Seminarium: Licentiatseminarier

Wind turbines over a hilly terrain: performance and wake evolution

Talare: Ann Hyvärinen
Organisation: KTH, Mekanik
Tidpunkt: 2018-02-23 10:15
Plats: D2


The aim of this licentiate thesis is to investigate wind-turbines placed in a complex-terrain environment. This is done by studying the flow around small- scale wind-turbine models placed over a landscape model with hills, and by comparing the results with corresponding data obtained over a flat terrain model. The studied flow features include the wind-turbine wake development and the turbine performance under different conditions, the effects from wake interactions, the influence of the ambient turbulence levels and the influence from a complex topography. Wind-tunnel measurements have been performed using particle image velocimetry and hot-wire anemometry to measure the velocity field. Additionally, numerical simulations, based on RANS modelling and actuator-disc techniques, have been made to support the experimental data and to gain further knowledge about the investigated flow cases. The results reveal that the hills promote a downward wake deflection be- hind the turbines and enhance the wind-turbine wake diffusion. As a consequence of this, and with the flow acceleration introduced by the hills, an improved power performance is seen for turbines exposed to wake-interference effects. A correlation is observed between the turbulence levels present in the flow, and the magnitude to which the hill-induced flow gradients influence the wake: Stronger wake deflections due to the hills are seen when the wind-turbine wake is more diffused. This is for instance the case when the wake of two tan- dem turbines is studied, or when higher ambient turbulence levels are present in the wind tunnel. A good qualitative agreement is seen when comparing the experimental and numerical results. The simulation results further indicate that the hills give rise to modulations of the wind-turbine wake. It is shown that these modulations can be reasonably captured by means of wake-superposition techniques, given that a wake model with sufficient accuracy is chosen.