Posters

Abstract

Reducing energy related CO₂ emissions is at the heart of the energy transformation [1]. IRENA estimated that the transition to electrified forms of transport and heat powered by renewable energy would provide 60% of the energy related CO₂ emissions reductions needed by 2050 [2]. From a 24% share in 2015, the contribution of renewable energies to electricity generation would reach 85% in 2050. Amongst renewables energies, wind energy is foreseen to have the largest impact with a 36% share [3]. In 2020, the levelized cost of electricity for wind project ranged on averaged between 0.037 USD/kW and 0.2 USD/kW depending on the regions [4]. Reducing these costs require having accurate estimates of the long term availability of the wind resource at hub height. Evaluating the wind resource is a two stage process. First, a prospective site is identified using wind atlases; then measurement are carried out to confirm the wind potential. Yet, setting up a measurement campaign is costly and time consuming [5]. Therefore, the accuracy, spatial availability and resolution of a wind atlas is of prime importance. Recently the Global Wind Atlas version 3.0 (GWA 3.0) was released by DTU Wind Energy department [6]. It provides wind speed (WS) estimates worldwide at 10m, 50m, 100m, 150m and 200m. Validation at a height of 35 m and for 35 sites showed a mean bias (MBE) on the WS of -1% and a standard deviation (std) of ±18% . To our knowledge, no additional validations were performed at larger heights up to this moment. The objective of this study is to calibrate the GWA 3.0 WS, hereafter named CGWA 3.0, at 100m and 80m using Tractebel's Wind Database (TBWD) and a kriging algorithm. In order to calibrate and validate the CGWA 3.0, Tractebel collected hundreds of long term averaged WS observations located worldwide with measurement heights ranging from 50m to 142m. TBWD is divided randomly into a 75% training and a 25% validation set. The training dataset is used to select the kriging parameters and the validation dataset is used to assess the accuracy of CGWA 3.0. First, all data from TBWD and GWA 3.0 are extrapolated vertically (VE) to a validation height using a power law based on wind shear computed with WS from the GWA 3.0 at 50 and 150 m. Second, the difference between the VE wind speed of GWA 3.0 and TBWD at 100m and 80m is computed for each site. Third, this discrete uneven bias map is interpolated spatially to a 250m grid using kriging interpolation methods [7]. Fourth, CGWA 3.0 is obtained by subtracting the estimated bias at each grid point from the GWA 3.0 wind estimates. We found at the validation station that MBE of GWA 3.0 and CGWA 3.0 were 8% and -0.4% whereas the std were 9% and 6%, respectively. The major implication of this study is to demonstrate the ability of the kriging algorithm to improve the WS estimates from the GWA 3.0 with observation having uneven spatial distribution.