You are on page 1of 24

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/269129167

Flutter Based Aeroelastic Optimization of an Aircraft Wing with Analytical


Approach

Conference Paper  in  Collection of Technical Papers - AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference · April 2012
DOI: 10.2514/6.2012-1796

CITATIONS READS
4 1,245

2 authors:

Melike Nikbay Pinar Acar


Istanbul Technical University Virginia Polytechnic Institute and State University
91 PUBLICATIONS   644 CITATIONS    90 PUBLICATIONS   288 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Optimization of curvilinear and spatially varying fiber paths in composite structures View project

Development of a Nonlinear Sonic Boom Prediction Software View project

All content following this page was uploaded by Melike Nikbay on 21 January 2015.

The user has requested enhancement of the downloaded file.


53rd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference<BR>20th AI AIAA 2012-1796
23 - 26 April 2012, Honolulu, Hawaii

Flutter Based Aeroelastic Optimization of an Aircraft


Wing with Analytical Approach

Melike Nikbay∗ and Pınar Acar†


Istanbul Technical University, Maslak, Istanbul, 34469, Turkey

We present an analytical flutter prediction methodology employing assumed mode tech-


nique for three dimensional wing and wing/store configurations. The flutter solution makes
use of Lagrange formulation for aeroelastic modeling and Theodorsen function for aero-
dynamic load calculation. An in-house flutter code is developed and validated by using
Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

benchmark problems, next applied to Goland and AGARD 445.6 wing models. Flutter
results in all cases are in excellent agreement with the reference data. The flutter code is
further enhanced to enable aeroelastic optimization and uncertainty based flutter analysis
of AGARD 445.6 wing/store configurations. Firstly, aeroelastic optimization study varying
input parameters such as taper ratio, sweep angle, spanwise elasticity and shear modulus
is performed to maximize flutter boundary of AGARD 445.6 wing and an optimum clean
wing model is ascertained. Next, the structural effects of designated external masses are
investigated for flutter of initial and optimized AGARD 445.6 wing models and optimum
configurations of store placement are determined. Finally, structural randomness such as in
spanwise elasticity and shear modulus of the wing are propagated through flutter analyses
and this uncertainty quantification is applied to initial and optimum AGARD 445.6 clean
wing models. Finally, for wing/store models, random parameters relevant to store masses,
store load placements are added to material property uncertainties and similarly flutter
boundary uncertainty is examined. In all analyses, uncertainty quantification is accom-
plished by Monte Carlo Simulation method with various Coefficient of Variation estimates.

Nomenclature
A Effective plate area
a Non-dimensional distance between elastic axis and center of mass
b Half chord distance
bR Half chord distance of reference station
Ey Elasticity modulus in spanwise direction
EI Bending rigidity
g Damping term
gw Damping term for bending motion
gθ Damping term for torsional motion
Gy Shear modulus in spanwise direction
GJ Torsional rigidity
Ip Polar moment of inertia
Isy Total moment of inertia of store load
I¯sy Total moment of inertia of store load per unit span
Iwy Total moment of inertia of clean wing
I¯wy Total moment of inertia of clean wing per unit span
Iy Mass moment of inertia
I¯y Mass moment of inertia per unit span
∗ Associate Professor, Department of Astronautical Engineering, Faculty of Aeronautics and Astronautics, Istanbul Technical

University, Maslak, Istanbul, Turkey, 34469; AIAA Member


† Research Assistant - PhD Student, Department of Astronautical Engineering, Faculty of Aeronautics and Astronautics,

Istanbul Technical University, Maslak, Istanbul, Turkey, 34469

1 of 23
Copyright © 2012 by Melike Nikbay. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
American Institute of Aeronautics and Astronautics
L Aerodynamic lift force
L Span distance
Lh ,Lα Lift coefficients of Theodorsen aerodynamics
Ls Span distance of store load
Mh ,Mα Moment coefficients of Theodorsen aerodynamics
My Aerodynamic pitching moment
m Total mass
m̄ Total mass per unit span
ms Mass of store load
m̄s Mass of store load per unit span
mw Mass of clean wing
m̄w Mass of clean wing per unit span
Q Generalized forces
q Generalized coordinates
Sy Static moment
s Set of optimization parameters
sL Lower bound of s
Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

sU Upper bound of s
T Kinetic energy
T Torsion
U Potential energy
Uf Flutter speed
w Bending displacement
wR Bending displacement of reference station
ys Store position
Z Complex variable for flutter solution
λ Taper ratio
Λ Sweep angle
ω Ratio of natural frequencies
ωf Flutter frequency
ωω Bending natural frequency
ωθ Torsional natural frequency
ρ Wing density
ρ∞ Free-stream density
φ Bending mode shape
θ Torsional displacement
θR Torsional displacement of reference station
ϕ Torsion mode shape

I. Introduction
Aeroelasticity, as a multidisciplinary research field, investigates the behavior of an elastic structure in
airstream and interaction of inertial, aerodynamic and structural forces. The most catastrophic aeroelastic
phenomenon, flutter, occurs when the structure extracts energy from air stream. Determination of flutter
boundary is a prerequisite for reliability of a aircraft design. Several analytical, numerical and experimental
techniques have been traditionally used for examination of flutter boundary. Analytical solutions are required
to recognize the physical background of a dynamic aeroelastic system. As stated by Shubov,1 the physical
meaning of flutter can not be completely understood unless an analytical solution procedure is applied. Ex-
perimental or numerical studies do not provide sufficient physical knowledge. An analytical flutter solution
of a wing model can be time, frequency and Laplace domain based. Time based approaches are known as
”Time Marching Methods” and based on a coupled form including correct estimations in both aerodynamics
and structural displacements.2 These approaches practise on time dependent step size calculations in the so-
lution domain. Laplace variable based studies employ a time independent solution,3 thus algebraic equations
are adequate to find flutter speed. Frequency based approaches such as V-g and p-k methods are the most

2 of 23

American Institute of Aeronautics and Astronautics


common techniques in aeroelastic modeling and flutter determination. The contemplated analytical method
of the present work is even based on a V-g frequency solution for flutter boundary determination while the
aeroelastic modeling is constituted by assumed mode technique. Assumed mode technique represents an
aeroelastic system in terms of mode shapes and generalized coordinates by using a definite mode number.4
Calculation of flutter boundary is strictly affected by aerodynamic modeling. Aerodynamic models can
be categorized by relating to flow conditions such as steady and unsteady regimes, and aeroelastic solution
method. A realistic flutter solution must consider unsteady aerodynamics, while several approaches such
as Wagner Function, Theodorsen Function, Rational Function Approach and Indicial Function Approach
can be employed to define the aerodynamic forces in flutter calculation. Theodorsen Function, which is
derived for thin airfoils in oscillations with small amplitudes, is frequently used in frequency based flutter
analysis. Wagner Function specifies the magnitude of lift and circulation around a wing with constant and
small angle of attack, and a speed value increasing impulsively from the beginning.5 Rational Function
Approximation represents generalized aerodynamic forces by using undetermined coefficients with mathe-
matical series approach.6 Indicial Function Approximation determines the unsteady aerodynamic effects in
time and frequency domains for flutter prediction.7
Traditional flutter analysis methods are based on deterministic aeroelastic simulation models but noth-
ing is exactly as designed.8 Uncertainties are unpreventable randomness in systems and their models. The
Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

sources of uncertainties can be various while the most common ones seen in aeroelastic systems are in
structural and aerodynamic models.9 Uncertainties such as in structural damping, mass distribution, flow
boundary conditions, geometry, material properties and aerodynamics have been studied in prior works in
literature.8, 10, 11 Monte Carlo Simulation (MCS) is a frequently used method in uncertainty quantification
in a stochastic framework however it becomes nonconvergent in computationally expensive problems. Mar-
ques12 considers MCS, perturbation and interval analyses in stability calculation of Goland wing based on
eigenvalues containing Euler aerodynamics effects. Kurdi13 determines flutter boundaries of heavy version of
Goland+ wing and wing/store configuration by considering uncertainties in structural dimensions. Random
variables are distributed with MCS while flutter speed is calculated by using the linear aerodynamic theory
of ZONA6 code.
The flutter solution method of the present work employs assumed mode technique for aeroelastic modeling
and V-g frequency domain solution for flutter boundary determination. The starting point of the aeroelastic
modeling is Lagrange equation while aerodynamic terms are defined in terms of Theodorsen function. The
solution is based on a station-by-station flutter analysis along wing span and the critical flutter speed is the
minimum of all calculated values. The proposed analytical method is validated by flutter analyses of two
benchmark problems from literature. An in-house solution code is developed for flutter analysis of Goland
and AGARD 445.6 wing models, and then further extended to compute deterministic and probabilistic
flutter boundaries of AGARD 445.6 wing/store configurations. The flutter solution code is then coupled
with the optimization software, modeFRONTIER, to perform aeroelastic design optimization of wing/store
models as well. The first application is the determination of optimum clean wing model in terms of input
variables such as taper ratio, sweep angle, elasticity and shear modulus along spanwise direction. Next, the
optimum store placement is presented along initial and optimum clean wing models. 3, 4 and 5-stations of
store placements are considered and optimum store locations are ascertained for each configuration. Finally,
a probabilistic flutter analysis is performed for these structures by regarding structural uncertainties such
as store masses and locations, and material properties of the clean wing model. This work presents a highly
accurate probabilistic flutter analysis since uncertainties are propagated by MCS by using a rather large
number of samples for each random parameter. Despite large number of simulations, the present work is
still advantegous in terms of computational efficiency since it makes use of only one solution code for flutter
calculation.

II. Flutter Analysis of Three Dimensional Wing Models


An analytical solution based on assumed mode technique for determination of flutter speed of a three
dimensional wing is presented. Assumed mode technique basically represents the displacements by the
product of mode shapes and generalized coordinates.
m
X
w(y, t) = φi (y) · w̄i (t) (1)
i=1

3 of 23

American Institute of Aeronautics and Astronautics


n−m
X
θ(y, t) = ϕi (y) · θ̄i (t) (2)
i=1

Definitions of kinetic and potential energies along spanwise direction are given in (3) and (4).
Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Figure 1. General Representation of a 3D Aeroelastic Model4

Zl  
1 1 2 1 2
T = mẇ − Sy ẇθ̇ + Iy θ̇ dy (3)
2 2 2
0

Zl 2 2 !
∂ 2 w(y, t)
 
1 ∂θ(y, t)
U= EI + GJ dy (4)
2 ∂y 2 ∂y
0

The first bending and the first torsional modes are considered in calculations since they have the major
effects on flutter boundary. Existence of flutter motion mostly affected by the first modes will be justified
by Goland and AGARD 445.6 wing applications.
Natural frequencies in bending and torsional motions have to be solved distinctly since the relevant
equations have different physical meanings and mathematical expressions. Euler-Bernoulli beam equations
are used to calculate natural frequencies. Equations of motion in bending and torsion are respectively
expressed in (5) and (6).4

∂2w ∂2 ∂2w
 
ρA 2 + 2 EI 2 = 0 (5)
∂t ∂y ∂y
∂T ∂2θ
= ρIp 2 (6)
∂y ∂t
Equations of flutter motion can be determined through Lagrange equations by using free vibration fre-
quencies and energy equalities.
     
d ∂T ∂T ∂U
− + = Qi (7)
dt ∂ q̇i ∂qi ∂qi

4 of 23

American Institute of Aeronautics and Astronautics


In Lagrange equations, generalized forces are related to aerodynamic terms which are lift per unit span
and pitching moment about elastic axis.

Z`
Qw = L(y, t)φ(y)dy (8)
0

Z`
Qθ = My (y, t)ϕ(y)dy (9)
0

Harmonic motion assumption is made to determine the displacements of reference station on flutter
boundary. Reference station is a station along wing span whose properties are considered to compute flutter
speed and it is generally 75% of span distance away from the wing root.4, 14 In the present work, flutter
speeds of three dimensional clean wing models are calculated by employing a station-by-station analysis.
The wing model is divided into 105 stations along the span and flutter speeds are determined for the
corresponding stations. The critical flutter speed is the minimum of all calculated values. The station-
by-station analysis even infers that the reference station is 75% of span distance away. Thus, the flutter
Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

boundary calculations of wing/store configurations in the next section will not be based on station-by-station
approach for computational time efficiency. Instead, the reference station will be assumed as if it is 75% of
span distance away from the wing root.

Figure 2. Representation of Station-by-Station Analysis

Aerodynamic loads must take account of sweep angle effects since the flutter boundary of AGARD 445.6
wing can strongly be affected from the non-zero parameter value. The aerodynamic modeling is based on
Theodorsen function and its associated aerodynamics. Since the regarding wings of the present work all
have thin airfoil sections with small thickness values, Theodorsen function is expected to provide a good
approximation for the aerodynamic model.
   
3 2 wR 1
L(y, t) = πρ∞ b ω cos Λ φ(y)Lh − θR ϕ(y) Lα − Lh +a (10)
b 2

5 of 23

American Institute of Aeronautics and Astronautics


M (y, t) = πρ∞ b4 ω 2 cos Λ − wbR φ(y) Mh − Lh 1
 
2 +a
h (11)
4 2 1
 1
2 i
+θR πρ∞ b ω cos Λϕ(y) Mα − (Mh + Lα ) 2 + a + Lh 2 +a

where Lh , Lα , Mh , Mα are functions of reduced frequency and Theodorsen function. They can be specified
by algebraic functions in terms of reduced frequency. Final equations of motion that lead to the solution
of flutter speed are determined after combining the structural and aerodynamic terms by counting on taper
ratio effect with respect to reference station. System of equations can finally be represented in basic forms
for flutter solution.
wR
A + BθR = 0 (12)
bR
wR
C + DθR = 0 (13)
bR
where A, B, C and D are coefficients of flutter determinant and their definitions are given in (14) to
(17).
"  2 # Z1
m b
Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

2
1 − ω2 Z +

A= Lh cos Λ φ(ỹ) dỹ (14)
πρ∞ b2R bR
0
"  3    # Z1
Sy b 1
B= − + Lθ − Lh +a cos Λ φ(ỹ)ϕ(ỹ)dỹ (15)
πρ∞ b3R bR 2
0
"  3    # Z1
Sy b 1
C= − + Mh − Lh +a cos Λ φ(ỹ)ϕ(ỹ)dỹ (16)
πρ∞ b3R bR 2
0
"  4    2 ! # Z1
Iy b 1 1
D= (1 − Z) + Mθ − (Mh + Lθ ) + a + Lh +a cos Λ ϕ2 (ỹ)dỹ (17)
πρ∞ b4R bR 2 2
0

In above equations, ỹ indicates dimensionless span distance whose definition is given in (18).
y
ỹ = (18)
L
The solution of the system uses artificial damping term based V-g technique to calculate the flutter speed
value. Thus, a complex variable, Z, comprising of these damping effects has to be defined by considering
gw = gθ = g for simplicity.
 ω 2
θ
Z= (1 + ig) (19)
ω
g = 0 is the sought after value for flutter boundary determination.

A. Validation of Flutter Analysis


The proposed flutter analysis method is applied to two benchmark wing models4 for validation purpose.
By regarding the design parameters given in Table 1, the flutter speeds and relative errors with respect to
reference data are calculated and shown in Table 2.
In the solution process, bending and torsional natural frequencies are not calculated distinctly since their
exact values are indicated by Bisplinghoff.4 Moreover, the flutter analysis is not based on a station-by-station
approach since the design properties of reference station are already implied in Table 1. The flutter results,
shown in Table 2, are in excellent agreement with the reference data. Thus, the proposed flutter solution
methodology is proven by these benchmark wing models and can be applied to more realistic wing structures
with more complex geometries.

6 of 23

American Institute of Aeronautics and Astronautics


Wing-1 Wing-2
Λ (degree) 30 45
m (slugs/ft) 0.0161 0.0138
ωω (rad/sec) 66π 44π
ωθ (rad/sec) 186π 184π
m/πρ∞ b2R 6.19 5.50
Sy /mbR -0.004 -0.224
Iy /mb2R 0.23 0.23
b = bR 0.333 0.333
a -0.02 0.20

Table 1. Design Parameters of Benchmark Wings4

Reference Calculated Relative Error


Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Wing-1 277 ft/s 279 ft/s 0.8%


Wing-2 270 ft/s 268 ft/s 0.7%

Table 2. Flutter Results for Benchmark Problems

Figure 3. Geometry of Goland Wing16

B. Flutter Analysis of Goland Wing


The presented flutter solution methodology is practised on a well-known aeroelastic benchmark problem,
Goland wing. The wing, which is treated as a cantilever beam, is firstly introduced in the work of Goland.15
Extensive research has been carried out to solve the flutter problem of Goland wing with various meth-
ods such as Rayleigh-Ritz analysis, Galerkin solution as analytical techniques beside computational ap-
proaches.13, 17 In the present work, flutter frequency and flutter speed of Goland wing are calculated with
the design parameters depicted in Table 1.15
The flutter analysis is now based on a station-by-station approach and it infers a reference station 75% of
span distance away from the wing root as expected. The results, which are presented in Table 4 and Figure

7 of 23

American Institute of Aeronautics and Astronautics


Parameter Value
L 20 ft
b 3 ft
EI/m 31.7 × 106 lbft3 /slug
GJ/Iy 1.23 × 106 lbft/slug
m̄ 0.746 slug/ft
I¯y 1.943 slugft2 /ft
Sy 0.447 slugft/ft
ρ 0.0001slugs/ft3

Table 3. Design Parameters of Goland Wing

Reference15 Calculated Relative Error


ωf 67.4 rad/s 65.5484 rad/s 2.75%
Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Uf 385 mph 374.7543 mph 2.66%

Table 4. Flutter Results for Goland Wing

Figure 4. Flutter Frequency-Damping Term Relation for Goland Wing

4, are satisfactory with respect to the work of Goland.15 Thus, the proposed flutter solution methodology is
validated by a three dimensional generic wing model and can be applied to a more realistic wing configuration
as in the next part. Henceforward, the wing model of interest will be AGARD 445.6.

8 of 23

American Institute of Aeronautics and Astronautics


III. Flutter Analysis of AGARD 445.6 Wing
The wing structure of major interest in this work is AGARD (Advisory Group for Aerospace Research and
Development) 445.6, which is the first aeroelastic configuration tested by Yates in the Transonic Dynamics
Tunnel (TDT) at NASA Langley Research Center. AGARD 445.6, which is made of laminated mahogany,
is a swept-back wing with a sweep angle of 45 degrees, taper and aspect ratios of 0.66 and 1.65 respectively.
The airfoil used in this wing is symmetrical NACA65A004 profile. The wing consists of two models as solid
and weakened models.18 Wall-mounted weakened model is considered in this work.
Studies in dynamic aeroelastic analysis and flutter calculations of AGARD 445.6 wing are extensive.
Several methods have been used to investigate the flutter boundaries. In the work of Beaubien,18 Computa-
tional Fluid Dynamics (CFD) is coupled with Computational Structural Dynamics (CSD) and time marching
simulations are performed by using Euler and Reynolds Averaged Navier Stokes (RANS) equations to calcu-
late flutter speed. Lee-Rausch19 performed linear stability analysis by calculating generalized aerodynamic
forces for various reduced frequencies. Flutter characteristics are determined by using V-g analysis which
is a similar approach with the present work. Allen20 showed that the flutter calculation of AGARD 445.6
wing using linear techniques provides reasonable results since the design and aerodynamics of the wing are
simple. Therefore, the linear flutter analysis method of the present work is expected to well-agree with the
Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

experimental results.
Flutter analysis of AGARD 445.6 wing is performed by using the pre-determined natural frequencies and
flutter equations. In flutter calculation procedure, the necessary design parameters of the wing are taken from
CAD model constructed in CATIA V5 by Nikbay et. al21 and also determined from the known geometrical
properties of the standard configuration. Material properties of weakened model and experimental results
for flutter analysis are given in the work of Yates.22

(a) Wing geometry (lengths in inches) (b) The solid model of the wing

Figure 5. AGARD 445.6 Wing Structure21

Natural frequency determination employs Euler-Bernoulli beam equations. Euler-Bernoulli solution was
previously investigated for AGARD 445.6 wing by Kamakoti,23 Kamakoti and Shyy,24 Kamakoti et. al.25
The modeling can be based on the use of plate/shell elements, however bending and torsional natural frequen-
cies calculated with beam assumption agree well with the results calculated with plate element assumption.23
Beam elements are chosen since they provide an advantegous solution with a simplified procedure24 while
still providing rather accurate results. The results determined by Kamakoti23 with 10 beam elements for the
first bending and the first torsion modes are almost equivalent to the results in the work of Yates26 which
employs 120 plate elements. Therefore, Euler-Bernoulli equations are used to calculate natural frequencies
of AGARD 445.6 in the present work.
Flutter speed of AGARD 445.6 wing is calculated for Mach number of 0.9011 by using analytically
determined natural frequency values and then compared to the experimental results stated by Yates22 and
the work of Kolonay27 in Table 5. The relative errors show the differences between the present work and
experiment while variation of flutter frequency with respect to artificial damping term, g, is shown in Figure
6.
Flutter frequency and flutter speed of analytical solution well-agree with the experimental results. Since
the analysis for standard configuration results with success, the same solution procedure can be extended
for the flutter analysis of AGARD 445.6 wing/store configurations as in Section V.

9 of 23

American Institute of Aeronautics and Astronautics


Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Figure 6. Flutter Frequency-Damping Relation for AGARD 445.6

Analytical Experimental22 Kolonay27 Relative Error


ωf (rad/s) 104.25 101.1 99.0 3.12%
Uf (m/s) 308.45 296.7 299.97 3.96%

Table 5. Flutter Results and Relative Errors

IV. Flutter Based Aeroelastic Design Optimization of AGARD 445.6 Wing


Aeroelastic design optimization of AGARD 445.6 clean wing is accomplished to enhance the flutter
boundary. The flutter solution code is revised such that it can present a flutter solution in terms of input
parameters which are defined as taper ratio, sweep angle, elasticity and shear modulus along spanwise
direction, and coupled with the optimization software, modeFRONTIER. The optimization problem is given
in (20) to (23) while the workflow is shown in Figure 7.

max Uf (s) (20)


s∈S

S = {s ∈ <, sL ≤ s ≤ sU } ; s = (λ, Λ, Ey , Gy ) (21)


0.66 < λ < 1.0; 0o < Λ < 60o ; (22)
2000M P a < Ey < 3000M P a; 200M P a < Gy < 300M P a (23)
NSGA-II (Non-Dominated Sorting Genetic Algorithm) is chosen as optimization algorithm with 1000
DoE (Design of Experiments). The optimization took about 11 hours and 13 minutes on a platform which
has Intel(R) Core(TM) 2 CPU 6400@2.13GHz processor and 2 GB of RAM on Microsoft Windows 7 64-bit
operating system. The optimum design with the flutter speed of 361.8843 m/s is determined among 100000
feasible solutions. Design parameters of optimum structure and initial configuration are given in Table 6.

10 of 23

American Institute of Aeronautics and Astronautics


λ Λ (Degree) Ey (MPa) Gy (MPa)
22
Initial Configuration 0.66 45 3671 409
Optimum Design 0.66 59.65 2020.85 299.02

Table 6. Design Parameters of Initial and Optimum AGARD 445.6 Wing Models
Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Figure 7. Workflow for Flutter Based Optimization of AGARD 445.6

Optimum design noteworthily improves the flutter boundary of AGARD 445.6 wing. Since flutter is a
catastrophic aeroelastic phenomenon, any increase in its boundary provides a more reliable flight. Optimum
flutter speed and improvement with respect to initial configuration are expressed in Table 7.

Initial Design Optimum Design


Flutter Speed 308.45 m/s 361.88 m/s
Improvement - %17.32

Table 7. Improvement of Flutter Speed in Optimum Design

V. Flutter Analysis of Wing/Store Configurations


The present section addresses deterministic flutter analysis of wing/store configurations carrying external
loads placed in various stations along the wing span. One of the main purposes of this work is to define a
general solution metholodology for the flutter analysis of wing/store configurations where the store loads can
be modeled as missiles, launchers or fuel tanks, etc. Thus, the flutter solution code is enhanced to examine
the structural effects of store loads such as inertias and masses. The parametric solution is expected to form
a basis for further analyses and optimization studies in various types of wing/store configurations ranging
from the simplest models to designs with high complexity levels such as fighter aircraft wings. An example
representation of a wing/store configuration is illustrated in Figure 8.
The structural effects of store masses are added through mass density values and the locations where the

11 of 23

American Institute of Aeronautics and Astronautics


Figure 8. General Representation of a Wing/Store Configuration28
Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

related masses place. The addition of inertia moments of the store loads to the solution process comprises
a similar approach. The flutter solution is based on a station-by-station analysis considering the sections
at which store loads are located. The store loads are assumed to be mounted at the center of gravity
of the relevant sections. The minimum flutter speed value among all calculations for regarded stations
is designated as the flutter speed of the wing/store configuration. The updated flutter coefficients of the
solution determinant with store loads effects are summarized in (24) to (27).

Z1 Z1  2
b
A = πρ∞ b2R 1 − ω 2 Z (mw + ms )φ2 dỹ + cos Λ Lh φ2 dỹ

(24)
bR
0 0

Z1 Z1  3   
b 1
B= −πρ∞ b3R Sy φϕdỹ + cos Λ Lθ − Lh +a φϕdỹ (25)
bR 2
0 0

Z1 Z1  3   
b 1
C= −πρ∞ b3R Sy φϕdỹ + cos Λ M h − Lh +a φϕdỹ (26)
bR 2
0 0

Z1 Z1  4    2 !
b 1 1
D= πρ∞ b4R (1 − Z) 2
(Iwy + Isy )ϕ dỹ +cos Λ Mθ − (Mh + Lθ ) + a + Lh +a ϕ2 dỹ
bR 2 2
0 0
(27)
where Iwy is total inertia moment of clean wing and Isy is inertia moment of store load for the corre-
sponding station in flutter analysis. Similarly, mw indicates the mass of clean wing while ms is the mass of
external load which is placed on the station of interest.
The remaining solution is the same with the presented procedure for flutter calculation of three dimen-
sional clean wing models. The effects of pylon structure and store aerodynamics are neglected. The solution
procedure is firstly validated by using Goland wing and an external store which is placed in different stations
along wing span as in the work of Fazelzadeh,29 and then implemented into aeroelastic design optimization
of AGARD 445.6 wing/store configurations with 3, 4 and 5 stations of store placements respectively.
Reference design parameters of the benchmark model are shown in Table 8 while experimental30 and
numerical29 flutter speed results are depicted and compared with the calculated results in Table 9. The
relative error, presented by Table 9, states the percentage error of the determined flutter speed with respect
to experimental result. Flutter speed-damping term plots are shown in Figure 9 to 12 for various store
positions, ys .

12 of 23

American Institute of Aeronautics and Astronautics


Parameter Value
L 1.2192 m
b 0.1016 m
EI 403.76N m2
GJ/Iy 198.58N m2
m̄w 1.2942kgm−1
I¯wy 0.0036kgm
Elastic axis 43.7 %
Center of Gravity 45.4%
ρ 1.224kgm−3
ms 1.578kg
Isy 0.0185kgm2

Table 8. Reference Design Parameters of Example Goland Wing/Store Model


Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Figure 9. Flutter Speed-Damping Term Relation for ys =0.2794 m

VI. Aeroelastic Design Optimization of Wing/Store Configurations


Aeroelastic design optimization of the present section aims to reach the best configuration for a wing/store
model with optimum store placements along wing span. The corresponding models are AGARD 445.6
wing/store structures with 3, 4 and 5 stations of store loads placements while the total external masses
are the same for each case. The clean wing models consist of initial design22 and optimum configuration
of Section IV. The store loads are modeled as point masses and their inertias are neglected due to lack of
technical information.

13 of 23

American Institute of Aeronautics and Astronautics


Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Figure 10. Flutter Speed-Damping Term Relation for ys =0.4318 m

Figure 11. Flutter Speed-Damping Term Relation for ys =1.1684 m

A. Flutter Based Optimization of Initial AGARD 445.6 Wing/Store Configuration


Flutter based design optimization is firstly applied to AGARD 445.6 wing/store configuration with initial
clean wing model to determine the optimum locations of store loads for the purpose of flutter speed maxi-
14 of 23

American Institute of Aeronautics and Astronautics


Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Figure 12. Flutter Speed-Damping Term Relation for ys =1.2192 m

Store Location Numerical Result29 Experimental Result30 Calculated Solution Relative Error
0.2794 m 100.89 m/s 98.75 m/s 96.0679 m/s 2.7160 %
0.4318 m 124.05 m/s 116.43 m/s 113.1926 m/s 2.7806 %
1.1684 m 112.17 m/s 112.17 m/s 121.7199 m/s 8.5138 %
1.2192 m 91.44 m/s 97.54 m/s 94.3449 m/s 3.2757 %

Table 9. Flutter Results of Example Goland Wing/Store Configuration

mization. The optimization studies are carried out by considering 3, 4 and 5 stations of store placements.
The total store mass is 1.25 kg and it is equally distributed on stations. The optimization problem involves
flutter speed maximization objective while design parameters are defined as distances of store stations mea-
sured from the root of the wing. Although the external loads are modeled as point masses for a preliminary
application, this approach is not realistic. Constraints are determined for distances between successive sta-
tions to place the related masses in a more realistic manner. General representation of the optimization
problem is given in (28) to (35) by regarding 5 stations of store placements. Constraint functions and design
variables can be defined in the same manner for 3 and 4 stations cases while the design objective is also same
for each optimization application.

max Uf (s) (28)


s∈S

S = {s ∈ <, sL ≤ s ≤ sU } (29)

s = {y1 , y2 , y3 , y4 , y5 } (30)

0 ≤ y1 , y2 , y3 , y4 , y5 ≤ 0.762 m (31)

15 of 23

American Institute of Aeronautics and Astronautics


x1 = y1 − y2 ≤ −0.04 m (32)

x2 = y2 − y3 ≤ −0.04 m (33)

x3 = y3 − y4 ≤ −0.04 m (34)

x4 = y4 − y5 ≤ −0.04 m (35)
where y1 , y2 , y3 , y4 , y5 define the distances for each station measured from the root of the wing while
x1 , x2 , x3 , x4 are the location constraints between successive stations. The optimization workflow is shown
in Figure 13.
Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Figure 13. Optimization Workflow of AGARD 445.6 Wing/Store Configuration

NSGA-II is used as optimization algorithm with 1000 DoE while the number of total designs are 100000 for
all cases. All optimization studies are performed on the same platform which has Intel(R) Core(TM) 2 CPU
6400@2.13GHz processor and 2 GB of RAM on Microsoft Windows 7 64-bit operating system. Optimum
design parameters with maximum flutter speeds are given in Table 10 for considered configurations.

B. Flutter Based Optimization of Previously Optimized AGARD 445.6 Wing/Store Config-


uration
Flutter based design optimization is applied to AGARD 445.6 wing/store configuration with optimum clean
wing model to determine the optimum locations of store loads. The optimization studies are again carried
out by considering 3, 4 and 5 stations of store placements.
The preferred optimization algorithm is again NSGA-II with 1000 DoE and 100000 total designs are
obtained for all cases. All optimization studies are performed on the same platform which has Intel(R)
Core(TM) 2 CPU 6400@2.13GHz processor and 2 GB of RAM on Microsoft Windows 7 64-bit operating
system. Optimum design parameters with maximum flutter speeds are given in Table 11 for regarded
configurations.

16 of 23

American Institute of Aeronautics and Astronautics


Parameter 3-Station 4-Station 5-Station
y1 0.68113 m 0.63283 m 0.57710 m
y2 0.72122 m 0.67948 m 0.61801 m
y3 0.76196 m 0.72057 m 0.65928 m
y4 - 0.76200 m 0.71182 m
y5 - - 0.76200 m
Uf 232.04 m/s 221.43 m/s 212.32 m/s

Table 10. Optimization Results for AGARD 445.6 Model with Initial Clean Wing

Parameter 3-Station 4-Station 5-Station


y1 0.68113 m 0.63283 m 0.57710 m
y2 0.72122 m 0.67948 m 0.61801 m
y3 0.76196 m 0.72057 m 0.65928 m
Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

y4 - 0.76200 m 0.71182 m
y5 - - 0.76200 m
Uf 314.46 m/s 288.89 m/s 265.11 m/s

Table 11. Optimization Results for AGARD 445.6 Model with Optimized Clean Wing

VII. Uncertainty Based Flutter Analysis of Wing and Wing/Store


Configurations
The present section aims to investigate the effects of structural uncertainties on the flutter boundary of
wing and wing/store configurations. Traditional uncertainty quantification methods in aeroelastic analysis
is based on choosing the best design among the model set by introducing the best distribution for random
parameters. Sources of uncertainties are various in fact they can be initial and boundary conditions, geo-
metric features, parametric variations in physical quantities and modeling errors.31 Deterministic methods
can be adequate for small variations, however increasing amount of uncertainties must require probabilistic
methods. Safety factor approach used in deterministic methods can cause design of heavy aircraft structures.
Design requirements in a deterministic model are defined strictly and any variation in design parameters can
probably violate the constraints, however system reliability can be increased with probabilistic analyses.
In the present work, uncertainty propagation is provided by MCS method while each random parameter
of AGARD 445.6 wing and wing/store configurations is modeled with 105 samples. Variations of random
variables are determined by COV=1% and COV=5% estimates where:
σx
COV = % (36)
µx
Many types of probability distributions can be used to model the uncertain parameters. Gaussian distri-
bution is preferred in the present work within the context of MCS since small variations of random variables
are considered due to high technology manufacturing possibilities of today’s world. MCS method favourably
inserts randomly generated exact values of uncertain variables for the corresponding simulation to reach the
solution instead of deriving any mathematical approximation. Thus, the accuracy of MCS is directly related
to the number of samples as defined in (37).31
1
M CSAccuracy = √ (37)
n
In the present work, the major point to use 105 samples for each random variable is to perform more
accurate aeroelastic analyses. The corresponding accuracy of MCS is 0.3162%. The uncertainty based flutter
analyses are firstly applied to AGARD 445.6 initial and optimum clean wing models, and then AGARD 445.6

17 of 23

American Institute of Aeronautics and Astronautics


wing/store configurations consisting of initial and previously optimized clean wing models with 3, 4 and 5
stations of store placements along wing span. Random variables are defined as elasticity and shear modulus
of the clean wing, and masses and locations of store loads. Minimum flutter speed among 105 results is
considered to achieve maximum reliability.

A. Uncertainty Based Flutter Analysis of AGARD 445.6 Clean Wing Models


In the present section, uncertainty based flutter analysis is performed for initial and previously optimized
AGARD 445.6 clean wing models. The uncertain parameters are assigned as elasticity and shear modulus
of the wing along spanwise direction. Initial deterministic values are compared to the results of uncertainty
based analyses. The results for uncertainty based flutter analysis of AGARD 445.6 clean wing models with
COV=1% and COV=5% estimates are given in Table 12 while Figure 14 shows the flutter speed distributions
for these estimates.

Casel Uf (det) Uf min Uf mean Uf max


Initial Wing (COV=1%) 308.4513 m/s 296.7518 m/s 308.4606 m/s 319.9008 m/s
Optimum Wing (COV=1%) 361.8843 m/s 339.0062 m/s 351.9817 m/s 366.2595 m/s
Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Initial Wing (COV=5%) 308.4513 m/s 338.0732 m/s 351.9625 m/s 366.3779 m/s
Optimum Wing (COV=5%) 361.8843 m/s 333.3994 m/s 352.0205 m/s 372.4808 m/s

Table 12. Results of Uncertainty Based Flutter Analysis of AGARD 445.6 Clean Wing Models

Figure 14. Uncertainty Based Flutter Analysis of Initial AGARD 445.6 Wing with COV=1% and COV=5%

B. Uncertainty Based Flutter Analysis of AGARD 445.6 Wing/Store Configurations


Uncertainty based flutter analysis is performed for AGARD 445.6 wing/store configurations consisting
of initial and previously optimized clean wing models with 3, 4 and 5 stations of store placements. The
uncertain parameters are material properties of clean wing models, and store masses and locations. Results
of uncertainty based analyses for COV=1% and COV=5% estimates are depicted in Table 13, 14 and 15.
Due to physical limitations of the placements, COV=0.25% estimation is used for station locations.

18 of 23

American Institute of Aeronautics and Astronautics


Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Figure 15. Uncertainty Based Flutter Analysis of Optimum AGARD 445.6 Wing with COV=1% and COV=5%

Casel Uf (det) Uf min Uf mean Uf max


3-Stations (Initial) (COV=1%) 232.04 m/s 227.8418 m/s 232.0408 m/s 235.7462 m/s
3-Stations (Optimum) (COV=1%) 314.46 m/s 308.2334 m/s 313.4027 m/s 318.6943 m/s
3-Stations (Initial) (COV=5%) 232.04 m/s 223.7533 m/s 232.0265 m/s 240.3017 m/s
3-Stations (Optimum) (COV=5%) 314.46 m/s 300.2594 m/s 313.3904 m/s 326.7064 m/s

Table 13. Results of Uncertainty Based Flutter Analysis of 3-Stations Wing/Store Model

Figure 16. Uncertainty Based Flutter Analysis of Initial 3-Stations Case with COV=1% and COV=5%

VIII. Conclusion
The present work introduces an analytical flutter analysis method for three dimensional wing and
wing/store configurations using assumed mode technique for enabling aeroelastic optimization and uncer-
19 of 23

American Institute of Aeronautics and Astronautics


Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Figure 17. Uncertainty Based Flutter Analysis of Optimum 3-Stations Case with COV=1% and COV=5%

Casel Uf (det) Uf min Uf mean Uf max


4-Stations (Initial) (COV=1%) 221.43 m/s 217.7585 m/s 221.4259 m/s 224.6774 m/s
4-Stations (Optimum) (COV=1%) 288.89 m/s 283.0084 m/s 287.9006 m/s 292.6584 m/s
4-Stations (Initial) (COV=5%) 221.43 m/s 213.8700 m/s 221.4256 m/s 228.8889 m/s
4-Stations (Optimum) (COV=5%) 288.89 m/s 274.8614 m/s 287.8993 m/s 299.7019 m/s

Table 14. Results of Uncertainty Based Flutter Analysis of 4-Station Wing/Store Model

Figure 18. Uncertainty Based Flutter Analysis of Initial 4-Stations Case with COV=1% and COV=5%

tainty based flutter analysis. The proposed analytical method is validated by two benchmark problems and

20 of 23

American Institute of Aeronautics and Astronautics


Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Figure 19. Uncertainty Based Flutter Analysis of Optimum 4-Stations Case with COV=1% and COV=5%

Casel Uf (det) Uf min Uf mean Uf max


5-Stations (Initial) (COV=1%) 212.32 m/s 208.7059 m/s 212.3180 m/s 215.5626 m/s
5-Stations (Optimum) (COV=1%) 265.11 m/s 259.5681 m/s 264.2754 m/s 269.5408 m/s
5-Stations (Initial) (COV=5%) 212.32 m/s 204.8917 m/s 212.3139 m/s 219.2612 m/s
5-Stations (Optimum) (COV=5%) 265.11 m/s 253.5033 m/s 264.2832 m/s 276.4995 m/s

Table 15. Results of Uncertainty Based Flutter Analysis of 5-Station Wing/Store Model

Figure 20. Uncertainty Based Flutter Analysis of Initial 5-Stations Case with COV=1% and COV=5%

then used to investigate the flutter boundaries of well-known aeroelastic models, Goland and AGARD 445.6

21 of 23

American Institute of Aeronautics and Astronautics


Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Figure 21. Uncertainty Based Flutter Analysis of Optimum 5-Stations Case with COV=1% and COV=5%

wings. An in-house code developed for flutter calculation is extended in a way to parametrically define the
solution steps in terms of input variables which are taper ratio, sweep angle, elasticity and shear modulus
of the wing, and coupled with an optimization framework to perform flutter based design optimization of
AGARD 445.6 clean wing model. Next, the flutter solution is enhanced to examine the flutter boundary of
wing/store configurations. Structural effects of store loads are examined while the updated methodology is
validated by an example using Goland wing with an external store placed on different stations along wing
span.
One of the main concern of this work is to construct the most efficient store loads placement based
on flutter criterion. Therefore, the optimum placement of store loads on initial and previously optimized
AGARD 445.6 clean wing models are investigated. The wing/store configurations comprise of 3, 4 and 5
stations of store placements however the most efficient case is determined as 3-stations loading.
Finally, uncertainty based flutter analyses of clean wing and wing/store models are performed to investi-
gate the effects of structural uncertainties. Random parameters are assigned as material properties of clean
wing structure, and locations and masses of store loads. Uncertainty quantification is accomplished by MCS
method with 1% and 5% of COV estimates. 105 samples for each random variable bring about the rather
accurate probabilistic results. Minimum flutter boundaries should be taken into account for maximum relia-
bility. The difference between deterministic and probabilistic results is a significant indication which points
out the necessity of probabilistic approach in aircraft design.
Developed flutter solution methodology forms a basis for the flutter analysis and flutter based optimiza-
tion of more complex structures and may be extended for military and civilian purposes and requirements
as future work. Structural and aerodynamic nonlinearities must be adapted to the solution procedure for
a more realistic application such as a fighter aircraft wing. In addition, all structural and aerodynamic
effects of store loads must be considered in calculations. Static aeroelastic responses and instabilities such
as divergence and control reversal may also be incorporated into design and optimization process of wing
and wing/store configurations for a more competent design.

References
1 Shubov, M. A., ”Mathematical Modeling and Analysis of Flutter in Bending-Torsion Coupled Beams, Rotating Blades

and Hard Disk Drives,” Journal of Aerospace Engineering, Vol. 17, 2004, pp. 256–269.
2 Goura, G., ”Time Marching Analysis of Flutter Using Computational Fluid Dynamics,” Ph.D. Dissertation, University

of Glasgow Department of Aerospace Engineering, 2001.


3 Dorf, R. C., Bishop, R. H., Modern Control Systems, Prentice Hall, 2008.
4 Bisplinghoff, R. L., Ashley, H. and Halfman, R. L., Aeroelasticity, Addison-Wesley Publishing Company, 1955.

22 of 23

American Institute of Aeronautics and Astronautics


5 Wagner, H., ”Über die Entstehung des dynamischen Auftriebes von Tragflgeln,” Journal of Applied Mathematics and

Mechanics, 1925, pp. 17–35.


6 Zouari, R., ”Détection Précoce d’Instabilité Aéroélastique des Structures Aéronautiques,” Ph.D. Dissertation,

L’Université de Rennes, 2008.


7 Marzocca P., Librescu, L., Kim, D-H., Schober S., ”Development of an Indicial Function Approach for the Two Dimen-

sional Incompressible/Compressible Aerodynamic Load Modeling,” Journal of Aerospace Engineering, Vol. 221, No. 3, 2007,
pp. 453–463.
8 Danowski, B. P., Chrstos, J. F., Klyde, D. H., Farhat, C., Brenner, M., ”Evaluation of Aeroelastic Uncertainty Methods,”

Journal of Aircraft, Vol. 47, No. 4, 2010, pp. 1266-1273.


9 Lind, R., ”A Presentation on Robust Flutter Margin Analysis and a Flutterometer,” NASA Tecnical Paper-97-206220,

1997.
10 Heinze, S., Ringertz, U., Borglund, D., ”Assessment of Uncertain External Store Aerodynamics Using µ-p Flutter

Analysis,” Journal of Aircraft, Vol. 46, No. 3, 2009, pp. 1062-1067.


11 Borglund, D., Ringertz, U., ”Efficient Computation of Robust Flutter Boundaries Using µ-k Method,” Journal of

Aircraft, Vol. 43, No. 6, 2006, pp. 1763-1769.


12 Marques, S., Badcock, K. J., Khodaparast, H. H., Mottershead, J. E., ”Transonic Aeroelastic Stability Predictions

Under the Influence of Structural Variability,” Journal of Aircraft, Vol. 47, No. 4, 2010, pp. 1229-1239.
13 Kurdi, M., Lindsley, N., Beran, P., ”Uncertainty Quantification of the Goland+ Wings Flutter Boundary,” AIAA

Atmospheric Flight Mechanics Conference and Exhibit, Hilton Head, South Carolina, 2007.
14 Dowell, E. H., Crawley, E. F., Curtiss, Jr. H. C., Peters, D. A., Scanlan, R. H., Sisto, F., A Modern Course in
Downloaded by Melike Nikbay on January 14, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-1796

Aeroelasticity,Kluwer Academic Publishers, 1995.


15 Goland, M., ”The Flutter of a Uniform Cantilever Beam,” Journal of Applied Mechanics, 12, 1945, pp. 197-208.
16 Khodaparast, H. H., ”Stochastic Finite Element Model Updating and its Application in Aeroelasticity,” Ph.D. Disser-

tation, University of Liverpool, 2010.


17 Qiangjun, Z., ”Modeling the Effect of Taper Ratio on Wing Flutter,” International Conference on Computer Application

and System Modeling, Shanxi, Taiyuan, 2010.


18 Beaubien, R. J., Nitzsche, F. and Feszty, D., ”Time and Frequency Domain Flutter Solutions for the AGARD 445.6

Wing,” International Forum on Aeroelasticity and Structural Dynamics, Munich, Germany, 2005.
19 Lee-Rausch, E. M. and Batina, J. T., ”Calculation of AGARD Wing 445.6 Flutter Using Navier-Stokes Aerodynamics,”

AIAA 11th Applied Aerodynamics Conference, Monterey, California, 1993.


20 Allen, C. B., Jones, D., Taylor, N. V., Badcock, K. J., Woodgate, M. A., Rampurawala, A. M., Cooper, J. E. and

Vio, G. A., ”A Comparison of Linear and Non-Linear Flutter Prediction Methods: A Summary of PUMA DARP Aeroelastic
Results,” Royal Aeronautical Society Aerodynamics Conference, London, 2004.
21 Nikbay, M., Fakkusoglu, N., Kuru, M. N., ”Reliability Based Multi-disciplinary Optimization of Aeroelastic Systems

with Structural and Aerodynamic Uncertainties,” 13th AIAA/ISSMO Multidisciplinary Analysis and Optimization (MAO)
Conference, Fort Worth, Texas, USA, 2010.
22 Yates, E., ”Standard Aeroelastic Configurations for Dynamic Response I-Wing 445.6,” AGARD Report No.765, 1985.
23 Kamakoti, R., ”Computational Aeroelasticity Using a Pressure-Based Solver,” Ph.D. Dissertation, University of Florida,

2004.
24 Kamakoti, R., Shyy, W., ”Fluid-Structure Interaction for Aeroelastic Applications,” Progress in Aerospace Sciences,

Vol. 40, No. 8, 2010, pp. 538-553.


25 Kamakoti, R., Lian, Y., Regisford, S., Kurdila, A., Shyy, W., ”Computational Aeroelasticity Using a Pressure-Based

Solver,” AIAA 40th Aerospace Sciences Meeting and Exhibit, Reno, NV, USA, 2002.
26 Yates, E., ”AGARD Standard Aeroelastic Configurations for Dynamic Response, Candidate Configuration I.-Wing

445.6,” NASA TM 100492.


27 Kolonay, R. M., ”Computational Aeroelasticity,” Presented in Technical Course Organized by The Applied Vehicle

Technology Panel (AVT) on Application of Adaptive Structures in Active Aeroelastic Control, METU, Ankara, Turkey, 2002.
28 Panagiotopoulos, E. E., Kyparissis S. D., ”CFD Transonic Store Seperation Trajectory Predictions with Comparison to

Wind Tunnel Investigations,” International Journal of Engineering, Vol. 3, No. 6, 2010, pp. 538-553.
29 Fazelzadeh, S. A., Mazidi, A., Kalantari, H., ”Bending-Torsional Flutter of Wings with an Attached Mass subjected to

a Follower Force,” Journal of Sound and Vibration, Vol. 323, No. 1, 2009, pp. 148-162.
30 Runyan, H. L., Watkins, C.E., ”Flutter of a Uniform Wing with an Arbitrarily Placed Mass according to a Differential

Equation Analysis and a Comparison with Experiment,” NACA TN 1848, 1949.


31 Choi, S. K., Grandhi, R. V., Canfield, R. A., Reliability-based Structural Design, Springer, 2007.

23 of 23

American Institute of Aeronautics and Astronautics

View publication stats

You might also like