IEA Annex 28 MexNext: Analysis of Wind Tunnel Measurements and Improvement of Aerodynamic Models

The project comprises the Danish contribution to the IEA Task 29 ”MexNext” coordinated by ECN from the Netherlands and with total 20 participating institutes from 11 different countries. The main content of MexNext has been a coordinated analysis of detailed aerodynamic measurements on a model rotor carried out within a previous EU project ”Model Experiments in Controlled conditions Mexico” back in 2006 with Denmark as one of the project participants. The experimental data comprise besides the more standard type measurements such ad strain gaug data from the blades and the tower, rotor power and rotor speed also;

1. Surface pressure measurements of the blade at 5 radial positions
2. PIV (Particle Image Velocity) measurements of the velocity field around the rotor and in wake.

In particular the last mentioned type of measurements is not common in experimental data sets on wind turbine rotors and thus makes this data set unique.

After the data processing and analysis, where Risoe DTU has constributed with detailed CFD computations investigating the influence of the tunnel walls on the flow, the data set has among other analyses been used for the following:

1. Detailed validation of design and analysis codes including the CFD program EllipSys3D devel oped in collaboration between Risoe DTU and DTU MEK; the aroelastic code HAWC2 developed at Risoe DTU and the aeroelastic code AL (Actuator Line) developed at DTU MEK
2. Improvement/development of the above mentioned programs
3. Development/validation of methods for determination of inflow angels on a rotor
4. Derivation of airfoil data for a rotor

Validation of computational tools is an important process both within research institutions and within industry. During validation of a code within a research institute there is a strong focus on identification of physical phenomena that could be described better through a further development of the code. Within the industry it is of big importance to get the design codes validated in order to know the uncertainty on the design loads. A reduction of his uncertainty will allow a reduction of the safety factors on the design loads and thus a cheaper turbine design. Both the EllipSys3D CFD code and the HAWC2 aeroelastic code are widely used within the industry and the certification institutes and they will thus be able to use the validation results from the project. Further data set from MexNext is also used directly by the industry when conducting their own validation of their design codes and for doing this it is of big importance to use a data set that already has been analyzed and evaluated against many different codes comprising also an evaluation of the quality of the data.

As mentioned above the detailed flow measurements around the rotor and the wake has been compared with the results from EllipSys3D and the AL model. In general a very good correlation is found between the results from the two codes and measurement and also on flow detailed (e.g. the velocity variations around the individual vortex sheets behind the rotor), which has not previously been possible to validate. However, two major deviations between model results and measurements are found (not only for the two mentioned models but generally for all models applied) and which has not been possible to explain;

1. In general the models over predict the aerodynamic forces on the rotor blades in contrast to the good correlation for the flow field (the flow field id the result of the applied forces)
2. The PIV measurements deviate considerably from the simulations on the inboard part of the blade bur unfortunately the PIV system could not cover the whole inboard part og the inboard blade.

Improvement of the codes, which has been used for comparison with the experimental results, has also been part of the project activities. This part comprises besides direct modifications in the codes also guidance on how the simulations should be set up properly and how they should be run. This could be the selection of time step in the simulations or the fineness of the CFD grid where the very detailed PIV pictures of the tip vortices have been utilized. A complex operation condition for a wind turbine rotor is operation with a yaw error which means that the incoming flow is no perpendicular to the rotor plane. Such test cases have been measures and in general a good correlation between simulations and measurements for both blade loads and wake velocities are found for the advanced CFD models where bigger deviations are seen for the simpler aerodynamic models which are part of the aeroelastic codes used by the industry. Adjustments of the parameters in the simple models can be considered on basis of these findings.

Further the project work has comprised development of models for derivation of the inflow angels along the rotor blades and derivation of airfoil data from measurements/computations of the aerodynamic loads on the blade. For the present model rotor the aerodynamic forces are derived at 5 radial positions where the surface pressure has been measured but on full scale rotors the load will normally be derived from a number of strain gaug measurements on the blade, e.g. conducted during prototype test of the turbine. The derivation of airfoil data is an important result because the airfoil data are used in the aeroelastic simulations of the turbine response. In the present work two simple methods to derive the inflow angle along the blade has been described and also how this procedure is used when deriving the airfoil data.

Finally, the project work has comprised investigations of some fundamental aerodynamic issues such as: 1) aerodynamic loading on the blades of a parked rotor; 2) 2D/3D airfoil data characteristics and 3) variation of the flow velocities in the rotor plane.