Posters

Abstract

Energy yield assessment (EYA) remains a challenge in complex sites with insufficient representative anemometry coverage. The standard approach of considering a single wind resource grid (WRG), that varies by direction only, to represent the average atmospheric stability conditions can present considerable cross-prediction errors when studied at a time series level.  The “average” directional WRG approach can show good performance by virtue of errors cancelling out and remaining “invisible” when viewed on average, but ignorance of temporal errors can cascade from wind speed modelling through the full EYA modelling chain to result in biases in performance when compared to real world SCADA data. It is suggested that seasonal and diurnal variations should be assessed in detail, especially in complex sites.  This work explores the use of multiple WRGs to allow the relevant wind field to be applied based on the atmospheric state, demonstrating improved cross-prediction behavior when compared to single “blended” WRGs. Multiple binning methodologies were assessed, using atmospheric stability indicators (Obukhov Length, Richardson Number, Planetary Boundary Layer Height,) and other (e.g., time-of-day, season) indicators.  The strategy was applied to results from industry-standard spatial modelling tools, Weather Research and Forecasting based models as well as full-year advanced meso-microscale coupling results (where data from a modified WRF model supplied the boundary conditions for the URANS microscale code, in a one-way, offline mode in the time domain). The coupled time series WRF-CFD approach captures the complex flow features that arise from local terrain and stability-driven effects. Impacts at the local turbine level, that would otherwise not be captured using solely a mesoscale or downscaled model, exhibit complex time varying behaviour. The findings support the need for a move to time-series EYA based on multiple WRGs to improve the cross-predictions over traditional (single WRG) methods and consequently more accurate EYA predictions. However, the micro-mesoscale coupled model did not always improve cross-prediction errors compared to microscale methods using idealized, steady-state conditions, which was attributed to the comparatively lower spatial resolution employed in the former to mitigate the increase in computational cost. The approach has been extended to 5 sites of varying complexity in the full EYA modelling chain and accuracy benefits of WRG switching demonstrated through detailed GAP analysis versus operational SCADA data.