This paper proposes to alter standard turbulence models in two ways. The first is to limit the eddy-viscosity in close proximity to the turbine rotor to reduce the initial mixing rate. The second is to add a source of turbulent kinetic energy in the near-wake to compensate for the fact that the actuator disk approach does not resolve the discrete vortices trailed from the rotor, therefore under-predicting the generation of turbulence that occurs when the vortices break-down. Comparing to two lab-scale experiments, these modifications showed significant improvement in predicting the wake recovery downstream of single turbines, and small arrays.
