Presentations

Abstract

Objective: To improve the estimation of ambient horizontal turbulence intensity (TI) from mesoscale modelling. It is important for the “Design Basis” phase of North Sea offshore wind projects, where TI drives site-specific fatigue load assessments, a key parameter relevant for uncertainty in turbine class selection and wind farm layout arrangement.   Methodology: We present a WRF-based TI estimation framework that explicitly accounts for turbulence anisotropy and the split between resolved and sub-grid variability in WRF. Simulations for the years 2010 to 2021 are performed with WRF v4.2 using the MYNN 2.5 Planetary-Boundary Layer scheme using 9 km and 3 km nested domains. Sensitivity tests at 1 km resolution have also been carried out. The key step is to estimate the horizontal wind-speed variation equivalent to cup anemometer measurements by combining (i) the resolved 10-min variability from the output of every time step and (ii) the sub-grid variance inferred from MYNN TKE (Turbulent Kinetic Energy), with the conversion governed by a vertical turbulence anisotropy ratio. Two complementary estimators are implemented: (1) an empirical, stability-conditioned anisotropy model based on data from the Lichteiland Goeree (LEG) campaign, using concurrent LEG lidar turbulence statistics and WRF stability diagnostics (Brunt–Väisälä frequency) to establish the relationship between WRF variables and turbulence anisotropy; and (2) a physics-based model derived from Kaimal turbulence theory under neutral condition. Validation: Robustness of the approach is assessed against a multi-site North Sea ensemble spanning different offshore regimes and sensor setups: FINO3, Dogger Bank West, and the IJmuiden masts (50–110 m). Across wind speeds 5–25 m s⁻¹, the proposed anisotropy-aware TI estimates reproduce observed wind-speed-binned TI with mean bias error typically below ~1% at most heights. Aggregated normalized TI errors across sites also show small biases on the order ~3-4% for the two proposed estimators, with standard deviations ~8%. The approach by Tai et. al. 2023, which only considers the variation of horizontal wind components, exhibits substantially larger positive bias, on the order of ~18%.   Impact: By embedding a LEG data-based, stability-dependent anisotropy correction (and a Kaimal neutral condition alternative) into a MYNN 2.5 WRF workflow, the method improves the precision of site-specific TI characterization relative to Tai et. al. 2023. This is expected to directly support reduced uncertainty in design-basis fatigue load inputs for large-scale offshore deployments, especially when extrapolating beyond limited mast records while retaining physically interpretable links to stability and boundary-layer structure. We expect the proposed method to be valid in other sea basins beyond the North Sea. References: Tai, S.-L. et. al.: Validation of turbulence intensity as simulated by the Weather Research and Forecasting model off the US northeast coast, Wind Energ. Sci., 8, 433–448, 2023, doi: 10.5194/wes-8-433-2023.