Figure 1 - Elevation map with Z-transect and met mast locations that takes place in the Alex17 benchmark

The ALEX17 benchmark: WRF-Meteodyn WT’s wind speed BIAS lower than 1%, with less than 2 hours computing time

The ALEX17 diurnal cycles in complex terrain benchmark (J. Sanz Rodrigo et al 2021 J. Phys.: Conf. Ser. 1934 012002) was the last experiment within the New European Wind Atlas (NEWA) project. Its objective was to identify the wind conditions upstream of the Alaiz Site in order to validate flow models. The site in question is a complex terrain one located near Pamplona, Spain. An elevation map of the site is displayed in figure 1.

This benchmark was intended for flow models that can reproduce wind conditions at the microscale level making it relevant for wind turbine siting and energy yield assessment. Realistic large-scale forcing characterized by mesoscale modeling was used. Meteodyn participated alongside Siemens Gamesa Renewable Energy, the National Renewable Energy Centre, Fraunhofer IWES, the Technical University of Denmark, UL Services and the Barcelona Supercomputing Centre.

The ALEX17 benchmark characteristics

The benchmark was focused on the assessment of wind predictions based on the analysis of hourly-averaged wind speed data from seven meteorological masts from a 4-day period of persistent northerly winds.

Several mesoscale and microscale methodologies using the Weather Research and Forecasting (WRF) model at the mesoscale level were employed:

  • WRF Mesoscale,
  • WRF-LES Multi-scale,
  • WRF-Alya (URANS and LES) with 1D tendencies,
  • WRF-Elypsis3D (URANS) with 3D tendencies,
  • WRF-Meteodyn WT (RANS) statistical downscaling.

Meteodyn’s methodology

“A statistical meso-micro coupling method for wind resource assessment application is developed in Meteodyn. WRF outputs at 9 km resolution are classified by wind direction and stability categories (based on the Monin-Obukhov length obtained from WRF and stability bins in Table 1) to obtain ensemble-averaged inlet and top boundary conditions for MeteodynWT RANS k-l model.

Three classes of atmospheric stability (unstable, neutral, and stable for 0-2) are simulated via the turbulence kinetic energy and the turbulence length scale [1]. A force term is added to relax the microscale flow to the mesoscale wind profiles in the meso zone that starts 500 m above ground level, vertical profiles of mesoscale model output being assimilated into the CFD model as a body force [2] [3] [4]. Below this level, no relaxation is applied because of the high topography resolution in the microscale CFD model. A 20 km microscale domain is simulated with 25 m horizontal resolution and 4 m vertical resolution resulting in a mesh size of 16 million points.” (J. Sanz Rodrigo et al 2021 J. Phys.: Conf. Ser. 1934 012002).

Read the complete paper

overall mean normalized bias Sanz_Rodrigo_2021_J._Phys. _Conf._Ser

Figure 2 - "Overall mean normalized bias" (J. Sanz Rodrigo et al 2021 J. Phys.: Conf. Ser. 1934 012002).

Initial results of the ALEX17 benchmark

The results confirm the added value of mesoscale-microscale coupling and CFD at reproducing non-conventional wind conditions in complex terrain. Those methodologies reduce the wind speed mean bias from 32%, at 3-km mesoscale, to +5% (J. Sanz Rodrigo et al 2021 J. Phys.: Conf. Ser. 1934 012002).

We are very proud of Meteodyn’s model results in this benchmark, reaching near-zero wind speed BIAS (-0.55 %).

The result details may be found in the last paragraph of page 9 and in the Conclusions section of the article.

Ask for a quote


[1] Meteodyn 2019 User Manual

[2] Zajaczkowski F J, Haupt S E and Schmehl K J 2011 J. Wind Eng.Ind. Aerodyn. 99 320-329

[3] Duraisamy V J, Dupont E and Carissimo B 2014 Journal of Physics: Conference Series 555 012031

[4] Buhr R, Kassem H, Steinfeld G, Alletto M, Witha B and Dörenkämper M 2021 Energies 14 ISSN 1996-1073 URL

Discover our news