Ain Shams Engineering Journal 9 (2018) 3459–3472

Contents lists available at ScienceDirect

Ain Shams Engineering Journal

journal homepage: www.sciencedirect.com

Computational analysis of two-dimensional wing aeroelastic flutter

using Navier-Stokes model
Magdy Saeed Hussin a,⇑, Ashraf Ghorab b, Mohamed A. El-Samanoudy b
a
Arab Organization for Industrialization, Cairo, Egypt
b
Mechanical Power Engineering Dept., Ain-shams University, Cairo, Egypt

a r t i c l e i n f o a b s t r a c t

Article history: All flying vehicles are not stiff enough, and its flexibility can interact with the flow around it to create self-
Received 22 September 2017 sustaining vibrations, so interaction between Computational Fluid Dynamic (CFD) and Computational
Revised 20 January 2018 Structure Dynamic (CSD) need to be considered. In the present work the aero-elastic model has been
Accepted 10 March 2018
implemented and evaluated for symmetrical NACA 0012. This is because of the abundance of section
Available online 11 December 2018
aero-elastic characteristics that leads to validation of the solution model. Navier Stokes Equations were
integrated in time for aerodynamic forces. Then, at each time step forces were coupled with airfoil equa-
Keywords:
tions of motion to numerically simulate its flutter. Compressible N-S equations were discretized using
Flutter
N-S Equations
finite volume method (FVM) and solved by Semi-Implicit Method for Pressure Linked Equations
SIMPLE algorithm (SIMPLE) Algorithm on unstructured grid by creating a MATLAB algorithm. In addition to that numerical
CFD method, classical methods have been used to investigate the flutter and the results have been compared.
Two-dimensional wing Ó 2018 Ain Shams University. Production and hosting by Elsevier B.V. This is an open access article under
the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction ing. Numerical methods of flutter analysis are time marching anal-
ysis and more accurate so, it can be used for analysis in transonic
Studying flutter conditions for a two-dimensional wing in sub- regime. The FVM is strictly conservative so, this method is pre-
sonic regime is the objective of this work because of the essential ferred in CFD. Another advantage of FVM is it can be implemented
role flutter plays in aircraft design. Flutter is a phenomenon that in the physical space on unstructured grid. Finally, by FVM, all
can occur to the structure which is subject to aerodynamic forces, boundary conditions can be easily implemented, since the
when becomes dynamically unstable, due to its extraction of unknown where variables are evaluated at the centroids of the vol-
energy from the air stream [1]. Aero-elasticity is the science that ume elements, not at their boundary faces. To study the structure’s
studies interactions between aerodynamic, stiffness, and inertial response to aerodynamic forces, Newmarks’s technique has been
forces on a structure; flutter is a phenomenon of Aero-elasticity used [3]. The advantage of using Newmark’s method over other
[2]. The analysis here assumed uniform flow along span-wise (infi- methods is that the method can be made unconditionally stable
nite span, and no span-wise bending). Historically methods of flut- and larger time steps can be used with better results.
ter analysis are classical and numerical. Nowadays classical
method usually only used as a first estimation of flutter bound-
aries; classical method has low accuracy especially in transonic 2. Airfoil equation of motion
regime due to limitation of using compressibility correction,
assumption of lift curve slope, and non-linear of structure stiffen- Fig. 1 shows a sketch of the airfoil which is considered as a rigid
section, supported by translational and rotational springs. So the
model is only two degrees of freedom. ’L’ (aerodynamic lift), ’M’
⇑ Corresponding author.
(aerodynamic moment), ’a’ (pitching), ’h’ (heave), ’c’ (chord length),
E-mail address: [email protected] (M.S. Hussin).
’xcg’ (distance between center of mass and center of support), ’Xcp’
Peer review under responsibility of Ain Shams University.
(center of pressure), ’X0’ (support), ’Kh’ (heave stiffness), ’Ka’ (pitch
stiffness), ’Ch’, ’Ca’ (damping), ’m’ (mass), X 0
X cp ¼ lc. Positive
heave is upward and positive moment is nose up.
Deriving the equations of motion by using La Grange’s Equa-
Production and hosting by Elsevier
tions [4–8]:

https://doi.org/10.1016/j.asej.2018.03.003
2090-4479/Ó 2018 Ain Shams University. Production and hosting by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
3460 M.S. Hussin et al. / Ain Shams Engineering Journal 9 (2018) 3459–3472

Fig. 1. NACA0012 airfoil section.

Airfoil kinetic energy ’K’:

1 _2 _ 1
mh
ha_ Sa þ Ia a_
2
K¼ ð1Þ
2 2
’Ia’, mass moment of inertia about support point, Sa ¼ mxcg mass
unbalance
Airfoil potential energy ’P’: Fig. 2.1. A finite volume grid near the airfoil.

1 1
Kh h þ Ka a2
2
P¼ ð2Þ
2 2 the fact that triangular elements are the most flexible shape to
Airfoil dissipative energy ’I’: fit any type of boundary, and easier in calculations (see Figs. 2).

1 _2 1
Ch h þ Ca a_
2
I¼ ð3Þ 3.1.2. Flow governing equations
2 2
By using Lagrange’s equation to get aero-elastic model:
@q

 (€)
 (€)
    Continuity : þ r:ðqvÞ ¼ 0 ð6Þ
m
Sa h Ch 0 h Kh 0 h L @t
þ þ ¼

Sa Ia €
a 0 Ca a_ 0 Ka a M
@
ð4Þ Momentum : ½qv þ r:fqvvg
@t
n o 2
n :: o h n : o ¼ r:flrvg - rP þ r: lðrvÞT
rðlr:vÞ þ f b ð7Þ
½M U þ C U þ ½K fUg ¼ fF g ð5Þ 3

@
U ¼ ½h a ;
T
F ¼ ½L M T Energy : ðqcP TÞ þ r:½qcP vT
@t
DcP DP 2
¼ r:½krT þ qT þ
lW þ lU þ qV
_ ð8Þ
3. Methods of problem solving Dt Dt 3

P
In this work, two methods have been used to investigate and Equation of state : q¼ ðideal gasÞ ð9Þ
predict flutter boundaries for this two dimensional airfoil. The first RT
method based on time domain analysis, where the aerodynamic q; v; T; P; cP ; l; k Are density, velocity, temperature, constant
model coupled with structure model and airfoil oscillations have pressure specific heat, dynamic viscosity, and thermal conductivity
been investigated on time domain at different flow velocities respectively. u; v Velocity components f b ; q_ are body forces and
[9,10]. For that purpose, FVM has been used to integrate and dis- heating source neglected in this problem, deriving these equations
cretize N-S Equation on the physical domain then SIMPLE algo- in [12,13]
rithm was used to get values of primitive variables of flow. After " 
 2  2  2 #
@u @ v @u @ v @v
2
that the pressure was integrated over surface of airfoil to get aero- @u
W¼ þ ; U¼ þ þ2 þ ð10Þ
dynamic forces which feed the structure model. @x @y @y @x @x @y
The second method is based on frequency domain analysis,
which uses classical matrix eigenvalue methods ðA - IkÞx ¼ 0, here
the real values of solution parameter (k) or the damping coeffi- 3.1.3. CFD modeling
cients are plotted against air speed and so flutter occurs at positive Flow governing Eqs. (6)–(9) were discretized by integrating
real value or zero damping respectively. The problem was investi- them over the two dimensional control volume to get three equa-
gated in frequency domain at different stiffness conditions, and all tions in numerical forms: discretized momentum equation, pres-
results were considered and compared with others. sure correction equation, and discretized energy equation. These
equations were solved iteratively by DILU Factorization pre condi-
3.1. Numerical model formulation tioner method. Then, SIMPLE Algorithim was used to get steady
solution [12,14]. SIMPLE Algorithim is implicit pressure-based
3.1.1. Grid generation scheme for N-S Equations. In this modeling FVM was used, where
Unstructured grid was generated in computational domain control volume balance does not limit cell shape; momentum,
around the airfoil [11].The grid was generated on the physical mass, energy conserved even on coarse grids. So, efficient and iter-
space where the flow governing equations were integrated ative solvers well developed. Nowadays FVM has the broadest
directly. The physical domain has been subdivided into suitable applicability (80%). Fig. 3 illustrates procedure of SIMPLE
number of triangles. This selection of triangular shape is due to Algorithm
 M.S. Hussin et al. / Ain Shams Engineering Journal 9 (2018) 3459–3472 3461

Fig. 2.2. A zoomed view of the grid near the leading edge of the airfoil.

Fig. 3. Flow chart of the SIMPLE algorithm.

3.1.4. Structural numerical model time as illustrated in Eqs. (11)–(17) and the solution procedure as
There are many methods used in analyzing response of the shown in the flow chart of Fig. 4. To avoid aliasing in CFD-CSD cou-
structure to aerodynamic forces in time domain. In this work New- pling, the selected time step for coupling should make sampling
mark’s method has been used because it is more stable as shown in frequency more than twice the maximum wave frequency (Harry
Table 1. Eq. (5) is the aero-elastic equation which was integrated in Nyquist theorem). In this problem, natural frequency of pitching
3462 M.S. Hussin et al. / Ain Shams Engineering Journal 9 (2018) 3459–3472

: ::
Table 1 Db
F t ¼ FtþDt
Ft þ a U þb U ð14Þ
Stability of structural dynamic response methods. t t

Method type Stability limits

Central difference method Explicit Conditionally stable DUt ¼ Db b
Ft=K ð15Þ
Forth order Rung Kutta Explicit Conditionally stable
method
The Wilson-h method2 Implicit It is possible to be unconditionally :: 1 1 : 1 ::
stable DU ¼ DU t - U - U ð16Þ
The Newmark-b method Implicit It is possible to be unconditionally
t bDt2 b D t t 2b t

stable
 
: c c : c ::
DU ¼ DU t
U þDt 1
U ð17Þ
t bDt b t 2b t
spring is = 20 Hz, and natural frequency of heave spring =5Hz,
so the minimum period is T = 0.048 s, which means coupling time
: : : ::
step of this problem should be 0.0244 s.
DUt ¼ UtþDt
Ut ; D U ¼ U
U ; D U
: t tþDt t t
n :: o n:o :: F0
C U
KU0 :: ::
½M U þ ½C U þ ½K fU g ¼ fF g; U¼ 0
ð11Þ ¼ U
U; c; b are Newmark parameters
tþDt t
0 M

b ¼KþC c þM 1
K ð12Þ
bDt bDt2 3.2. The classical methods

1 c 1 c These methods are frequency domain analysis where eigenval-

a¼M þC ; b¼M þ Cð
1ÞDt ð13Þ
bDt b 2b 2b ues solutions were used to obtain the natural frequencies and
damping ratios of the system at different airspeeds.

Fig. 4. Flow chart of aero-elastic coupling procedure.

 M.S. Hussin et al. / Ain Shams Engineering Journal 9 (2018) 3459–3472 3463

 ( :: )
   
Table 2 m - Sa h Kh 0 h @CL a
Properties of the airfoil [10]. þ ¼ q1 c ð19Þ

Sa Ia
::
a 0 Ka a @ a lca
Geometry Profile:NACA0012 airfoil (symmetric)
Inertia 2 In free vibration, an initial condition is imposed and motion
m = 51.5 kg, Ia ¼ 2:275 kgm , x0 ¼ 0:4 m,xcg ¼ 0:0429
Stiffness Kh ¼ 50828:463 N=m,Ka ¼ 35923:241 Nm=rad then occurs. The motion takes the form of a non-oscillatory or
Damping 1h ¼ 0:01, oscillatory decay; the latter corresponds to the low values of
1a ¼ 0:01Ca ¼ 5:71 Nms=radCh ¼ 32:358 Ns=m damping normally encountered in aircraft, so we can assume that
Initial h0 ¼ 0:01 m, a0 ¼ 0 rad the motion can take the form of exponential functions h ¼ ho expkt ,
displacements
Initial velocities dh=dtÞ0 ¼ 0:001 m=s, da=dtÞ0 ¼ 0:01 rad=s
a ¼ ao expkt




   
m - Sa Kh 0 @Cl 0
1 h0 0
k2 þ þ q1 c ¼

Sa Ia 0 Ka @ a 0 - ðlcÞ a0 0
3.2.1. Direct eigenvalue
ð20Þ
This method does not include structure or aerodynamic damp-
ing and assumes linear aerodynamic model and linear structural
 

model. Where lift curve slope was prescribed by the classical,
h0 @Cl 0
1
K þ ½A þ k2 ½M ¼ 0; ½A ¼ q1 c
incompressible, non-viscous aerodynamic theory. The method a0 @ a 0 - ðlcÞ ð21Þ
was applied at different stiffness parameters for airfoil section ðAerodynamic stiffness AerodynamicstiffnessÞ
and the results have been considered. The onset of flutter is when
real parts of solution become positive. Deriving equations of the 
det ½K þ ½A þ k2 ½M ¼ 0 ðeigenvalue problemÞ ð22Þ
system as in [5,15,16]:
By solving Eq. (22) flutter velocity of the system can be
@CL @Cl 1
L ¼ q1 c a; M ¼ q1 c 2 al ¼ q1 c2 Cm ; q1 ¼ q V2 ð18Þ obtained.
@a @a 2 1 1 The parameters ’q1’, ’V1’, ’c’, ’Cm’, a are free stream density,
When substituting lift, moment expressions in airfoil Eq. (4) speed of free stream, chord length, lift coefficient, moment coeffi-
and neglecting structure damping get: cient, and angle of attack (pitching angle) respectively

Fig. 5.1. Lift coefficient ’CL’ vs. pitching ’’a’’

Fig. 5.2. Leading edge moment coefficient ’CM’ vs. pitching ’’a’’.
3464 M.S. Hussin et al. / Ain Shams Engineering Journal 9 (2018) 3459–3472

Fig. 5.3. Coefficient of pressure ’Cp’ at a = 0°,

Fig. 5.4. Coefficient of pressure ’Cp’ at a = 2°,

3.2.2. Direct eigenvalue with compressibility correction damping ratio becomes zero. Deriving equations of the system
This method is the same as the previous classical method are as follows:
except the compression equation is included in the system equa-  2 
@CL @CL c
tions to address the error of flow compression especially when L ¼ q1 c aeff M ¼ LðlcÞ
q1 c a_ ð23Þ
Mach > 0.5
@a @ a 16V1
 
1 _ 1 3
aeff ¼ a
hþ c
X0 a_ ð24Þ
3.2.3. P-K method V1 V1 4
In this method, damping from viscous structure and unsteady
Substituting Eqs. (23) and (24) into Eqs. (5) we can get:
aerodynamics are integrated in airfoil equation of motion and " ( ) ( )  
  €
Theodorsen’s function is used to model the changes in amplitude h h  h_ 0
and phase of the unsteady aerodynamic forces [5,17]. Airfoil equa- ½½K þ ½A þ M þ C þ ½CA ¼ ð25Þ
a a€ a_ 0
tion of motion can be put in state space form; state space form can
be solved in time domain or in frequency domain. Here frequency " #
matching was used to get frequencies and damping ratios at differ- @Cl
1
V1

3=4c
X
V1
0

½CA  ¼ q1 c ðAerodynamic dampingÞ

ent air speed of interest. The resulting set: frequency, damping, and @a ðlcÞ
- ðlcÞð3=4c
X0 Þ 2
þ 16V
c
V1 V1 1
air speed have been plotted; the flutter onset speed is when the
ð26Þ
 M.S. Hussin et al. / Ain Shams Engineering Journal 9 (2018) 3459–3472 3465

Fig. 6.1. Time histories of heave motion at the support point of airfoil (free speed 186.5 m/s).

Fig. 6.2. Time histories of twist motion at the support point of airfoil (free speed 186.5 m/s).

n o h iT 3.3. Theodore’s unsteady function

fx1 g ¼ fUg ¼ ½h a , fx2 g ¼ U_ ¼ h_ a_ ; fxg ¼
T
Putting
  h iT n o
x1
¼ h a h_ a_ get fx_ 1 g ¼ fx2 g ¼ U_ Theodorsen’s functions is a complex function C(k) used to
x2 remedy phase lag between airfoil motion and the unsteady aerody-
Eq. (25) Can be redefined using the previous expressions: namic forces [5,17].
2 3 ð2 Þ

0 0 1 0 H1 ðkÞ
CðkÞ ¼ ; Hðn2Þ : Hankel function of second kind
6 7 ð2Þ
H1 ðkÞ þ iH0 ðkÞ
ð2 Þ
fx_ g ¼ 4 0 0 0 1 5fxg
ð27Þ
-1 -1
- ½M ½K þ A - ½M ½C þ CA  ð29Þ
 fx_ g ¼ ½Sfxg ðState spaceÞ
Hðn2Þ ðkÞ ¼ J n ðkÞ
iY n ðkÞ
Assume fx_ g ¼ kfxg ðHankel function formed in Bessel function of first and second kindÞ

det½½S - k½I ¼ 0 ðeigenvalue problemÞ ð28Þ CðkÞ ¼ F ðkÞ þ iGðkÞ ð30Þ

3466 M.S. Hussin et al. / Ain Shams Engineering Journal 9 (2018) 3459–3472

Fig. 6.3. Time histories of aerodynamic lift (free speed 186.5 m/s).

Fig. 7.1. Time histories of heave motion at the support point of airfoil (free speed 185 m/s).

cx
k ¼ 2V 1
, k is the reduced frequency (main parameter), xis the vibrating airfoil has been numerically simulated. The results were
frequency obtained from eigenvalue. generated for unit width and unit chord length of the airfoil. In this
simulation, the airfoil is considered as a rigid section, supported by
translational and rotational springs at a point in its chord. There-
½AModified ¼ CðkÞ½A ðModified aerodynamic stiffnessÞ ð31Þ
fore, only heave and pitch degrees of freedom are permitted at
the point of support, and the spring mass system is subjected to
½CA Modified ¼ CðkÞ½CA  ðModified aerodynamic dampingÞ ð32Þ initial displacements and initial velocities. All calculations were
executed by using Matlab codes. The stiffness and damping prop-
erties of springs were chosen so that flutter instability occurs in
4. Results and discussions the subsonic regime; in classical methods stiffness and damping
values have been doubled to check the response at speeds close
In this work, flutter predictions were presented for symmetric to cruising speed of commercial passenger aircrafts. The airfoil
NACA 0012 cross section which is described in Fig. 1. The interac- properties were uniform along span. The properties of the airfoil
tion of the two-dimensional compressible viscous flow and the are given in Table 2
 M.S. Hussin et al. / Ain Shams Engineering Journal 9 (2018) 3459–3472 3467

Fig. 7.2. Time histories of twist motion at the support point of airfoil (free speed 185 m/s).

Fig. 7.3. Time histories of aerodynamic lift (free speed 185 m/s).

4.1. Steady results 4.2. Time-domain results (CFD-CSD coupling)

The pressure distribution on the airfoil surface and the corre- In this analysis the speed was gradually changed to reach a flut-
sponding coefficient of lift and moment were calculated for differ- ter clear case: simple harmonic motion, self-oscillation case, and
ent angle of attack at Mach = 0.5 as shown in Figs. 5.1 and 5.2. damped case.
These results were obtained to validate the present CFD code with Figs. 6.1 and 6.2 show that when free velocity reaches 186.5 m/
the available results. In Fig. 5.3 at ’a = 00’, the net area enclosed by sec heave and twist angle oscillate unboundedly and system dis-
Cp curve was zero, so no aerodynamic force was acting on airfoil. In placements increase exponentially with time and the system is
Fig. 5.4 at ’a = 20 the stagnation point creeps towards the lower unstable, and so the lift increase unboundedly as illustrated in
surface and the area enclosed by Cp curve increases with increas- Fig. 6.3.
ing angle of attack. Figs. 5.1 and 5.2 also illustrate that lift coeffi- Figs. 7.1 and 7.2 show that when free velocity decreases to
cient and leading edge moment coefficient are nearly linear. reach 185 m/s the system reaches the stable point where both
3468 M.S. Hussin et al. / Ain Shams Engineering Journal 9 (2018) 3459–3472

Fig. 8.1. Time histories of heave motion at the support point of airfoil (free speed 184 m/s).

Fig. 8.2. Time histories of twist motion at the support point of airfoil (free speed 184 m/s).

heave and twist motions are simple harmonic motion and their figures it can be seen that, as the air speed increases, the two fre-
amplitudes remain constant with time, and so lift history is quencies (heave and pitch) approach one another but the damping
bounded as illustrated in Fig. 7.3; the free stream speed at this of both modes remain at zero. The two frequencies become equal
point is called flutter speed of this airfoil. at velocity near 190 m/s. At this point, the heave and pitch modes
Figs. 8.1 and 8.2 show the system response at free speed 184 m/ coalesce and one of the damping ratios is going to be positive and
s.This response illustrates that as the speed of free stream the other is going to be negative. Hence the system becomes unsta-
decreases below the flutter speed as the amplitudes of heave and ble, which is the flutter condition. The Figures also show that, if the
pitch motions oscillation decays with time, and so lift history stiffness of the two springs doubled, the flutter condition will delay
decays with time as illustrated in Fig. 8.3 to reach 268 m/s.
Figs. 10.1 and 10.2 plot free speed versus the imaginary and real
4.3. Frequency-domain results parts of the solution obtained from eigenvalue problem, Eq. (22),
but compressibility correction was included in lift coefficient. On
Figs. 9.1 and 9.2 plot free speed versus the imaginary and real figures it can be seen that flutter start at speed 175 m/s which
parts of the solution obtained from eigenvalue problem, Eq. (22), is less than of previous system. This is expected because of
with lift coefficient without compressibility correction. From these compressibility correction improves the lift coefficient which in
 M.S. Hussin et al. / Ain Shams Engineering Journal 9 (2018) 3459–3472 3469

Fig. 8.3. Time histories of aerodynamic lift (free 184 m/s).

Fig. 9.1. Plot system frequencies vs. free speed for un-damped case (solid line for doubled stiffness).

turn expedite flutter condition. These figures also show that, if the vibration will be damped out, whereas at higher speed, one of
stiffness of the two springs doubled, the flutter condition will delay the damping ratios becomes positive and the other negative. Hence
to reach 230 m/s the system becomes unstable, which is the flutter condition. The
Figs. 11.1 and 11.2 plot free speed versus the imaginary and real Figs. also show that, if the stiffness of the two springs doubled,
parts of the solution obtained from eigenvalue problem Eq. (28) the flutter condition will delay to reach 244 m/s
updated by Theodorsen’s function. The imaginary parts of the roots
indicate the circular frequencies (radians/sec) of heave and pitch,
while the real parts indicate decay or increase of amplitudes with 5. Conclusions
time. From Fig. 11.2, it is clear that, the flutter speed is 189 m/s
which is higher than flutter speed of previous system. This is In the present work, we tried to detect flutter onset by time-
expected because of the presence of structural and aerodynamic domain and frequency-domain methods. The analysis has been
damping. At speeds below this, any initial dynamic structural implemented on symmetrical NACA0012 for specified geometrical
3470 M.S. Hussin et al. / Ain Shams Engineering Journal 9 (2018) 3459–3472

Fig. 9.2. Plot motion amplitude vs. free speed for un-damped case (solid line for doubled stiffness).

Fig. 10.1. Plot system frequencies vs. free speed for un-damped case with compressibility correction (solid line for doubled stiffness).

and structural parameters. In time-domain analysis CFD model and to zero. CFD-CSD technique assumes quasi-steady flow condition,
CSD model were coupled; the response was investigated in time at and the model implemented by MATLAB code. In this CFD model,
different free stream speeds. It was found that at certain speed the computational grid is attached to the airfoil surface and moves
system oscillates steadily (simple harmonic motion) and this speed with it accordingly where pitch angles are assigned to the bound-
is the critical speed. At any free speed higher than critical speed, ary conditions. There is a big difference between flutter condition
system motions increase exponentially (flutter conditions). At free obtained by CFD-CSD model and those obtained by classical meth-
speed less than critical speed, system motions decay and converge ods. This is because classical methods have many assumptions on
 M.S. Hussin et al. / Ain Shams Engineering Journal 9 (2018) 3459–3472 3471

Fig. 10.2. Plot motion amplitude vs. free speed for un-damped case with compressibility correction (solid line for doubled stiffness).

Fig. 11.1. Plot system frequencies vs. free speed for P-K case (solid line for doubled stiffness).

lift coefficient and neglect flow viscosity, so classical method is low tic behavior of aircrafts in transonic regimes [5]. Flutter results of
accuracy especially at high speeds; the highly non-linear nature of present analysis (VFM based N-S 185 m/s) was compared with
the transonic regime results in large variations of the forces and result of [18] (FEM based N-S 184.55 m/s) and there is small differ-
moments with small changes of the aerodynamic configuration. ence which may be the result of different methods of solution.
As a result, classical method does not predict accurately aero elas- Results of classical methods illustrate that increasing stiffness
3472 M.S. Hussin et al. / Ain Shams Engineering Journal 9 (2018) 3459–3472

Fig. 11.2. Plot motion amplitude vs. free speed for P-K case (solid line for doubled stiffness).

Table 3 [9] M.S.U.K.A.I. Akhtar, Modeling the Aerodynamic Lift Produced by Oscillating
Flutter points obtained by the present work. Airfoils at Low Reynolds Number, arxiv, 2015, pp. 1–20.
[10] Chen Xiangying, Zha Ge-Cheng, et al. Numerical simulation of flow induced
Method Flutter velocity vibration based on fully coupled-structural interactions. AIAA; 2004. p. 1–24.
[11] Hoffmann Klaus A, Chiang Steve T. Computational fluid dynamics. 4th
Direct eigenvalue, without damping 190 m/s, 268 m/s
ed. USA: Engineering Education System; 2000.
(2*stiffness)
[12] Moukalled F, Mangani L, Darwish M. The finite volume method in
Direct eigenvalue, without damping, with 175 m/s, 230 m/s computational fluid dynamic. Springer; 2016.
compressibility correction (2*stiffness) [13] Feistauer Miloslav, Felcman Jiri, Straskraba Ivan. Mathematical and
P-K method ,with Theodorsen’s function C(k) 189 m/s, 244 m/s computational methods for compressible flow. Oxford: Clarendon Press; 2003.
(2*stiffness) [14] Hirsch C. Numerical computation of Internal and external flows. John Wiley &
FVM based N-S 185 m/s Sons; 2007.
[15] Raymond L, Bisplinghoff Ashley. Aeroelasticity. Addison-Wesley Publication;
1957.
[16] Mukherjee Sdmenath, Manjuprasad M, Onkar Amit Kumar, et al. Time domain
delay flutter onset; but this is on the expense of weight which is a simulation of airfoil in the subsonic regime using fluid structure coupling
critical factor in aircraft design. The analysis of this work can be through panel method. NAL Struct Technol Division 2008:1–38.
[17] Hodges Dewey H, Alvin Pierce G. Introduction to structural dynamics and
extended to include transonic regime. Flutter points obtained by aero-elasticity. 2nd ed. Cambridge University Press; 2011.
this work are illustrated in Table 3. [18] Onkar Amit Kumar, Arun Kumar A, et al. Flutter prediction of an airfoil using
fluid structure interaction in time domain through finite element method
based Navier-Stokes solver. NAL Technical Memorandum; 2011. TM-ST-11 06.
Acknowledgements

This work is supported by Faculty of Engineering, Ain Shams Magdy Saeed Hussin: Bachelor of Aerospace Engineer-
University, Cairo, Egypt ing, Holds a Diploma in Aviation Engineering, Master of
Mechanical Engineering, Work in Arab Organization for
References Industrialization, Aircraft factory, Cairo, Egypt. I am now
studying my PhD degree in Ain Shams University,
International personal Passport, green
[1] Anumula Mahender Reddy, Daniel Antony A. Numerical simulation of airfoil
vibration. Int J Mech Prod Eng Res Devel 2016;4(6):53–8.
[2] Bisplinghoff Raymond L, Ashley Holt, Halfman Robert L.
Aeroelasticity. Canada: General Publishing Company; 1983.
[3] Géradin Michel, Rixen Daniel J. Mechanical vibrations theory and application
to structural dynamics. John Wiley & Sons; 2015.
[4] Dowwell Earl H et al. A modern course in aero-elasticityKluer. Kluer Academic
Publisher; 2005.
[5] Right Jan R, Cooper Jonathan E. Introduction to aircraft aero-elasticity and Ashraf Ghorab: Bachelor of Mechanical Engineering, Master of Mechanical Engi-
loads. Wiley, John & Sons; 2007. neering, PhD in Fluid, Assistant Professor, Faculty of Engineering, Ain Shams
[6] Alonso Juan J, Jameson Antony. Fully-implicit time-marching aero-elastic University, Cairo, Egypt.
solutions. AIAA; 1994. p. 10–3.
[7] Janarthanan B, Bruce J. Ralphin Rose, Study and Analysis on the Influence of Mohamed A. El-Samanoudy: Bachelor of Mechanical Engineering, Master of
Flutter Frequency on Airplane Stability. IJTARME 2013;2(3):90–6.
Mechanical Engineering, PhD in Fluid, Professor, Faculty of Engineering, Ain Shams
[8] Gharali Kobra, Johnson David A. Dynamic stall simulation of a pitching airfoil
University, Cairo, Egypt.
under unsteady freestream velocity. J Fluid Struct 2013:229–43.