You are on page 1of 6

ENOC 2011, 24-29 July 2011, Rome, Italy

Modeling and experimental identication of an aeroelastic wing with pitch free-play nonlinearity
Rui Vasconcellos , Abdessattar Abdelke , Flavio Marques and Muhammad Hajj Laboratory of Aeroelasticity, University of S o Paulo, S o Carlos, Brazil. a a Department of Engineering Science and Mechanics,MC 0219, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061, USA.
Summary. In this work, the freeplay nonlinearity is approximated by a polynomial expansion and a quasi-steady approximation of the
aerodynamic loads is used. The databases for the identication are generated from analytical and experimental investigations of a rigid airfoil supported by a nonlinear torsional spring. The coefcients of the polynomial expansion representing the freeplay nonlinearity are identied. The normal form of Hopf bifurcation is derived and used to characterize the systems stability.

Introduction
Nonlinear aspects of aeroelastic systems can arise from different sources including aerodynamic nonlinearities, structural deformations and the interaction of the structure with the surrounding ow eld. Additionally, freeplay of moving surfaces or external stores can lead to complex nonlinear responses including limit cycle oscillations (LCO) or chaotic motions [1-4]. The nonlinearities can arise from unsteady aerodynamic sources, large structural deections, and/or partial loss of structural or control integrity [5]. Furthermore, nonlinearities associated with moving surfaces or external stores are inevitable. In these cases, freeplay is one of the most common nonlinearity. Freeplay can arise from worn hinges and loose of attachments which are generally related with aircraft aging. The combination of freeplay nonlinearities with the aerodynamic and structural nonlinearities can lead to changes in the basic response of the system such as changing it from supercritical to subcritical responses. These changes can result in catastrophic damages. To that end, many researchers investigated the effect of freeplay nonlinearity on aeroelastic systems [6-12]. The discontinuous representation of freeplay nonlinearity is quite difcult to analyze mathematically and the inclusion in numerical simulations can be time consuming, since there is a need to locate and integrate at the exact discontinuity point. Consequently, mathematical representations or models that can be used to determine the impact of such nonlinearities on the systems response are helpful. The objective of this work is to determine whether the representation of a freeplay nonlinearity by a polynomial approximation is valid. This is achieved by performing system identication in a pitch-plunge airfoil using data from experiments where freeplay nonlinearity is introduced in the torsional spring. The system identication makes use of pitch-plunge governing equations with quasi-steady approximation of the aerodynamic loads. Once modeled, the normal form that represents the observed LCO is derived from the governing equations and is used to identify the type of bifurcation associated with the freeplay nonlinearity of the experiment. It is shown that the cubic nonlinearity can be used as a representation of the freeplay nonlinearity for the range of motions obtained in this experiment. The normal form of Hopf bifurcation was then derived and the instability was determined to be of the subcritical type.

Experimental setup
The experimental apparatus has been designed to simulate a typical wing section in a two-dimensional incompressible ow. As shown in Fig. 1(a), a freeplay nonlinearity () that is equal to 2.25 degrees is introduced in the pitch motion. Since the typical section is two-dimensional, all parameters of the system are dened per unit span and are assumed to be uniformly distributed. The airfoil is composed of a foam-wood made N ACA0012 rigid wing that is mounted vertically at the 1/4 of the chord from leading edge by using an aluminium shaft. At the edges, the shaft is connected through bearings to the support of the plunge mechanism that is a bi-cantilever beam made of two steel leaf-springs. The torsional pitch stiffness and freeplay mechanism comprises a steel leaf spring inserted tightly into a slot in the main shaft at the top of the wing section. The free end of the leaf spring is placed into a support that allows for freeplay variations. A picture of the experimental setup and a schematic of the freeplay mechanism are presented in Figs. 1(b) and 1(c), respectively. The aeroelastic tests were carried out using an open-circuit blower-like wind tunnel. To induce motions of the wing, an initial displacement in the plunge was applied. This process was repeated starting with low speeds of about 2 m/s. It was observed that all disturbances with the same set of initial displacement would damp at speeds less than 8.4 m/s. Above this speed, limit cycle oscillations (LCO) were observed as shown in Fig. 2. In addition, we observed that LCO could be initiated at lower speeds than 8.4 m/s if larger initial displacements are introduced. This indicates that the system has a subcritical behavior. We have also observed that the pitch angle increases dramatically when the freestream velocity is larger than 9 m/s. Clearly, aerodynamic stall became relatively important. Because the effort here is to determine how freeplay nonlinearity can be identied. The identication is performed over the region up to a freestream velocity of 9 m/s so that aerodynamic nonlinearities are excluded.

ENOC 2011, 24-29 July 2011, Rome, Italy

Torque

0 (deg)

(a)

(b)

(c)

Figure 1: (a) Freeplay nonlinearity respresentation (b) Experimental setup (c) Pitch freeplay mechanism

(a)

(b)

(c)

Figure 2: Experimental time histories for different freestream velocities (a) U = 8.44 m/s, (b) U = 8.62 m/s and (c) U = 8.94 m/s

Governing equations
We model the wing section as an aeroelastic system that is allowed to move with two degrees of freedom namely the, plunge (w) and pitch () motions as shown in Fig. 3. The equations of motion governing this system have been derived in many references [13,14], and are written in the following form
mT m w x b m w x b I w + cw 0 0 c w + kw 0 0 k () w = L M (1)

In the above equations, mT is the total mass of the wing with its support structure, mw is the wing mass alone, I is the

Figure 3: Schematic of an aeroelastic system under uniform airow

mass moment of inertia about the elastic axis, b is the semichord length, x is the nondimensional distance between the center of mass and the elastic axis. Viscous damping forces are described through the coefcients cw and c for plunge and pitch motions, respectively. In addition, L and M are the aerodynamic lift and moment about the elastic axis and are assumed to be given by the quasi-steady representation:
L = U 2 bcl eff M = U 2 b2 cm eff (2) (3)

where U is the freestream velocity and cl and cm are the aerodynamic lift and moment coefcients. The effective angle

ENOC 2011, 24-29 July 2011, Rome, Italy

of attack due to the instantaneous motion of the airfoil is given by.


eff = + h 1 + ( a)b U 2 U (4)

The two spring stiffness for plunge and pitch motions are represented by kw and k , respectively. In this experimental setup, the sole source of nonlinearity is the freeplay pitch nonlinearity. For the analysis, we approximate the freeplay nonlinearity by a polynomial expansion, i.e, we write
k () = k0 + k1 + k2 2 + ... (5)

The equations are expressed in the state space form as:


X = B(U )X + Q(X, X) + C(X, X, X) (6)

where Q(X, X) and C(X, X, X) are, respectively, the quadratic and cubic vector functions of the state variables which describe the freeplay pitch nonlinearity. The matrix B(U ) has a set of four eigenvalues i , i = 1, 2..., 4. These eigenvalues are complex conjugates (2 = 1 and 4 = 3 ). The real parts of these eigenvalues correspond to the damping coefcients and the positive imaginary parts are the coupled frequencies of the aeroelastic system. The solution of the linear part is asymptotically stable if the real parts of the i are negative. In addition, if one of the real parts is positive, the solution of the linearized system is unstable. The speed, Uf , for which one or more eigenvalues have zero real parts, corresponds to the onset of the linear instability and is termed as the utter speed.

Linear model vs experimental results


In this section, we validate the coupled natural frequencies and the utter speed as determined from the analysis against the experimental values. The plotted curves in Fig. 4(a) and 4(b) show the variation of the damping terms and the coupled frequencies as function of the freestream velocity. These plots were obtained using the parameters values obtained from experiment. Using these curves, we determine an analytical value of 8.42 m/s for the utter speed which is close to the measured values between 8.35 m/s and 8.43 m/s. Values of the coupled natural frequencies and the utter frequency as predicted from the linear analysis of Eq. (6) are shown in the second column of Table 1. Experimental values of the coupled frequencies as well as utter frequency are determined and presented in the third column of Table 1. A comparison of the values in the second and third columns of Table 2 shows that the predicted coupled and utter frequencies are within an acceptable margin of the experimental values.
Table 1: Comparison between theoretical and experimental results
Theoretical results 2.69 5.02 4.73 8.42 Experimental results 2.80 5.10 4.98 8.35 < Uf < 8.43

Pitch coupled natural frequency (Hz) at U = 0 m/s Plunge coupled natural frequency (Hz) at U = 0 m/s Flutter frequency (Hz) Uf (m/s)

20

2 0 Re i 2 Im i

4
Magnitude (dB)

10

2 0 2

10

20

4 4 0 2 4 6 U 8 10 12 14 0 2 4 6 U 8 10 12 14

30

40

50 0

10

15

20

25

30

35

40

45

50

frequency (Hz)

(a)

(b)

(c)

Figure 4: Variation of (a) the damping terms (b) the coupled frequencies as function of the freestream velocity and (c) Experimental power spectrum when U = 8.62 m/s

Identication of the nonlinear torsional spring coefcients


As discussed before, the freeplay nonlinearity is approximated by a polynomial expansion that includes quadratic, cubic and higher order terms. To identify the terms in this approximation that are mostly associated with the freeplay nonlinearity, we present in Fig. 4(c) the power spectrum of the measured pitch motion when U = 8.62m/s. The spectrum shows peaks at the response frequency of f = 5.85 Hz and its third superharmonic and subharmonic.

ENOC 2011, 24-29 July 2011, Rome, Italy

1 1 The presence of these peaks at 3f and 3 f and absence of strong peaks at 2f and 2 f indicate that of all terms in the expansion the cubic term is the most relevant. As such, only cubic nonlinearity, k2 , is considered to model the freeplay nonlinearity and determine the associated type of bifurcation using the normal form. Fig. 5(a) shows the variation of the pitch angle as a function of the cubic nonlinear term k2 when the freestream velocity U is set to 8.62 m/s. We note that increasing the coefcient of the cubic nonlinearity results in a decrease of the amplitude of the pitch angle. Therefore, we can easily identify the value of this coefcient to obtain the experimentally measured value of the pitch amplitude. Using this gure, it is concluded that this coefcient should be equal to 0.11 N m to obtain pitch angle of 8.5 deg.
14 12 10 deg 8 6 4 2 0

20 15 deg 10 5 0
0.2 0.4 k2 0.6 0.8 1.0

10 5 deg 0 5 8.2 8.4 8.6 U ms 8.8 10 0.0 0.2 0.4 ts 0.6 0.8 1.0

(a)

(b)

(c)

10 5 deg 0 5 10 0.00 0.05 0.10 0.15 0.20 0.25 0.30 ts


(d)

Figure 5: (a) Variation of the pitch angle as function of the cubic nonlinear coefcient k2 when U = 8.62 m/s, (b) Variation of the pitch angle with freestream velocity for k2 =0.11 N m and (c,d) Time histories of the pitch angle for k2 =0.11 N m and U = 8.62 m/s: experimental data in red, numerical simuluations in black.

Figure 5(b) shows the variation of the pitch angle as a function of the freestream velocity when k2 = 0.11 N m. The solid curve presents the analytical results while the red points show the measured values. Figs. 5(c) and 5(d) show the time histories of the pitch angle for one second and 0.3 s, respectively, when U = 8.62 m/s. We note that there is a slight shift in the frequency between the experimental and theoretical results. This behavior is associated with a change in the pitch stiffness that is caused by higher angles of attack which stressed the spring wire as discussed by Conner et al [9].

Nonlinear normal form: Subcritical behavior


In this section, we determine the type of bifurcation that is associated with the utter of the wing. For this, we compute the normal form of the Hopf bifurcation of the aeroelastic system near the utter speed Uf . To derive the nonlinear normal form, we rst add a perturbation term, 2 U Uf , to the utter speed (U = Uf + U 2 Uf ) which leads to the appearance of the secular terms at the third order. Taking account of the perturbation term of the utter speed, the matrix B(U ) is written as the sum B(Uf ) + 2 B1 (Uf ) where
0 0 B1 (Uf ) = 0 0 0 d1 Uf 0 d3 Uf
2

0 2 2k1 Uf 0 2 2k2 Uf

0 d2 Uf 0 d4 Uf

Eq. (6) is then written as


X = B(Uf )X + U B1 (Uf )X + C(X, X, X) 0 C2 C= 0 C4 (7)

Where

Letting P be the matrix whose columns are the eigenvectors of the matrix corresponding to the eigenvalues j1 1 and j2 of B(Uf ), we dene a new vector Y such that X = P Y and rewrite Eq. (7) as

ENOC 2011, 24-29 July 2011, Rome, Italy

P Y = B(Uf )P Y +

U B1 (Uf )P Y + C(P Y)

(8)

Multiplying Eq. (8) from the left with the inverse P

of P , we obtain
U KY + P 1 C(Y) (9)

Y = JY +
1

where J = P B(Uf )P is a diagonal matrix whose elements are the eigenvalues j1 1 and j2 of B(Uf ) and K = P 1 B1 (Uf )P . We note that Y2 = Y1 , Y4 = Y3 , and hence Eq. (9) can be written in component form as
5

Y1 = j1 Y1 1 Y1 + Y3 = j2 Y3 +

U
1 5

K1i Yi + N1 (Y, Y, Y)

(10)

U
1

K3i Yi + N3 (Y, Y, Y)

(11)

where the Ni (Y, Y, Y) are tri-linear functions of the Y. Because we considered only cubic nonlinearities, the solution of Y1 decays to zero. Consequently, we retain only the non-decaying solution (Y3 ). Moreover, to compute the normal form of the Hopf bifurcation of Eqs. (10 and 11) near U = Uf , we follow Nayfeh and Balachandar[15] and search for a third-order approximate solution of Eq. (11) in the following form
Y3 = Y31 (T0 , T2 ) +
2

Y32 (T0 , T2 ) +

Y33 (T0 , T2 ) + O( 4 )

(12)

where Tn =

t. In terms of the Ti , the time derivative can be expressed as


d = + dt T0
2

= D0 + T2

D2

(13)

Substituting Eqs. (12 and 13) into Eq.(11). These new equations must hold for all values of , therefore the coefcients of like powers of must satisfy these new equations. This leads to two different set of relations corresponding to and 3 as follows Order( )
D0 Y31 j2 Y31 = 0 (14)

Order( )
D0 Y33 j2 Y33 = D2 Y31 + U (K33 Y31 ) + N (Y31 , Y31 , Y31 ) + cc + N ST (15)

where N ST stands for terms that do not produce secular terms and cc stands for the complex conjugate of the preceding terms. The solution of Eq. (14) can be expressed as
Y31 = A(T2 )ej2 T0 D2 A = U K33 A + e A2 A (16)

Substituting Eq. (16) into Eq. (15), eliminating the terms that lead to secular terms, we obtain the modulation equation
(17)

The effects of the cubic nonlinearity (k2 ) on the system are expressed through the expression of e . For convenience, we write Eq. (17) as
D2 A = A + e A2 A (18)

where = U K33 . Letting A(T2 ) = 1 aei(T2 ) and separating the real and imaginary parts in Eq. (18), we obtain the following nonlinear 2 normal form of Hopf bifurcation:
a = r a + = i + 1 er a3 4 1 ei a2 4 (19) (20)

where a is the amplitude and is the shifting angle of the periodic solution. Eq. (19) has generally three equilibrium solutions which are:
a = 0, a = 4r er

a = 0 which corresponds to the xed points (0,0). The other two solutions are the nontrivial ones. The origin is asymptotically stable for r < 0 or r = 0 and er < 0 , unstable for r > 0 or r = 0 and er > 0. For the nontrivial solutions, they exist when r er < 0. Furthermore, they are stable (supercritical Hopf bifurcation) for r > 0 and er < 0 and unstable (subcritical Hopf bifurcation) for r < 0 and er > 0. Upon substituting the physical parameters, the expressions of the linear and nonlinear coefcients of the normal form are given by
r = 1.19U er = 0.00166k2 (21)

ENOC 2011, 24-29 July 2011, Rome, Italy

Based on the numerical integrations, the cubic nonlinear coefcient k2 is set equal to 0.11 N m. Consequently, because (er > 0), this system has a subcritical instability. This result conrms the experimental observation of the existence of (LCO) for larger initial displacements at freestream velocities that are smaller than the utter speed. The ability to model and identify the type of bifurcation associated with the freeplay nonlinearity provides a capability that can be used to avoid subcritical responses such as the one observed in these experiments.

Conclusion
In this work, the identication of pitch freeplay nonlinearity by polynomial expansion in an aeroelastic system was performed . The system is composed of a pitch and plunge airfoil with nonlinear torsional spring. The databases for the identication are generated from experimental investigations. Once modeled, a numerical integration is performed to validate the model and to identify the cubic term of the expansion. It is shown that the cubic nonlinearity can be used as a representation of the freeplay nonlinearity for the motions presented in the experiment. The normal form of the representative model was then derived and it was determined that the ensuing Hopf bifurcation is of the subcritical type. Although the representation of freeplay nonlinearity by a polynomial expansion is valid here, this model may not be accurate under other conditions.

Acknowledgment
The authors acknowledge the nancial support of the State of S o Paulo Research Agency (FAPESP), Brazil (grant a 2007/08459-1) and the Coordination for the Improvement of Higher Education Personnel (CAPES), Brazil (grant 0205109).

References
[1] E.H. Dowell, D. Tang, Nonlinear aeroelasticity and unsteady aerodynamics, AIAA Journal 40 (2002) 1697-1707. [2] H.C. Gilliat, T.W. Strganac, A.J. Kurdila, An investigation of internal resonance in aeroelastic systems, Nonlinear Dynamics 31 (2003) 1-22. [3] A. Raghothama, S. Narayanan, Non-linear dynamics of a two-dimensional air foil by incremental harmonic balance method, Journal of Sound and Vibration 226 (1999) 493-517. [4] L. Liu, Y.S. Wong, B.H.K. Lee, Application of the centre manifold theory in non-linear aeroelasticity, Journal of Sound and Vibration 234 (2000) 641-659. [5] T. ONeil, Nonlinear aeroelastic response - analyses and experiments, 34th AIAA, Aerospace Sciences Meeting and Exhibit, 1996. [6] M.D. Conner, Nonlinear aeroelasticity of an airfoil section with control surface freeplay, PhD thesis, Department of Mechanical Engineering and Materials Science, Duke University (1996). [7] D.S. Woolston, An investigation of effects of certain types of structural nonlinearities on wing and control surface utter, Journal of the Aeronautical Sciences, (1957) 9936-9956. [8] B.H.K. Lee, S. Price, E.H. Dowell, Y. Wong, Nonlinear aeroelastic analysis of airfoils: bifurcation and chaos, Progress in Aerospace Sciences 35 (1999 ) 205-334. [9] M.D. Conner, D.M. Tang, E.H. Dowell, L.N. Virgin, Nonlinear behavior of a typical airfoil section with control surface freeplay: a numerical and experimental study, Journal of Fluids and Structures 11 (1997) 89-109. [10] R.M.G. Vasconcellos, F.D. Marques, M.H. Hajj, Time series analysis and identication of a nonlinear aeroelastic section, In 51stAIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference, Orlando, FL, April 12-15, 2010. [11] L. Daochun, G. Shijun, X. Jinwu, Aeroelastic dynamic response and control of an airfoil section with control surface nonlinearities, Journal of Sound and Vibration 239 (2010) 4756-4771. [12] Y.C. Fung, An introduction to the theory of aeroelasticity, Wiley NY, 1955. [13] E.H. Dowell, A modern course in aeroelasticity, Kluwer, Dordrecht, 1995. [14] T.W. Strganac, J. Ko, D.E. Thompson, A.J. Kurdila, Identication and control of limit cycle oscillations in aeroelastic systems, Proceedings of the 40th AIAA/ASME/ASCE/AHS/ASC Structrures, Structural Dynamics, and Materials Conference and Exhibit, AIAA Paper No. 99-1463, 1999. [15] A.H. Nayfeh, B. Balachandran, Applied nonlinear dynamics, Wiley series in nonlinear science, NY, 1994.

You might also like