We Are Intechopen, The World'S Leading Publisher of Open Access Books Built by Scientists, For Scientists
We Are Intechopen, The World'S Leading Publisher of Open Access Books Built by Scientists, For Scientists
6,800
Open access books available
183,000
International authors and editors
200M Downloads
154
Countries delivered to
TOP 1%
most cited scientists
12.2%
Contributors from top 500 universities
Research on Aeroelasticity
Phenomenon in Aeronautical
Engineering
Hoang Thi Kim Dung and Nguyen Phu Khanh
Abstract
1. Introduction
1
Aerodynamics
2
Research on Aeroelasticity Phenomenon in Aeronautical Engineering
DOI: http://dx.doi.org/10.5772/intechopen.91748
structure at a low speed with four different wing models such as two rectangular
and two trapezoid 3D-shape wings, each 3D-shape wing had symmetric and asym-
metric airfoil, respectively.
The subsonic aeroelastic stability of a two-dimensional panel resting on a
continuous elastic foundation was investigated in Ref. [12]. Tests were conducted
experimentally on a 104 24 0.018 in. rectangular aluminum panel in a low-
speed wind tunnel. Comparison of experiment and theory showed a good agree-
ment in flutter speed and wavelength but poor agreement in wave speed and
frequency at flutter. This discrepancy was attributed to the limitations in the test
setup as well as to the general difficulty of predicting the wave speed and frequency
as accurately as the flutter speed. Reference [13] tested the first AGARD standard
aeroelastic configuration for dynamic response, 445.6 wing, in the 16 foot transonic
dynamics tunnel at the National Aeronautics and Space Administration (NASA)
Langley Research Center. Several models of the wing were tested in the transonic
dynamic tunnel including full-span and semispan models over a range of Mach
number from 0.338 to 1.141. The NASA conducted experiments in wind tunnel to
estimate the aeroelastic characteristics of new and advanced flight vehicles, includ-
ing fixed-wing, rotary-wing, and space-launch configurations. Reviews and assess-
ments were made regarding available facilities, measurement techniques, and other
means and devices useful in testing. The needs and requirements for advances
and improvements in testing capabilities for future experimental research and
development programs are described [14].
In [15], aeroelastic concepts for increased aircraft performance were mentioned.
Active aeroelastic concepts as well as robust analysis concepts aiming at efficient
analysis were carried out using numerical models with uncertain or varying model
parameters. A high aspect ratio wing in wind tunnel testing conditions was consid-
ered for exploitation of fluid-structure interaction of active aeroelastic structures.
The structural flexibility was exploited by using multiple control surfaces such that
the deformed wing shape gives minimum drag for different flight conditions. Two
different drag minimization methods were carried out: one was to reduce induced
drag based on numerical optimization techniques, and another was to reduce
measured total drag using real-time optimization in the wind tunnel experiment.
An approach for the prediction of dynamic modal transient response and flutter
characteristics of structures with unknown system parameters, such as stiffness and
mass, using experimental modal parameters was remarked in [16]. A finite element
model was created by using the actual material properties of the structure to study
the correlation of the results. The computed transient responses and flutter veloci-
ties by the proposed method using experimental modal parameters observed that
material properties were not a prerequisite.
In Ref. [17], transonic flutter characteristics of AGARD 445.6 wing between the
numerical method [5] and the experimental data of NASA’s experiment were com-
pared [13]. The comparison provided the basis for developing the numerical setup
and the experimental setup to check out subsonic flutter characteristics of some
simple wing structures (e.g., a thin plate). Aeroelasticity on airplane wing, which
had supercritical airfoil, was carried out in [18]. The model wings were made from
different materials and dimensions. Hence, varied wing structures were created to
accomplish a comprehensive analysis, and the flutter velocity was also restricted to
appropriate values within the working range of experiment devices. At the same
time, infinite element method with the help of the ANSYS software was also
conducted to simulate the phenomena on the same model wings as a verification for
the precision of the experimental models.
3
Aerodynamics
ð1Þ
ð2Þ
where:
w: structural displacement at any time instant and position.
q: generalized displacement vector.
[M]: generalized mass matrice.
[C]: damping matrice.
[K]: stiffness matrices.
ϕ: normal modes of the structure.
N: total number of modes of the structure.
F: generalized force vector, which is responsible for linking the unsteady aero-
dynamics and inertial loads with the structural dynamics.
Eq. (1) shows that there are distinct terms representing the structure, aerody-
namic, and dynamic disciplines. This equation was solved numerically by integrated
CFD-CSD tool on the ANSYS software. This method was also called the two-way
fluid-solid interaction (FSI). Aerodynamic loads were first calculated by CFD
solver. Then, these loads were used to calculate the structural response of the wing
structure through the fluid-solid interface. By using CSD solver, the structural
deflection was estimated, and the mesh in each time step was deformed. The
simulation of the two-way FSI is presented in Figure 1 [10].
Figure 1.
Two-way FSI algorithm.
4
Research on Aeroelasticity Phenomenon in Aeronautical Engineering
DOI: http://dx.doi.org/10.5772/intechopen.91748
Following [8], the first four modes of vibration were sufficient to accurately
model the wing structure response. So, in order to determine the time step of the
unsteady problem, a modal method was applied to estimate first the natural fre-
quency of the first four modes and then calculate the time step using the following
formula:
ð3Þ
where:
Δt: time step
f: natural frequency
The flutter velocity was estimated from the vibration of the wing tip position,
the most dangerous position of the wing [5, 8, 9, 13]. From the variation of this
position, the damping coefficient except the influence of structural damping was
measured. It means that there was an effect of aerodynamic damping coefficient on
the vibration of the wing structure. The aerodynamic damping coefficient was
calculated as follows:
ð4Þ
where:
Xi: ith peak of vibration.
n: number of periods.
ζ: aerodynamic damping coefficient.
The damping coefficient increased/decreased when the air velocity decreased/
increased. The type of vibration was distinguished from the value of damping
coefficient:
• ζ > 0: The vibration was convergent. The wing structure was stable.
• ζ < 0: The vibration was divergent. The wing structure was unstable.
• ζ = 0: The vibration was harmonic oscillation. The wing was in critical state.
The air velocity was in the flutter velocity.
AGARD 445.6 wing with aspect ratio of 1.65, taper ratio of 0.66, and 45o sweep
angle at quarter chord line was studied as seen in Figure 2. The cross section of the
wing was NACA65A004 airfoil in the stream-wise direction. This NACA65A004
airfoil was a symmetric airfoil with a maximum thickness of 4% of the local chord.
The dimensions of the wing were root chord of 0.558 m, tip chord of 0.368 m, and
semispan of 0.762 m. The wing model used in aeroelastic experiments [13] was
constructed by laminated mahogany which was modeled as an orthotropic material
with different material properties in different directions. The properties of the
laminated mahogany are given in Table 1. The modal analysis was performed using
mechanical APDLs solver to evaluate the accuracy of the constructed model.
The AGARD 445.6 wing was modeled at a zero attack angle and at altitudes of
9.65 and 14 km, the same conditions of experimental study in [13]. The wing was
5
Aerodynamics
Figure 2.
Semispan AGARD 445.6 wing model.
Table 1.
Mechanical properties for the weakened AGARD 445.6 wing.
Figure 3.
Computational grids. (a) AGARD 445.6 wing grids. (b) Fluid computational grids.
meshed in 9257 nodes and 1350 elements (Figure 3a). The fluid domain in CFD
problem was meshed in 67,949 nodes and 279,535 elements (Figure 3b).
The free-stream air velocity was from 0.29 to 0.59 M at an altitude of 9.65 km
and from 0.47 M to 0.73 M at an altitude of 14 km.
2.3 Results
The mode shapes are obtained from the finite element analysis of the modeled
wing. The deflection contours between the modal analysis and experiment [13]
were compared as shown in Figure 4. The natural frequencies between the devel-
oped solution, experiment [13], and other researches were also compared as shown
in Table 2. It could be concluded that the obtained results were in good agreement
6
Research on Aeroelasticity Phenomenon in Aeronautical Engineering
DOI: http://dx.doi.org/10.5772/intechopen.91748
Figure 4.
Comparison of the mode shapes. (a) Mode 1. (b) Mode 2. (c) Mode 3. (d) Mode 4.
with the experimental results in [4, 6, 13] within an error relative of 8%. The
frequency of the first mode was around 9.96 Hz. Following Eq. (3), the time step of
the unsteady problem was estimated about 0.005 s.
At altitude 9.65 km, the Mach number of air velocity was varied from 0.29 to
0.59 (M = 0.29; 0.35; 0.41; 0.47; 0.53; 0.59). From Eq. (4), the aerodynamic
damping coefficient was calculated and presented in Figure 5a. The zero damping
7
Aerodynamics
coefficients were interpolated at Mach number 0.46. While the experimental zero
damping was 0.499 in [13], the difference between simulation results and results of
[13] was about 8%.
Table 2.
Natural frequencies of modal analysis.
Figure 5.
Aerodynamic damping coefficients. (a) At altitude 9.65 km. (b) At altitude 14 km.
8
Research on Aeroelasticity Phenomenon in Aeronautical Engineering
DOI: http://dx.doi.org/10.5772/intechopen.91748
At altitude 14 km, the Mach numbers of air velocity were varied from 0.47 to
0.73 (M = 0.47; 0.53; 0.59; 0.65; 0.67; 0.73). From Eq. (4), the aerodynamic
damping coefficient was calculated and presented in Figure 5b. The zero damping
coefficients were interpolated at Mach number 0.58. While the experimental zero
damping was 0.678 in [13], the difference within simulation results was about 14%.
Figure 6.
Vibration of wing at altitude H = 9.65 km. (a) M = 0.29 – ζ = 0.0071. (b) M = 0.47 – ζ = 0.000125.
(c) M = 0.53 – ζ = 0.006.
9
Aerodynamics
In both the two considered altitudes (9.65 and 14 km), the numerical results
agreed well with the experimental results of [13] with a relative error about 14%.
This difference would be from the computation such as the quality of mesh and
order of model in CFD and CSD.
For more details of the stability of the wing structure, the vibrated value of the
wing tip position at three Mach numbers were plotted as shown in Figure 5.
At a Mach number smaller than the flutter value, the damping coefficient was
positive, and the wing was stable (Figure 6a).
At a Mach number near the flutter value, the damping coefficient was zero, and
the vibration was harmonic oscillation (Figure 6b).
At a Mach number greater than the flutter value, the damping coefficient was
negative, and the vibration was divergent (Figure 6c). This divergent vibration
Figure 7.
Comparison of Cp distribution at M = 1.141, α = 0°. (a) 26 % semispan. (b) 75.5 % semispan.
10
Research on Aeroelasticity Phenomenon in Aeronautical Engineering
DOI: http://dx.doi.org/10.5772/intechopen.91748
would create the damage of the wing such as loss of control for the flap, aileron,
fracture of wing, etc.
Figure 8.
Wing-tip oscillation depends on the selected time step.
Figure 9.
Wing-tip oscillation of flutter point at M = 0.499.
11
Aerodynamics
Table 3.
Flutter characteristics at M = 0.499.
This remark illustrated that the developed solution could be used to specify the
transonic flutter characteristics with errors less than 10%.
3. Experimental method
The test model was set in AF6116 (M = 0.1) subsonic wind tunnel located at
the Hanoi University of Science and Technology, which was of a blowdown type
with a closed test section (0.4 0.5 1.0 m3). The wind speed could be arbitrarily
varied up to 30 m/s, where the Reynolds number based on wing root chord was 106,
which was driven by an 8 kW electric motor.
Flutter characteristics were determined with the help of the frequency meter
and load cell, which allowed to specify the flutter frequency and root wing force,
respectively. The oscillated frequency was measured by the DT-2234C+ frequency
meter. The signal of measured frequency was averaged by five measurements. The
force applied to the wing was measured by load cell system. In the experimental
aeroelastic analysis, the flutter frequency and the flutter amplitude were measured
at different velocities ranging from 10 to 30 m/s using an oscillator generator
system.
Two wing models with the parameter and dimension are shown in Figure 10
and Table 4. The non-structure wing had only balsa wood, while the structure wing
had balsa wood for skin and carbon rod and hard wood for the inner parts.
3.3 Results
12
Research on Aeroelasticity Phenomenon in Aeronautical Engineering
DOI: http://dx.doi.org/10.5772/intechopen.91748
Figure 10.
Experimental models. (a) Wing model. (b) Wing with support. (c). Broken wing.
Table 4.
Wing models.
Attack angle (0) Velocity of first oscillation (m/s) Flutter velocity (m/s) Frequency (Hz)
0 20 22 37.38
Table 5.
Flutter characteristics at different attack angles—Non-structure wing.
Figure 11.
Force at the wing root of the non-structure wing—Attack angle 10°.
13
Aerodynamics
velocity of non-structure wing was found at 19.5 m/s. The load at the wing root of
this wing at flutter velocity is shown in Figure 12. The maximum force was 5.44 N,
and the minimum force was 4.32 N.
With structure wing, different modes of vibration appeared depending on the
characteristics of the structure as remarked in [8]. With the help of the oscillator
generator system, the specific oscillation frequencies of the first four modes were
estimated as shown in Table 6.
Figure 12.
Force at the wing root of the non-structure wing—Attack angle 10° and velocity 19.5 m/s.
1 24.3 23.5
2 67.0 63.0
3 103.6 115.3
4 132.0 139.0
Table 6.
Specific oscillation frequency of the non-structure and structure wings.
Figure 13.
Amplitude of oscillation of the structure wing—Mode 1.
14
Research on Aeroelasticity Phenomenon in Aeronautical Engineering
DOI: http://dx.doi.org/10.5772/intechopen.91748
The frequencies of the first and second modes of the non-structure wing were
higher than those of the structure wing, while the frequencies of third and fourth
mode of structure wing were higher than those of non-structure wing (Table 6).
Considering both wings in the first mode of oscillation, when the force was applied
to the wing, the amplitude of the non-structure wing was higher than that of the
structure wing (Figure 13). In conclusion, the non-structure wing was easier to
resonate than the structure wing.
4. IBM method
4.1 Methodology
Fluid flow and deformation of structure were governed by the following equa-
tions with assumption of linear elastic structure [11]:
ð5Þ
ð6Þ
ð7Þ
ð8Þ
ð9Þ
ð10Þ
where:
u: fluid velocity vector
p: fluid pressure
f: force that affected on wing
Re: Reynolds number
Uc: displacement velocity
ωp: angular velocity
xc: center of gravity
θp: rotation of wing r
mp: mass of wing
Ip: inertial moment of wing
F: force created by fluid passing through the wing
T: moment created by fluid passing through the wing
To solve out these equations using IBM method, the most important was that the
velocity of fluid at fluid-solid interface was equal to the velocity of the wing. It
means that the interaction force (f) between the wing and fluid was calculated such
that the boundary condition of fluid was satisfied on the surface of the wing.
IBM method used the Cartesian grid and immersed boundary that were illus-
trated in Figure 14, in which the moving surface of wing was described by
15
Aerodynamics
Figure 14.
Cartesian grid and immersed boundary.
Lagrangian points (rounded points) and fixed points in fluid were called Eulerian
points. Parameters of Lagrangian points were noted as capitalization.
Discrete partial derivative of velocity over time in Eq. (5) with denote interme-
diate velocity at zero force of Lagrangian point, Ûk, force F of Lagrangian points
were estimated as follows:
ð11Þ
where:
k: time step.
Uk+1: identified from the moving surface of wing, so this velocity was known
as U(b).
Force was created from displacement and affected on the fluid element. Force
was calculated using the following interpolation formula:
ð12Þ
where:
x: coordinate of Eulerian point.
N: set of Lagrangian points round Eulerian point l.
xl: coordinate of Lagrangian point.
∆Ul: volume of effect corresponding to Lagrangian point l.
δh: 3D delta function identified as follows:
ð13Þ
ð14Þ
16
Research on Aeroelasticity Phenomenon in Aeronautical Engineering
DOI: http://dx.doi.org/10.5772/intechopen.91748
1 1 3
8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
> 5 3j r j 1 3ð1 jrjÞ2 ≤ jrj ≤
6 2 2
>
>
>
>
>
φðrÞ ¼ 1 1 3jrj2 jrj ≤ 1
< qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
(15)
>
> 3 2
: 0 jrj ≥ 3
>
>
>
>
2
ð16Þ
where:
k: step calculation of Runge-Kutta method (k = 1, 2, 3)
αk: coefficient of kth step calculation
γk: coefficient of kth step calculation
ζk: coefficient of kth step calculation
υ: kinematic viscosity
Apply this instantaneous velocity to calculate Lagrangian velocity on the surface
of the wing:
ð17Þ
This Lagrangian velocity was combined with wing velocity, U(b)xl, which was
calculated from the dynamic equation of wing, to calculate forces of Lagrangian
points (F) following Eq. (11). Then, apply this force of Lagrangian points to Eq. (12)
to calculate force of Eulerian points (f).
• Step 2: Solve out Navier-Stokes Eqs. (5) using calculated forces of Eulerian
points to estimate the effect of flutter of the wing into the velocity field around
the wing:
ð18Þ
ð19Þ
• Step 3: Solve Eq. (17), and calculate velocity and pressure at kth step of the
Runge-Kutta method:
ð20Þ
17
Aerodynamics
ð21Þ
ð22Þ
ð23Þ
After calculating the velocity of center of gravity (Uck) and angular velocity of
wing (ωpk), coordinates of Lagrangian points were estimated by the same
expressions.
Four different wing models were carried out in order to analyze the effect of the
wing structure. There were two rectangular and two trapezoid 3D-shape wings, and
each 3D-shape wing had symmetric (NACA65A004) and asymmetric (supercriti-
cal) airfoil, respectively. The wings had the same area of 450 cm2 and the same
semispan-wise length of 30 cm. Therefore, the rectangular wing had a chord length
of 7.5 cm, while the trapezoidal wing has no sweep angle, and the leading edge line
had a tip chord length of 5 cm and root chord length of 10 cm. The wings were made
of aluminum (Figure 15).
Experiments were performed using a low-speed blowdown wind tunnel, which
belongs to the Department of Aerospace Engineering at the Hanoi University of
Science and Technology, Vietnam. This wind tunnel had a maximum free-stream
velocity in empty test section of 30 m/s that corresponded to Reynolds number 106.
The wind tunnel was operated continuously by an 8 kW electric motor. The turbu-
lence level in test section was slightly less than 1%. Free-stream velocity was kept
constant in test section within 2%. Total pressure of free-stream and dynamic
pressures were measured by pitot tube within 2%. Ambient temperature was
measured within 1%. Both experimental and numerical researches were
performed at air velocity of 20 m/s and attack angle of 5°.
Figure 15.
Wing models. (a) Rectangular wing. (b) Trapezoidal wing.
18
Research on Aeroelasticity Phenomenon in Aeronautical Engineering
DOI: http://dx.doi.org/10.5772/intechopen.91748
For the experimental study, 160 pressure taps were applied on the wing model
(Figure 15). These pressure taps were connected to an external digital manometer
via stainless and silicon tubes. Each pressure tap was measured one time with
Figure 16.
Instant displacement of wings. (a) NACA65A004 – Rectangular wing. (b) NACA65A004 – Trapezoidal
wing. (c) Supercritical – Rectangular wing. (d) Supercritical – Trapezoidal wing.
19
Aerodynamics
waiting time of 5 s (average of about 1000 instant values) using the Keyence
pressure measurement. The standard deviation of the Keyence pressure measure-
ment errors was within 0.001 Pa. Moreover, flutter of wing was captured with
help of high performance camera.
4.3 Results
The results of IBM method were analyzed at three different instants (Figure 16):
The wing deformation was maximum at the tip of the wing and decreased
gradually into the root of the wing over time. However, the normal stress was found
to have an opposite tendency in comparison with deformation. The maximum
normal stress was observed at the root of the wing, while the minimum normal
stress was found at the tip of the wing (Figures 17 and 18). It could be explained by
the fixed support with fuselage at the root of the wing and free support at the tip of
the wing [13]. These remarks were in well agreement with the experimental results
within a relative error less than 10% (Table 7).
Figure 17.
Instant normal stress—Rectangular wing.
20
Research on Aeroelasticity Phenomenon in Aeronautical Engineering
DOI: http://dx.doi.org/10.5772/intechopen.91748
At T0 instant, the normal stress had important value near the wing tip. During
flutter behaviors of wing, this important normal stress propagated from the tip of
the wing to the root of the wing. The maximum value of the normal stress was
found out at the root of the wing and at T2 instant.
With the same airfoil, the rectangular wing was found to be more distorted and
have higher maximum deformation and higher maximum normal stress than the
trapezoid wing. Thus, 3D-shape wing contributed significantly to the deformation
of wing when aeroelasticity phenomenon occurred (Table 7).
With the same 3D-shape wing, the maximum deformation and maximum nor-
mal stress of NACA65A004 rectangular wing were higher than those of the rectan-
gle supercritical wing. Meanwhile, the maximum deformation and maximum
normal stress of NACA65A004 trapezoidal wing were less than those of the super-
critical trapezoidal wing. It could be concluded that the 3D shape of wing played an
important role in the durability of the structure (Table 7).
Figure 18.
Instant normal stress—Trapezoidal wing.
Wing IBM method (mm) Experiment method (mm) Relative error (%)
Table 7.
Maximum deformation of the wing tip.
21
Aerodynamics
5. Conclusions
The flutter phenomenon of AGARD 445.6 wing was determined by (a) a modal
approach for a structural response; (b) an aerodynamic damping coefficient to
predict the appearance of flutter phenomenon; (c) a strongly coupled FSI method to
predict the aeroelastic response for subsonic and transonic flutter characteristics;
(d) an experiment method to predict the aeroelastic response for subsonic flutter
characteristics with wing structure; (e) and an IBM method to improve the inter-
face between the fluid and solid of aircraft wing.
In brief, the major results could be summarized as follows:
Acknowledgements
A part of this work was supported by the Ministry of Science and Technology in
Vietnam through the bilateral and multilateral research project HNQT/SPĐP/12.19.
22
Research on Aeroelasticity Phenomenon in Aeronautical Engineering
DOI: http://dx.doi.org/10.5772/intechopen.91748
xc center of gravity
θp rotation of wing
mp mass of wing
Ip inertial moment of wing
F force that created by fluid go pass through the wing
T moment that created by fluid go pass through the wing
X coordinate of Eulerian point
N set of Lagrangian points round Eulerian point l
xl coordinate of Lagrangian point
∆Ul volume of effect corresponded to Lagrangian point l
δh 3D delta function
αγζ coefficient of step Runge-Kutta calculation
υ kinematic viscosity
E11 normal stress following x-coordinates
E22 normal stress following y-coordinates
E33 normal stress following z-coordinates
G shear stress
ρ density
Cp coefficient of pressure
α attack angle
Author details
© 2020 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.
23
Aerodynamics
References
[9] Liu F, Cai J, Zhu Y, Tsai HM, [17] Nguyen PK, Mori K, Hoang TKD,
Wong ASF. Calculation of wing flutter Nguyen VH, Nguyen HA. Research on
by a coupled fluid-structure method. simulation and experiment of dynamic
Journal of Aircraft. 2001;38(2):334-342 Aeroelastic analysis on wing structure.
Applied Mechanics and Materials. 2015;
[10] Hoang TKD, Nguyen PK, Tran NM. 798:541-545. DOI: 10.4028/www/
Study of wing flutter by a coupled scientific.net/AMM.798.541
24
Research on Aeroelasticity Phenomenon in Aeronautical Engineering
DOI: http://dx.doi.org/10.5772/intechopen.91748
25