Back to the programme printer.gif Print

Wednesday, 12 March 2014
09:00 - 10:30 Aerodynamics and rotor design
Science & Research  

Room: Llevant
Session description

The session is oriented to show recent computational and experimental findings on aerodynamic phenomena in horizontal (HAWT) and vertical access wind turbines (VAWT), as well as on new developments on system identification techniques related to the aeroelastic behaviour of wind turbine rotors and new aerodynamic design trends for very large wind turbines.

Learning objectives:

Delegates will learn about:

  • recent computational and experimental findings on aerodynamic phenomena in HAWT and VAWT
  • new developments on system identification techniques related to the aeroelastic behaviour of wind turbine rotors
  • innovative design trends for the aerodynamics of very large wind turbines
Lead Session Chair:
Alvaro Cuerva, Universidad Politécnica de Madrid, Spain

Sandrine Aubrun, Univ. Orléans, PRISME Laboratory, France
Carlo Bottasso Technische Universität München, Germany
Stefano Cacciola (1) F P Carlo L Bottasso (2)
(1) Politecnico di Milano, Milano, Italy (2) Technische Universität München, München, Germany

Printer friendly version: printer.gif Print

Presenter's biography

Biographies are supplied directly by presenters at EWEA 2014 and are published here unedited

Carlo L. Bottasso received a Ph.D. in Aerospace Engineering from the Politecnico di Milano in Italy. Since May 2013, Dr. Bottasso holds the Chair of Wind Energy at the Technical University of Munich, Germany. He has taught flight mechanics and wind energy at the Department of Aerospace Science and Technology of the Politecnico di Milano, Italy, reaching the rank of Full Professor. Dr. Bottasso has held visiting positions at numerous institutions, including Aalborg University in Denmark, the Georgia Institute of Technology, the National Renewable Energy Laboratory (NREL), the Lawrence Livermore National Laboratory, NASA Langley, Rensselaer Polytechnic Institute and others.


Estimation of wind turbine model properties - towards the validation of comprehensive high-fidelity multibody models


Experimental tests performed on wind turbines or parts of them are an essential step in the process of validation and verification of designed machines. Several kinds of measurements, which may include as example natural frequencies, mode shapes, static deflections, rotor aerodynamic loads, are typically achievable by means of laboratory, wind tunnel or field testing. This work is concerned to the use of such measurements for estimating the properties of wind turbine models in order to provide validated comprehensive multibody models, for both design and certification and for wind turbine active control applications.


Modern comprehensive aero-servo-elastic models of wind turbines are based on first principles, and include sophisticated mathematical models of the various components, as blades, tower and drive-train, coupled with aerodynamics, electrical parts and controllers. Hence, wind turbine analysis codes are typically based on complex, non-linear, multi-field models, [1]. It is expected that the design of future very large turbines will surely increase the importance of having adequate mathematical models, which can correctly capture the relevant physics of such systems. The fidelity of overall models depends on the fidelity of sub-models and of their couplings, whereas the accuracy of sub-models depends on their ability to capture the relevant physics and on the correct tuning of their parameters. Since modern multibody models, are actually able to reproduce the relevant physics and the couplings between sub-systems, great attention must be deserved to the tuning of the model parameters.
The discipline of system identification, [2, 3] , which finds applicability in the validation and verification of mathematical models for both design and certification and in wind turbine active control applications, [4], can satisfy such need. Parameter estimation for wind turbines is still in its infancy and much needs to be done before suitable techniques can be effectively used by industry. It appears that, given the complexity of turbines, a practical approach would be to use a divide-and-conquer method to incrementally estimate the parameters of the various sub-models using specific experimental observations, as intuitively depicted in Fig. 1.

The parameter estimation is formulated as a maximum-likelihood constrained optimization [5], solved using a Sequential Quadratic Programming (SQP) approach [6], in which measurements and constraints are chosen on the basis of the specific problem. For each sub-model the procedure to follow is summarized here below:
1. Develop a multibody model [7], whose outputs depend on the parameters to-be-identified.
2. Collect experimental data from specific tests.
3. Form the likelihood function and the constraints (if any).
4. Solve the constrained optimization problem, i.e. obtain the unknowns such that the cost function is minimized.
5. Validate the estimated model using a different set of measurements.

Main body of abstract

Within this general framework, this work is concerned with two sub-models identification. More specifically, the main objective is to describe the use of the parameter estimation for updating first the structural beam models and second the aerodynamic properties of wind turbine blades from experimental measurements.
First it is described the estimation of the structural and the inertial distributed properties of a blade.
A real wind turbine blade of 7.5 meter length was tested with the purpose of updating its beam model. The blade was bolted at its root and equipped with three saddles. Each saddle allowed the application of flap-wise, edge-wise and torsional loads. In the aggregate, nine separate load cases, involving mainly flap-wise, mainly edge-wise and mainly torsional responses, were considered. For each tests, the displacements of 19 sections along blade span have been computed by processing high fidelity laser-scanner data as well described in [8]. The set of measurements was completed also with the first 4 vibrating modes, the blade weight and its center of gravity position.
The flap-wise and edge-wise stiffness, the torsional rigidity and the mass distribution along blade are estimated from experimental data using the proposed approach. Fig. 2 shows the comparison between the measurements and the outputs of nominal and identified models: triangle symbols indicate the experimental data, dashed lines the outputs of the nominal model while solid lines those of identified model. The quality of the estimated properties, not displayed here for the conciseness, is demonstrated by the excellent agreement obtained.

Second, it is described the estimation of the aerodynamic properties from wind tunnel tests.
The proposed approach was used for the calibration of lift and drag characteristics of the rotor of an aeroelastically scaled and actively controlled wind turbine model, described in more detail in [9].
The rotor has a diameter of 2 m, with a solidity of about 0.04. To account for the small Reynolds due to the reduced size of the model, the blade was designed using two special low-Reynolds airfoils AH79-100C and WM006, the former airfoil being used in the inboard and the latter in the outboard section of the blade.
Rotor aerodynamic performance was measured by testing the model in the low-turbulence aeronautical test section of the wind tunnel at the Politecnico di Milano, for varying wind speed, rotor speed and blade pitch settings. Testing conditions were optimized to cover a wide range of angles of attack, while at the same time keeping the Reynolds number as high as possible and similar across the various test points. Power and thrust coefficients as well as tip speed ratio (TSR) for each test points, corrected for blockage, were readily obtained from data provided by the wind tunnel instrumentation and turbine sensors (e.g. strain gages and encoders). See [9, 10] for the details.
The estimation problem is shown to be difficult and typically ill-posed, because of low observability and collinearity of the unknowns. To overcome this issue, the problem was reformulated according to the singular values decomposition improving dramatically the goodness of the estimates [10].
Fig. 3 and 4 report respectively the agreement between the power and thrust coefficients measured and predicted: triangle symbols indicate the experimental data, dashed lines the outputs of the initial model and solid lines those of the identified model. Also in this case, the estimated model is well correlated with the experimental measures.


Several conclusions can be stated from the results of this work:
• A method for updating and validating high fidelity wind turbine multibody models was developed.
• The proposed methods were tested with the help of two problems of great interest: the estimation of blade structural and aerodynamic properties. Both problems have been solved efficiently.
• The use of constrained optimization and of multiple sources of information greatly improves the identification problem which results to be more robust with respect to the standard and widespread unconstrained problem. As example, the blade mass distribution is now identifiable in a non destructive way.
• The use of the singular values decomposition is essential for solving ill-conditioned or low-observable estimation problem, as that related to the estimation of blade sectional aerodynamic properties.
• The developed procedures can be used effectively by industries for updating and validating their wind turbine models for several applications such that design, certification and active control synthesis.

In terms of possible extensions of the present work, the most interesting one is concerned to the tuning of the properties of other wind turbine sub-models.
In particular, rather than the estimation of tower properties, which might result in an application of the same technique here described, the estimation of the properties characterizing inflow and wake models represents a challenge of great importance. Dealing with the current trend in wind engineering research, having adequate inflow and wake models will be a key aspect in wind farm control or acoustic noise reduction problems.
The full paper will describe in detail both the formulation and the examples relating to structural and aerodynamic properties estimation.

Learning objectives
The learning objectives are:
1. The importance of parameters identification in the design, certification and for wind turbine active control applications.
2. How to perform suitable system identification processes
3. Methods and goals of structural and aerodynamic properties estimation for wind turbine models.

[1] Bottasso CL, Croce A, Savini B, Sirchi W, Trainelli L. Aero-servo-elastic modeling and control of wind turbines using finite element multibody procedures. Multibody System Dynamics 16(3):291–308, 2006.
[2] Ljung L. Perspectives on system identification. In Plenary talk at the Proceedings of the 17th IFAC World Congress, Seoul, South Korea, 2008.
[3] Kerschen G, Worden K, Vakakis AF, Golinval JC. Past, present and future of nonlinear system identification in structural dynamics. Mechanical Systems and Signal Processing, 20(3):505–592, 2006.
[4] Wright AD, Kelley ND, Osgood RM. Validation of a model for a two-bladed flexible rotor system: Pro-gress to date. Technical Report NREL/CP-500-25514, National Renewable Energy Wind Laboratory (NREL), 1617 Cole Boulevard, Golden, Colorado 80401-3393, November 1998. Presented also at AIAA/ASME Wind Energy Symposium. Reno, Nevada. January 11–14 1999.
[5] Jategaonkar RV. Flight Vehicle System Identification: a Time Domain Methodology. AIAA, Progress in Astronautics and Aeronautics, Reston, VA, USA, 2006.
[6] Barclay A, Gill PE, Rosen JB. SQP methods and their application to numerical optimal control. Technical Report NA 97–3, Department of Mathematics, University of California, San Diego, CA, USA, 1997.
[7] Bauchau OA, Bottasso CL, Nikishkov YG. Modeling Rotorcraft Dynamics with Finite Element Multibody Procedures. Mathematics and Computer Modeling 33:1113-1137, 2001.
[8] Bottasso CL, Cacciola S, Croce A. Estimation of blade structural properties from experimental data. Wind Energy 16(4): 501–518, 2013.
[9] Bottasso CL, Campagnolo F, Petrović V. Wind tunnel testing of scaled wind turbine models: beyond aerodynamics. Renewable Energy. (Under review).
[10] Bottasso CL, Cacciola S, Iriarte X. Calibration of Wind Turbine Lifting Line Models from Rotor Loads. Journal of Wind Engineering & Industrial Aerodynamics. (Under review).