Improving Boundary-layer Wind Energy Forecasts by Representing Wind Farms in Mesoscale Models: Validation and Improvements using WFIP2 Observations

Forecasting wind power production remains a great challenge for utilities and power operators, especially in geographic areas with numerous interacting wind farms or within large wind farms in which wake effects reduce the power production of waked turbines. The mesoscale numerical weather prediction Weather Research and Forecasting (WRF) model incorporates a Wind Farm Parameterization (WFP) that uses elevated drag and turbulent kinetic energy production to represent the effect of wind turbines on the atmosphere, but this parameterization has not yet been validated in an onshore wind farm in complex terrain. As the WFIP2 field study area is home to nearly 5 GW of wind deployment, with numerous distinct wind farms, we use the extensive meteorological observations in the WFIP2 domain to identify case studies for tests of the WFP.

Given the relatively wide spacing between turbines in the region – most of the turbines are located 10-20 rotor diameters (D) from each other in the direction of most frequent flow – this region presents an ideal test of the mesoscale WFP. Therefore, we identify numerous cases with wind speeds varying from 8-12 m s-1 in unstable and stable conditions. When flow is from the east, downwind wind speed and direction profiles from instrumentation located at Wasco, OR, can be used to quantify far-downwind wake behavior and assess the performance of the WFP. When winds are from the west, as is most common, the power measurements from the individual turbines are compared to those predicted by the WFP to assess the performance of WFP. Scanning lidar measurements from the Arlington field site document individual wakes from turbines in both easterly and westerly flow conditions.

Beyond documenting the benefits of forecasting with the WFP as opposed to simulations with no accounting for the effects of turbines on the flow, we test a number of improvements to the WFP that have been motivated from other studies in both real and idealized scenarios. These improvements include reductions in the amount of turbulence generated by turbines in the simulations and incorporating a stability-dependent turbulence generation in the WFP as suggested by recent large-eddy simulations. While our simulations use 1-km horizontal resolution, we also test the value of high vertical resolution with the WFP, including nominally 10-m and 20-m resolution in the lowest 200 m of the domain.