Farm level power curves for wind integration studies: considering turbine wake and complex terrain effects

The increasing penetration of variable energy vector such as wind comes with enormous challenges. To surmount this, one of the important tasks in wind integration studies is the accurate forecasts of energy yield both in long and short terms to ensure the security of network and supply, and ultimately a penalty-free electricity trading. These activities require accurate knowledge of local wind climate at micro-scale level. However, microscale-quality data are seldom available and on-site wind measurement takes time. Wind dataset from met stations are usually collected at 10 metres height which is quite poor to be extrapolated to wind farm site and to boundary layer. Regional wind dataset on the other hand is collected at a resolution that is too coarse to capture short-scaled and short-lived local effects such as terrain and/or wake-induced turbulence.This work combines an industry-standard microscale model by WindSim with WRF mesoscale model at 3 km resolution to obtain a downscaled, terrain- and wake-modified dataset with 300 metres resolution suitable for local wind climate predictions. The coupled model uses eleven years WRF dataset (2000-2010) as initials and inlet boundary conditions (BCs), fixed pressure as the top BC, and the site’s digital terrain model (DTM) dataset as the bottom BCs to drive CDF simulation in WindSim. The study site is 7kmx7km covering the entire Braes of Doune wind farm in Scotland. The refined dataset is probably the first attempt to downscale the recently developed regional WRF mesoscale model for a specific site in the UK by taking cognizance of terrain and wake effects. The results from the coupled model are verified in two steps- first, against the actual production data from ELEXON, and then against the previous power curve derived by fitting warranted power curve to the WRF dataset. To investigate the directional dependecy of power curves, a set of representative power curves are derived for 12 sectors at 30° interval. In addition, the vertical and directional profiles of wind speed, inflow angle, wind shear/exponent, turbulent kinetic energy, and turbulence intensity are derived for the study site.The fidelity of the downscaled dataset to model the actual production pattern is demonstrated, and a set of normalized directional power curves and vertical profiles of micro-scale variables are predicted as representative of the wind farm site and could be used to predict power output, tendencies for turbine loading, and wind farm feasibility studies specific to the site