Wind farm interactions with the surrounding airflow leads to a reduction in velocity greater than the linear sum of single turbine inductions and is known as global or upstream blockage. The mechanisms and magnitude of global blockage effect are not yet fully understood. Models to simulate upstream blockage to improve efficiency estimates and better understand global blockage have thus far not been refined. The aim of this research is to investigate the sensitivity of upstream blockage to the numerical configuration of CFD simulations to improve model accuracy and understanding of global blockage.

RANS simulations are executed under steady state conditions with a k − ϵ turbulence model. A neutral stability, pressure-driven atmospheric boundary layer is modelled with fully developed uni-directional flow. Wind farms are modelled as actuator discs with 5 rows of turbines on flat terrain. Streamwise and spanwise spacing is set to 7 turbine diameters (D) and 5D respectively. Global blockage is measured against the induction of a single turbine at 2.5D upstream. Variables investigated include the domain height, lateral extent, inlet and outlet distances from the wind farm.

Domain heights ranging from 5D to 25D are investigated for change in magnitude and scale of upstream blockage for a laterally infinite wind farm. A clear trend of increasing blockage with domain height is observed. At domain heights of less than 15D, upstream velocity is increased by a maximum of 0.18% (5D). Larger domain heights produce a maximum velocity reduction of 0.23% (25D). The shape of upstream blockage is independent of domain height.

A finite wind farm of 5 columns and lateral extents ranging from 2.5D to 20D on each side are utilized to investigate the impact on blockage. Wider domains of 5D to 20D display increasing blockage with width, while a domain of 2.5D exhibits behaviour similar to a laterally infinite wind farm. Blockage ranging from 0.13% (10D) to 0.31% (20D) reduction in velocity is shown to be highest at the center column of turbines and decreases toward the outer columns.

Inlet and outlet distances ranging from 15D to 100D are modelled. Upstream blockage for inlet distances of 50D to 100D produce consistent upstream blockage magnitude and extent of 0.22% and 30D respectively. Shorter inlet distances result in decreased upstream blockage with a minimum of 0.12% (15D). The shape of blockage remains consistent through all inlet ranges. Outlet distance have no identifiable impact on upstream blockage magnitude and extent.

Changes to the numerical configuration show a clear correlation of increased blockage with cross sectional area of the domain. Constraining the domain in the vertical and lateral directions constricts flow resulting in reduced blockage. Blockage becomes independent of inlet distances at values of 50D and higher. Outlet distance has no identifiable impact on upstream blockage. Choosing a numerical configuration with adequately sized domain boundaries is pertinent in producing realistic upstream blockage.