties when modelling cluster wakes, DNV and RWE have embarked on a validation exercise, for the two operational wind farms of Amrumbank (German Bight) and Triton Knoll (UK), with a focus on wind directions where the object wind farm is in the partial wake of neighbouring clusters. In the validation, pattern of production (PoP) from two CFD models (RWE CFD, DNV CFD using WRF informed boundary conditions) and two engineering models (WindFarmer EVM + LWF, and RWE VV model) have been compared with operational data. Some initial results from this analysis, for conditions on the plateau of the thrust curve, were presented in [2]. This showed that the higher fidelity CFD and VV delivered normalised PoP agreeing best with the observations. In the case of Triton Knoll, the analysis confirmed that the observed PoP is affected by wakes of a wind farm cluster located some 30 km upstream, without limiting the analysis to stable conditions. In the current contribution, we are going into some of the details regarding the various models' performance and their respective ability to capture effects that are a manifestation of blockage, internal or external wakes. The sensitivity of the model results to the characteristics of the modelled atmospheric conditions (stability, boundary layer height) will be reviewed. The impact of non-homogeneity in the background flow, due to coastal transition, and its potentially confounding effect, when interpreting observed PoP, will also be discussed, as will the choices made when processing the SCADA data. Evidence is emerging that atmospheric stability plays an increasingly important role at larger scales with respect to static and dynamic effects. Discussed is the potential impact of this and the increasing risk associated with the lack of understanding and measurements of the marine boundary layer for the purposes of parameterising turbine interaction models.#_msocom_4 Finally, the predicted total turbine interaction losses (aggregated wakes and blockage) from all models will be compared. To avoid divulging confidential information about the wind farms, the aggregated losses will be calculated for a reasonably realistic, yet hypothetical wind rose and wind speed distribution. This comparison between models will allow to draw conclusions on the effect that the choice of model, its setup and site-specific stability distribution input can have on the predicted AEP. References#_msocom_6 1. Hasager, C.B.; et al., Energies 2015, 8, 5413-5439. https://doi.org/10.3390/en8065413. 2. C. Rodaway et al, poster presentation, ACP 2023, Austin, https://acp2023rt.eventscribe.net/posters/posterWall.asp.