1 RUZGEM 2013 Proceedings of the Conference on Wind Energy Science and Technology RUZGEM 2013 October 3-4, 2013, Ankara, TURKEY RZGM2013-24 A COMPREHENSIVE NUMERICAL MODEL FOR HORIZONTAL AXIS WIND TURBINES AEROELASTICITY Angelo Calabretta 1 Roma Tre University Rome, Italy Marco Molica Colella 2 Roma Tre University Rome, Italy Luca Greco 3 CNR-INSEAN Rome, Italy Giulio Dubbioso 4 CNR-INSEAN Rome, Italy Claudio Testa 5 CNR-INSEAN Rome, Italy Massimo Gennaretti 6 Roma Tre University Rome, Italy ABSTRACT This paper deals with a computational aeroelastic tool aimed at the analysis of performance, response and stability of horizontal axis wind turbines. It couples a nonlinear beam model for blades structural dynamics with an unsteady state-space sectional aerodynamic load model taking into account dynamic stall and inflow effects induced by rotor wake. An extension of 2D static coefficients for high angles of attack is provided to characterize operations in deep stall regime. The Galerkin method is applied to the aeroelastic differential system, with the introduction of a novel approach for the spatial integration of the additional aerodynamic states related to wake vorticity and dynamic stall. Periodic blade responses are determined by a harmonic balance approach and a standard eigenproblem is solved for the stability analysis. Validation of the applied unsteady, sectional aerodynamics model is performed through comparisons with experimental data concerning NACA0012 and S809 airfoil undergoing oscillatory pitch motion. Further, results obtained by the aeroelastic code including dynamic stall modeling applied to the NREL/NASA Ames Phase VI two-bladed rotor in axial flow are presented, with comparisons to available experimental and numerical data. NOMENCLATURE C DOF f = = = Chord length [m] Degrees Of Freedom Point of separation’s function k M = = , reduced frequency Mach number q = , non-dimensional pitch rate s = , non dimensional time 1 PhD Student, Department of Engineering, [email protected]2 Fellow Researcher, PhD, Department of Engineering, [email protected]3 Researcher, CNR-INSEAN Italian Ship Model Basin, [email protected]4 Researcher, CNR-INSEAN Italian Ship Model Basin, [email protected]5 Researcher, CNR-INSEAN Italian Ship Model Basin, [email protected]6 Professor, Department of Engineering, [email protected]
12
Embed
A Comprehensive Numerical Model for Horizontal Axis Wind Turbines Aeroelasticity
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1 RUZGEM 2013
Proceedings of the Conference on Wind Energy Science and Technology RUZGEM 2013
October 3-4, 2013, Ankara, TURKEY
RZGM2013-24
A COMPREHENSIVE NUMERICAL MODEL FOR
HORIZONTAL AXIS WIND TURBINES AEROELASTICITY
Angelo Calabretta1
Roma Tre University Rome, Italy
Marco Molica Colella2
Roma Tre University Rome, Italy
Luca Greco3
CNR-INSEAN Rome, Italy
Giulio Dubbioso4
CNR-INSEAN Rome, Italy
Claudio Testa5
CNR-INSEAN Rome, Italy
Massimo Gennaretti6
Roma Tre University Rome, Italy
ABSTRACT
This paper deals with a computational aeroelastic tool aimed at the analysis of performance,
response and stability of horizontal axis wind turbines. It couples a nonlinear beam model
for blades structural dynamics with an unsteady state-space sectional aerodynamic load
model taking into account dynamic stall and inflow effects induced by rotor wake. An
extension of 2D static coefficients for high angles of attack is provided to characterize operations in deep stall regime. The Galerkin method is applied to the aeroelastic differential
system, with the introduction of a novel approach for the spatial integration of the additional
aerodynamic states related to wake vorticity and dynamic stall. Periodic blade responses are
determined by a harmonic balance approach and a standard eigenproblem is solved for the
stability analysis. Validation of the applied unsteady, sectional aerodynamics model is
performed through comparisons with experimental data concerning NACA0012 and S809
airfoil undergoing oscillatory pitch motion. Further, results obtained by the aeroelastic code
including dynamic stall modeling applied to the NREL/NASA Ames Phase VI two-bladed
rotor in axial flow are presented, with comparisons to available experimental and numerical
x’ = , aerodynamic states derivatives with respect to non-dimensional time
α = Angle of attack [rad]
= Angle of attack at ¾ chord point [rad]
ω = Motion pulsation [rad/s]
INTRODUCTION
This paper presents a comprehensive computational tool for the aeroelastic analysis of horizontal axis
wind turbines. The effectiveness of a wind turbine relies on the capacity of rotor blades to extract energy from
the incoming wind with high levels of reliability, sustainability and safety standards. Hence, multidisciplinary
tools able to examine rotor aerodynamics, aeroelasticity and control strategies with good levels of accuracy at
reduced computational costs are of great interest for wind turbines designers. This issue is particularly crucial for
very large wind turbines and is the subject of many research projects worldwide (see e.g. the EU-FP6 project
UPWIND in [1]).
Key issues for the successful analysis and design of wind turbines are rotor aerodynamics and structural
dynamics modeling, along with the strategies adopted for the spatial/temporal integration of the aeroelastic
system [2]. In order to predict rotor operative equilibrium conditions and stability boundaries, tools able to take
into account rotor unsteady operating conditions induced by both onset wind variations and structural dynamics, are mandatory. The structure of currently applied wind turbine blades is quite stiff, and thus almost all blade
models are based on linear beam theories. However, such an approach cannot take into account blade
geometrical non-linearities. Moreover, as wind turbines size increases and blades get more flexible, the use of
non-linear structural models should be considered.
Concerning aerodynamic simulation, state-of-the-art aeroelastic codes for wind turbine design relies on
the well established engineering methods based on the Blade Element Momentum Theory (BEMT) technique
[3]. It is enhanced through tuned-up corrections such as blade tip flow corrections [4], dynamic inflow [5] and
dynamic stall models [6,7] to take into account three-dimensional flow, spatial/temporal variability of wind
speed and direction, transient loads induced by unsteady wake effects, flow separation and centrifugal effects.
More recently, CFD tools based on RANSE, DES and LES aerodynamics solvers have been proposed to achieve
a physically consistent description of viscosity and turbulence effects on wind turbines blades. Nevertheless the
high computational costs limit their use to the prediction of isolated wind turbines performance and to the
detailed analysis of blades local flow. Thus their application to the aeroelastic analysis for practical preliminary
design use is still a future scenario [8]. Alternative to BEMT models, three-dimensional inviscid-flow methods
allow to describe wind turbines flow at still limited computational costs [9,10]. In this framework, solvers based
on a Boundary Element Method (BEM) may efficiently predict rotor loads and blades wake induced inflow. This
kind of prediction tools have been extensively validated in the past by some of the authors for the analysis of rotorcraft [11] and marine propellers [12], whereas applications to wind and tidal turbines are more recent [13]. A
drawback of these methods is that viscosity effects on blade loads can be only approximately described, thus for
aeroelastic and performance analysis concerning off-design operations, the inclusion of further modeling to
account for unsteady flow separation is necessary. From the structural point of view, state-of-the-art aeroelastic
tools often relies on Finite Element Method (FEM) approaches, which allow the description of complex blades
deformation states. Nevertheless, in the framework of design and optimization applications, modal (spectral)
approaches provide a computationally efficient alternative way to predict the structural dynamics of turbines
within aeroelastic solution tools.
The present work proposes a rotor aeroelastic model where generalized aerodynamic loads are
evaluated through a sectional formulation taking into account unsteady effects due to both shed vorticity and
dynamic stall onset [6,7]. Three-dimensional wake inflow correction is provided by a BEM solver for
incompressible, potential, attached flows. An extension of 2D static coefficients at high blade angles of attack is
presented to characterize operations in deep stall regime. Centrifugal effects are considered by Snel’s correction
[14]. This unsteady aerodynamic model is coupled with a nonlinear, integro-differential formulation for the
structural dynamics analysis of bending-torsional blades subject to moderate deformations. A Galerkin approach
for the spatial integration of the aeroelastic system is applied, using a novel technique for the spectral description
of the additional states appearing in the sectional aerodynamics model to describe wake vorticity and dynamic stall. The time response may be evaluated either by a harmonic-balance technique (suitable for steady operative
condition) or by a time-marching solution algorithm.
In the following, first, structural and aerodynamics models are briefly outlined, then the techniques
applied for numerical integration of the fully coupled aeroelastic system are illustrated. Furthermore, some
results concerning validation performed for the unsteady sectional dynamic stall model with experimental data
(carried out on NACA0012 and S809 airfoils) are presented. Finally, a detailed aeroelastic analysis of the
3 RUZGEM 2013
NREL/NASA Ames Phase VI two-bladed rotor in axial flow [15] is shown, along with comparisons with
available experimental and numerical data.
ROTOR AEROELASTIC MODELING
Rotor Structural Model Beam-like model are applied to describe the structural dynamics of rotor blades; these are based on the
nonlinear bending-torsion formulation described in [16], that is valid for straight, slender, homogenous,
Denoting with , vector state-space B-L model can be written in the form
and the resulting aerodynamic forces are
(1)
where CN and CC are normal and chord force coefficients, CM is the pitching moment coefficient, is the angle
of attack and q is the pitch rate.
Dealing with attached flow conditions, and take into account the effects of shed vorticity from trailing edge, thus being fully equivalent to the circulatory terms of the Theodorsen theory [19], as shown in
Fig.1 where the hystheresis loop of normal force coefficient is presented. The first order ODEs governing these
states are forced by the effective angle of attack at airfoil three-quarters chord point ( ) which can be
expressed as a function of the downwash ( ) as follows
(2)
where VW is the free-stream velocity.
Figure 1. Normal force coefficient on a NACA0012 airfoil undergoing pitch motion (k=0.099):
comparison of B-L and Theodorsen formulations for unsteady attached flow
The next six states reproduce non circulatory loads accounting for flow compressibility. This aspect is
crucial: at very low values of free stream velocity this formulation tends to provide singular values for
4 RUZGEM 2013
aerodynamic loads, due to Mach number appearing at denominator of some terms of Eq. (1) [18]. Under these
conditions, the present formulation may be substituted by a simplified one for incompressible flows where
unsteady loads are described by two additional aerodynamic states providing circulatory loads added to the non
circulatory terms described by the Theodorsen theory. The analysis of normal force coefficient hysteresis loop
for a NACA0012 airfoil undergoing purely harmonic pitch motion between 0° and 20° (Fig.2) shows that for
the original B-L model converges, whereas for lower Mach it is unstable.
(a) M = 0.3 (b) M = 0.0015 (c) M = 0.001
Figure 2. Comparison between B-L model for compressible and incompressible flow
As already mentioned, separated flow conditions are considered through the remaining four aerodynamic
states of the B-L model. Three aerodynamic states are introduced to simulate the effects of trailing edge
separation. Specifically, for an airfoil undergoing unsteady motion such that the angle of attack is close to or
greater than the static stall angle, flow separation onset is modeled through a time lag between separated and
fully attached force coefficients (state . This leads to the definition of a quasi-steady separation point characterized by a time lag with respect to the instantaneous angle of incidence. Progressive flow separation at
trailing edge is governed by the dynamics of the boundary layer which leads to a time lag of the unsteady
separation point (states and ). The last aerodynamic state represents leading edge separation (state )
and models the dynamic stall effect on airfoil unsteady loads [6,18]; this phenomenon occurs when the leading
edge pressure reaches a critical value causing the shedding of a vortex travelling over the airfoil. The lift increase
effect of the travelling vortex on airfoil loads is shown in Fig.3 where the normal force coefficient for a
NACA0012 airfoil undergoing harmonic pitch motion with amplitude 10° around a mean value of 10° is
predicted by B-L model with and without the activation of the state .
Figure 3. Comparison between B-L model and experimental data (k=0.099)
Dealing with rotating blades aerodynamics, it is necessary to extend 2D static aerodynamic coefficients
between the range -180° < α < +180° for operations at high angles of attack that occur in deep stall regimes,
where no experimental data are available [20]; thus, Viterna-Corrigan model [20], that predicts lift and drag
coefficients beyond 2D static polars, is used:
(3)
5 RUZGEM 2013
where
(4)
is the angle of attack where the maximum of lift coefficient is observed, and denote lift and
drag coefficients at stall angle , respectively and is the maximum drag coefficient (at α = 90°). Within
the framework of Viterna-Corrigan stall model, depends on the aspect ratio λ (defined as the ratio
between blade span and mean chord) as follows:
(5)
The case λ > 50 may be interpreted as an infinite blade whose aerodynamic properties are fully described
by 2D aerodynamic coefficients, whereas the case λ 50 also considers 3D (finite span) effects.
It has been observed that, for rotating blades, stall regime starts at higher angles of attack as compared
with wind tunnel measurements on static aerofoils [20]. This has been attributed to centrifugal pumping effects,
reducing the adverse pressure gradient that leads to aerofoil stall. This effect is more evident at the blade root than blade tip. A widely used empirical description of this effects based on the laminar boundary layer theory is
given by Snel [14].
The B-L model is implicitly two-dimensional. Dealing with rotating blades aerodynamics, an
enhancement of this approach is herein proposed: to overcome the limitations in the description of unsteady
attached flow and to take into account flow three-dimensional effects, a modification of the boundary condition
at the airfoil three-quarters chord (downwash) point is introduced. In detail, the perturbation velocity induced by
the rotor wake is evaluated at the rear neutral point of each blade section by a Boundary Element Method. This
approach is based on a boundary integral formulation for potential flows around lifting bodies which is suitable
to describe strong three-dimensional and unsteady flow effects and blade wake shape evolution [11,12,13]. Note
that the shed vorticity generated by blade sections is already taken into account by the unsteady airfoil theory,
thus its contribution to downwash shall be neglected. Following [12], once the solution in terms of velocity
potential on the blade surface is evaluated by the BEM approach, wake induced inflow ( ) can be determined. Finally, the downwash induced by the wake on blade sections is combined with blade motion to obtain the
effective angle of attack as follows
(6)
where , , is the rotor angular velocity and is the blade local twist (Fig.4).
Figure 4. Effective angle of attack at three-quarter chord point
Integration of the Turbine Aeroelastic System Coupling the equations governing blade structural dynamics with the unsteady aerodynamic loads model
yields the aeroelastic integro-partial differential equations to be integrated. The space integration is performed
through the Galerkin approach [21].
6 RUZGEM 2013
For this purpose there is the need of knowing forcing terms distribution in the added states equations. The
method here applied consists of projecting them on the respective trial functions, that are the same as those used
for the structure. Applying the principle of superposition, added states equation ( - ) are decomposed into
bending (in-plane, v and out-of-plane, w since they have same trial function), torsion () and rigid motion contributions, each forced by the corresponding distribution of the forcing terms:
(7)
where
(8)
In this way, all terms are presents and appear only once in the equations and coupling of the aerodynamic
Jacobian matrix with the structural one is straightforward [21]. This approach allows to express dynamic stall
variables through the same set of functions used for the structure.
Remaining states equations ( - ), piecewise continuous due to the action of the separation point
function, f (equal to 1 for attached flow sections and 0 for fully separated flow sections), are solved differently.
Along the rotor blade span a discrete number of characteristic sections are defined, and on each of them the
equations for - are independently integrated; an interpolation between the different sections yields the
radial distribution of states - .
The aeroelastic system can be solved by using the harmonic balance approach as described in [21].
Alternatively, a time marching solution can be obtained by a standard Newmark-β algorithm.
RESULTS
The proposed aeroelastic model is applied to the analysis of the NREL Phase VI wind turbine tested at
NASA Ames wind tunnel. This experimental dataset is described in [15] and, despite the design constraints due
to scientific needs, it is representative of modern commercial wind turbines. In detail, a 10.58 m diameter two-
bladed turbine rotor in axial uniform inflow is considered (upwind S configuration). Blade shape is characterized
by a S809 airfoil with linear chord and non-linear twist spanwise distributions (Fig.5). Moreover, a cylindrical
and a linear transition zones characterize the blade up to 25% of the radius. A constant pitch value of 3° is
considered and the blade cone angle is set to 0°. Further details on the rotor mechanical and geometrical
characteristics can be found in [15] and [22].
Figure 5. CAD rendering of NREL Wind Turbine blade
In the following, validation of the B-L unsteady aerodynamics model is firstly addressed by comparing
present computations with available experimental and numerical data. Specifically, a NACA0012 and the S809
airfoils are considered. Then, a thorough analysis of NREL Phase VI wind turbine is performed.
Validation of B-L model applied to 2D airfoil The B-L model is herein validated through the analysis of a 2D airfoil undergoing harmonic pitch motion
about the quarter-chord point. Numerical results are compared to experimental data [18,23]. Figure 6 presents the
aerodynamic normal force and moment coefficients hysteresis loops for the NACA0012 airfoil [18] for the
motion law where k=0.099. Figure 7 shows similar results for the S809 airfoil [23]
oscillating about a mean angle of 7°, with amplitude of 10° and k=0.077. For both test cases, numerical
predictions show that the proposed implementation of the B-L model reproduces CN and CM magnitude with
reasonable accuracy in the considered range of α. Slight overestimation on CN is observed in the reattachment
7 RUZGEM 2013
phase during the downstroke for the normal force coefficient, whereas major discrepancies appear on CM, even if
the general quality of the numerical data is still reasonable. Greater differences in the hystheresis loop of the
moment coefficient are evidenced on the S809 aerofoil. A devoted tailoring of the semi-empirical coefficients
appearing in the aerodynamic formulation should improve the numerical predictions: the present analysis has
been performed by using the same set of parameters provided by [18] for the NACA airfoil.
Figure 6. Normal force and moment coefficients for the NACA0012 airfoil
Figure 7. Normal force and moment coefficients for the S809 airfoil
3D BEM wake inflow In this section, the wake inflow determined by the BEM solver to evaluate blade effective angle of attack
is examined. Specifically, two different operating conditions are considered = 5 m/s and 8 m/s. For these conditions, Table 1 summarizes BEM predicted turbine thrust and torque along with the corresponding measured
data and the relative percentage errors.
Table 1. Comparison between numerical and experimental data for turbine thrust and torque
The influence of three-dimensional effects on blade loading can be evaluated by the analysis of blade
circulation, which is directly correlated to the intensity of wake vortices. To this aim, Figure 8 shows the
spanwise distribution of circulation for the considered operating conditions. Excluding the irregular trend in the
cylindrical and transition zone of the blade (r/R < 0. 25), blade circulation tends to increase smoothly along the
span highlighting 3D flow effects that rapidly increase near the tip. Moreover, the variation of circulation along the blade increases with wind velocity (and hence with blade loading).
8 RUZGEM 2013
Figure 8. Spanwise distribution of circulation for two different operating conditions
This kind of blade behavior is responsible for the distribution of wake induced axial velocity shown in
Fig. 9: as expected, at higher blade loading greater values of are predicted. Finally, the effective angle of
attack, evaluated by Eq. 6, is shown in Fig. 10.
Figure 9. Axial component of rotor wake induced velocity Figure 10. Blade effective angle of attack
Turbine performance and stability analysis The proposed aerodynamic model is herein used to predict rotor performance. First, a rigid blade
configuration is considered. A crucial issue when wake induced velocity coming from a BEM approach is used
to enhance a 2D aerodynamic solver, is the range of wind speed (and thus of blade sections angle of attack) for
which values can be considered reliable. As a matter of fact, as the wind speed increases, larger portions of the
blades start to face flow separation and the BEM approach provides overestimated airloads (Fig.11) resulting in
unphysical high values of induced velocity. The analysis of BEM predicted rotor performance (and of blade
sections angle of attack, Fig.10) shows that fully attached conditions are achieved only when the wind speed is
less than 8 m/s.
In the present approach, for blade sections experiencing an angle of attack lower than 20°, the sectional aerodynamics formulation is corrected by the wake inflow evaluated through a BEM approach; for blade
sections where the angle of attack exceeds 20°, the 2D aerodynamics coefficients are given by Viterna-Corrigan
formulas for a flat-plate, with the addition of a centrifugal pumping variation according to Snel’s model.
Figure 11. Turbine thrust (left) and torque (right): predictions by BEM compared to experiments
9 RUZGEM 2013
Once the above described tuning process is performed, the proposed model is then used to evaluate rotor
performance. Figure 12 shows the comparison of the proposed approach with measured data and numerical
[24], which is based on a Blade Element Momentum Theory approach and
includes a 3D post-stall correction to airfoil behavior. This figure concerns thrust and torque generated by the
rotor as function of the wind velocity. Compared to BEM solution, for wind speeds above 8 m/s, the present
approach provides reasonable results and closer to experimental ones. In particular, regarding thrust, solution deviates from experimental data in the same way as that obtained from code PROPID, while torque prediction
reproduces experimental data in a better way.
Figure 12. Turbine thrust (left) and torque (right): predictions by present model compared to
Next, blade elasticity effects have been introduced in order to assess their influence on predicted rotor
performance. The corresponding computed thrust and torque are compared with those obtained under rigid-blade
rotor assumption in Fig.13. It demonstrates that, under the examined conditions, the predicted rotor performance
is barely affected by blade deformation effects. This is a consequence of blade stiffness characteristics provided
in [15].
Figure 13. Turbine thrust (left) and torque (right): influence of blade elasticity
Figure 14. Rotor aeroelastic eigenvalues (zoom of the low-damping region is shown in the right picture)
10 RUZGEM 2013
Then, the aeroelastic stability of the turbine has been analyzed. To this aim, first the linearized time-
invariant state-space approximation of the small-perturbation rotor aeroelastic equations (about a reference trim
configuration identified through the harmonic balance approach) has been derived, and then the corresponding
eigenproblem has been solved, for a given set of wind speeds. Figure 14 presents the turbine aeroelastic
eigenvalues predicted in the most critical condition, that corresponds to the maximum wind speed (Vw=25m/s). A
stable response to perturbation is observed (as detailed by the zoom in the right hand side picture), with high modal frequencies due to blade stiffness (indeed, small trim blade tip deformations have been evaluated for
5m/s≤ Vw ≤25m/s, as shown in Fig.15). In Fig.14 the zero-frequency eigenvalues are related to the additional
aerodynamic states, whereas the remaining ones concern the first flap, lag and torsion modes (from lower to
higher frequency).
Figure 15. Blade tip deflections for the reference trim configuration: influence of wind speed
Finally, the time marching response to a perturbation of the complete, nonlinear aeroelastic system has
been carried out. In order to have a comprehensive tool for turbines aeroelastic analysis, such a capability is of
great importance indeed, in that operating conditions might be such to generate non periodic rotor response,
strongly affected by the contribution from nonlinear terms (as it might happen because of gust occurrence or non
uniform wind distribution). Moreover, the parameters appearing in the equations governing some of the
additional aerodynamic states of the B-L model, are characterized by discrete variations. Thus, the transition
from attached to separated conditions may be simulated only through a time marching solution. Figure 16
presents the turbine response to an arbitrary perturbation in terms of the bending and rigid components of the
first additional state. The rigid component, uncoupled with respect to the structural dynamics, shows a first-order
behavior, whereas the bending component shows a damped oscillatory response. The application of the
logarithmic decrement rule to the time signals allows the identification of the damping coefficient that may be
compared with the one determined through the eigenvalue analysis. Table 2 compares the perturbed response
dynamics characteristics as determined from the eigenanalysis and from the time marching solution for state x1. Although not coincident, these appears to be in good agreement. However, note that for the examined test case,
only trailing edge separation occurs, while leading-edge one does not appear: the presence of leading edge
separation would increase the difference between the results from the two stability prediction strategies, thus
making the application of the time marching approach necessary for system stability assessment.
Figure 16. Bending and rigid components time marching solution of added state x1
11 RUZGEM 2013
Table 2. Frequency and damping results of the fully coupled system
Eigenvalues analysis Post processing of time marching solution
Frequency Damping Frequency Damping
=9.13
-2 +11
CONCLUSIONS
A fully coupled aeroelastic tool for the analysis of horizontal axis wind turbines in axial flow is
presented. Specific focus is posed to the rotor aerodynamics model which combines the B-L dynamic stall model
with the induced inflow velocity provided by a three-dimensional Boundary Element Method. A Galerkin
approach is applied for the spatial integration of the aeroelastic system, using a novel approach for the spectral
description of the additional states appearing in the sectional aerodynamic model to describe wake vorticity and
dynamic stall. The numerical analysis conducted has led to the following results:
- The B-L aerodynamic model, as expected, is fully adequate in predicting the unsteady airloads acting on the
two pitching airfoils considered for validation;
- The correction of the 2D aerodynamic solver through a 3D BEM approach to evaluate rotor wake induced
velocity due to trailed vorticity yields quantitatively reasonable turbine performance predictions in the
considered wind range, especially in terms of generated torque. Modeling of blades sections transition from
attached to separated flow conditions remains a critical issue when dealing with performance predictions;
- For the present test case, blades elasticity and deformation do not affect turbine performance. Moreover, the
stability analysis, performed both through a time marching solution of the complete non linear aeroelastic system and by a linearized eigenvalues approach, has indicated a stable behavior of the rotor in the
considered wind range.
REFERENCES
[1] ”UpWind - Design limits and solutions for very large wind turbines”, http://www.upwind.eu, 2011.
[2] Hansen, M.O.L., Sorensen, J.N., Voutsinas, S., Sorensen, N., Madsen, H.Aa., “State of the art in wind
turbine aerodynamics and aeroelasticity”, Progr. in Aerospace Sciences, Vol. 42, no. 4, 2006, pp. 285-330.
[3] Leishman, J.G., “Principles of Helicopter Aerodynamics”, Cambridge University Press, 2006.
[4] Shen, W.Z., Mikkelsen, R., Sørensen, J.N., Bak, C., “Tip loss corrections for wind turbine computations,”
Wind Energy, Vol. 8, no. 4, 2005, pp. 457 – 475.
[5] Leishman, J.G., “Challenges in Modelling the Unsteady Aerodynamics of Wind Turbines,” Wind Energy,
vol. 5, no. 2 – 3, 2002, pp. 85 – 132. [6] Hansen, M.H., Gaunaa, M., and Madsen, H.Aa., “Beddoes-Leishman Type Dynamic Stall Model in State-
Space and Indicial Formulations,” Risoe Technical Report no. 1354(EN), 2004.
[7] Leishman, J.G., and Beddoes, T.S., “A Semi-Empirical Model for Dynamic Stall,” J. Am. Helicopter Soc.,
[16] Hodges D.H., Dowell E.H., “Nonlinear equation for the elastic bending and torsion of twisted nonuniform
rotor blades”, NASA TN D-7818, 1974.
[17] Hodges D.H., Ormiston R.A., “Stability of elastic bending and torsion of uniform cantilever rotor blades
in hover with variable structural coupling”, NASA TN D-8192, 1976.
[18] Leishman, J.G., and Crouse, G.L. Jr., “State-Space Model for Unsteady Airfoil Behavior and Dynamic
Stall,” AIAA Paper 89-1319, 1989, pp. 1372-1383. [19] Theodorsen, T., “General theory of aerodynamic instability and the mechanism of flutter,” NACA Report
496, 1935.
[20] Martinez J., Bernabini L., Probst O., Rodriguez C., “An improbe BEM model for the Power Curve
Prediction of Stall-regulated Wind Turbines”, Wind Energy 2005, 8:385-402.
[21] Bernardini G., Serafini J., Molica Colella M., Gennaretti M., “Analysis of a structural-aerodynamic fully
coupled formulation for aeroelastic response of rotorcraft”, Aerospace science and technology.
[22] Giguère, P. and Selig, M.S., “Design of a tapered and twisted blade for the NREL combined experiment
rotor,” NREL/SR-500-26173, Golden (Colorado), USA, 1999.
[23] Gupta, S., Leishman, J.G., “Dynamic Stall Modelling of the S809 Aerofoil and Comparison with