Description

The expansion of offshore wind energy increases the need for considering the variable
meteorological background climate over larger areas as well as for taking farm wakes from neighbouring wind farms into account. Mesoscale meteorological models equipped with a wind farm parameterisation (WFP) have been identified as a useful tool for that (Fischereit et al., 2020; Veers et al., 2019).

At the same time, individual turbines cannot be resolved with a typical resolution of mesoscale models for this topic of about 1 to 3 km. High-resolution wake models, i.e. both simple engineering-type wake models as well as CFD-RANS models, are designed for this purpose. However, their simplified assumptions of horizontally uniform inflow are not always met over larger areas. Therefore, high resolution wake models might be less suitable for modelling farm-to-farm wakes and long-distance wakes.

With this in mind, the aim of this study is two-fold. Firstly, to evaluate for intra-farm wakes the performance of a WFP in a mesoscale model with a typical horizontal resolution against mast measurements, SCADA data from Rødsand II and high resolution wake models. Secondarily, to compare farm-to-farm and long distance wakes of a WFP and high resolution wake models.

For this study, we employ the Weather Research and Forecasting (WRF) mesoscale model (Skamarock et al., 2008) with the Explicit Wake Parameterisation (EWP, Volker et al., 2015). As high-resolution CFD-RANS model, we use PyWakeEllipSys
(https://topfarm.pages.windenergy.dtu.dk/cuttingedge/pywake/pywake_ellipsys), which is based on the EllipSys3D CFD flow solver (Michelsen 1992, Sørensen 1994), and different simple engineering-type wake models included in the pyWake suite (Pedersen et al. 2019). The models are applied to the Fehmarn Belt area with the wind farms Nysted and Rødsand II. The nested WRF domain with the two wind farms is shown in Figure 1. We simulate two one-month long periods with WRF with high availability of measurement data and SCADA; one before commissioning of Rødsand II and one after commissioning of Rødsand II, respectively. After the evaluation of the performance of WRF and WRF-
EWP in the two periods against mast measurements, flow cases are defined for the high-resolution wake models for the post-commissioning period.

The evaluation of the WRF simulation before commissioning shows that WRF is able to simulate the background meteorological conditions in the Fehmarn belt area. After commissioning, WRF-EWP performs better than a simulation with WRF without WFP. The average wake effect during this period is about 1 m/s.

Since SCADA data are only available for Rødsand II, one flow case with eastern wind and wind speeds of 10±1 m/s has been chosen for the high-resolution wake models. Initial results for the comparison for intra-farm wakes of WRF-EWP and SCADA data indicates that WRF-EWP captures the wind speed reduction inside the farm along the flow (Figure 2) but overestimates reductions perpendicular to the flow.

In the presentation, we will show results for this and other flow cases comparing WRF-EWP with different high-resolution wake models for intra-farm, farm-to-farm and long distance wakes and discuss possible improvements for the models.
Period25 May 2021
Event titleWind Energy Science Conference
Event typeConference
LocationHannover, Germany