Search

This report is a continuation of the Wind Turbine Airfoil Catalogue [1] which objective was, firstly to provide a database of aerodynamic characteristics for a wide range of airfoil profiles aimed at wind turbine applications, and secondly to test thetwo-dimensional Navier-Stokes solver EllipSys2D by comparing its results with experimental data. In the present work, the original two-dimensional results are compared with three-dimensional calculations as it was surmised that the two-dimensionalassumption might be in some cases responsible for discrepancies between the numerical flow solution and the actual fluid flow, and thereby the incorrect prediction of airfoil characteristics. In addition, other features of the flow solver, such astransition and turbulence modelling, and their influence onto the numerical results are investigated. Conclusions are drawn regarding the evaluation of airfoil aerodynamic characteristics, as well as the use of the Navier-Stokes solver for fluid flowcalculations in general.

For modern rotor blades with their very large aspect ratio, the blade tip is a very limited part of the overall rotor, and as such of limited importance for the overall aerodynamics of the rotor. Even though they may not be very important for the overallpower production, the tip noise can be very important for the acoustics of the rotor [15], and the blade tips can as well be important for the aerodynamic damping properties of the rotor blades [13]. Unfortunately, not many options exists for predictingthe aerodynamic behavior of blade tips using computational methods. Experimentally it is di±cult to perform detailed measurements in the form of pressure and velocity measurements in natural wind conditions on modern large scale turbines due to theinherent unsteadiness in the natural wind. The present study describes the application of four different Navier-Stokes solvers to tip shape studies, and shows that these codes are well suited to study the flow around different tip shape geometries, andcan predict the pressure distributions at the blade tip quite accurately.

The problem of the aeroelastic stability of wind turbine blades is addressed in this report by advancing the aerodynamic modelling in the beam element type codes from the engineering-type empirical models to unsteady, 2D or 3D, Navier-Stokes solvers. Inthis project, structural models for the full wind turbine blade have been combined with 2D and 3D unsteady Navier-Stokes solvers. The relative disadvantage of the quasi-3D approach (where the elastic solver is coupled with a 2D Navier-Stokes solver) isits inability to model induced flow. The lack of a validation test case did not allow for quantitative comparisons with experimental data to be carried out; instead the results of the advanced aeroelastic tools are qualitatively cross-compared. Allinvestigated methods predicted qualitatively similar results. They all resulted in positive aerodynamic damping values for the flap mode, in a decrease in damping with the increase of wind speeds and in a minimum value for the damping for wind speedaround 15~m/s. The eigenvalue analyses resulted in steeper distributions for this mode. The agreement in aerodynamic damping decrease with the increase of wind speed is also observed in the distributions for the lead-lag mode. In perspective, theuncoupled, linear method results in higher values of aerodynamic damping compared to the 3D aeroelastic tool. The quasi-3D tool results in lower aerodynamic damping values in the higher wind speeds and in lower damping values in the lower wind speedregime. Apart from the computations for the full blade, 2D computations for the so-called “typical section” have been carried out. The 2D aeroelastic tools resulted in similar aerodynamic damping values. Qualitative agreement was better for the lead-lagmode. The presence of roughness tapes has a small, rather negligible impact on aeroelastic stability as depicted by the results of both aeroelastic tools. On the other hand, in conformity to the inability of the adopted computational model to successfullypredict the corresponding test cases under work package 2 of the project, the aeroelastic tools are not capable to predict the correct physical trends when the blade is equipped with stall strips. The resulting decrease in damping in the computations byboth tools is not in accordance with experience.

This paper deals with Danish Government Foreign Borrowing and Debt Administration in the period 1959 to 1991 the Nationalbank succeeded the Ministry of Finance as the administrator.

In the EC project KNOW-BLADE a work package has been defined to investigate the possibility to numerically model aerodynamic accessories in existing Navier-Stokes solvers. Four different aerodynamic accessories have been investigated. Firstly, thepotential of applying active flow control by means of a pulsating jet placed at the leading edge in order to enhance mean lift. The general trend is that increased pulsation frequency is beneficial, in that it reduces the oscillation amplitude and raisesthe mean lift level while lowering the mean drag level. An increased jet exit velocity has a tendency to increase the oscillation amplitude, which is not very attractive for load control on wind turbines. Secondly, the effect of vortex generators hasbeen modelled using two phenomenological vortex generator models. The models have been applied to three airfoil configurations. For all cases investigated the models shows qualitatively the correct behaviour, even though there are a considerable spread inthe degree of success. Thirdly, the influence of adding a stall strip for changing the airfoil characteristics was investigated. Stall strips at three different positions were directly modelled by changing the airfoil geometry. In general the 7mm stallstrips placed at P00 and P-02 had the greatest effect on the max lift followed by stall strip P02. Unfortunately, there was not sufficient agreement between the experimental results and the simulations to draw any conclusions of optimum position andgeometry of the stall strip. Finally, the effect of surface roughness was modelled by either modifying the boundary condition of the turbulence model or by modifying the airfoil geometry. Using the roughness model gave relatively good agreement withmeasurements and it must be concluded that the effect of using roughness tape can be better predicted with a roughness model compared to using a modified airfoil surface.

3D rotor computations for the Greek Geovilogiki (GEO) 44 meter rotor equipped with 19 meters blades are performed. The lift and drag polars are extracted at five spanvise locations r/R= (.37, .55, .71, .82, .93) based on identification of stagnationpoints between 2D and 3D computations. The inner most sections shows clear evidence of 3D radial pumping, with increased lift compared to 2D values. In contrast to earlier investigated airfoils a very limited impact on the drag values are observed.

The present work describes a thorough investigation of 2D computations of the flow around a NACA 63-415 aerofoil fitted with stall strips (SS). A mesh study as well as a time step study is carried out and all computations are compared with experiments.Two different SS, 5mm and 7mm are investigated at several positions. Furthermore the influence of laminar to turbulent transition and the effect of a rounded SS were tested. There is not sufficient agreement between the experimental results and thesimulations to draw any conclusions of optimum position and geometry of the SS. The 7mm SS's placed at P00 and P-02 has the greatest effect on the max lift followed by SS P02. The 5mm SS's does, as in the experiment, not change the lift curve noticeably.Even though this investigation does not conclusively succeed in verifying the experimental results with CFD, many useful conclusions can be drawn from the results. It is observed in the experiment that the vertical force fluctuates at higher angles ofattack. This indicates that small bubbles are being shed off the profile causing the force to vary. This property is observed when transition is included in the model and also when the tip of the SS is rounded. From this result it could be concluded thatthe level of turbulence produced on the tip of the SS is very important for the development of the flow downstream. In the sharp tip calculations using fully turbulent computations, this is most likely too high, which resulted in the fine structures beingdamped out, with only one bubble appearing.

A new simplified transition model for wind turbine blades is described along with the implementation in the EllipSys3D code. The method is based on a sectional treatment of the turbine blade under the assumption of chordwise flow, and lookup tables oftransition point location computed by external 2D programs. The coupling of the 2D transition point location and the 3D sectional flow is performed through the stagnation point location. The method is applied to a single rotor case, the NORDTANK 500/41rotor with LM19.1 blades. The transitional computations show improved agreement with measurements for wind speeds between 11 and 15 m/s. For higher wind speeds, the validity of the transition location computed by the 2D XFOIL code is questionable, and theresults cannot be trusted. Analysis of the results comparing fully turbulent and transitional spanwise distributions of tangential forces, reveal that the decrease in power production when applying the transition model is mainly a consequence of thedecrease in driving force on the inboard part of the blade between 5 and 12 meter radius. The results are very encouraging, and further studies of other rotors are needed for further validation.

The complex three-dimensional flow around three different tip shapes on a rotating wind turbine blade is investigated and analyzed using Computational Fluid Dynamics. Differences in production, flapwise bending moments and forces are discussed. A methodfor determining the local inflow angle of attack is presented and further analysis is performed on lift and drag coefficients. It is shown that the original Standard tip results in a more concentrated tip vortex leading to a steeper gradient on bothtangential and normal forces when approaching the tip, whereas the two tapered tips show a more flat behavior. This again leads to lower flapwise bending moments and lower production for the Standard tip compared to the two tapered tips. At 12 m/s,though, the Swept tip shows a separation pattern on the surface. This separation causes a decrease in normal force and an increase in tangential force. The Taper tip keeps the higher loading causing the flapwise bending moment to be higher as seen inmeasurements. To determine the radial variation of lift and drag coefficients the local inflow angle of attack is determined. It is shown that the Standard tip experiences a slightly larger angle of attack at the tip compared to the two tapered tips. Thelift coefficients are kept at a more constant level for the two tapered tips due to the decrease in chord, while the drag coefficients actually decrease for the two tapered tips, especially for the Swept tip. For the Swept tip at 12 m/s both lift and dragcoefficients changed considerably due to the separation. Differences in aerodynamic damping of the three tips were investigated using HAWCDAMP. The Standard tip seems to be slightly less damped with respect to the edgewise vibrations.

This report sets up an evaluation of the two-dimensional Navier-Stokes solver EllipSys2D in its present state. This code is used for blade aerodynamics simulations in the Aeroelastic Design group at Risø. Two airfoils are investigated by computing theflow at several angles of attack ranging from the linear to the stalled region. The computational data are compared to experimental data and numerical results from other computational codes. Several numerical aspects are studied, as mesh dependency,convective scheme, steady state versus unsteady computations, transition modelling. Some general conclusions intended to help in using this code for numerical simulations are given.