Effects of turbulence on the performance and wake of a model-scale wind turbine

Abstract

This thesis work presents experimental results on the effects of different ambient turbulence conditions on the performance and wake development of a model-scale wind turbine, focusing on two main parameters to characterise the nature of the incoming nature: these are its intensity I∞ and its spectral distribution, represented by the integral time scale of the free-stream T0. The power generated by the model-scale turbine is seen to increase with the intensity of the velocity fluctuations, as well as with the free-stream integral time scale; the former finding is in line with established literature, albeit the magnitude of the increase is higher than what expected, while the latter confirms some recent works that model the turbines as low-pass filters, more apt to harvest energy from low-frequency fluctuations. The wake evolution under turbulence is also observed to be highly affected by the atmospheric conditions: wakes developed under higher turbulence intensity evolve more rapidly under the condition that the free-stream T0 is low; in other words, for the same I∞, conditions that are favourable for power harvesting generate longer wakes, unfavourable from the point of view of a wind farm. The predictions of some analytical wake models are compared to the obtained wake, highlighting how fine-tuning of the model parameters can result in very accurate predictions. The results obtained stress the need to include a virtual origin in the wake models, a practice customary for bluff-bodies but seldom employed for wind turbines; this quantity has been seen to relate to the stability of the helical tip-vortex structure, and thus conveys a physical meaning. The rate at which the mean velocity field evolves in the streamwise direction has been related to the Reynolds shear stress in the turbine wake by means of an analytical relation; this allows to predict the mean velocity field in the turbine wake with knowledge limited to the second-order statistics. This last finding has been leveraged to formulate a framework using proper orthogonal decomposition to predict the full-scale wake from limited probe data; preliminary results show that this is possible with acceptable results, correctly predicting both the intensity of the turbulence in the turbine wake from limited data and the mean wake velocity.

More information

Identifiers

ORCID for Bharathram Ganapathisubramani: orcid.org/0000-0001-9817-0486

Catalogue record

Export record