You are on page 1of 6

Effect of Structural Nonlinearity on the Dynamic

Response of a Coupled Acoustic-Aeroelastic System


Nicholas F. Giannelis 1,* and Gareth A. Vio 1
1

School of Aeronautical, Mechanical and Mechatronic Engineering, Building J11, The University of Sydney, NSW, 2006, Australia
*
nicholas.giannelis@sydney.edu.au

Abstract
Aircraft are inherently exposed to acoustic fields during operation. Various studies in active flutter
suppression have shown that acoustic excitation of an aeroelastic system can be tuned to delay the
presence of the flutter instability. This paper aims to investigate the influence of an acoustic field on the
flutter stability of a simple aeroelastic model. The finite element method is used to couple the acoustic
and aeroelastic systems. Linear flutter analysis is conducted through eigenvalue analysis of a state space
formulation of the aeroelastic equations of motion. Control surface freeplay is modelled and nonlinear
flutter analysis is conducted through numerical integration. In both the linear and nonlinear analyses, the
coupled aero-acoustic system is found to exhibit a flutter velocity on the order of 9% in excess of the
aeroelastic system. Further, the inclusion of acoustic harmonics introduces interesting complexities to the
nonlinear responses of the structural degrees-of-freedom.
Keywords: Nonlinear Aeroelasticity, Acoustics, Flutter

Introduction
Aircraft unavoidably operate in the presence of acoustic fields, and given structural frequencies of
comparable order, aero-acoustic interactions may act to alter the inherent aeroelastic stability of a flight
system. In [1], this phenomenon is examined experimentally, where acoustic excitation is successfully
employed to achieve active utter suppression. A number of studies on acoustic flutter suppression in
supersonic cascades have also yielded promising results [2-4]. More recently, developments in the
formulation of coupled aero-acoustic systems have been seen, with [5, 6] detailing a unified Boundary
Element/Finite Element coupling scheme. In [7], the Finite Element Method is used to couple a simple
aeroelastic system with an acoustic field, with linear flutter analysis revealing improved stability in the
presence of the acoustic medium. Linear aeroelastic analysis is however limited in its ability to model
complex phenomena, and [8] shows that even in simple aeroelastic systems, nonlinear Limit Cycle
Oscillations (LCOs) and bifurcations can present to provide insight to the underlying system stability.
In this paper, the nonlinear aeroelastic response of a simple coupled aero-acoustic system is investigated.
The aeroelastic, acoustic and coupled systems are detailed, followed by the methodology in determining
the linear and nonlinear stability boundaries. The results of the linear flutter analysis are then presented,
along with a discussion of the bifurcation behaviour, time response and restoring force characteristics of
the nonlinear systems. To conclude, the key results are summarised.

Model
Aeroelastic System
The structural model investigated in the present work is an extension of the Hancock wing proposed in
[9]. The 3 degree-of-freedom (DoF) system, illustrated in Figure 1, consists of a rigid wing with flap
angle , twist angle and control surface deflection angle . Bending, torsion and control surface DoFs
are constrained by three discrete springs of stiffness ,
and
respectively. The particular wing
considered exhibits a 2 m chord, 12 m span, aerodynamic centre (ac) at 25% chord, flexural axis ( ) at
40% chord and hinge axis ( ) at 80% chord. The wind-off natural frequencies of the flap, pitch and
control modes are imposed at 7 Hz, 15 Hz and 22 Hz respectively.

16th Australian Aerospace Congress, 23-24 February 2015, Melbourne

Fig. 1: 3 DoF Hancock Model


Although the inertial definitions of the structure pertain from a uniform flat plate, the aerodynamic model
assumes a conventional aerofoil section with an ideal lift curve slope of 2 . Modified strip theory is
employed to model the aerodynamics under a quasi-steady assumption, where the aerodynamics of a
system experiencing variable oscillatory motions are equal to the instantaneous values of an equivalent
system under constant motion. To better approximate the actual flutter mechanism under the quasi-steady
assumption, aerodynamic pitch damping ( ) and control damping ( ) derivatives are also included.
The coupled aeroelastic system is represented by the second order ordinary differential equation:

where

(1)

and
are the aeroelastic inertial, damping and stiffness matrices respectively,
are the physical displacements and and are the velocities and accelerations of
each structural DoF respectively. See [10] for full derivation and definition of the component matrices in
Equation 1, noting that in the present analysis no structural damping is considered.
Acoustic System
The acoustic field is modelled through the finite element method by a single four node quadrilateral
isoparametric element encompassing the planform area of the wing. Analogous to the aeroelastic system,
the acoustic field is represented by the second order differential equation:

(2)

where
and
are the acoustic inertial and stiffness matrices respectively (see [11] for complete
definitions),
is a vector of nodal pressures, is a vector of the second time derivatives of nodal
pressures,
is the fluid density (1.225 kg/m3) and is a spatial coupling matrix defined as:

(3)

where
and
are matrices of the structural and acoustic shape functions respectively,
is the
boundary between the fluid and the structure and is a transformation matrix between the acoustic and
structural degrees of freedom. The matrix of acoustic shape functions is derived from the definition of the
four node isoparametric element (see [12] for a complete derivation). The matrix of structural shape
functions is computed in physical coordinates via modal analysis. The displacement is determined at
each of the acoustic nodes by:
(

where represents the node number of the acoustic element.

16th Australian Aerospace Congress, 23-24 February 2015, Melbourne

(4)

Aeroelastic-Acoustic Coupling
Coupling between the aeroelastic and acoustic systems proceeds following the method described in [11]
and is achieved through the unsymmetric system of equations given by:
[
where

]{ }

]{ }

]{ }

(5)

is the speed of sound in air at sea level (343 m/s).

Linear Flutter
To determine the linear utter stability of the aeroelastic and coupled aero-acoustic systems, a state space
method is employed to determine the frequency and damping of the systems described by Equations 1 and
5 across a range of airspeeds. The systems are transformed to first order form:
[

(6)

where ,
and
are the mass, damping and stiffness matrices respectively of the system under
consideration. For each case, the damping of the structural modes is tracked across the velocity range,
with the flutter condition given by the first crossing of zero damping.
Freeplay Nonlinearity
The effect of a hinge stiffness dead-band on the flutter stability of the aeroelastic and aero-acoustic
systems is determined by including a nonlinear restoring force function (
) given by:
{

(7)

where
is the spring stiffness and
to
defines the freeplay region. The time responses of the
original and acoustic systems are computed through numerical integration under an explicit Runge-Kutta
(4, 5) scheme. To determine the flutter stability no external forcing function is applied, however an initial
condition of +5 control surface deflection is introduced to examine the systems response. The time
responses are computed across an analogous range of airspeeds to the linear analysis, with a time step of
0.0001 s and
points. From the final 20% of the response signal, the amplitude of control surface
deflection is extracted at the zero velocity crossings to determine whether the system is decaying, unstable
or exhibiting LCOs.

Results
Linear Flutter
The frequency and damping for each of the structural degrees of freedom across a range of airspeeds is
given in Figure 2. The results are in qualitative agreement with the findings of [7], where the inclusion of
an acoustic field delays the onset of flutter. A 9.3% increase in stability is observed in the acoustic case,
with flutter speed rising from 323 m/s to 353 m/s. In both the original and acoustic systems, instability is
a product of coalescence in the flap and pitch modes, with the pitch DoF becoming unstable at the flutter
point. It is important to note that due to the inherent stiffness of the structural model, the linear flutter
speed is observed in the supersonic regime. This paper proceeds under the assumption that low speed
aerodynamics remain valid, as the principle concern is the effect of the acoustic field on the structural
dynamics.
The increase in stability observed in the acoustic system can be explained by the introduction of the
acoustic harmonics, which act to shift the wind-off natural frequency of the pitch DoF from 14.3 Hz to
15.9 Hz. The increase in frequency spacing between the fluttering modes necessitates a greater energy
input before the instability occurs. Further, the transition to flutter in the presence of the acoustic field is
more gradual, with the pitch damping of the acoustic system exhibiting a shallower gradient through the
zero damping condition than that of the original system.
16th Australian Aerospace Congress, 23-24 February 2015, Melbourne

Fig. 2: V-g & V- Plots of Linear Systems


Bifurcation Behaviour
The linear flutter analysis is limited in scope due to its inability to capture LCOs. The presence of such
LCOs may act to delay the onset of flutter, and as such, a time domain analysis is performed to compare
the nonlinear responses of the original and acoustic systems. In Figure 3, the bifurcation plots of both
models are given. The presence of the acoustic field yields significant differences in the topology of the
bifurcation branches when compared to the original system. For the original system represented in Figure
3a, the structure exhibits a decaying response to the initial control surface deflection up to the point of the
linear flutter velocity. A single low amplitude LCO is then observed until 348 m/s, where a second
equilibrium branch appears. Beyond 355 m/s two stable low amplitude LCOs persist in the system up to
383 m/s, where apparent period doubling begins. Flutter in the nonlinear original system occurs at 397
m/s, 22.9% greater than the flutter speed predicted through linear analysis.

(a) Original System

(b) Acoustic System

Fig. 3: Bifurcation Diagrams of the Control Surface DoF


In Figure 3b, the bifurcation diagram of the coupled acoustic system is presented. The acoustic system
exhibits very low amplitude LCOs at velocities below the linear flutter speed of 353 m/s, with purely
decaying responses observed from 334 m/s. The amplitude of the acoustic LCOs increases marginally up
to 387 m/s, where the system exhibits a purely decaying response. Above 387 m/s, four stable LCO
branches exist centred about a positive control surface deflection, with a single branch centred about a
negative control deflection equilibrium. The acoustic system further shifts to higher amplitude LCOs
above 405 m/s, with amplitudes increasing up to the flutter point of 435 m/s, 23.2% greater than the linear
flutter speed.

16th Australian Aerospace Congress, 23-24 February 2015, Melbourne

From the bifurcation diagrams of Figure 3, several interesting observations can be made. Where the
original system conveys symmetry about the equilibrium control surface deflection (0), this behaviour is
not evident in the acoustic responses. Under positive initial control deflection, the acoustic LCO orbits
appear to centre about a positive fixed point. This asymmetry is further investigated by performing an
equivalent analysis with negative deflection as an initial condition. In Figure 3a, the topology of the
bifurcation branching remains consistent regardless of the sense of the initial condition. Conversely, in
Figure 3b, the stable orbits now centre about negative control surface deflection, highlighting an increased
sensitivity to initial conditions in the presence of an acoustic field. Nonetheless, the coupled acoustic
system exhibits significantly greater inherent stability, both in the absolute flutter speed and in the
amplitude of oscillations sustainable prior to flutter. The nonlinear flutter velocity of the acoustic system
exceeds that of the original system by 9.6%, whilst prior to flutter a 30 LCO is sustained, 376% larger
than the 6.3 LCO of the aeroelastic model.
The ability of the acoustic system to sustain a variety of stable LCOs at a particular airspeed is further
illustrated in Figure 4. With an initial control deflection of 5 at 390 m/s, the acoustic system is permitted
to settle to a low amplitude LCO. An impulsive force of negative sense is then introduced, and following
the decay of transients, the response is observed to settle to a higher amplitude LCO. The response
highlights the presence of multiple equilibrium orbits in the acoustic system, and the persistence of
stability at velocities well in excess of the flutter point predicted through linear analysis. The particular
attractor to which an orbit is drawn is highly dependent on the initial conditions; however responses may
be forced to orbit about a different attractor, given appropriate excitation.

Fig. 4: Time Response of Acoustic System at 390 m/s


Restoring Force
The increased complexity introduced to the system through the acoustic field is evident in the restoring
force plots of Figure 5, which highlight the restoring force magnitude and characteristics in the nonlinear
DoF of the systems. The dead-band zone is clearly visible in both systems, with no hysteresis arising from
the presence of the acoustic coupling. There is an obvious shift in the centre of oscillation for the acoustic
signal, with attraction to the
fixed point, and repulsion by the
fixed point. The original system
oscillates symmetrically around the
points, as also observed in Figure 3a.
The original system presents a high restoring force for the same angle of oscillation, when compared to
the coupled acoustic system. This can be explained by the extra energy required to energise the higher
order frequency, clearly present in the response of the coupled system. Whereas the orbits of the original
system in Figure 5a are of type period-1, the presence of acoustic harmonics in the coupled system
increases the periodicity of the orbits to period-6, hence producing the complex response of Figure 5b.

Conclusions
In this paper, the influence of an acoustic field on the aeroelastic behaviour of a simple 3 DoF system has
been investigated. Linear analysis finds an increase in flutter speed of 9.3% in the coupled system relative
to the original, a result of the acoustic harmonics increasing the wind-off frequency spacing of the
fluttering modes. With the inclusion of a control hinge freeplay nonlinearity, interesting behaviour is
16th Australian Aerospace Congress, 23-24 February 2015, Melbourne

observed in the acoustic system. A higher sensitivity to initial conditions is present, with apparent
asymmetry in the bifurcation diagram. Multiple LCO branches are also seen to persist at any given
airspeed, and the acoustic system is found to capable of sustaining very high amplitude oscillations prior
to instability. The flutter point of the acoustic system determined through nonlinear analysis is found to
exceed the linear flutter speed by 23.2%, and remains 9.6% greater than the nonlinear flutter speed of the
original system. The presence of an acoustic medium is also shown to significantly increase the
periodicity and complexity of the control surface DoF orbits.

(a) Original System

(b) Acoustic System

Fig. 5: Restoring Force in Control Surface DoF at 340 m/s

Acknowledgements
The authors would like to acknowledge Kok Hwa Yu and Harijono Djojohihardjo for their assistance in
clarifying the foundations and methodology in their work.

References
1.
2.
3.
4.
5.
6.
7.
8.

9.
10.

11.
12.

Huang, X.Y. Active control of aerofoil flutter. AIAA journal, 25(8):1126-1132, 1987.
Nagai, K. and Namba, M. Effect of acoustic control on the flutter boundaries of supersonic cascade. In
Unsteady Aerodynamics and Aeroelasticity of Turbomachines, pages 165-179. Springer, 1998.
Lu, P.J., Pan, D. and Yu, Y.D. Acoustic flutter control of three-dimensional transonic rotor flow. Journal
of propulsion and power, 18(5):1003-1011, 2002.
Lu, P.J. and Chen, S.K. Evaluation of acoustic flutter suppression for cascade in transonic flow. Journal of
engineering for gas turbines and power, 124(1):209-219, 2002.
Djojodihardjo, H. Unified BE-FE aerodynamic-acoustic-structure coupling scheme for acoustic effects on
aeroelastic structures. ICAS Paper, 7(7.5), 2008.
Djojodihardjo, H., Akbar, M., Gunawan, L. and Kusni, M. Unified aerodynamic-acoustic formulation for
aero-acoustic structure coupling. ICAS Paper, 9(10.1), 2010.
Yu, K.H., Djojodihardjo, H. and Kadarman, A.H. Acoustic effects on binary aeroelastic model. IIUM
Engineering Journal, 12(2):123-130, 2011.
Dimitriadis, G. Complete bifurcation behaviour of aeroelastic systems with freeplay. In Proceedings of the
52nd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference, Denver,
Colorado, number 2142. AIAA, 2011.
Hancock, G.J., Wright, J.R. and Simpson, A. On the teaching of the principles of wing flexure-torsion
flutter. Aeronautical Journal, 89(888):285-305, 1985.
Wright, J.R., Wong, J., Cooper, J.E. and Dimitriadis, G. On the use of control surface excitation in flutter
testing. Proceedings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineering,
217(6):317-332, 2003.
Davidsson, P. Structure-acoustic analysis; finite element modelling and reduction methods. PhD thesis,
Lund University, 2004.
Coskuner, L. Combined direct-adjoint approximations for large-scale design-oriented structural acoustics
finite-element analysis. Master's thesis, The University of Washington, 2004.
16th Australian Aerospace Congress, 23-24 February 2015, Melbourne

You might also like