The prediction of unsteady three-dimensional aerodynamics on wind turbine blades.
PhD thesis, University of Glasgow.
Full text available as:
After introducing the main features of the aerodynamics of wind turbines, a review of key theoretical studies and aerodynamic modelling methods provides the opportunity to focus on predictive methods and the main technical challenges associated with the aerodynamics of HAWTs. The basic aerodynamic method adopted in this study, a classic Blade Element Momentum theory model, BEM, is described next and its extension to yawed flow is detailed for completeness. Analysis then focuses on how stall delay due to three-dimensional effects can be predicted on a HAWT. Implementation of a semi-empirical stall delay model shows sensitivity to blade geometry but no dependency on wind velocity or rotational speed. This seems to be physically incorrect and suggests that a deeper understanding of 3-D effects is still needed if better algorithms are to be developed.
The work then examines the onset of dynamic stall. A 2-D semi-empirical correlation of vortex stall onset, developed previously at Glasgow University, is implemented and validated through available field data from the NREL turbine Phases II and IV. The comparison of measured and predicted locations of dynamic stall onset highlights some interesting features of the three-dimensionality of the process; after the local inception, earlier dynamic stall appears to be triggered in adjacent stations. An attempt to study how 3-D stall delay interacts with the onset of dynamic stall, shows that stall delay appears not to influence the inception of dynamic stall in the way it does static stall. Moreover, the firsts signs of dynamic stall onset are generally best characterised by the correlation when it assumes locally 2-D flow. This is a significant result, as it demonstrates that the earliest signs of dynamic stall onset on wind turbines can be correctly predicted using 2-D tools. A closer examination of the discrepancies between the predictions and measurements has highlighted the particular aerodynamic characteristics of the S809 aerofoil, utilised as the blade section of the NREL turbines. The unusual stalling characteristics of this aerofoil bring into question the significance of the static stall angle in relation to dynamic stall. It is show that other features of the static behaviour may provide a more appropriate link to dynamic stall for some aerofoils.
Finally, the phenomenon of tower shadow on a downwind turbine is studied. Unaveraged pressure measurements and integrated normal force coefficients from tests conducted at Glasgow University are analysed. The analysis highlights many interesting features of the tower shadow response. In particular, as the blade enters the tower shadow region, there is a rapid reduction in normal force due to the tower wake velocity deficit. As the blade leaves the tower shadow, the recovery is consistently slower and more progressive and apparently extends further than the edge of the velocity deficit region. These observations are then used in a examination of tower shadow modelling. A steady model, based on a cosine shaped velocity deficit is evaluated by comparison with the wind tunnel measurements. Unfortunately neither the phase nor the intensity of the response is adequately captured. This leads to the implementation of a new model, based on classic unsteady thin aerofoil theory that accounts for the aerofoil wake induced velocities. The unsteady model captures, in a satisfactory manner, the global response of the blade through the tower shadow region, with a negligible computational cost.
Actions (login required)