Professional Documents
Culture Documents
Aerospace Structures and Materials (Liu - 2016)
Aerospace Structures and Materials (Liu - 2016)
Volume 1
Aerospace Structures and
Materials
Edited By
Yucheng Liu
Department of Mechanical Engineering
Mississippi State University
MS 39762
USA
Frontiers in Aerospace Science
Volume # 1
Aerospace Structures and Materials
This is an agreement between you and Bentham Science Publishers Ltd. Please read this License Agreement
carefully before using the ebook/echapter/ejournal (“Work”). Your use of the Work constitutes your
agreement to the terms and conditions set forth in this License Agreement. If you do not agree to these terms
and conditions then you should not use the Work.
Bentham Science Publishers agrees to grant you a non-exclusive, non-transferable limited license to use the
Work subject to and in accordance with the following terms and conditions. This License Agreement is for
non-library, personal use only. For a library / institutional / multi user license in respect of the Work, please
contact: permission@benthamscience.org.
Usage Rules:
1. All rights reserved: The Work is the subject of copyright and Bentham Science Publishers either owns the
Work (and the copyright in it) or is licensed to distribute the Work. You shall not copy, reproduce, modify,
remove, delete, augment, add to, publish, transmit, sell, resell, create derivative works from, or in any way
exploit the Work or make the Work available for others to do any of the same, in any form or by any
means, in whole or in part, in each case without the prior written permission of Bentham Science
Publishers, unless stated otherwise in this License Agreement.
2. You may download a copy of the Work on one occasion to one personal computer (including tablet,
laptop, desktop, or other such devices). You may make one back-up copy of the Work to avoid losing it.
The following DRM (Digital Rights Management) policy may also be applicable to the Work at Bentham
Science Publishers’ election, acting in its sole discretion:
● 25 ‘copy’ commands can be executed every 7 days in respect of the Work. The text selected for copying
cannot extend to more than a single page. Each time a text ‘copy’ command is executed, irrespective of
whether the text selection is made from within one page or from separate pages, it will be considered as a
separate / individual ‘copy’ command.
● 25 pages only from the Work can be printed every 7 days.
3. The unauthorised use or distribution of copyrighted or other proprietary content is illegal and could subject
you to liability for substantial money damages. You will be liable for any damage resulting from your misuse
of the Work or any violation of this License Agreement, including any infringement by you of copyrights or
proprietary rights.
Disclaimer:
Bentham Science Publishers does not guarantee that the information in the Work is error-free, or warrant that
it will meet your requirements or that access to the Work will be uninterrupted or error-free. The Work is
provided "as is" without warranty of any kind, either express or implied or statutory, including, without
limitation, implied warranties of merchantability and fitness for a particular purpose. The entire risk as to the
results and performance of the Work is assumed by you. No responsibility is assumed by Bentham Science
Publishers, its staff, editors and/or authors for any injury and/or damage to persons or property as a matter of
products liability, negligence or otherwise, or from any use or operation of any methods, products instruction,
advertisements or ideas contained in the Work.
Limitation of Liability:
In no event will Bentham Science Publishers, its staff, editors and/or authors, be liable for any damages,
including, without limitation, special, incidental and/or consequential damages and/or damages for lost data
and/or profits arising out of (whether directly or indirectly) the use or inability to use the Work. The entire
liability of Bentham Science Publishers shall be limited to the amount actually paid by you for the Work.
General:
1. Any dispute or claim arising out of or in connection with this License Agreement or the Work (including
non-contractual disputes or claims) will be governed by and construed in accordance with the laws of the
U.A.E. as applied in the Emirate of Dubai. Each party agrees that the courts of the Emirate of Dubai shall
have exclusive jurisdiction to settle any dispute or claim arising out of or in connection with this License
Agreement or the Work (including non-contractual disputes or claims).
2. Your rights under this License Agreement will automatically terminate without notice and without the
need for a court order if at any point you breach any terms of this License Agreement. In no event will any
delay or failure by Bentham Science Publishers in enforcing your compliance with this License Agreement
constitute a waiver of any of its rights.
3. You acknowledge that you have read this License Agreement, and agree to be bound by its terms and
conditions. To the extent that any other terms and conditions presented on any website of Bentham Science
Publishers conflict with, or are inconsistent with, the terms and conditions set out in this License
Agreement, you acknowledge that the terms and conditions set out in this License Agreement shall prevail.
CHAPTER 7 FATIGUE LIFE OPTIMIZATION OF LASER PEENED AIRCRAFT COMPONENTS ....... 302
Anoop Vasu and Ramana V. Grandhi
NOMENCLATURE ............................................................................................................................................. 303
INTRODUCTION ................................................................................................................................................ 303
INFLUENCING FACTORS FOR FATIGUE LIFE PREDICTION .............................................................. 308
Applied Load .................................................................................................................................................. 309
Laser Peening Induced Residual Stresses ...................................................................................................... 310
Residual Stress Relaxation ............................................................................................................................. 311
Fatigue Life .................................................................................................................................................... 312
Re-Peening ..................................................................................................................................................... 314
OPTIMIZATION PROCEDURE....................................................................................................................... 314
Design Variables ............................................................................................................................................ 315
Re-Peening Schedule .............................................................................................................................. 315
Mid-Span Duration ................................................................................................................................ 316
Peak Pressure ....................................................................................................................................... 317
Percentage Overlap .............................................................................................................................. 317
Optimization strategy ................................................................................................................................... 319
RESULTS AND DISCUSSION ......................................................................................................................... 320
Compressive Residual Stress Optimization Results ..................................................................................... 320
Optimization of the Re-Peening Schedule ................................................................................................... 322
CONCLUSION ................................................................................................................................................. 323
CONFLICT OF INTEREST ............................................................................................................................. 324
ACKNOWLEDGEMENTS ............................................................................................................................... 324
REFERENCES ................................................................................................................................................... 324
PREFACE
This book covers a variety of aspects of design, analysis, and manufacturing of aerospace
structures and materials, which include design/failure criteria for aerospace structures,
materials selection, manufacturing methods, engineering design and analysis of structural
assemblies and components, concurrent engineering, and interdisciplinary design technology.
Challenges that current aerospace engineers are faced with are as follows. First, they must
understand the new and existing materials and the level of engineering already applied in their
creation. Second, they need to develop a working design by combining material behavior and
structural function to meet all the design requirements. Third, an optimum design has to be
conducted based on the original working design to address manufacturability and economical
constraints. Finally, the durability and longevity of the final structure should be continually
assessed.
This book equips aerospace engineers with the necessary knowledge and practical skills to
tackle these challenges in an industrial or R&D environment. Readers of this book will
acquire knowledge of a broad range of the entire process in the context of the design,
manufacturing, and analysis of aerospace structures and components. The readers will also be
exposed to advanced research and development in fields directly relevant to aerospace
engineering, which will improve our capability to design the next generation of aerospace
structures and materials.
Interesting topics that will be demonstrated through this book include: structural dynamics
and impact simulation, acoustic and vibration testing and analysis, fatigue analysis and life
optimization, reversing design methodology, non-destructive evaluation, remotely piloted
helicopter, surface enhancement of aerospace alloys, manufacturing of metal matrix
composites, aerospace applications of carbon nanotubes, carbon fiber reinforcements, variable
stiffness composites, aircraft material selection, etc. Furthermore, besides aerospace
engineering, the topics included in this book also impact a broad variety of engineering areas
such as mechanical engineering, materials engineering and science, civil engineering,
automotive engineering, computer engineering and science, etc.
The book is intended to serve as a reference for researchers, engineers, engineering faculty, as
well as graduate students. Finally, the editor would like to extend cordial appreciation to all
contributors for their great support and invaluable contributions.
Yucheng Liu
Department of Mechanical Engineering
Mississippi State University
MS 39762, USA
ii
List of Contributors
Department of Mechanical and Materials Engineering, Wright State University,
Anoop Vasu
Dayton, USA
Axel Herrmann Composite Technology Center GmbH (An Airbus Company), Stade, Germany
Georg Lonsdorfer Composite Technology Center GmbH (An Airbus Company), Stade, Germany
CHAPTER 1
NOMENCLATURE
Sα, Sβ Static mass moment of the fin about the elastic axis.
T Temperature.
Tα Pitching moment.
u Flow velocity.
v Unit direction vector of the air velocity.
w Displacements along the normal vector of each surface.
x Vector of dynamic states.
𝑥̅ ,𝑧̅ Position of the center of gravity related to elastic axis.
α Pitch angle.
β Flap angle.
αf, βf, hf Free-plays of the corresponding degrees of freedom.
γ Specific heat ratio.
ν Prandtl-Meyer function.
θ Elastic rotation vector.
θ Natural mode shapes of the elastic rotations.
ρ Density.
ω Natural frequency.
ψ Oblique wave angle.
ξ Generalized modal coordinates.
INTRODUCTION
Abbas et al. [15] used the third-order piston theory along with a 2-DOF
wing section model to analyze the supersonic flutter and post flutter
behaviors under the effects of cubic stiffness nonlinearities and free-plays
in both plunging and pitching motions. They observed LCOs and chaotic
behaviors over a large variety of configurations of the aeroelastic system.
Tang and Dowell [16] performed theoretical and experimental aeroelastic
studies to investigate the effect of nonzero initial angle of attack on the
flutter, LCO and gust response of a typical wing section with control
surface free-play in an incompressible flow.
In this chapter, modeling the unsteady aerodynamic loads acting on thick
supersonic fins are estimated by the local application of the piston theory.
To this aim the steady state properties of the flow over the surface of the
fin are obtained using the shock/expansion wave equations and the piston
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 9
where qi denotes the ith generalized coordinate and Qi is the ith generalized
force. Further, D, P and KE stand for the Rayleigh dissipation function,
potential and kinetic energies, respectively.
N
e en n t (2)
n 1
N
n n (t ) (3)
n 1
(4)
where ξ and f are vectors containing the generalized modal coordinates and
forces, respectively and Ks is the diagonal structural stiffness matrix,
defined as
where ωi is the natural frequency of the ith mode. The generalized forces
due to the unsteady aerodynamic pressure p over the fin’s surface S are
calculated as follows
1 K ,1 c
2 b 2 b
Tf
Kh ,ch
Ka ,ca
1K ,1c
2 b 2 b
It is assumed that free-play exists in all DOFs and static frictional torque is
applied on the pitch direction. The Rayleigh dissipation function (D) as
well as the potential (P) and kinetic (KE) energies of the system are
obtained as follows:
1
D (ch h 2 c 2 c 2 ) (7)
2
12 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
1
P ( K h h2 K 2 K 2 ) (8)
2
Substitution of Eqs. (7-9) into Eq. (1), one obtains the following governing
equations of motion
M s x Cx fs f a 0, (10)
where x = [h α β]T is the vector of dynamic states. Also, Ms and Cs are the
structural mass and damping matrices being defined as
m S S
M s S I zz I xz 2mxz (11)
S I xz 2mxz I xx
ch 0 0
Cs 0 c 0 (12)
0 0 c
where S mx and S mz are the static mass moment of the fin about
the elastic axis.
The vectors fs and fa in Eq. (10) represent the structural and aerodynamic
forces and moments acting on the fin which are defined as
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 13
fs Fs (h) Ts ( ) Tf ( ) M s ( )
T
(13)
fa La Ta Ma
T (14)
K h (h hf ) h hf
Fs (h) 0 hf h hf (15)
K (h h ) h hf
h f
K ( f ) f
Ts ( ) 0 f f (16)
K ( ) f
f
K ( f ) f
M s ( ) 0 f f (17)
K ( ) f
f
F (h ), M (b )
T (a)
Kh Kb Ka
-ac , -bc - hc
h (t ), a(t ), b(t )
hc ac , bc
According to first order the piston theory, the unsteady pressure over a flat
plate is expressed in terms of its transverse deflection w as follows:
u 2 1 M2 2
p (w w) (18)
M 2 1 u M 2 1
Where the over-dot represents the temporal derivation and the prime
symbol indicates spatial derivation, thus w is the slope of the transverse
deformation in the flow direction. Also M, ρ and u are the Mach number,
air density and airspeed over the surface, respectively.
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 15
For thin fins in zero angle of attack, weak compression and expansion
waves occur over the surface and thus the variations of flow properties (e.g.
Mach number, temperature and density) on the surface are negligible. But,
if the thickness of the fin is considerable or it is oriented in a nonzero angle
of attack, the flow properties change drastically behind the shock and
expansion waves. Having this in mind, one must obtain the flow properties
over the fin's surfaces and apply the piston theory locally on each surface.
The flow properties behind shock and expansion waves over the fin
surfaces can be obtained numerically using steady CFD solvers. However,
such computational tools are usually time consuming due to their
limitations for grid generation and iterative numerical schemes, especially
for 3D geometries. The shock/expansion wave theory is one of the simplest
methods which furnishes analytical relations for the solution of steady two
or three dimensional supersonic flows. Recalling that these waves are
inherently two dimensional in nature, a procedure is purposed to solve the
steady flow over the fin.
Fig. (3) shows a sketch of surfaces of a thick fin in supersonic flow. The
upper and lower surfaces of the fin are numbered individually and the
Mach number, density and temperature of the ith surface are denoted by
M i , i and, Ti respectively. The angle αi between the two following
surfaces i-1 and i is obtained as
The upstream air velocity on the i-1th surface can be decomposed into
tangential and normal components with respect to the edge Oi-1 Oʹi-1, where
the corresponding Mach components are shown by M ti1 and M i n
1.
O'i
Oi-1
i ( M ) ( M in1 ) (21)
and is the Prandtl-Meyer function being defined as
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 17
1 1 1 2
(M ) tan ( M 1) tan 1 M 2 1 (22)
1 1
with γ describing the specific heat ratio. Also, ψi is the oblique wave angle
that is determined by solving the following equation.
(M in1 )2 sin 2 ( i ) 1
tan i 2cot i (23)
(M in1 )2 ( cos 2 i ) 2
By determining the Mach number over the ith surface, the density,
temperature and velocity direction over the surface are evaluated. Beyond
an oblique shock the air density and temperature are obtained as
i ( 1) M i21 sin 2 i
(24)
i 1 2 ( 1) M i21 sin 2 i
i 2 ( 1) M i21 1
1
( ) (26)
i 1 2 ( 1) M i2
Ti 2 ( 1) M i21
(27)
Ti 1 2 ( 1) M i2
Also, the unit vector of the air velocity v i on the ith surface is defined as:
M it1t i 1 M (ni t i 1 )
vi (28)
( M it1 ) 2 ( M ) 2
18 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
where ti-1 is the unit vector of the edge Oi 1Oi1. The procedure continues
to any surface similarly. For this purpose, the tangential and normal
components of the Mach vector with respect to the edge Oi Oi are
obtained as:
Assuming very small elastic deflections of the fin, the velocity and slope of
the transverse deformations on each surface of the fin is obtained as
w (n ).v (31)
w n.e (32)
Substituting Eqs. (31) and (32) into Eq. (18), the localized piston theory
can be formulated over the fin’s surfaces as
u 2 1 M2 2
p
u M 2 1
(n ).v n.e (33)
M2 1
The unsteady aerodynamic pressure acting on the fin’s surfaces is obtained
in terms of the generalized modal coordinates after substitutions of the
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 19
modal expansions from Eqs. (2 and 3) into Eq. (33) which can be
summarized as:
N N
p u 2 Cn n (t ) u Cen n (t ) (34)
n 1 n 1
M2 2 (35)
Cen en .n
(M 2 1)1.5
(n n ).v (36)
Cn
M 1 2
Substituting Eq.(34) into Eq.(6), yields the following expression for the
generalized aerodynamic forces
N N
f m K mn
a
n Cmn
a
n (37)
n 1 n 1
where Kamn and Camn are respectively the elements of the aerodynamic
stiffness and damping matrices which are obtained by calculating the
following integrals over the whole surfaces of the fin.
In order to obtain the aerodynamic stiffness and damping matrices, the fin’s
surface is discretized into small elements as shown in Fig. (4) and the
integrals in Eqs. (38 and 39) are evaluated numerically. To this aim, the
integrands of Eqs. (38 and 39) are approximated using their nodal values
and element shape functions within each surface element. The Gauss
quadrature technique can be used for the numerical integration.
20 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
Z X
Bifurcation Theory
Phase Plots
Chaos
Lyapunov Exponents
The Lyapunov exponents are a powerful tool for diagnosing whether or not
an oscillation is chaotic. It actually characterizes the rate of separation of
infinitesimally close trajectories. Quantitatively, two trajectories in phase
space with initial separation δZ0 diverge at a rate given by:
Z (t ) et Z0 (40)
where λ is the Lyapunov exponent. The rate of separation can be different
for different orientations of initial separation vector. Thus, there is a variety
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 23
1 Z (t )
lim lim ln (41)
t Z0 0 t Z0
Numerical Results
1.35
CMD
CL D
0.2
1.2
1.05 0.1
0.05 0.1 0.15 0.2 0.25 0.05 0.1 0.15 0.2 0.25
t/c t/c
(a) (b)
Fig. (5). Comparison of the local piston theory and Euler method in estimating CL and CM versus the
thickness ratio at M=4 and zero angle of attack.
0.4
Local piston theory Local piston theory
1.8 Euler (2D) Euler (2D)
Euler (3D) Euler (3D)
0.3
1.6
CMD
CL D
0.2
1.4
0.1
1.2
1 0
0 5 10 15 20 0 5 10 15 20
D0(deg) D0(deg)
(a) (b)
Fig. (6). Comparison of the local piston theory and Euler method in estimations of CL and CM versus
the initial angle of attack at M=4 and t/c = 10%.
Fig. (5) shows the obtained results for different thickness to chord ratios
at Mach number 4 and zero angle of attack. The variation of the
aerodynamic lift and moment derivatives with respect to the initial angle
of attack for a fin with t/c=10% at Mach number 4 is depicted in Fig. (6).
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 25
To clarify the effect of the fin’s thickness and angle of attack on the lift
coefficients CL , CL , CL and moment coefficients CM , CM , CM several
h h
case studies were carried out. The fin is considered to be embedded in a
flow stream with the following characteristics
u10-5 u10-3
0
M=12
M = 12
M=3 -1
1.5
-0.5
.
-2 .
CLD
CLD
CL h
1
-3 M=3
-1
M = 12
-4
0.5 M=3
-1.5
0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25
t/c t/c t/c
-1 -5
u10 u10 u10-4
4
0
M =12 M = 12
3 -0.5
-1
M=3
CM D
. .
CMh
CMD
-1
2
-2 M=3
M=12
-1.5
1
M=3
-2 -3
0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25
t/c t/c t/c
Fig. (7). Variation of the aerodynamic derivatives versus thickness ratio at zero angle of attack.
26 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
u10-5 u10-3
0
M=12
M = 12
M=3 -1
1.5
-0.5
.
-2 .
CLD
CLD
CL h
1
-3 M=3
-1
M = 12
-4
0.5 M=3
-1.5
0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25
t/c t/c t/c
-1 -5
u10 u10 u10-4
4
0
M =12 M = 12
3 -0.5
-1
M=3
CM D
. .
CMh
CMD
-1
2
-2 M=3
M=12
-1.5
1
M=3
-2 -3
0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25
t/c t/c t/c
Fig. (8). Variation of the aerodynamic derivatives versus initial angle of attack at t/c = 10%.
To verify the aeroelastic model, a clamped delta wing as shown in Fig. (9)
is considered. An aeroelastic analysis is performed and the obtained
results are compared with those from the wind tunnel tests carried out at
NASA Langley wind tunnel by Tuovila and McCarty [21]. The
experiment was carried out at Mach 3 with the air density of 0.37kg/m3.
The wing is considered to be made of magnesium plate with 0.86mm
thickness, elastic modulus of 42.8GPa, density of 1766kg/m3 and Poisson
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 27
ratio of 0.35.
6"
45
Clamped
Fig. (9). Geometry of the delta wing in supersonic flow.
Using a finite element code, the first three natural modes and frequencies of
the wing are obtained and then the flutter speed and frequency of the wing
are determined using the proposed model.
300
200
Frequency (cps)
Damping
100 200
100
-100
500 1000 1500 2000 2500 500 1000 1500 2000 2500
Velocity (ft/s) Velocity (ft/s)
(a) (b)
Fig. (10). Variation of the aeroelastic damping and frequency versus the airspeed for the delta wing.
Table 2. The obtained flutter results for the delta wing in comparison with experiment.
Fig. (12) shows the flutter and divergence speed of the fin versus the
thickness ratio. The results illustrate that there is a remarkable reduction in
the aeroelastic stability margin when the thickness increases, particularly in
the low angles of attack.
Flutter
4 D = 0
q
Divergence
3.5
Mach Number
5q
2.5
q
10
2
0.05 0.1 0.15 0.2
t/c
Fig. (12). Variation of the flutter and divergence Mach number for a 2-DOF rigid fin versus
thickness ratio for different angles of attack.
30 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
Flutter
3.3 t/c=10% Divergence
Mach Number
15%
2.7 10%
20%
2.4
15%
2.1
20%
0 2 4 6 8 10
D(deg)
Fig. (13). Variation of flutter and divergence Mach number for a 2-DOF rigid fin with the angle of
attack at various thickness ratios.
10
Flutter
9 Divergence
8
Mach Number
t/c=10%
7
15%
6
20%
5
2
-5 0 5 10 15
Sweep angle (deg)
Fig. (14). Variation of flutter and divergence Mach number for a 2-DOF rigid fin versus sweep
angle.
Flutter
Divergence
t/c=10%
5
Mach Number
15%
4 20%
10%
15%
3
20%
2
0 0.2 0.4 0.6 0.8 1
Taper ratio
Fig. (15). Variations of the flutter and divergence Mach number for a 2-DOF rigid fin versus taper
ratio.
150mm
Rigid base
10
0 5
10
45mm
85 15mm
125mm
150mm
Fig. (16). Geometry and dimensions of the elastic supersonic fin.
Carbon-epoxy Foam
E1(GPa) 140 2.0
E2(GPa) 10 -
V12 0.3 0.3
G12(GPa) 5.2 -
G13(GPa) 5.2 -
G23(GPa) 3.4 -
1550 100
(kg / m3 )
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 33
The fin is clamped to a rigid base and is made up of a composite shell with
uniform thickness of 0.5mm, which forms the outer aerodynamic surfaces
and is filled by foam. The fin surface is considered to be a four layered
Carbon/epoxy composite with symmetric cross-ply orientation. Table 3
shows the materials properties of the fin.
The aeroelastic model is developed based on the first six natural vibration
modes of the fin which are extracted using a finite element model and are
shown in Fig. (17).
Fig. (17). The first six natural mode shapes of the elastic fin and their corresponding natural
frequencies.
34 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
Fig. (18) shows the variations of damping and frequency of the aeroelastic
modes with Mach number for flight at sea level. The obtained results show
that the flutter Mach number for the considered configuration is 5.97.
u10 2 u10 4
1 1.6
0 1.4
Frequency
Damping
-1 1.2
-2 1
-3 0.8
-4 0.6
3 4 5 6 7 8 3 4 5 6 7 8
Mach Number Mach Number
(a) (b)
Fig. (18). Plots of aeroelastic damping and frequency for the elastic fin at zero angle of attack.
Finally, for the last linear investigation, variation of the flutter Mach
number versus the initial angle of attack is shown in Fig. (19) which proves
that the stability of the fin decreases at higher angles of attack.
6
5.9
Mach Number
5.8
5.7
5.6
5.5
5.4
0 5 10 15 20 25
D(deg)
Fig. (19). Variation of the flutter Mach number of the elastic fin versus the initial angle of attack.
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 35
500
400
-5
Frequency
Damping
-10
300
-15
Mode 1
Mode 2
200
-20 Mode 3
Fig. (20). Variations of the aeroelastic damping and frequency versus the Mach number for a 3D fin.
36 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
In the next step, results from the nonlinear dynamics of the same fin is of
interest. As the plunging, flapping and pitching responses are quite similar
in behavior, the graphical illustrations are restricted only to the pitching
time-histories. Fig. (21a) shows the dynamic behavior of the fin as a
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 37
1 2 3 4
1
Pitch (Min/Max) Angle (deg)
0.8
0.6
0.4
3 4 5 6 7
3
Lyapunov Exponent
-3
-6
3 4 5 6 7
0.03
0.02
T(s)
T=0 T
0.01
0
3 4 5 6 7
Mach Number
Fig. (21). Variation of (a) pitch (min/max) angle of the fin, (b) Lyapunov exponent and (c) period
of the pitch time series versus the Mach number.
38 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
Finally, it seems that chaotic response starts at region four where M 7.2 .
The Lyapunov exponent is presented in Fig. (21b) at various Mach
numbers. The magnitude of the exponent changes to positive for
M 7.2 which confirms chaotic response in the fourth region. Fig. (21c)
depicts period of the oscillation at different Mach numbers. The periods of
oscillation are 0, 0.013, 0.027 and infinite for the consequent regions,
respectively. It means that at M 6.3 , a period doubling phenomenon
occurs. To have a better investigation, time histories and the phase-plane
diagrams for Mach numbers 3, 5.5, 6.5 and 7.5 are plotted in Figs. (22 to
25). These diagrams are obtained after the transient response, so the
responses converge to steady state behavior.
400 60
200
200 30
100
(deg/s)
h (mm/s)
(deg/s)
0 0 0
-100 -30
-200
-200
-60
-400
-4 -3 -2 -1 0 0 0.5 1 1.5 2 2.5 -0.6 -0.4 -0.2 0
h (mm) (deg) (deg)
-0.4
-2.1
1.4
-2.4 -0.5
1.6
(deg)
h (mm)
(deg)
-2.7 -0.6
1.8
-3 2 -0.7
0 5 10 15 0 5 10 15 0 5 10 15
Time (s) Time (s) Time (s)
Fig. (22). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at
Mach 3.
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 39
300 200
60
150 100
30
(deg/s)
h (mm/s)
(deg/s)
0 0 0
-30
-150 -100
-60
-300 -200
-4.5 -4.2 -3.9 -3.6 2.4 2.64 2.88 3.12 -0.72 -0.66 -0.6 -0.54 -0.48
h (mm) (deg) (deg)
2.4
-0.5
-3.6
2.6
-3.9
h (mm)
(deg)
(deg)
-0.6
2.8
-4.2
3 -0.7
-4.5
Fig. (23). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at
Mach 5.5.
800
100
400
400
50
200
(deg/s)
h (mm/s)
(deg/s)
0 0
0
-200 -50
-400
-400
-100
-800
-6.6 -5.5 -4.4 -3.3 2.5 3 3.5 4 4.5 -0.84 -0.72 -0.6 -0.48
h (mm) (deg) (deg)
2.1
-3
-0.84
2.8
-4 -0.72
(deg)
h (mm)
(deg)
-5 3.5 -0.6
-6 -0.48
4.2
Fig. (24). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at
Mach 6.5.
40 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
-160
2000 1200
-80
1000
600
(deg/s)
h (mm/s)
(deg/s)
0
0
0
-1000
-600 80
-2000
-1200 160
-1
7.5
0
-0.75
5
-4
h (mm)
(deg)
(deg)
-0.5
2.5
-8
-0.25
0
-12
Fig. (25). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at
Mach 7.5.
Effects of Thickness
0.69
0.46
0.23
6 9 12 15 18
Lyapunov Exponent
-2
-4
-6
6 9 12 15 18
0.03
0.02
T(s)
T=0 T
0.01
0
6 9 12 15 18
t/c(%)
Fig. (26). Variations of (a) pitch (min/max) angle of the fin, (b) Lyapunov exponent and (c) period of
the pitch time series versus the thickness at Mach 5.
42 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
80
400
400
40
200
(deg/s)
h (mm/s)
(deg/s)
0
0 0
-400
-200 -40
-800 -400
-80
0 2 4 6 -4 -3 -2 -1 0 0 0.2 0.4 0.6
h (mm) (deg) (deg)
2.4 -3
0.72
-2.7
3
0.64
(deg)
h (mm)
(deg)
-2.4
3.6 0.56
-2.1
0.48
4.2
0 10 20 30 0 10 20 30 0 10 20 30
Time (s) Time (s) Time (s)
Fig. (27). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at
t/c=8%.
400
4
80
200 2
40
h (mm/s)
(deg/s)
(deg/s)
0 0
0
-200 -2
-40
-400 -4
-80
-5 -4 -3 -2 -1 1.75 2 2.25 2.5 -0.59 -0.58 -0.57 -0.56 -0.55
h (mm) (deg) (deg)
-0.54
1.6
-1.5
-0.56
2
(deg)
h (mm)
(deg)
-3
2.4 -0.58
-4.5
Fig. (28). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at t/c
=14%.
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 43
400 6
60
200 3
30
h (mm/s)
(deg/s)
(deg/s)
0 0 0
-200 -30 -3
-400 -60 -6
-0.52
-1
1.6
-2 -0.54
1.8
(deg)
h (mm)
(deg)
-3
-0.56
2
-4
2.2
-0.58
Fig. (29). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at t/c
=16%.
1000 -100
600
500 -50
300
(deg/s)
h (mm/s)
(deg/s)
0 0
0
-500 50
-300
-1000 100
-600
-1.7
4 -0.75
-3.4
3
(deg)
h (mm)
(deg)
-0.6
-5.1
2 -0.45
-6.8
1 -0.3
Fig. (30). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at t/c
=18%.
44 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
1 2
0 3 6 9 12 15
0.03
T(s)
0.02
T=0
0.01
0
3 6 9 12 15
(deg)
Fig. (31). Variations of (a) pitch (min/max) angle of the fin and (b) period of the pitch time series
versus α0 at Mach 5.
The influence of initial angle of attack on the dynamic response of fins with
t/c = 10% loaded at a flow stream just as the previous case is depicted in
Fig. (31a). As can be seen, the fin remains stable in the first region. At α0 =
10.2° bifurcation occurs and after that stable LCOs are encountered. Fig.
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 45
80
400
400
40
200
(deg/s)
h (mm/s)
(deg/s)
0
0 0
-400
-200 -40
-800 -400
-80
0 2 4 6 -4 -3 -2 -1 0 0 0.2 0.4 0.6
h (mm) (deg) (deg)
2.4 -3
0.72
-2.7
3
0.64
(deg)
h (mm)
(deg)
-2.4
3.6 0.56
-2.1
0.48
4.2
0 10 20 30 0 10 20 30 0 10 20 30
Time (s) Time (s) Time (s)
Fig. (32). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at α0
= 7°.
46 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
400
4
80
200 2
40
h (mm/s)
(deg/s)
(deg/s)
0 0
0
-200 -2
-40
-400 -4
-80
-5 -4 -3 -2 -1 1.75 2 2.25 2.5 -0.59 -0.58 -0.57 -0.56 -0.55
h (mm) (deg) (deg)
-0.54
1.6
-1.5
-0.56
2
(deg)
h (mm)
(deg)
-3
2.4 -0.58
-4.5
Fig. (33). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at α0
= 12°.
400 6
60
200 3
30
h (mm/s)
(deg/s)
(deg/s)
0 0 0
-200 -30 -3
-400 -60 -6
-0.52
-1
1.6
-2 -0.54
1.8
(deg)
h (mm)
(deg)
-3
-0.56
2
-4
2.2
-0.58
Fig. (34). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at α0
= 15°.
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 47
500
0.5
400
0
Frequency
Damping
-0.5 300
-1 Mode 1
Mode 2 200
Mode 3
Fig. (35). Variation of the aeroelastic damping and frequency versus Mach number at x/c=0.55.
48 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
0.5
500
0
400
Frequency
Damping
-0.5 300
Mode 1
Mode 2 200
-1 Mode 3
Fig. (36). Variation of the aeroelastic damping and frequency versus Mach number at x/c=0.57.
Fig. (37). Variation of (a) pitch (min/max) angle of the fin, (b) Lyapunov exponent and (c) period
of the pitch time series versus the hinge position at Mach 5.
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 49
800 400 80
200 40
400
h (mm/s)
(deg/s)
(deg/s)
0 0
0
-200 -40
-400
-400 -80
1.8
-0.4
-2.8
2.25 -0.5
h (mm)
(deg)
(deg)
-3.5
-0.6
2.7
-4.2
-0.7
3.15
0 5 10 15 0 5 10 15 0 5 10 15
Time (s) Time (s) Time (s)
Fig. (38). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at
x/c=0.55.
200 80
300
100 40
150
(deg/s)
h (mm/s)
(deg/s)
0 0 0
-3
3
0.7
-2.7
3.5
h (mm)
(deg)
(deg)
0.6
-2.4
4
0.5
-2.1
Fig. (39). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at
x/c=0.58.
50 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
800 100
600
400 50
300
(deg/s)
(deg/s)
h (mm/s)
0
0
0
-50
-400
-300
-100
-6 -4.5 -3 -1.5 1 2 3 4 -0.9 -0.75 -0.6 -0.45
h (mm) (deg) (deg)
0
-6.4 -0.4
1.5
-4.8
-0.6
h (mm)
(deg)
(deg)
3
-3.2
-0.8
4.5
-1.6
Fig. (40). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at
x/c=0.628.
2000 1400
100
1000 700
h (mm/s)
(deg/s)
(deg/s)
0 0 0
-1000 -700
-100
-2000 -1400
-10 -5 0 5 10 -4 0 4 8 -1 -0.5 0 0.5 1
h (mm) (deg) (deg)
-10
-6
0.6
-5 -3
h (mm)
(deg)
(deg)
0 0
0
3
5 -0.6
Fig. (41). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at
x/c=0.65.
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 51
Also, it seems at this point that the unstable mode shifts from plunging to
pitching Figs. (35 and 36). In the third region, 0.58 x / c 0.67 different
types of behavior (periodic and chaotic) occur. Fig. (37b) shows variations
of the Lyapunov exponent versus x/c. Based on this figure, periodic
response is considered at 0.627 x / c 0.63 (position A). Also, a chaotic
behavior is predicted for the fin in the remaining positions of the third
region. Similar to other investigated parameters, in Fig. (37c) the period of
oscillation as a function of the hinge position is shown. Finally, time
histories and the phase-plane diagrams at different hinge positions are
illustrated in Figs. (38-41).
FreeStream
Fig. (43). Variation of (a) pitch (min/max) angle of the fin, (b) Lyapunov exponent and (c) period
of the pitch time series versus the sweep angle at Mach 5.
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 53
60
300
270
150 30
(deg/s)
h (mm/s)
(deg/s)
0
0 0
-270
-150
-30
-540
-300
0 1.5 3 4.5 -3 -2 -1 0 0 0.2 0.4 0.6
3.9 0.48
-2.6
3.6 0.51
-2.4
h (mm)
(deg)
(deg)
0.54
3.3
-2.2
0.57
3
-2
0.6
0 5 10 15 0 5 10 15 0 5 10 15
Time (s) Time (s) Time (s)
Fig. (44). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at
sweep angle -10°.
60
200
100
100 30
50
(deg/s)
h (mm/s)
(deg/s)
0 0 0
-50
-100 -30
-100
-200 -60
-4 -3.8 -3.6 -3.4 2.34 2.47 2.6 2.73 2.86 -0.7 -0.65 -0.6 -0.55 -0.5
h (mm) (deg) (deg)
-0.48
-4
2.4 -0.54
-3.8
h (mm)
(deg)
(deg)
2.56 -0.6
-3.6
-0.66
2.72
-3.4
-0.72
Fig. (45). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at
sweep angle 0°.
54 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
500 300
80
250 150
40
h (mm/s)
(deg/s)
(deg/s)
0 0 0
-40
-150
-250
-80
-300
-500
3 3.5 4 4.5 5 -3.2 -2.8 -2.4 -2 0.4 0.5 0.6 0.7 0.8
h (mm) (deg) (deg)
-3.3 0.8
3
-3
0.7
3.5
(deg)
h (mm)
(deg)
-2.7
0.6
4
-2.4
0.5
4.5
-2.1
0.4
29.7 29.8 29.9 30 29.7 29.8 29.9 30 29.7 29.8 29.9 30
Time (s) Time (s) Time (s)
Fig. (46). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at
sweep angle 3°.
1600
2000
200
(deg/s)
0 0
0
-100
-1000
-800
-200
-2000
-12
-5 1
-6 0.5
h (mm)
(deg)
(deg)
0
0
0
-0.5
5
6
-1
Fig. (47). Phase plane diagram and time histories of the fin in pitch, flap and plunge direction at
sweep angle 15°.
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 55
̅
𝐼 xx Mass moment of inertia 400kg cm2
𝐼 zz̅ Mass moment of inertia 100 kg cm2
̅
𝐼 xz Mass moment of inertia -20 kg cm2
M Fin's mass 1.5kg
A Location of the elastic axis, positive 0.08
rearward
B Semi-chord at the root 7.5 cm
x α = e x - αx Static unbalance parameter 0.0027
56 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
10
0 200 Zh
Frequency(Hz)
ZE
Damping ZD
-10
100
-20
-30 0
6 7 8 9 6 7 8 9
Mach Number Mach Number
Fig. (48). Variation of the aeroelastic damping and frequency versus the Mach number for the
linear model.
40
-20
-40
Plung (Max/Min) displacement (m)
0.4
0.2
-0.2
-0.4
6
Pitch (Max/Min) Angle (deg)
-2
-6
2 3 4 5 6 7 8
Mach Number
Fig. (50). Phase plane diagram and time histories of the fin with Tf ( ̇ ) = 0.5 N.mm in pitch
direction at Mach 2.5.
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 59
Fig. (51). Phase plane diagram and time histories of the fin with Tf ( ̇ ) = 0.5 N.mm in pitch
direction at Mach 4.5.
60 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
Fig. (52). Phase plane diagram and time histories of the fin with Tf ( ̇ ) = 0.5 N.mm in pitch
direction at Mach 7.65.
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 61
Fig. (53). Phase plane diagram and time histories of the fin with Tf ( ̇ ) = 0.5 N.mm in pitch
direction at Mach 8.
CONFLICT OF INTEREST
The authors confirm that they have no conflict of interest to declare for
this publication.
62 Frontiers in Aerospace Science, Vol. 1 Firouz-Abadi et al.
ACKNOWLEDGEMENTS
Declared none.
NOTES
1
Part of this chapter has been previously published in Journal of
Fluids and Structures. Volume 40, July 2013, Pages 163–184.
doi:10.1016/ j.jfluidstructs.2013.03.019.
REFERENCES
[1] D. D. Liu, Z. X. Yao, D. Sarhaddi and F. R. Chavez, "From piston theory to a
unified hypersonic/supersonic lifting surface method.," Journal of Aircraft, vol.
34, pp. 304-312, 1997.
[http://dx.doi.org/10.2514/2.2199]
[2] W. Zhang, Z. Ye and C. Zhang, "Analysis of supersonic aeroelastic problem
based on local piston theory method," 45th AIAA Aerospace Sciences Meeting
and Exhibit, Vols. AIAA Paper 2007--1072., 2007.
[http://dx.doi.org/10.2514/6.2007-1072]
[3] W. Zhang, Z. Ye, C. Zhang and F. Liu, "Supersonic flutter analysis based on a
local piston theory," AIAA Journal, vol. 47, pp. 2321-2328, 2009.
[http://dx.doi.org/10.2514/1.37750]
[4] R. D. Firouz-Abadi and S. M. Alavi, "Effect of thickness and angle of attack on
the aeroelastic stability of supersonic fins," The Aeronautical Journal , vol. 116,
pp. 777-792, 2012.
[http://dx.doi.org/10.1017/S0001924000007272]
[5] L. Librescu, P. Marzocca and W. A. Silva, "Linear/non-linear supersonic panel
flutter in a high-temperature field.," Journal of Aircraft, vol. 41, pp. 918-924,
2004.
[http://dx.doi.org/10.2514/1.679]
[6] L. Tang, R. E. Bartels, P. -C. Chen and D. D. Liu, "Numerical investigation of
Linear/Non-Linear Aeroelastic Response Frontiers in Aerospace Science, Vol. 1 63
CHAPTER 2
*Corresponding author Yucheng Liu: Department of Mechanical Engineering, Mississippi State University,
MS, 39762, USA; Tel/Fax: (001)662-325-1536/662-325-7223; E-mail: yliu@me.msstate.edu.
INTRODUCTION
developed for a 2DOF linear system, and then extended for a MDOF
system developed as a finite element model. Wind tunnel test is then
conducted to obtain the physical dynamic parameters. The finite element
model and the force determination method are then validated by
comparing the analytical results with the experimental results.
LITERATURE REVIEW
Analytical Study
treated as a flexible thin plate in the flow and the one-way coupling and
two-way coupling were discussed separately. The coupled fluid-structure
method was also used by Liu et al. [2] in calculation of wing flutter.
Fabunmi [3] presented a method of using the pseudo-inverse technique to
determine the operating vibratory loads on a structural system. In his
work, two types of uniform beam were studied: the cantilever beam and
the free end beam. Results of numerical calculations showed that the
accuracy of the determined forces correlates well with a parameter that is
related to the number of modes contributing to the response of the
structure at a given frequency. Shyy and Kamakoti [4] studied the
interactions between rigid and flexible structures and fluid based on an
airfoil system model. A loosely coupled approach was used to perform the
combined fluid and structure computations. Two different airfoil
configurations were employed to obtain the displacement of the airfoil
and the pressure on it. A suitable interfacing technique was incorporated
to couple and synchronize the flow and structure solver. Liu and Shepard
[5] addressed an approach of dynamic force identification based on
enhanced least squares and total least-squares schemes in the frequency
domain. This approach is effective in reducing the random errors that
occur in structural response signals. Two regularization filters, named as
the truncated singular value decomposition (TSVD) filter and the
Tikhonov filter, were employed in conjunction with the conventional
least-squares scheme at specific frequencies. A new least-squares form of
the Morozov’s discrepancy principle was formulated to aid in selecting
the optimum regularization parameter for these filters at each frequency.
The accuracy in using conventional least-squares, TSVD-based
least-squares, and Tikhonov filter-based least-squares schemes were
compared analytically and numerically in their study. E. Parloo et al. [6]
demonstrated a method of identifying dynamic force by means of
in-operation modal models. The applicability of the sensitivity-based
normalization approach for force identification on the basis of output-only
data was evaluated. The quality of the reconstructed FRF data and the
advantages of using an iterative weighted pseudo-inverse approach over a
classic pseudo-inverse one were discussed through experiments
Vibratory Force for Aircraft Wings Frontiers in Aerospace Science, Vol. 1 69
Experimental Study
dominant. Lee [20], Joshi [21], Sun [22], and Nakade [23] developed
different experimental methods to measure the unsteady aerodynamic
loadings and investigate the interaction between the loadings and the
object’s dynamic structural response.
Here
1 k ke 0 1
x m 0 c dC L
x , b , M 0 I , C K , K ke k ke 2 , H 1 c0 ,
c a 2 d 0 c 4 c (2)
3 c
c 0
c dC L
1
L 4 c
2 d 1 c 0 2 1 c 0 3 c 0 8 dC L
c
4 c c
/
4 c 4 c d
where m and I are mass and mass moment of inertia about the gravity
center, respectively; fa is the excitation force from the aileron; α is the
proportional damping constant; k and kθ are translational and rotational
structural stiffness, respectively, e is the distance between the gravity and
the elastic center; U denotes air speed; ρ denotes air density; c denotes
chord length; c0 is the distance between the leading edge and the elastic
center; and CL is the local lift coefficient. The derivative dCL/dθ is
assumed to be constant, with a theoretical value of 2π for incompressible
flow.
72 Frontiers in Aerospace Science, Vol. 1 Bi and Liu
k x
fa
U
c
k
co e ca
c
Rewrite Eq. (1) into the state space form and we have:
AY Bf
Y (3)
a
Here
0 I
Y [ x, x , , ] T , A 1 1 ,B [0,M 1 ] T (4)
M (K U H) M (C UL)
2
where A and B are the system and input matrices, respectively. The
system states then can be determined from Eq. (3) in the frequency
domain as:
Y(ω) = (jωI – A)-1BFa(ω) (5)
The eigenvalues λi of the system matrix A are functions of the air speed U,
which are related to the system’s natural frequencies ni and damping
ratios i as:
i i ni jni 1 i
2
(6)
As shown in Fig. (3a), when U increases, n1 increases and n2 decreases
steadily. The damping ratio of the second mode ζ2 decreases with an
increasing airspeed U and it reaches zero at the critical speed Ucr. At air
speed U > Ucr, ζ2 < 0, and hence the system becomes unstable.
Vibratory Force for Aircraft Wings Frontiers in Aerospace Science, Vol. 1 73
In the time domain, the equations of the motion for the system (Fig. 2)
can be written as:
or
From above equations, it can be seen that in order to determine the force
F(ω), the H(ω) and X(ω) has to be constructed or measured through
experiments.
74 Fro
ontiers in Aerospa
ace Science, Vol. 1 Bi and Liu
The objective
o of this
t study is to identify th the distributeed aerodynammic force
and moment
m actin
ng on an airrcraft wing during the fflight. By em mploying
the in
ndirect forcee identificatiion for this problem, thhe wing’s fr frequency
response matrix H(ω) has to o be obtaineed from a gground vibraation test
(GVTT), where bo oth the excitaation and ressponse are m measurable. NNext, the
outpu
ut structural response X(ω)
X is meassured and thhe F(ω) cann then be
calcullated from Eq.
E (9) or (10).
( Differeent force dettermination methods
used in
i this study are demonstrated in folllowing sections.
80
(a)
70
Natural Frequency
mode 1
60
50 mode 2
40
30
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
U/Ucr
0.2
(b)
0.15 mode 1
Damping Ratio
0.1
mode 2
0.05
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
U/Ucr
Fig. (3)). (a) Natural frequencies, (b) damping ratios as a function of airrspeed.
Forcee Determina
ation Metho
ods
H ()
i, j 1
ij
100
X(ω) = X(ω) + Nx, and N x X()
i1
(14)
where γ is the noise level. The frequency response matrix then can be
estimated from the measured data by [6]:
H1 = GFXGFF-1 (15)
H2 = GXXGXF-1 (16)
76 Frontiers in Aerospace Science, Vol. 1 Bi and Liu
Here “H1” and “H2” are the estimated frequency response matrices, GFF
and GFX are auto-power spectrum of force (F) and cross-power spectrum
of force (F) and displacement (X), respectively. Likewise, GXX and GXF
are auto-power spectrum of displacement (X) and cross-power spectrum
of displacement (X) and force (F). The frequency response matrices
determined by Eq. (15) or (16) are transposed and inverted, and then are
substituted into Eq. (9) or (10) to calculate the forces F(ω). Such method
is called “H1” and “H2” force determination method.
Eqs. (17 and 18) calculate the inverted frequency response matrix directly,
which can reduce the error introduced during the inversion of the
frequency response matrix. The forces then can be calculated by
substituting Eqs. (17), (18) into Eq. (9) or (10).
pseuddo-inverse method,
m wherre H11 is thee very first response fuunction in
the fo
orce responsee matrix [H]. The actual and estimateed force andd moment,
as weell as the perrcentage erro
ors are also displayed inn Figs. (5 annd 6). As
shown n from these figures, by using th the pseudo-iinverse metthod, the
percentage errors of the estimmated force and momennt are very low; with
average values at 0.085% and d 0.047%, resspectively.
-4
x 10
2
Frequency Response Function H(w)
0
2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
Seconndly, the freequency resp ponse functiion, actual aand estimateed forces
and moments,
m as well as percentage errorrs are re-calcculated by ussing “H1”
and “H
“ 2” force determination
d n methods. The sample rate to calcculate H1
and H2 is 20 Hz, and the perriod is 0.05 seconds. Beecause the eexcitation
force and noise added in this study are the ranndom numbbers, the
simullation resultss will be sliightly differrent each tim
me the calcuulation is
perforrmed. Thus, we have to redo the caalculation forr a certain am mount of
times to obtain th he average values
v of thee results andd use those vvalues to
comppare and validate the diffferent force ddeterminatioons. In this sstudy, the
calcullation is repeated 100 tiimes so that the obtained average reesults are
78 Fro
ontiers in Aerospa
ace Science, Vol. 1 Bi and Liu
0.15
0.1
0.05
0
2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.15
0.1
0.05
0
2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
% Force Determination Error
0.4
0.2
0
2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.005
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.01
Estimated Moment
0.005
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
% Moment Determination Error
0.2
0.1
0
2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
-4
x 10
0
2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.3
0.2
0.1
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
Estimated Translational Force
0.4
0.3
0.2
0.1
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
% Force Determination Error
15
10
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.01
Actual Moment
0.005
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.01
Estimated Moment
0.005
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
% Moment Determination Error
30
20
10
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
-4
X 10
2
Frequency Response Function H(w)
0
2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.4
0.2
0.1
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.3
0.2
0.1
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
40
30
20
10
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
Finallly, actual, esstimated forcces and momments, and tthe percentaage errors
are reecalculated using
u the “F1” and “F2” m
methods andd plotted in (Figs. 13
to 16)). The displaayed results are
a obtainedd at the airspeed of 7.6666 m/s and
82 Fro
ontiers in Aerospa
ace Science, Vol. 1 Bi and Liu
noise level of 5% %. Based on n Figs. (13 aand 14), thee average peercentage
errorss for the forcce and moment yielded ffrom the “F1” method arre 2.69%
and 0.47%,
0 whilee these valu ues are 8.90% % and 0.47%% given by the “F2”
metho od, according g to the Figss. (15 and 166).
Actual Moment 0.015
0.01
0.005
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.01
Estimated Moment
0.005
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
% Moment D e termination E rror
40
30
20
10
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.4
0.2
0.1
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.3
0.2
0.1
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
15
10
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.005
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.01
Estimated Moment
0.005
0
0 2 4 6 8 10 12 14 16 18 20
% Moment Determination Error
Frequency (Hz)
30
20
10
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.4
0.2
0.1
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.3
0.2
0.1
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
40
30
20
10
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.01
0.005
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
0.001
Estimated Moment
0.005
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
% Moment Determination Error
40
30
20
10
0
0 2 4 6 8 10 12 14 16 18 20
Frequency (Hz)
The presented force determination methods (Eqs. 9 to 18) are then applied
for multi-degree-of-freedom wing models.
Wn fa
qn
vn
Elasticity center w3
line fa
q3
U
v3
w2 Gravity center
line f a
q2
v2
w1
fa l
q1
v1
co yq ca
c c 0 0 0 0
c c 0 0 0 0
0 0 12 d 6ld 12 d 6ld (20)
Ki
0 0 6ld 4l 2 d 6ld 2l 2 d
0 0 12 d 6lb 12 d 6ld
0 0 6ld 2l 2 d 6lb 4l 2 d
Here a = (ρIpl)/6, b = (ρAl)/420, c = GJ/l, d = EI/l3, yθ is the Y-distance
between the wing’s gravity and elastic center; ρ is the material’s density;
Ip and I are the polar and regular moment of inertia of the wing model’s
cross section, respectively; A is the cross-sectional area; l denotes the length
of single element; G denotes the shear modulus of elasticity; J represents
the polar moment of inertia; and E is the Young’s modulus (see Fig. 17).
The input vector b and the proportional damping matrix C are given as:
b = [-ca, -ca, …, -ca, 1, 1, …, 1]; C = αM + βK (21)
where ca is the Y-distance between the trailing edge and the gravity center,
and α and β are the proportional damping constant.
The displacement vector x (assume this model has n elements):
x = [θ, x]T = [θ1, …, θn, x1, …, xn]T (22)
th
where θi is the i element’s rotational displacement and xi is its
translational displacement. The system’s mass and stiffness matrices can
be written as:
2a a 0 156 by 0 54by 0 0 0
a 4a a 54by 0 312 by 0 54by 0 (23)
0 a 4a 0 0 54by 0 312 by 0
M 156 by 54by 0 156 b 22lb 54b 13b 0 0
0 0 0 22lb 4l2b 13lb 3l2b 0 0
54by 312 by 54by 54lb 13lb 312 b 0 54b 13lb
0 0 0 13lb 3l2b 0 8l2b 13lb 3l2b
0 54by 312 by 0 0 54b 13lb 312 b 0
0 0 0 0 0 13lb 3l2b 0 8l2b
Vibratory Force for Aircraft Wings Frontiers in Aerospace Science, Vol. 1 87
c c 0 0 0 0 0 0 0
c 2c c 0 0 0 0 0 0 (24)
0 c 2c 0 0 0 0 0 0
K 0 0 0 12 d 6 ld 12 d 6 ld 0 0
0 0 0 6 lb 4l2 d 6 ld 2l2 d 0 0
0 0 0 12 d 6 ld 24 d 0 12 d 6 ld
0 0 0 6 ld 2l2 d 0 8l2 d 6 ld 2l2 d
0 0 0 0 0 12 d 6 ld 24 b 0
0 0 0 0 0 6 lb 2l2 d 0 8l2 d
0 1 0
1 c0
0 c( ) 1
0 c dCL 4 c (25)
H
2 d 0 0
1 c
c( 0 )
4 c
3 c
1 c( 0 ) 0
4 c
1 c 1 c 3 c 8 dCL 3 c
c dCL c( 0 ) c2[( 0 )( 0 ) 1 c( 0 )
(26)
L 0
2 d
4 c 4 c 4 c d 4 c
1 c0 2 1 c0 3 c0 8 dCL
0 c( ) c [( )( ) 1
4 c 4 c 4 c d
Boundary conditions are added on the global system, which assume that
the initial bending, torsional, and translational displacements are zero.
Similarly, for this multi-degree-of-freedom system, the governing
88 Fro
ontiers in Aerospa
ace Science, Vol. 1 Bi and Liu
, x n , 1,...,, n , x 1,..., x n ] T
Y [1,...,n , x1,..., (27)
Fig. (18). depicts the relationship betweenn the naturaal frequencies and the
airspeeed, and Figg. (19) displays the rellationship beetween the damping
ratioss and the airrspeed. From
m those figuures, it can bbe found thhat as the
airspeeed U increaases, the dam
mping ratio oof the first m
mode ζ2 decreeases and
becom mes zero at the
t critical speed
s Ucr. A
At air speed U > Ucr, ζ2 < 0, and
hencee the system becomes un nstable. The same phenomenon was observed
from the two degrrees-of-freed dom system.
The first
f and second mode shapes of the aircraft wing model at the
airspeeed U = 0 aree plotted in Figs.
F (20 andd 21), wheree the first moode shape
referss to bending and the secoond mode shhape refers too torsion. Tabble 1 and
2 list the relation
nships betweeen natural ffrequencies and dampinng rations
and thhe airspeed (including
( th
he first six m
modes).
40
35 (a)
30
Natunal Frequency (Hz)
25
20
15
10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
U/Ucr
(b)
0.4
0.35
0.3
0.25
Damping Ratio
0.2
0.15
0.1
0.05
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
U/Ucr
0
(a)
-0.005
Displacement for the wing
-0.01
-0.015
-0.02
-0.025
-0.03
-0.035
20
10 20
15
10
5
0 0
Chord direction
Span direction
0.03 (b)
0.02
Displacement for the wing
0.01
-0.01
-0.02
-0.03
20
10 20
15
10
5
0 0
Chord direction Span direction
-4
x 10
1
0.9
0.8
Frequency respones function H(w)
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
Fig. (22). FRF H11 at U/Ucr = 0 with 5% noise using traditional inverse method, based on
multi-degree-of-freedom model.
92 Fro
ontiers in Aerospa
ace Science, Vol. 1 Bi and Liu
0.2
0.15
Actual Force
0.1
0.05
0
0 10 20 30 40 50 60 70 80 90 100
Frequecy(Hz)
0.2
0.15
Estimated Force
0.1
0.05
0
0 10 20 30 40 50 60 70 80 90 100
Frequecy(Hz)
2
%Force determination error
1.5
0.5
0
0 10 20 30 40 50 60 70 80 90 100
Frequecy(Hz)
0.08
Actual moment
0.06
0.04
0.02
0
0 10 20 30 40 50 60 70 80 90 100
Frequency (Hz)
0.1
Estimated moment
0.08
0.06
0.04
0.02
0
0 10 20 30 40 50 60 70 80 90 100
Frequency (Hz)
% Moment determination error
0
0 10 20 30 40 50 60 70 80 90 100
Frequency (Hz)
Next, the frequency response function H11, actual, estimate forces and
moments, as well as percentage errors are calculated by “H1” and “H2”
methods. The noise is added through Eqs. (13 and 14). The results plotted
in following figures are formulated at air speed of 9.34 m/s and noise
level of 5%. As shown in Figs. (25 and 26), the average percentage errors
for force and moment calculated from H1 method are 17.36% and 15.80%.
Nevertheless, the percentage errors given by H2 method are 22.42% and
20.74%, as displayed in Figs. (27 and 28).
0.1
0.08
Actual Force
0.06
0.04
0.02
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
0.1
0.08
Estimated Force
0.06
0.04
0.02
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
%Force determination error
100
50
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
Fig. (25). Actual, estimated translational force and percentage error, U/Ucr = 2/3 with 5% noise
using H1 method, based on multi-degree-of-freedom model.
94 Frontiers in Aerospace Science, Vol. 1 Bi and Liu
0.01
0.008
oment
0.006
Actual M
0.004
0.002
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
0.01
ent
0.008
atedMom
0.006
0.004
Estim
0.002
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
inationerror
150
ent determ
100
50
%Mom
10 20 30 40 50 60 70 80 90
Frequency(Hz)
Fig. (26). Actual, estimated rotational moment and percentage error, U/Ucr = 2/3 with 5% noise
using H1 method, based on multi-degree-of-freedom model.
0.08
Actual Force
0.06
0.04
0.02
10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
0.1
0.08
atedForce
0.06
Estim
0.04
0.02
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
inationerror
150
Forcedeterm
100
50
%
0
10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
Fig. (27). Actual, estimated translational force and percentage error, U/Ucr = 2/3 with 5% noise
using H2 method, based on multi-degree-of-freedom model.
Vibratory Force for Aircraft Wings Frontiers in Aerospace Science, Vol. 1 95
0.01
0.008
0.004
0.002
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
0.01
Estimated Moment
0.008
0.006
0.004
0.002
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
%Moment determination error
250
200
150
100
50
0
0 10 20 30 40 50 60 70 80 90
Frequency(Hz)
Fig. (28). Actual, estimated rotational moment and percentage error, U/Ucr = 2/3 with 5% noise
using H2 method, based on multi-degree-of-freedom model.
Finally, actual, estimated forces and moments, and the percentage errors
are recalculated using the “F1” and “F2” methods and plotted in Figs. (31
to 34). The displayed results are obtained at the airspeed of 9.34 m/s and
noise level of 5%. According to Figs. (29 and 30), the average percentage
errors for the force and moment yielded from the “F1” method are 8.77%
and 10.62%, while these values are 14.62% and 15.72% yielded from the
“F2” method, as shown in Figs. (31 and 32).
96 Frontiers in Aerospace Science, Vol. 1 Bi and Liu
0.1
0.08
Actual Force
0.06
0.04
0.02
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
0.1
0.08
atedForce
0.06
Estim
0.04
0.02
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
150
inationerror
100
Forcedeterm
50
%
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
Fig. (29). Actual, estimated translational force and percentage error, U/Ucr = 2/3 with 5% noise
using F1 method, based on multi-degree-of-freedom model.
0.01
0.008
oment
0.006
ctual M
0.004
A
0.002
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
0.01
0.008
atedMoment
0.006
stim
0.004
E
0.002
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
120
inationerror
100
80
ent determ
60
40
Mom
20
%
10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
Fig. (30). Actual, estimated rotational moment and percentage error, U/Ucr = 2/3 with 5% noise
using F1 method, based on multi-degree-of-freedom model.
Vibratory Force for Aircraft Wings Frontiers in Aerospace Science, Vol. 1 97
0.1
0.08
Actual Force
0.06
0.04
0.02
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
0.1
0.08
EstimatedForce
0.06
0.04
0.02
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
150
%Forcedeterminationerror
100
50
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
Fig. (31). Actual, estimated translational force and percentage error, U/Ucr = 2/3 with 5% noise
using F2 method, based on multi-degree-of-freedom model.
0.01
0.008
oment
0.006
Actual M
0.004
0.002
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
0.015
oment
0.01
EstimatedM
0.005
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
inationerror
100
80
ent determ
60
40
om
20
%M
0
10 20 30 40 50 60 70 80
Frequency(Hz)
Fig. (32). Actual, estimated rotational moment and percentage error, U/Ucr = 2/3 with 5% noise
using F2 method, based on multi-degree-of-freedom model.
98 Frontiers in Aerospace Science, Vol. 1 Bi and Liu
DISCUSSIONS
In above sections, the forces and moments estimated from the traditional
pseudo-inverse method, “H1” and “H2” methods, as well as “F1” and “F2”
methods are compared to the actual values. From these comparisons, it is
found that the errors caused by the pseudo-inverse method are much
lower than those errors yielded from other methods, which is because in
that method, the mass, stiffness and damping matrices are directly
obtained from the two degrees-of-freedom and multi-degree-of-freedom
models. However, in reality, those matrices are difficult to be obtained
from a real complicated model. In the “H1”, “H2”, “F1”, “F2” methods
which use cross-spectrum and auto-spectrum, the frequency response
function is estimated from the input force and the corresponding structural
response. Therefore, such methods are easier to apply for force
determination in real problems.
Compared to the “H1” and “H2” methods, the “F1” and “F2” methods yield
lower errors in identifying the force and moment values. This is because
the direct determination methods directly calculate the inverted frequency
response matrix, therefore eliminate the error introduced during the
inversion of the frequency response matrix. It is also found from our
study that when the airspeed and noise level increase, the force
determination errors also rise.
EXPERIMENT TESTINGS
Experiment Model
A 3D wing model with dimensions fitting the wind tunnel test was
designed using the commercial CAD package Solid Edge (Fig. 33). The
CAD model was then manufactured in the rapid prototype laboratory in
University of Louisville (Fig. 34). The airfoil shape of this model is
NACA-65A004 and its dimensions are listed in Table 3.
rectangular cutouts were created at the side surfaces of both ends of the
boom. The cables that connect the accelerometers and signal amplifier
were fed through the interior of this model in order to eliminate
aerodynamic influence. Thus, both the wing and boom are hollow, the
thickness of the wing model is 1mm, and an 8.2 mm × 8.5 mm space is
available inside the boom.
At the root of the wing, a plate is attached to fix the wing to the wind
tunnel wall. In this plate, a helicoids hole is created to connect the model
with the wind tunnel. The material of the wing model is Dura Form
polyamide (PA), with Young’s modulus 1600 MPa, Poisson’s ratio 0.4,
and density 590 kg/m3.
Table 3.
3 Dimensions of wing model.
Diameter of bo
oom 1 7.27 (mm)
Grou
und Vibratio
on Test
Experriment Setup
p
The purpose
p of th
he ground vibration
v testt is to obtainn the airfoill model’s
frequeency respon nse functionss (FRFs) byy measuring the excitatiion force
and itts behavior of
o response. Fig. (35) shoows the flow w chart of thiis test. In
this test,
t a calibrrated impactt hammer ((with a rubbber tip) wass used to
excitee the model and two acccelerometerrs were employed to meeasure its
responses (Fig. 36 6). In that figure, a sensoor was instaalled at the end of the
impacct hammer to o measure th he excitationn force, and its sensitivitty is 2.25
mv/NN. During thee test, the wiing model w was rigidly m mounted by cclamping
Vibratorry Force for Aircrraft Wings Frontiers in A
Aerospace Sciencee, Vol. 1 101
it onto a steel tubbe (Fig. 33). Next, 22 tesst points weere excited separately
by the impact haammer and the t model’s responses w were recordeed by the
acceleerometers. These
T test points
p are eiither locatedd at the leadding and
trailin
ng edges, at accelerometter locations,, or at the wiing-to-boom m joint, as
shown n in Fig. (37
7). The choosing of an am mount of tesst points enssured that
the ob btained mod de shapes weere related too the entire wwing model.. In order
to miinimize the errors, 10 tests were reepeated at eevery test pooint. The
resoluution of the data
d acquisittion system is 0.5 Hz, thhe sampling time is 2
sec, and
a the samp pling frequenncy is 5000 HHz.
Moda
al Analysis Methods
M
The SDOF
S metho
od is develo
oped based oon frequenccy domain m
models of
Vibratorry Force for Aircrraft Wings Frontiers in A
Aerospace Sciencee, Vol. 1 103
the frequency resp ponse functions (FRFs) eevaluated inn the neighboorhood of
a singgle mode, which
w is an approximatte method thhat ignores complex
conjuugate contrib bution and can solve for both m modal frequeency and
residu
ue. The direect approach h identifies the continuuous-time m model and
parammeters directly from th he dynamic response ddata of the system.
Becau use the paarameters of o the moddel can ooffer clear physical
interp
pretation, thiis method haas been incrreasingly appplied in receent years
especcially in the area of ind dustrial appllication. Leaast squares complex
exponnential (LSCCE) method is a time doomain modaal analysis m method. It
exploores the relattionship betwween the immpulse respoonse functionns (IRFs)
of a multi-degree
m e-of-freedomm system andd its compleex poles andd residues
througgh a compleex exponential. The soluution of thiss model usinng LSCE
algoriithm leads too the establiishment of a polynomiall whose rootts are the
comp plex roots of the system.
The modal
m analyssis results caalculated by these three methods aree listed in
Table 4. Figs. (388 and 39) display the freequency respponse functiion H11,11
measu ured from thhe experimen nt and the syynthesized H11,11 by assuuming the
airfoil system a lin
near two deg grees-of-freeedom model.
H 11.11
0.12
0.1
0.08
Magnitude
0.06
0.04
0.02
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
Fig. (38). The frequency response functtion H11,11 measuured from the expperiment.
104 Frontiers in Aerospace Science, Vol. 1 Bi and Liu
r=1 14.36 14.79 10.67 0.0121 0.01 0.0189 1 1.1370 1 1.129 1 0.909
1
r=2 31.82 31.57 31.11 0.0033 0.0057 0.0043 1 -0.9898 1 -1.0267 1 -1.0166
r=1 14.34 14.76 10.68 0.0116 0.0161 0.0102 1 1.0760 1 1.1950 1 1.3187
2
r=2 31.82 31.61 31.12 0.0078 0.0045 0.0044 1 -0.9598 1 -0.9793 1 -0.9665
r=1 14.4 14.1 10.74 0.0117 0.0261 0.0095 1 1.1650 1 1.1349 1 1.3135
3
r=2 31.81 31.62 31.12 0.0077 0.0043 0.0044 1 -0.9698 1 -0.9759 1 -0.9690
r=1 14.14 14.66 13.44 0.0124 0.0396 0.0094 1 1.490 1 1.3362 1 0.9876
4
r=2 31.77 31.59 31.1 0.007 0.0043 0.0043 1 -0.9731 1 -0.9769 1 -0.9847
r=1 13.85 13.85 13.43 0.0136 0.0883 0.0095 1 1.1630 1 0.9259 1 1.1606
5
r=2 31.74 31.58 31.08 0.0068 0.0045 0.0045 1 -0.9765 1 -0.9766 1 -1.0162
r=1 13.76 14.16 13.4 0.0135 0.1011 0.0096 1 1.1532 1 0.9508 1 1.2752
6
r=2 31.74 31.57 31.05 0.0066 0.0045 0.004 1 -0.9757 1 -0.9757 1 -0.8965
r=1 13.74 13.66 13.39 0.0144 0.086 0.0085 1 1.1599 1 0.8799 1 1.2246
7
r=2 31.7 31.54 31.05 0.0067 0.0046 0.0049 1 -0.9755 1 -0.9797 1 -0.9822
r=1 13.76 13.58 13.38 0.0145 0.0155 0.0098 1 1.159 1 0.9978 1 1.4757
8
r=2 31.83 31.57 31.07 0.0068 0.005 0.0049 1 -0.9761 1 -0.9779 1 -0.9862
r=1 13.74 13.59 13.35 0.0165 0.0103 0.0099 1 1.1606 1 1.1091 1 1.2032
9
r=2 31.74 31.59 31.08 0.0071 0.005 0.0051 1-0.9768 1 -0.9784 1 -0.9854
r=1 13.71 13.49 13.36 0.0174 0.0313 0.018 1 1.1667 1 1.120 1 1.0956
10
r=2 31.72 31.74 31.23 0.0048 0.0048 0.005 1 -0.9763 1 -0.9795 -0.9787
Vibratory Force for Aircraft Wings Frontiers in Aerospace Science, Vol. 1 105
(Table 4) contd.....
r=1 13.72 13.67 13.36 0.0167 0.0451 0.012 1 1.1568 1 1.0871 1 1.1234
11
r=2 31.77 31.69 31.18 0.0054 0.0048 0.0052 1 -0.9782 1 -0.9787 1 -0.9838
r=1 13.72 13.46 13.27 0.0265 0.0317 0.0122 1 1.1739 1 1.2358 1 1.1591
12
r=2 31.72 31.75 31.24 0.0059 0.0053 0.0054 1 -1.1035 1 -0.9815 1 -0.9830
r=1 13.71 13.61 13.33 0.0191 0.0309 0.0193 1 1.1656 1 1.1811 1 1.1681
13
r=2 31.72 31.82 31.31 0.0059 0.0048 0.005 1 -0.8974 1 -0.9806 1 -0.9766
r=1 13.69 13.68 13.36 0.0165 0.0201 0.011 1 1.1587 1 1.1021 1 1.1172
14
r=2 31.73 31.78 31.28 0.0066 0.0047 0.0051 1 -0.9210 1 -0.9802 1 -0.9730
r=1 13.69 13.6 13.37 0.0154 0.0101 0.0102 1 1.1567 1 1.0605 1 1.2017
15
r=2 31.72 31.82 31.32 0.0055 0.0045 0.0048 1 -0.8032 1 -0.9792 1 -0.9700
r=1 13.7 13.8 13.35 0.0156 0.0153 0.0138 1 1.1515 1 1.0459 1 1.1893
16
r=2 31.71 31.83 31.33 0.0043 0.0033 0.0048 1 -0.9073 1 -0.9769 1 -0.9611
r=1 13.71 13.93 13.37 0.0149 0.2111 0.0102 1 1.1471 1 1.0505 1 1.1230
17
r=2 31.71 31.79 31.35 0.005 0.0042 0.0047 1 -0.8324 1 -1.2415 1 -0.9723
r=1 13.72 13.92 13.36 0.0155 0.124 0.0103 1 1.1489 1 1.1298 1 1.1156
18
r=2 31.72 31.89 31.4 0.0046 0.0045 0.0049 1 -0.8012 1 -0.9772 1 -1.0046
r=1 13.74 13.92 13.63 0.0149 0.0254 0.0228 1 1.1404 1 1.1888 1 1.0942
19
r=2 31.73 31.88 31.38 0.0074 0.0044 0.0047 1 -0.7548 1 -0.9765 1 -0.9957
r=1 14.34 12.8 13.45 0.0119 0.4737 0.0132 1 1.1683 1 1.5013 1 1.3084
20
r=2 31.75 31.62 31.13 0.0082 0.0038 0.0043 1 -0.7887 1 -0.9716 1 -0.9663
r=1 13.95 16.35 13.43 0.0151 0.03 0.0101 1 1.1648 1 1.3731 1 0.8575
21
r=2 31.72 31.86 31.35 0.0081 0.0037 0.0047 1 -0.7630 1 -0.9759 1 -0.9980
r=1 13.85 16.78 15.45 0.0139 0.008 0.0103 1 1.1099 1 1.2207 1 1.5981
22
r=2 31.72 31.72 31.89 0.0081 0.0028 0.0019 1 -0.9406 1 -0.9401 1 -0.6932
106 Frrontiers in Aerosp
pace Science, Vol. 1 Bi and Liu
H 11.11
0.14
0.12
0.1
0.08
Magnitude
0.06
0.04
0.02
0
0 10 20 30 40 50 60 70 80 90 100
Frequency(Hz)
From the Table 4, the results of naturaal frequencyy, damping rratio and
modee shape obtaiined by the three
t methodds are fairly closed to eaach other.
By comparing
c the frequeency responnse functions measureed from
experriment and obtained from f synthhesized calcculation, it can be
concluuded that the airfoil system can an be repreesented as a linear
two-d
degrees-of-frreedom systeem and solvee numericallyy.
Wind
d Tunnel Tesst
Theorretical Back
kground
The purpose
p of th
his experimeent is to idenntify the air-ddynamic forrce acting
on thee wings. Asssuming no exxcitation forcce, the equattion of motioon of this
systemm can be wrritten as Eq. (1), whose terms on thhe right handd side are
the aeerodynamic force which h are called aactual forcess on the system. The
goal of
o the experriment is to identify thee actual forcces using thee data of
displaacement meaasured during g the aircraft
ft flight periood.
Vibratory Force for Aircraft Wings Frontiers in Aerospace Science, Vol. 1 107
where H(ω) is the frequency response function of the system and F(ω) is
the actual force in the frequency domain. H(ω) is obtained from ground
vibration testing, X(ω) is the airfoil response measured during flight,
which can be measured by the two accelerometers mounted on the wing.
Thus, the air-dynamic forces acting on the wing during flight can be
identified by the experiment results.
Experiment Setup
The experiment setup for the wind tunnel test is the same as the previous
test and the data were collected using the same way (Fig. 40). The speed
of wind was measured by a Pitot gage as shown in Fig. (41). Table 5 lists
the seven wind speeds measured by the Pitot gage in the wind tunnel.
Table 5. Measured wind speeds and average errors of measured aerodynamic forces under
different speed.
speeds and the errors increased with the air speed. The maximum
percentage error for moments is 21.14%.
CONCLUSION
In the experimental analysis part, two tests, ground vibration test and
wind tunnel test were performed on a rapid prototyped airfoil model. The
model’s natural frequencies, damping ratios, mode shapes were calculated
based on the data measured from the ground vibration test. The
aerodynamic loadings acting on this model under different wind speeds
were measured from the wind tunnel test and validated by comparing to
the actual force values.
CONFLICT OF INTEREST
The authors confirm that they have no conflict of interest to declare for
this publication.
ACKNOWLEDGEMENTS
Declared none.
110 Frontiers in Aerospace Science, Vol. 1 Bi and Liu
REFERENCES
[1] R.M. Kirby, Z. Yosibash, and G.E. Karniadakis, "Towards stable coupling
methods for high-order discretization of fluid-structure interaction: Algorithms
and observations", J. Comput. Phys., vol. 223, no. 2, pp. 489-518, 2007.
[http://dx.doi.org/10.1016/j.jcp.2006.09.015]
[2] F. Liu, J. Cai, Y. Zhu, H.M. Tsai, and A.S. Wong, "Calculation of wing flutter by
a coupled fluid-structure method", J. Aircr., vol. 38, no. 2, pp. 334-342, 2001.
[http://dx.doi.org/10.2514/2.2766]
[5] Y. Liu, and W.S. Shepard Jr, "Dynamic force identification based on enhanced
least squares and total least-squares schemes in the frequency domain", J.
Sound Vibrat., vol. 282, no. 1/2, pp. 37-60, 2005.
[http://dx.doi.org/10.1016/j.jsv.2004.02.041]
[7] I. Lee, and S-H. Kim, "Aeroelastic analysis of a flexible control surface with
structural nonlinearity", J. Aircr., vol. 32, no. 4, pp. 868-874, 1995.
[http://dx.doi.org/10.2514/3.46803]
[8] S-H. Kim, and I. Lee, "Aeroelastic analysis of a flexible airfoil with a freeplay
nonlinearity", J. Sound Vibrat., vol. 193, no. 4, pp. 823-846, 1996.
[http://dx.doi.org/10.1006/jsvi.1996.0317]
[9] B.H. Lee, S.J. Price, and Y.S. Wong, “Nonlinear aeroelastic analysis of
Vibratory Force for Aircraft Wings Frontiers in Aerospace Science, Vol. 1 111
airfoils: bifurcation and chaos", Prog. Aerosp. Sci., vol. 35, no. 3, pp. 205-334,
1999.
[http://dx.doi.org/10.1016/S0376-0421(98)00015-3]
[10] I.D. Roy, and W. Eversman, "Adaptive flutter suppression of an unswept wing",
J. Aircr., vol. 33, no. 4, pp. 775-783, 1996.
[http://dx.doi.org/10.2514/3.47014]
[12] G. Dimitriadis, and J.E. Cooper, "A method for identification of non-linear
multi-degree-of-freedom systems", Proceedings of the Institution of Mechanical
Engineers, Part G: Journal of Aerospace Engineering, vol. 212, no. 4, pp.
287-298, 1998.
[http://dx.doi.org/10.1243/0954410981532441]
[13] M.D. Conner, D.M. Tang, E.H. Dowell, and L.N. Virgin, "Nonlinear behavior
of a typical airfoil section with control surface freeplay: a numerical and
experimental study", J. Fluids Structures, vol. 11, no. 1, pp. 89-109, 1997.
[http://dx.doi.org/10.1006/jfls.1996.0068]
[14] G. Schewe, H. Mai, and G. Dietz, “Nonlinear effects in transonic flutter with
emphasis on manifestations of limit cycle oscillations", J. Fluids Structures, vol.
18, no. 1, pp. 3-22, 2003.
[http://dx.doi.org/10.1016/S0889-9746(03)00085-9]
[16] C.H. Chen, "Determination of flutter derivatives via a neural network approach",
J. Sound Vibrat., vol.263, no. 4, pp. 797-813, 2003.
[http://dx.doi.org/10.1016/S0022-460X(02)01279-8]
112 Frontiers in Aerospace Science, Vol. 1 Bi and Liu
[17] D. Tang, and E.H. Dowell, "Experimental and theoretical study on aeroelastic
response of high-aspec--ratio wings", J. Aircr., vol. 39, no. 8, pp. 1430-1441,
2001.
[18] S. Potter, and R. Lind, "Developing uncertainty models for robust flutter
analysis using ground vibration test data", Proc. 42th Structures, Structural
Dynamics, and Materials Conference, 2001 Seattle, WA, USA
[http://dx.doi.org/10.2514/6.2001-1585]
[19] C.R. Pickrel, "Airplane ground vibration testing: nominal modal model
correlation", S.V. Sound and Vibration, vol. 36, no. 11, pp. 18-23, 2002.
[20] S. Lee, "Unsteady aerodynamic force prediction on a square cylinder using k-ε
turbulence models", J. Wind Eng. Ind. Aerodyn., vol. 67-68, pp. 79-90, 1997.
[http://dx.doi.org/10.1016/S0167-6105(97)00064-0]
[21] M.V. Joshi, and N.M. Reddy, "Aerodynamic force measurements over missile
configurations in IISc shock tunnel at M∞ = 5.5", Exp. Fluids, vol. 4, no. 6, pp.
338-340, 1986. [http://dx.doi.org/10.1007/BF00266300]
[22] M. Sun, and J-H. Wu, "Aerodynamic force generation and power requirements
in forward flight in a fruit fly with modeled wing motion", J. Exp. Biol., vol.
206, no. Pt 17, pp. 3065-3083, 2003.
[http://dx.doi.org/10.1242/jeb.00517] [PMID: 12878674]
CHAPTER 3
Computational and Analytical Investigation of
Lateral Impact Behavior of Pressurized
Pipelines
Yangqing Dou* and Yucheng Liu
*Corresponding author Yangqing Dou: Department of Mechanical Engineering, Mississippi State University,
MS, 39762, USA; Tel: 337-704-8410; E-mail: yd120@msstate.edu.
those functions are verified through several techniques. The outcomes of this study
have potential benefits on research of safety and reliability of pressurized pipelines
in hydraulic system of aerospace and development of advanced pipeline materials.
NOMENCLATURE
INTRODUCTION
PROBLEM DESCRIPTION
Wf = Wl + Wg (1)
Lateral Impact Behavior of Pressurized Pipelines Frontiers in Aerospace Science, Vol. 1 117
Original position
of pipeline
wg
wf
wl
Final position of
deformed pipeline
Fig. (1). Wf, Wl, and Wg for the idealized deformed cross-section of a pipeline.
MATERIAL PROPERTIES
In creating FEA models used for impact analysis that involves large strain
and high strain rate problems, the material properties have to be carefully
defined to correctly capture the impact. Here, elastic plastic material law
with kinematic isotropic hardening is chosen to model the pipeline
material (cold drawn steel). Dynamic effects of strain rates are taken into
account by scaling static yield stress with a factor with the assumption of
Cowper-Symonds relation [12]:
1/ P
0 0 1 (2)
C
where ε is the strain rate and Eq. (2) shows that during dynamic crushing,
the original yield strength σ0 should be replaced by the dynamic flow
stress σʹ0 due to the strain rate effects. C and P are constants of
Cowper-Symonds relation, which are used to define the influence of
material strain rate sensitivity in this study.
The pipelines are made of cold drawn steel, whose material properties are
118 Frontiers in Aerospace Science, Vol. 1 Dou and Liu
listed in Table 1 and were implemented into the FEA models through the
card *MAT_PLASTIC_KINEMATIC. The indenter is assumed to be rigid
during the impact simulation and was defined using *MAT_RIGID.
Table1. Plastic kinematic material properties for the LS-DYNA pipeline models.
72 FEA models were created for the pressurized pipelines and one for the
rigid knife-edge indenter. The pipelines were modeled using 4-node
Belytschko-Tsay shell elements with five integration points through the
thickness. The indenter was modeled using rigid elements with rigid
material properties assigned and its mass is set as 17.48 kg. It has the
same shape and mass as the indenter used by Jones [9]. The contact
between the pipelines and the indenter was defined through automatic
surface to surface contact algorithm. Within the defined contact pair, the
impact face on the pipeline was chosen as master contact surface and the
impact face on the indenter the slave surface. An initial speed of 10 m/s
was applied on the impact, at which it impacted onto the pipeline.
Two loading curves were developed to define the internal pressure and
gravity, which were applied onto the pipeline model as loading conditions.
In applying the internal pressure curve, its direction was carefully selected
so that the internal pressure was applied onto the inner side of the pipeline
Lateral Impact Behavior of Pressurized Pipelines Frontiers in Aerospace Science, Vol. 1 119
The generated FEA models include 1000 to more than 25,000 elements.
More than 300 impact simulations were then performed on those models
after applying appropriate boundary conditions (constraints on both ends
of the pipelines), loading conditions (internal pressure on the pipelines),
and initial conditions (initial velocity on the indenter). Figs. (2 and 3)
present two impact scenarios, where the indenter impacted a pipeline
model with outside diameter of 60 mm and internal pressure of 60 bar at
its center and one quarter span, respectively.
Fig. (2). FEA model showing an indenter impacts on center of a pressurized pipeline.
120 Frontiers in Aerospace Science, Vol. 1 Dou and Liu
Fig. (3). FEA model showing an indenter impacts on one quarter of a pressurized pipeline.
FEA Results
After the impact simulations, important FEA results were collected and
compared, from which typical lateral impact response of the pressurized
pipelines would be described. (Figs. 4 and 5), for example, plot the
deformed configurations of the FEA models presented in Figs. (2 and 3),
respectively. Some significant FEA results are recorded in Tables 2 and 3
for pipeline models which were struck at middle and one quarter span
position, respectively.
Table 2. FEA results for the pipes struck at the middle position.
D p Wf Wl Wg Fmax Fm vi vr Ep
No.
(mm) (bar) (mm) (mm) (mm) (kN) (kN) (m/s) (m/s) (kJ)
1 22 0 34.31 25.37 8.94 47.14 34.65 9.93 2.02 0.83
2 22 30 34.26 25.28 8.98 47.42 34.74 9.93 2.02 0.83
122 Frontiers in Aerospace Science, Vol. 1 Dou and Liu
(Table 2) contd.....
D p Wf Wl Wg Fmax Fm vi vr Ep
No.
(mm) (bar) (mm) (mm) (mm) (kN) (kN) (m/s) (m/s) (kJ)
3 22 60 34.21 25.20 9.01 47.45 34.86 9.93 2.03 0.83
4 22 90 34.15 25.12 9.03 47.64 34.95 9.93 2.03 0.83
5 22 120 34.06 25.02 9.04 47.77 35.04 9.93 2.04 0.83
6 22 150 33.98 24.92 9.06 47.86 35.07 9.93 2.05 0.83
7 42 0 24.60 20.19 4.41 50.09 36.19 9.93 3.14 0.78
8 42 30 24.08 19.68 4.46 51.00 36.78 9.93 3.19 0.77
9 42 60 23.70 19.19 4.51 51.89 37.10 9.93 3.24 0.77
10 42 90 23.31 18.78 4.53 52.36 37.44 9.93 3.29 0.77
11 42 120 21.31 16.75 4.56 52.39 37.89 9.93 3.34 0.77
12 42 150 20.61 16.05 4.58 52.75 37.87 9.93 3.37 0.76
13 60 0 18.05 15.73 2.33 56.80 36.47 9.93 4.18 0.71
14 60 30 16.05 13.75 2.32 59.45 38.04 9.93 4.46 0.69
15 60 60 14.71 12.43 2.28 61.80 38.87 9.93 4.68 0.67
16 60 90 13.59 11.33 2.26 64.35 39.85 9.93 4.83 0.66
17 60 120 12.83 10.58 2.25 65.61 40.49 9.93 4.94 0.65
18 60 150 12.07 9.80 2.27 67.68 41.13 9.93 5.02 0.64
19 80 0 16.83 15.63 1.20 58.37 34.51 9.93 4.92 0.65
20 80 30 12.45 11.42 1.03 64.18 37.36 9.93 5.58 0.59
21 80 60 10.42 9.46 0.96 69.80 39.29 9.93 5.93 0.56
22 80 90 8.98 8.00 0.98 74.10 41.24 9.93 6.16 0.53
23 80 120 8.36 7.30 1.06 76.16 42.45 9.93 6.29 0.52
24 80 150 7.62 6.47 1.15 76.19 43.15 9.93 6.37 0.51
25 100 0 15.90 15.22 0.68 59.30 32.15 9.93 5.30 0.61
26 100 30 10.05 9.56 0.49 68.00 36.62 9.93 6.29 0.51
27 100 60 7.60 7.09 0.51 74.68 39.51 9.93 6.70 0.50
28 100 90 6.56 5.98 0.58 80.57 42.53 9.93 6.89 0.55
29 100 120 5.57 5.01 0.56 85.81 43.66 9.93 7.02 0.62
30 100 150 5.11 4.43 0.68 86.54 45.75 9.93 7.02 0.79
31 120 0 16.12 15.51 0.61 54.71 30.09 9.93 5.47 0.60
32 120 30 8.12 7.75 0.37 66.35 35.28 9.93 6.59 0.49
Lateral Impact Behavior of Pressurized Pipelines Frontiers in Aerospace Science, Vol. 1 123
(Table 2) contd.....
D p Wf Wl Wg Fmax Fm vi vr Ep
No.
(mm) (bar) (mm) (mm) (mm) (kN) (kN) (m/s) (m/s) (kJ)
33 120 60 6.25 5.87 0.38 79.10 38.41 9.93 7.00 0.43
34 120 90 5.37 4.88 0.49 86.63 41.61 9.93 7.21 0.41
35 120 120 4.83 4.22 0.61 91.95 44.72 9.93 7.26 0.40
36 120 150 4.48 4.71 0.77 89.86 49.10 9.93 7.33 0.39
Table 3. FEA results for the pipes struck at one quarter span position.
No. D p Wf Wl Wg Fmax Fm vi vr Ep
(mm) (bar) (mm) (mm) (mm) (kN) (kN) (m/s) (m/s) (kJ)
1 22 0 21.71 13.80 6.91 55.22 39.35 9.93 1.88 0.83
2 22 30 21.61 15.79 6.90 53.93 39.38 9.93 1.89 0.83
3 22 60 21.50 14.58 6.92 56.42 39.41 9.93 1.90 0.82
4 22 90 21.41 14.49 6.92 57.16 39.46 9.93 1.91 0.82
5 22 120 21.31 14.39 6.92 54.85 39.50 9.93 1.92 0.82
6 22 150 21.24 15.31 6.93 56.19 39.53 9.93 1.93 0.81
7 42 0 19.20 16.10 3.10 59.17 42.09 9.93 2.71 0.80
8 42 30 18.56 15.47 3.09 60.56 42.47 9.93 2.83 0.77
9 42 60 17.95 14.83 3.12 62.24 42.73 9.93 2.90 0.75
10 42 90 17.43 14.31 3.13 63.13 42.94 9.93 2.96 0.74
11 42 120 16.98 13.84 3.14 64.10 43.11 9.93 3.01 0.73
12 42 150 16.56 13.42 3.14 64.89 43.31 9.93 3.07 0.73
13 60 0 19.00 17.15 1.85 61.38 40.38 9.93 3.34 0.76
14 60 30 17.11 15.30 1.81 65.92 42.47 9.93 3.66 0.71
15 60 60 15.76 13.98 1.78 69.63 44.06 9.93 3.89 0.68
16 60 90 14.64 12.89 1.75 72.20 45.18 9.93 4.09 0.67
17 60 120 13.72 11.97 1.75 73.57 45.92 9.93 4.23 0.67
18 60 150 12.81 11.04 1.77 76.15 46.60 9.93 4.35 0.68
19 80 0 18.76 18.63 1.13 60.63 37.40 9.93 3.80 0.74
20 80 30 13.27 12.17 1.10 68.61 40.65 9.93 4.50 0.69
124 Frontiers in Aerospace Science, Vol. 1 Dou and Liu
(Table 3) contd.....
No. D p Wf Wl Wg Fmax Fm vi vr Ep
(mm) (bar) (mm) (mm) (mm) (kN) (kN) (m/s) (m/s) (kJ)
21 80 60 12.95 11.97 0.98 75.48 43.02 9.93 4.89 0.65
22 80 90 11.01 10.12 0.89 78.49 44.92 9.93 5.15 0.63
23 80 120 10.04 9.17 0.88 81.48 46.10 9.93 5.34 0.61
24 80 150 9.02 8.14 0.88 84.71 47.48 9.93 5.49 0.60
25 100 0 18.39 17.46 0.93 58.91 34.39 9.93 4.09 0.72
26 100 30 12.36 11.87 0.49 68.72 34.43 9.93 5.21 0.63
27 100 60 9.99 9.56 0.48 79.35 43.41 9.93 5.75 0.57
28 100 90 7.72 7.22 0.50 84.12 45.87 9.93 6.09 0.54
29 100 120 6.79 6.32 0.47 86.52 48.23 9.93 6.28 0.52
30 100 150 5.78 5.20 0.58 89.78 50.09 9.93 6.36 0.51
31 120 0 18.04 17.23 0.81 54.79 31.84 9.93 4.45 0.68
32 120 30 10.52 10.05 0.47 68.21 38.90 9.93 5.73 0.58
33 120 60 7.68 7.25 0.43 78.82 43.13 9.93 6.45 0.50
34 120 90 4.63 4.45 0.18 89.96 46.33 9.93 6.78 0.46
35 120 120 4.21 3.82 0.39 101.1 51.74 9.93 7.08 0.42
36 120 150 4.10 3.74 0.36 101.1 51.74 9.93 7.08 0.42
Some of the listed FEA results obtained from this study were compared
with several previously published experimental results [8, 9] and a good
agreement was achieved. Thus, the accuracy of the developed FEA
models and proposed computational approach was validated.
Impact force
80000
70000
60000
50000
Force(N)
40000
30000 middle span
20000 quarter span
10000
0
1093
1275
1457
1639
1821
2003
2185
2367
2549
2731
2913
3095
3277
3459
3641
3823
183
365
547
729
911
1
Step
Fig. (6). Comparison of impact force yielded from models presented in Figs. (2 and 3).
2.50E-02
2.00E-02
1.50E-02
Wf(m)
1.00E-02
middle span
5.00E-03
quarter span
0.00E+00
47
93
139
185
231
277
323
369
415
461
507
553
599
645
691
737
783
829
875
921
967
1
Step
Fig. (7). Comparison of deformation yielded from models presented in Figs. (2 and 3).
126 Frontiers in Aerospace Science, Vol. 1 Dou and Liu
J Absorbed Energy
9.00E+02
8.00E+02
7.00E+02
6.00E+02
5.00E+02
4.00E+02
3.00E+02
middle span
2.00E+02
1.00E+02 quarter span
0.00E+00
1006
1207
1408
1609
1810
2011
2212
2413
2614
2815
3016
3217
3418
3619
3820
202
403
604
805
1
Step
Fig. (8). Comparison of absorbed energy yielded from models presented in Figs. (2 and 3).
80-100
60-80
100 40-60
20-40
80
Fmax(KN)
0-20
60
40
150
20 120
90
0 60
22 30
42 60 80 0
100
120
(a)
80-100
60-80
100 40-60
20-40
80
0-20
Fmax(KN)
60
40
150
20 120
90
0 60
22 30
42 60 80 0
100
120
(b)
Fig. (9). Maximum impact force for different impact position at (a) center (b) one quarter.
128 Frontiers in Aerospace Science, Vol. 1 Dou and Liu
30-40
40
20-30
30
10-20
Wf(mm)
20
0-10
150
10 120
90
0 60
22 30
42
60 0
80
100
120
(a)
30-40
40
20-30
30
10-20
Wf(mm)
20
150 0-10
10 120
90
0 60
22 30
42 60 80 0
100
120
(b)
Fig. (10). Maximum deformation for different impact position at (a) center (b) one quarter.
Lateral Impact Behavior of Pressurized Pipelines Frontiers in Aerospace Science, Vol. 1 129
0.8-1
0.6-0.8
1 0.4-0.6
0.8 0.2-0.4
0-0.2
0.6
Ep(kJ)
0.4
150
0.2 120
90
0 60
22 30
42 60 80 0
100
120
(a)
0.8-1
0.6-0.8
1 0.4-0.6
0.8 0.2-0.4
0-0.2
0.6
Ep(kJ)
0.4
150
0.2 120
90
0 60
22 30
42 60 80 0
100
120
(b)
Fig. (11). Maximum absorbed energy for different impact position at (a) center (b) one quarter.
130 Frontiers in Aerospace Science, Vol. 1 Dou and Liu
Fig. (13). Maximum impact force of pipeline models struck at one quarter.
Lateral Impact Behavior of Pressurized Pipelines Frontiers in Aerospace Science, Vol. 1 131
Fig. (17). Maximum absorbed energy of pipeline models struck at one quarter.
From Figs. (6 to 17) it can be observed that the different impact positions
did not evidently alter the characteristic lateral impact response of the
pressurized pipelines. The response surfaces of maximum impact force,
displacement, and absorbed energy obtained from both impact scenarios
showed similar shapes (Figs. 9 to 11). The influences of internal pressure
and diameter on the impact response of those pipelines also rendered
similar tendencies under both mid-position impact and one
quarter-position impact (Figs. 12 to 17).
significantly lower than those resulted from the mid-position impact (as
shown in Figs. (14 and 15), which can also be well explained through the
classic cantilever beam theory. However, as reflected from (Figs. 16 and
17), the impact energy absorbed during the mid-position impact was
slightly lower than that absorbed during the one quarter-position impact.
Those computational findings agree well with the phenomena observed
experimentally by Jones and his coworkers [1-3].
As for the influences of the internal pressure and outside diameter, Figs.
(12 and 13) reveal that when the internal pressure increased, the
maximum impact force also increased, even such tendency was not
apparent when the out diameter was low. It is believed that the internal
pressure will enhance impact resistance of the pipeline model, therefore
leads to higher impact forces. Meanwhile, those figures also tell us that
under the same pressure, the pipeline models with larger outside diameter
may be subjected to higher impact force.
From Figs. (14 and 15) it is found that the maximum transverse
displacement decreased with an increasing internal pressure. Similar to
the maximum impact force, such tendency was not clear when the outside
diameter was low. This phenomenon can also be explained through the
improved impact resistance of the pipelines due to the increased internal
pressure. On the other side, the total permanent transverse displacement
Wf decreased when the outside diameter D increased. However, on
inspecting the results listed in Tables 1 and 2, it can be seen that the
permanent local displacement Wl did increase as the outside diameter
increased, which can be perfectly explained using the theoretical model
produced by Oliveira et al. [10].
Figs. (16 and 17) show that when the internal pressure increased, the
impact energy absorbed by the pipelines decreased (still, such tendency
was not that evident for pipelines with a small outside diameter). Similar
finding was also reported in Ng and Shen’s work, which were obtained
through experiments [8]. Since the internal pressure enhances the rigidity
of the pipeline, that way the indenter will rebound at a comparatively high
134 Frontiers in Aerospace Science, Vol. 1 Dou and Liu
speed after impacting the pipeline and less impact energy will be
absorbed by the pipeline. Meanwhile, from both figures it is observed that
similar to the displacement case, the absorbed impact energy decreased
when the outside diameter increased.
ANALYTICAL ANALYSIS
In this study, the impact response of the pipelines (it can be the impact
force, deformation, and absorbed energy) is approximated using a series
of the basic functions in a form of
n
y(x) F( p, D), W ( p, D), orE ( p, D) β i i (p, D) (3)
i 1
where n represents the number of basic functions φi(p, D). In this paper,
the polynomials are used to build up these basic functions to formulate F,
W, and E.
In Eq. (3), the βi, known as the regression coefficients, are estimated using
the method of least squares. Suppose we have m (m > n) observations
(obtained from FEA) for the yielded response yi (y1 – ym) based on the m
Lateral Impact Behavior of Pressurized Pipelines Frontiers in Aerospace Science, Vol. 1 135
sampling design points (p, D)i, the least squares function is therefore
expressed as
2
m m n
L εi y i β j j (p, D)
2
(4)
i 1 i 1 j1
where the design points (p, D)i are selected from the specified design
space, εi is the error between the response yi observed at these points, and
the RS approximation at that point. Afterwards, the coefficient vector b =
(β1, β2, …, βn)T can be determined by L / 0 , which is
b = (ФTФ)-1ФTy (5)
where Ф denotes the matrix consisting of basic functions evaluated using
m sampling points, which is
1 (p, D)1 n (p, D)1
Φ
(6)
1 (p, D) m n (p, D) m
By substituting Eq. (5) into (3), the response surface model is created and
the response functions (F(p, D), W(p, D), and E(p, D)) then can be fully
determined.
m
SS T (y i y i ) 2 (9)
i 1
(SS T SS E )/n
F (10)
SS E /(m n 1)
SS E
R2 1 (11)
SS T
m 1
2
R adj 1 (1 R 2 ) (12)
mn
SS E
RMSE (13)
m n 1
According to the classical RSM theory, the larger the values of R2 and
R2adj, and the smaller the value of RMSE, the better the model fit.
As shown in Tables 2 and 3, six pressures and diameters were selected for
modeling and simulation and totally 36 combinations were presented.
Thus, in Eq. (4) the matrix Φ has 36 rows (m = 36), which corresponds to
the 36 combinations of p and D (No. 1 to 36 in Tables 1 and 2). In this
study, quartic polynomial is used as the basic function because it provides
the best fitness to the real problems [12]. Thus, the basic functions φi are
terms in a full quartic form, which are 1, p, D, p2, pD, D2, p3, p2D, pD2,
Lateral Impact Behavior of Pressurized Pipelines Frontiers in Aerospace Science, Vol. 1 137
D3, p4, p3D, p2D2, pD3, D4. Substituting p and D values into Eq. (6), a 36
× 15 matrix Φ then can be created.
Mid-span impact
(a)
(b)
(c)
Fig. (18). Quartic response surfaces of (a) Fmax; (b) Wf; (c) Ep under mid-span impact.
140 Frontiers in Aerospace Science, Vol. 1 Dou and Liu
(a)
(b)
(c)
Fig. (19). Quartic response surfaces of (a) Fmax; (b) Wf; (c) Ep under one quarter span impact.
Lateral Impact Behavior of Pressurized Pipelines Frontiers in Aerospace Science, Vol. 1 141
PARAMETRIC STUDY
22D
22D
100 42D
42D
60D
100 60D
80D
90 80D
100D
100D
120D 90 120D
80
80
70
Fmax (KN)
Fmax (KN)
70
60
50 60
40 50
30 40
0 30 60 90 120 150 0 30 60 90 120 150
Pressure (bar) Pressure (bar)
(a) (b)
Fig. (20). Effects of p on maximum impact force, (a) mid-span impact (b) one quarter span impact.
142 Frontiers in Aerospace Science, Vol. 1 Dou and Liu
22D
40 42D
35 60D
30 80D
25 100D
W (mm)
20 120D
15
10
0
0 30 60 90 120 150
Pressure (bar)
(a)
22D
25 42D
60D
20
80D
100D
15
W (mm)
120D
10
0
0 30 60 90 120 150
Pressure (bar)
(b)
Fig. (21). Effects of p on maximum permanent deformation, (a) mid-span impact (b) one quarter
span impact.
Lateral Impact Behavior of Pressurized Pipelines Frontiers in Aerospace Science, Vol. 1 143
22D
42D
60D
0.8
80D
100D
120D
0.7
0.6
Ep(KJ)
0.5
0.4
0.3
0 30 60 90 120 150
Pressure (bar)
(a)
22D
42D
0.9
60D
80D
100D
0.8
120D
0.7
Ep (KJ)
0.6
0.5
0.4
0.3
0 30 60 90 120 150
Pressure (bar)
(b)
Fig. (22). Effects of p on absorbed energy, (a) mid-span impact (b) one quarter span impact.
144 Frontiers in Aerospace Science, Vol. 1 Dou and Liu
From above figures, it was found that the internal pressure affects the
behavior of pressurized pipelines under both mid-span and one quarter
span impact in similar ways. As shown in (Figs. 20a and 20b), the
maximum impact forces yielded in the one quarter span impact are
apparently higher than those generated during the mid-span impact. This
is because that the “one quarter position” is closer to the fixture and
therefore is more rigid than the “middle position”, which leads to higher
impact force values. However, as reflected in (Figs. 22a and 22b), the
change of impact positions does not evidently alter the energy absorbed
during the impacts.
Figs. (20 to 22) also reveal that when the internal pressure increases, the
maximum impact force also increases, even such tendency is not evident
when the outside diameter is small. Thus, it can be deduced that the
internal pressure can enhance the impact resistance of the pipeline and
therefore leads to higher impact forces. On the contrary, as the pressure
increases, the impact energy absorbed by the pipelines decreases, still
such tendency is not that clear when the outside diameter of the pipelines
is small. Since the internal pressure improves the overall impact
resistance of the pipeline model, therefore after impacting a pipeline with
higher internal pressure, the indenter will rebound at a higher speed and
only small portion of impact energy will be absorbed by the pipeline itself.
These findings agree very well with the results presented in ng and shen’s
work [8], who drawn similar conclusions through a series of impact tests
and experimental analysis.
0bar
100
30bar
90
60bar
80
90bar
70
120bar
60
Fmax (kN) 150bar
50
40
30
20
10
0
0 20 40 60 80 100 120
Diameter (mm)
(a)
0bar
120
30bar
100 60bar
90bar
80 120bar
Fmax (kN)
150bar
60
40
20
0
0 20 40 60 80 100 120
Diameter (mm)
(b)
Fig. (23). Effects of D on maximum impact force, (a) mid-span impact (b) one quarter span
impact.
146 Frontiers in Aerospace Science, Vol. 1 Dou and Liu
0bar
40
30bar
35
60bar
30 90bar
W (mm) 25 120bar
150bar
20
15
10
5
0
0 20 40 60 80 100 120
Diameter (mm)
(a)
0bar
25
30bar
60bar
20
90bar
120bar
15
W (mm)
150bar
10
0
0 20 40 60 80 100 120
Diameter (mm)
(b)
Fig. (24). Effects of D on maximum permanent deformation, (a) mid-span impact (b) one quarter
span impact.
Lateral Impact Behavior of Pressurized Pipelines Frontiers in Aerospace Science, Vol. 1 147
0bar
0.9
30bar
0.8
60bar
0.7
90bar
Ep (KJ) 0.6 120bar
0.5 150bar
0.4
0.3
0.2
0.1
0
0 20 40 60 80 100 120
Diameter (mm)
(a)
0bar
0.9
30bar
0.8
60bar
0.7
90bar
0.6 120bar
Ep (KJ)
0.5 150bar
0.4
0.3
0.2
0.1
0
0 20 40 60 80 100 120
Diameter (mm)
(b)
Fig. (25). Effects of D on absorbed energy, (a) mid-span impact (b) one quarter span impact.
148 Frontiers in Aerospace Science, Vol. 1 Dou and Liu
At first, it was observed that just like the internal pressure, the outside
diameter affects the behavior of the pipeline models impacted at two
positions in similar ways. From above figures, it was found that the
internal pressure affects the behavior of pressurized pipelines under both
mid-span and one quarter span impact in similar ways. As illustrated in
Figs. (20 and 22), it is verified in Figs. (23 and 25) that the maximum
impact forces yielded in the one quarter span impact are higher than those
generated during the mid-span impact, and the energy absorbed in both
cases are very close to each other.
From Fig. (23a and 23b), it can be seen that as the outside diameter
increases, the maximum impact forces also increases for most internal
pressures. Exceptions were however observed at low internal pressures.
For mid-space impacts, when the internal pressure is lower than 30 bar,
the maximum impact force value goes down when the outside diameter
exceeds 100 mm. For the one quarter span impacts, the increasing
tendency of the maximum force value becomes much less apparent when
the internal pressure is 30 bar and the outside diameter passes 80 mm, and
when the internal pressure is 60 bar and the outside diameter reaches 100
mm. Furthermore, for pipelines without internal pressure, the maximum
force starts decreasing when the outside diameter reaches 60 mm. As
reflected in Fig. (25a and 25b), when the outside diameter increases, the
energy absorbed by the pipelines decreases and the decreasing tendency is
much more obvious than the decreasing caused by the increasing internal
pressure (Fig. 22a and 22b). In summary, the influence of the outside
diameter on pressurized pipelines’ impact behavior is similar to that of the
internal pressure, both of them will enhance impact resistance of the
pipeline models.
CONCLUSION
CONFLICT OF INTEREST
The authors confirm that they have no conflict of interest to declare for
this publication.
ACKNOWLEDGEMENTS
Declared none.
REFERENCES
[1] N. Jones, S.E. Birch, R.S. Birch, L. Zhu, and M. Brown, "An experimental
study on the lateral impact of fully clamped steel pipes", In: Proceedings of
the Institution of Mechanical Engineers., vol. 206, pp. 111-127, 1992.
[http://dx.doi.org/10.1243/PIME_PROC_1992_206_207_02]
[2] N. Jones, and W.Q. Shen, "A theoretical study of the lateral impact of fully
clamped pipelines", In: Proceedings of the Institution of Mechanical
Engineers, vol. 206, pp. 129-146, 1992.
[http://dx.doi.org/10.1243/PIME_PROC_1992_206_208_02]
[3] N. Jones, and R.S. Birch, "Influence of internal pressure on the impact
behavior of steel pipelines", ASME Journal of Pressure Vessel Technology, vol.
118, pp. 464-471, 1996.
[http://dx.doi.org/10.1115/1.2842215]
[4] S.Y. Shin, B. Hwang, S. Lee, N.J. Kim, and S.S. Ahn, "Correlation of
microstructure and charpy impact properties in API X70 and X80 line-pipe
steels", Mater. Sci. Eng. A, vol. 458, pp. 281-289, 2007.
[http://dx.doi.org/10.1016/j.msea.2006.12.097]
[5] B.N. Leis, R.B. Francini, R. Mohan, D.L. Rudland, and R.J. Olson,
"Pressure-displacement behavior of transmission pipelines under outside
forces towards a serviceability criterion for mechanical damage", In:
Proceedings of the 8th International Offshore and Polar Engineering
Conference, Canada, 1998, pp. 60-67.
Lateral Impact Behavior of Pressurized Pipelines Frontiers in Aerospace Science, Vol. 1 151
[6] K.S. Chen, and W.Q. Shen, "Further experimental study on the failure of fully
clamped steel pipes", Int. J. Impact Eng., vol. 21, no. 3, pp. 177-202, 1998.
[http://dx.doi.org/10.1016/S0734-743X(97)00083-3]
[7] W.Q. Shen, and K.S. Chen, “An investigation on the impact performance
of pipelines", Intern. J. Crashworthiness, vol. 3, no. 2, pp. 191-209, 1998.
[http://dx.doi.org/10.1533/cras.1998.0070]
[8] C.S. Ng, and W.Q. Shen, "Effect of lateral impact loads on failure of
pressurized pipelines supported by foundation", In: Proceedings of the
Institution of Mechanical Engineers, Part E: J. Process Mech. Eng., vol. 220,
pp. 193-206, 2006.
[http://dx.doi.org/10.1243/0954408JPME97]
[9] N. Jones, and R.S. Birch, "Low-velocity impact of pressurized pipelines", Int.
J. Impact Eng., vol. 37, no. 2, pp. 207-219, 2010.
[http://dx.doi.org/10.1016/j.ijimpeng.2009.05.006]
[10] J.G. de Oliveira, T. Wierzbicki, and W. Abramowicz, "Plastic behavior of
tubular members under lateral concentrated loading", Technical Report
82-0708, Det norske Veritas, Research Division, 1982.
[11] H. Oktem, T. Erzurumlu, and H. Kurtaran, "Application of response surface
methodology in the optimization of cutting conditions for surface roughness",
J. Mater. Process. Technol., vol. 170, pp. 11-16, 2005.
[http://dx.doi.org/10.1016/j.jmatprotec.2005.04.096]
[12] Y-C. Liu, "Optimum design of thin-walled box section beams for
crashworthiness analysis", Finite Elem. Anal. Des. vol. 44, no. 3, pp. 139-147,
2008.
[http://dx.doi.org/10.1016/j.finel.2007.11.003]
[13] Y-C. Liu, “Design optimization of tapered thin-walled square tubes", Intern. J.
Crashworthiness, vol. 13, no. 5, pp. 543-550, 2008.
http://dx.doi.org/10.1080/13588260802222102]
[14] Y-C. Liu, "Crashworthiness design of multicorner thin-walled columns",
Thin-walled Struct., vol. 46, no. 12, pp. 1329-1337, 2008.
152 Frontiers in Aerospace Science, Vol. 1 Dou and Liu
[http://dx.doi.org/10.1016/j.tws.2008.04.003]
[15] Y-C. Liu, "Crashworthiness design of thin-walled curved beams with box and
channel cross sections", Intern. J. Crashworthiness, vol. 15, no. 4, pp.
413-423, 2010.
[http://dx.doi.org/10.1080/13588261003647410]
[16] Y-C. Liu, "Thin-walled curved hexagonal beams in crashes – FEA and design",
Intern. J. Crashworthiness, vol. 15, no. 2, pp. 151-159, 2010.
[http://dx.doi.org/10.1080/13588260903094400]
[17] Y-C. Liu, "“Optimization of the crushing performance of tubular structures“,
HKIE (The Hong Kong Institution of Engineers)", Transactions, vol. 17, no. 1,
pp. 16-25, 2010.
Frontiers in Aerospace Science, 2016, Vol. 1, 154-239 153
CHAPTER 4
1
Department of Mechanical & Industrial Engineering, Louisiana State
University, Baton Rouge, LA 70803, USA
2
Department of Mechanical Engineering, Southern University, Baton
Rouge, LA 70813, USA
*Corresponding author Guoqiang Li: Department of Mechanical & Industrial Engineering, Louisiana State
University, Baton Rouge, LA 70803, USA; Department of Mechanical Engineering, Southern University, Baton
Rouge, LA 70813, USA; Tel/Fax: 001-225-578-5302/001-225-578-5924; E-mail: lguoqi1@lsu.edu
INTRODUCTION
Background
During the World War I, many steel structures failed at low temperature.
English aeronautical engineer, A. A. Griffith, provided a theory to explain
the failure of brittle materials and built up the fundamental research of the
Linear Elastic Fracture Mechanics [5]. A major limitation of Griffith’s
theory is that it assumes infinite stress at the initial crack tip point as the
radius “r” approaches zero. This is impossible in reality. Therefore, in
1954, Irwin modified the Griffith theory and proposed a plastic zone
developed from the initial crack tip. As the applied load increases, the
plastic zone increases in size until the crack grows and the material
behind the crack tip unloads. The plastic loading and unloading cycle near
the crack tip leads to the dissipation of energy as heat. Hence, a
dissipative term has to be added to the energy balance relation devised by
Griffith for brittle materials. In physical terms, additional energy is
needed for crack growth in ductile materials when compared to brittle
materials [6]. However, one basic assumption in Irwin's linear elastic
fracture mechanics is that the size of the plastic zone is small compared to
the crack length. This assumption is quite restrictive for certain types of
failure in structural steels though such steels can be prone to brittle
fracture, which has led to a number of catastrophic failures [6].
In the following three decades, numerous studies had been contributed to
failure problems of bonded joints [7]. During this stage, most efforts were
focused on the classical linear elastic fracture mechanics (LEFM). The
advantage of LEFM lies in the obvious simplicity and decent accuracy,
especially for relatively brittle materials and interfaces. Within the
framework of LEFM, the remote loadings can be correlated to the critical
conditions (crack growth) by a local parameter: stress intensity factor (SIF)
or a global parameter: strain energy release rate (ERR). Despite the huge
success of LEFM, the assumption of small-scale yielding beyond the crack
tip has been identified as one of the major limitations. Specifically, with the
increased use of modern toughened adhesives, the cohesive fracture
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 157
associated with plastic zones along the adhesive interlayer, in many cases,
could be comparable to or even larger than the thickness of the adherends.
The stresses vary with the relative surface separations according to the
assumed interface law during the fracture process.
J-Integral
Also during this period of time, Rice [10] proposed a path independent
integral method to calculate the energy release rate (ERR), which is
referred to as the well-known J-integral:
158 Frontiers in Aerospace Science, Vol. 1 Ji et al.
u
J (Wdy T ds) (1.1)
x
where W(x, y) is the strain energy density; x and y are the coordinate
directions, T= n is the traction vector; n is the normal to the curve or
path Γ, σ is the Cauchy stress, and u is the displacement vector [10].
This method can describe the case when there is sufficient crack-tip
deformation and the part no longer obeys the linear-elastic approximation.
Rice's analysis, which assumes non-linear elastic (or monotonic
deformation-theory plastic) deformation ahead of the crack tip, is
designated as the J integral [10]. This analysis is limited to some
situations. First, plastic deformation at the crack tip cannot extend to the
furthest edge of the loaded part. Second, the assumed non-linear elastic
behavior of the material is a reasonable approximation in shape and
magnitude to the real material's load response. Also note that the J
integral approach can be reduced to the Griffith theory for linear-elastic
behavior [10]. This method can measure the ductile material with plastic
deformation. Therefore, it can be used in this study to measure the
fracture energy of adhesive layer.
The Mode II, or in-plane interface shear fracture, loading mode is also of
particular importance for adhesive joints [33]. Mode II loading may be
induced when a cracked adhesive joint or a layered composite is subjected
to bending. The various experimental fracture mechanics approaches to
Mode II usually utilize some form of test specimen which is subjected to
applied bending loads in order to determine the values of the Mode II
toughness GIIC [52].
techniques.
One of the main problems that have been encountered with Mode II
loading has been the poor reproducibility of the values of the measured
Mode II toughness GIIC [54]. The experimental loading-unloading cycling
test conducted by Russel and Street [53] indicated a maximum error of
around 2% in GIIC in composites if friction was ignored. More recently,
Davidson and Sun [55] and Davidson et al. [56] considered the effects of
friction in the ENF specimen and 4-ENF tests were conducted on
composites. They concluded that friction accounted for only about 2%
and 5%, respectively, of the measured values of GIIC from their tests. The
effects of friction in the specimen could be a possible reason for this poor
reproducibility.
Some local tests have been conducted to measure the critical CZM
parameters. For instance, with J-integral based form, the local interfacial
traction-separation laws were experimentally tested under pure Mode-I
and pure Mode-II loading conditions [47, 48, 50]. It is noted that these
studies focused on the interface fracture of bonded joints with identical
adherends through standard fracture test methods (for instance, double
cantilever beam (DCB) and end notched fracture (ENF). Recently, an
experimental study was also conducted to investigate the local mixed
mode interfacial constitutive behaviors [27]. Most recently, Zhu et al. [51]
conducted a local fracture test to extract the loading-rate dependent
traction–separation laws. These recent efforts on the local test of interfacial
traction-separation and traction-slipping laws across the adhesive interlayer
provided valuable information.
In load-bearing structures, the adhesive layers are usually weaker than the
adherends. Therefore, the adherends can be treated as rigid constraints on
the adhesive layer. This means that the region where a crack can
propagate is almost in a state of prescribed deformation [49]. During
crack propagation, a zone where the adhesive is damaged precedes the
crack tip. This zone is denoted as the damage zone. With a tough
engineering adhesive, it is substantially longer than it appeared in a bulk
test and typically several times larger than the thickness of the layer [49].
Thus, fracture behavior depends on the thickness of the adhesive layer.
The fracture energy for a specific adhesive can reach a maximum value at
a certain layer thickness [3, 40]. Therefore, the effect of adhesive
thickness is a very important parameter to design the safety factor of
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 163
structure.
Research Objectives
THEORETICAL BACKGROUND
Theoretical Derivations
in which
E xi hi E xi hi3
Ai ; Di
1 v 2zxi 12(1 v 2zxi )
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 167
Ni and Mi are the axial force and bending moment per unit width of the
beam i (i=1, 2), respectively; Ai and Di are the axial and bending stiffness
of the beam i (i=1, 2) per unit width under the plane strain condition.
The equilibrium equations of each beam within the bonded region are
written in the conventional way as
dN1 dN 2
( x) ; ( x) (2.8)
dx dx
dQ1 dQ2
( x) ; ( x) (2.9)
dx dx
dM 1 h dM 2 h
Q1 1 ( x) ; Q2 2 ( x) (2.10)
dx 2 dx 2
where σ(x) and τ(x) are the interface normal and shear stress, respectively;
N1 and N2 are the internal axial forces in beam 1 and 2, respectively; M1
and M2 are the internal bending moments in beam 1 and 2, respectively;
Q1 and Q2 are the internal shear forces in beam 1 and 2, respectively (see
Fig. 2.3). The resultant forces and moment in the two beams are denoted
as follows:
N1 N1 N T ; Q1 Q2 QT ; M1 M 2 M T (2.11)
By substituting Eq. (2.7) into the derivative of Eq. (2.6), it can be derived
that
h N h N (2.12)
1 M1 1 ( 2 M 2 2 )
2 D1 A1 2 D2 A2
With Eq. (2.7) and note that w=w1-w2, we have follows:
M1 M 2
w w1 w2 ( ) (2.13)
D1 D2
1 dM 1 1 dM 2
w w1 w2 ( ) (2.14)
D1 dx D 2 dx
With Eqs. (2.4) through (2.14), the following two equations can be
derived that,
168 Frontiers in Aerospace Science, Vol. 1 Ji et al.
h1 D2 h2 D1 h h 1 1 (h h ) 2
w 1 2 QT [( ) 1 2 ] (2.15)
2( D1 D2 ) 2( D1 D2 ) A1 A2 4( D1 D2 )
h1 h2 h12 h22 1 1
( Q1 Q2 ) ( ) (2.16)
2D1 2D2 4D1 4D2 A1 A2
By further taking the derivative of both sides of Eqs. (2.15 and 2.16), the
follows can be derived, respectively,
h1 D2 h2 D1 ( 4) h h dQT 1 1 (h h ) 2
w 1 2 [( ) 1 2 ] (2.17)
2( D1 D2 ) 2( D1 D2 ) dx A1 A2 4( D1 D2 )
h1 h h2 h2 1 1
( 2 ) ( 1 2 ) (2.18)
2D1 2D2 4D1 4D2 A1 A2
The resultant shear force QT=Q1+Q2 in the two adherends can be
determined by global analysis with the given boundary conditions. For the
sake of simplicity, in the current theoretical model, we only consider the
conditions that QT is constant or piecewisely constant.
Governing Equations
h 2 h2
QT (2.24)
2D A 2D
By substituting Eqs. (2.22 and 2.23) into Eq. (2.19), we obtain
D d 4w
( w) (2.25)
2 dx 4
It is worth noting that Eq. (2.24) represents the governing equation of
interfacial shear behaviors, while Eq. (2.25) represents the governing
equation of interfacial peel behaviors. It is also noted that if only: (i) the
resultant shear force QT in the two adherends is a constant or piecewisely
constant within the bonded length; and (ii) the two adherends are
identical.
Strictly, the adhesive material must be nonlinearly elastic during the test.
However, for a monotonic loading process (no unloading occurs), the
cohesive separation as well as plastic dissipation in the adhesive layer
170 Frontiers in Aerospace Science, Vol. 1 Ji et al.
where W(x, y) is the strain energy density; x and y are the coordinate
directions, T=n σ is the traction vector; n is the normal to the curve or path
Γ, σ is the Cauchy stress, and u is the displacement vector.
Let us consider an arbitrary tangential cohesive law τ=τ(δ). Note
Q1+Q2=QT is a constant along the x coordinate. Let us integrate both sides
of the governing Eq. (2.24) as follow:
h 2 h2
d 2D
QT d ( )d
A 2D
(2.27)
h 2 h2
(
)d
2D
QT d ( )d
A 2D
(2.28)
1 h 2 h2 δ
[( ) 2 ( m ) 2 ] QT ( m ) ( )d (2.29)
2 2D A 2D δm
where δʹm represents the derivative of δ=0, and δm represents the value of
δ when dδ/dx=0.
It should be noticed that the integral limits on both sides of Eq. (2.29)
must be corresponded to each other. However, when the joint’s overlap
length and the length between the support and concentrated middle load
are sufficiently long relative to the cohesive zone, we may approximate
the solution with δm→0 and δʹm →0.
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 171
d 3 w dw 1 d 2 w 2 2 w
( ) ( w)dw 0 (2.32)
dx 3 dx 2 dx 2 D 0
One may verify the equivalence between Eqs. (2.25 and 2.32) by taking
derivative of Eq. (2.32) with respect to x and then comparing it to Eq.
(2.25).
w0 1 D1 D2 M 1 M 2 2 D1 D2 P P
JI
0
( w)dw [
2 D1 D2 D1
D2
] ( 1 2 ) 0
D1 D2 D1 D2
(2.33)
1 1
[ M 1 M 2 ] 2 ( P1 P2 ) 0
J I ( w) 4D 2
( w) (2.35)
w w
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 173
1 h
( ) 2 QT 0
2 0 2D
2
2 h
J II ( ) A 2 D
( ) (2.36)
We shall discuss more details regarding how to measure the cohesive laws
in the next section. Meanwhile, it is noted that this interface constitutive
relationship is the equivalent interface cohesive law, not necessarily the
intrinsic cohesive law. This is because in addition to the intrinsic cohesive
separation, possible plastic deformations in the adhesive layer contribute
to the entire separation between the two adherends during the fracture test.
By this point, the general theoretical models have been derived for the
interface peel behavior or normal separation (Mode-I fracture) and the
interface shear behavior or tangential slip (Mode-II fracture). It is worth
noting that the derived model can be also applied to the in-plane mixed
174 Frontiers in Aerospace Science, Vol. 1 Ji et al.
mode fracture (Mode I/II fracture) only if the two adherends are identical.
However, as mentioned before, the specific formula might be different for
different fracture test configurations due to the different boundary
conditions. Therefore, in the work, three common standard fracture test
methods: double cantilever beam (DCB), end notched flexure (ENF) and
single leg bending (SLB) are adopted for Mode-I, Mode-II and in-plane
mixed Mode I/II, respectively.
DCB specimens have been widely adopted to investigate the pure Mode-I
interface fracture of bonded joints in previous studies [2, 3, 16-18, 21, 26,
41-48, 61].
Consider a typical double cantilever beam (DCB) specimen as shown in Fig.
(2.3). It is assumed that the adherends are linearly elastic during the entire
fracture test process. A thin adhesive layer is used to bond the two
adherends.
P
h
ha
h
P
a L
Fig. (2.3). Schematic of DCB test specimen.
P1=-P2=P). Note that the relationship between the relative rotation and each
beam’s rotation at the crack tip is θ0=2θ10=-2θ20. Therefore, Eq. (2.34) can
be rewritten for DCB specimen as follow:
w0 P2a2
J I ( w)dw 2 P10 (2.37)
0 D
w0 Pa 2
J I ( w)dw 2 P [ 10 ] (2.38)
0 2D
The first term (Pa2/2D) in the bracket reflect the variation of rotation in
each beam, the sum in the bracket can be replaced as follow:
Pa 2
10 1P (2.39)
2D
where θ1P is the upper beam’s rotation at the cross section of loadline. In
addition, we may denote the lower beam’s rotation at the cross section of
loadline by θ2P. One may readily see that θ1P =-θ2P. We further introduce a
concept: relative rotation θP, which reflects the relative rotation between
the two beams at the cross section of loadline.
Eq. (2.41) indicates that the energy release JI for pure Mode-I fracture of
DCB specimen equals the product of peel force P and relative rotation θP
between the two beams at the loadline.
According to Eq. (2.35), the interfacial normal stress σ of DCB specimen
176 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Eq. (2.42) indicates that once the crack tip opening w0 and JI are
simultaneously measured (or JI-w0 curve is measured), the interface
normal stress σ is experimentally determined.
Obviously, with the ENF configuration, the resultant shear force QT in the
section from left support to the mid-span loading P is a constant. Let’s
assume that the shear forces in the upper and lower beam at the cross
section of crack tip are Q1 and Q2, respectively. Evidently, Q1 and Q2 will
not vary within the region from the left support to the cross section of the
crack tip.
Since the axial forces N1=N2=0 at the cross section of crack tip, with Eq.
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 177
Note that for the SLB specimen, the resultant MT at the cross section of
crack tip is still MT = M1 + M2 = QT .a = Pa/2 (QT is still equal to the
support reaction force, and QT=P/2), although M1 = Pa/2 and M2=0,
respectively. The energy release rate component JII of SLB specimen can be
expressed by,
1 ha P 2 h P
δ0
( ) 0
J II ( 0 ) ( )d 2 2 D 2 2D 2 (2.46)
0 2 h2
A 2D
Similarly, the interfacial shear stress τ(δ0) of SLB specimen (at the crack
tip) can be determined as follow:
1 ha P 2 h P
( ) 0
2 2D 2 2D 2
2
2 h
J II ( 0 ) A 2D
( ) (2.47)
0 0
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 179
w0 1 1
J I ( w)dw [QT a]2 (QT ) 0 (2.48)
0 4D 2
With the similar approach to the above section, the lower beam’s rotation
at the crack tip is equal to that at the loadline (θ20=θ2P) because there are
no bending moment in the lower beam within the region from left support
to the crack tip. Thus the Mode-I energy release rate component JI can be
expressed as follow:
QT QT a 2 Q Q a2
JI [ 0 ] T [ T 10 20 ]
2 2D 2 2D
2
Q Q a Q (2.49)
T [ T 10 2 P ] T [1P 2 P ]
2 2D 2
QT
P
2
where θP is the relative rotation between the two beams at the loadline. It
is worth noting that in a real experimental SLB, it is not convenient to
measure the rotation of lower beam’s rotation at the loadline due to the
presence of the support, neither at the crack tip. However, one may simply
choose any location between the crack tip and support where is more
convenient, since the lower beam’s rotations within this region are
identical.
Since P=2QT, with Eq. (2.49), Mode-I energy release rate component JI
can be expressed as follow:
w0 P
J I ( w)dw P (2.50)
0 4
180 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Eqs. (2.46, 2.47, 2.50 and 2.51) indicate that once the crack tip slip δ0,
loading force P, and the rotation of the two adherends are simultaneously
recorded, JI and JII can be calculated (or JI-w0 and JII-δ0 curves can be
obtained). Then, the interface shear stress τ (δ0) and interface normal
stress σ (w0) can be experimentally determined, simultaneously.
EXPERIMENTAL METHOD
Raw Material
Adhesive Material
Adherends Material
General purpose 1018 low carbon steel bars with yield strength of 372.3
MPa and elastic modulus of 209 GPa were used to fabricate metal based
adherends of Double Cantilever Beam (DCB), End Notched Flexure (ENF),
and Single Leg Bending (SLB) specimens. Before the adhesive was applied,
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 181
the surface of the adherends were grounded by sand paper, and then cleaned
by using acetone.
Fabrication of Specimen
There are two different methods to manufacture the DCBs. One is that the
adherends are cut before the adhesive curing processes. The other one is
that the two adherends with adhesive experience the curing processes and
then are cut to small DCB specimens. The major problem of this method is
that during the milling and cutting processes, the adhesive layer could be
damaged by metal chips. Therefore, this study cut the adherends before
bonding them with adhesive.
The thickness of the adherends was measured before the adhesive was
applied and compared with the thickness of the specimen after curing. In
order to obtain different and desirable thickness of the adhesive layer, six
different thickness feeler gage inserts were inserted in between the two
adherends. The feeler gages can also control the edge of the adhesive and
help in keeping the adherends parallel. It is also noted that the surface of the
feeler gauges were sprayed with a super-thin layer of mold release agent so
that the feeler gauges could be easily removed after the curing.
A very thin mylar tape with thickness of 0.035 mm was carefully inserted
from the edge of adhesive layer by 2 mm at the middle height of the
adhesive layer immediately after the application of the adhesive layer to
create a sharp initial crack, as shown in Fig. (3.1).
182 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Fig. (3.1). microscope image shows the inserted 0.035 mm thick mylar sheet located at the middle
plane of the adhesive layer.
The prepared specimens were then pressed by the same weights and cured
for 24 hours at room temperature. The specimens were further put into an
oven at 60.5oC (141oF) for 1 hour for post-curing. After that, they were
cooled down to room temperature before test.
Fig. (3.2). DCB specimen bonded with polished/cleaned adherends and uniform thickness of
adhesive layer through standard feeler gauge.
Since the distance between the loadline and the edge of the adhesive layer is
50 mm, the total initial crack length “a” is 52mm. The geometry
information of each group of specimens was given in Table 3.1. After the
adhesive was applied, a steel pin with the same diameter as the threaded
hole was inserted into the hole to hold the adherends in place.
In the current study, low carbon steel bar were used to fabricate the 9.5mm
thick, 25.4mm wide and 609.6 mm long adherends of enf specimens, as
shown in Fig. (3.3). A total of five groups of specimens were numbered
184 Frontiers in Aerospace Science, Vol. 1 Ji et al.
from group 1 to group 5 for the average adhesive thicknesses of 0.1 mm,
0.2mm, 0.4mm, 0.6mm, and 0.8mm, respectively. Each group had 3
effective specimens, and a total of 15 effective specimens were fabricated.
The specimen geometries were given by Table 3.2.
After curing, the wire shape feeler gages with relevant thickness were
inserted to the end of the specimen. Therefore, during the three-point
bending test, the upper and lower adherends can rotate with the same angle.
Steel Based Single Leg Bending Specimen for Mixed Mode I/II Test
In the current study, steel based adherends has the same geometry as
laminated composite based adherends. A total of five groups of specimens
were numbered from ha=0.1mm to ha=0.8mm with the average adhesive
thicknesses of 0.1 mm, 0.2mm, 0.4mm, 0.6mm, and 0.8mm, respectively.
In order to consider the effect of the initial crack length, three initial crack
length were selected from a=20mm, 50mm, and 80mm. Specimen design
was shown in Fig. (3.4).
In order to maintain coaxial peel force during the testing, two self-aligned,
free-rotating ball pins were designed and fabricated using tool-grade steel,
as schematically shown in Fig. (3.5).
186 Frontiers in Aerospace Science, Vol. 1 Ji et al.
One end of the ball pin with threads was mated with the prefabricated
threads within the holes in the test specimen, and the other end was
connected with the MTS machine. The ball pin had an ability to rotate 360o
in the XZ plane and 30o to -30o in the XY and YZ planes.
The MTS 810 machine was used to conduct the test and collect the loading
force data “P” and the displacement “Δ” of the DCB specimens at the
loading point. The fracture test was conducted under the displacement
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 187
controlled mode. The loading rate was set as 1 µm/s and the data collecting
frequency was 1 Hz.
In order to measure the rotating angle “θ” of the adherends� during the�peel
test, a digital inclinometer and sensor were attached at the free end of the
adherends to collect the data during the test. The accuracy of the
inclinometer is 0.01o. The test range is from -70o to 70o. The data
acquisition frequency is 1 Hz.
Fig. (3.6). DCB specimen attached with inclinometer during the peel test.
188 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Fig. (3.8). Mixed Mode I/II fracture test of two adhesively bonded steel joint with mini inclinometer
and high resolution CCD camera.
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 189
This section includes two parts: one is the results and discussion based on
the Mode I test data of steel adherends DCB specimens; the other is based
on the data of laminated composite DCB Specimens.
The loading rate (at loading point) was equal to 1 µm/s. The preliminary
studies showed that this rate could lead to a very stable descending branch
when the crack was propagated, and it could be safely treated as a
quasistatic loading condition. In the current study, a total of six groups of
specimens which were numbered from group 1 to group 6 for the average
adhesive thicknesses of 0.09 mm, 0.2mm, 0.4mm, 0.6mm, 0.8mm, and
1mm, respectively, were prepared and tested. Each group had 5 effective
specimens, and a total of 30 effective specimens experienced the double
cantilever beam (DCB) test by using the MTS 810 machine. The geometric
parameters of each specimen have been given in Table 3.1. Based on the
analytical results of Ouyang and Li [2], with the current geometric
configuration and the experimental data, it was found that all adherends
(tool-grade steel) remained linear elastic during the entire test process
without any plastic deformation. This indicates that all energy dissipations
were contributed by cohesive fracture and the plastic dissipations in the
adhesive layer. A typical force vs. displacement curve (a specimen in group
2 with an adhesive thickness of 0.2 mm) at the loading point is shown in Fig.
(4.1).
The force linearly increased until the local damage occurs. The load
nonlinearly increases until the crack initiation process is completed. After
that, the crack tip starts propagating. A continuous drop in the peel force
was seen as the crack continuously propagated until the crack reached the
DCB specimen’s end.
190 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Fig. (4.1). A typical relationship between loadline displacements Δ and loadline peel force P for
Group 2 (ha=0.2mm).
The data of the rotation angel at the loading point were collected by the
inclinometer. A typical experimental curve between the rotation angle θP at
the loadline and the displacement Δ at the loadline (a specimen in group 2
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 191
with thickness of 0.2 mm) is shown in Fig. (4.2). With Eq. (2.41), the
experimental energy release rate J is determined by combining the
measured θP and P (or combining Figs. 4.1 and 4.2). A typical relationship
(a specimen in group 2 with thickness of 0.2 mm) between global energy
release rate J and loadline displacement Δ is given in Fig. (4.3). In a real test,
we found that it was fairly difficult to exactly define when the initial crack
tip was propagated. Therefore, a characteristic strain energy release rate J0
was defined which represented the J value when the maximum peel force P
was reached. Note that with the growth of the crack, the global strain energy
release rate keeps increasing as shown in Fig. (4.3) (J-Δ curve). This
implies that the plastic dissipations beyond the crack tip must keep
increasing during this stable growth process. However, the increase rate of J
becomes slower and slower as the crack grows, which seems nearly stable
even by the end of the test. Therefore, an approximate asymptote is added in
Fig. (4.3) to estimate the interfacial toughness of the bonded joints. And this
estimated asymptotic value is denoted by JC or fracture energy.
Fig. (4.2). A typical relationship between loadline displacements Δ and loadline rotation of adherend
θP for Group 2 (ha=0.2mm).
192 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Fig. (4.3). A typical relationship between loadline displacements Δ and energy release rate J for
Group 2 (ha=0.2mm).
In this study, crack tip local deformations along the entire adhesive layer
were recorded using high resolution CCD camera. It was also observed
that a whitening zone appeared near the initial crack tip and was
becoming more visible as the load was increased, as shown in Fig. (4.4).
Fig. (4.6). A typical relationship between energy release rate J and local crack tip separation δ for
Group 2 (ha=0.2mm).
With the global test results of different groups, the average maximum peel
loads Pcr of the five specimens in each group (with error bar) are plotted as
a function of the adhesive thicknesses (0.09 mm, 0.2 mm, 0.4 mm, 0.6
mm, 0.8 mm, and 1.0 mm) in Fig. (4.7). It can be observed that the
average peak load was increased from 578.8 N (Group 1) to 791.12 N
(Group 6) when the thickness of the adhesive layer increased (from 0.09
mm to 1.0 mm). These critical loads correspond to the characteristic
energy release rate J0. The average estimated fracture energy JC (by the
approximate asymptotes) and the average characteristic J0 (at the
maximum peel load) of the five specimens in each group (with error bar)
were plotted as a function of the adhesive thickness in Fig. (4.8). One may
see that the fracture energy JC at the thickness of 0.09 mm is increased by
approximately 160% when the adhesive thickness becomes 1.0 mm. The
significant increase of J0 and JC are responsible for the increased load
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 195
Fig. (4.7). The average maximum peel load Pcr at different adhesive thicknesses ha.
196 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Fig. (4.8). The estimated fracture energy JC and the characteristic energy release rate J0
(corresponding to Pcr) with different adhesive thicknesses ha.
Fig. (4.9). The ratio of ΔJ (ΔJ=JC─J0) over J0 as a function of the adhesive thickness ha.
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 197
Fig. (4.10). Typical shape of the equivalent interfacial traction-separation laws with different
thicknesses of adhesive layer.
Fig. (4.11). The shapes of all equivalent interfacial traction-separation laws with different
thicknesses of adhesive layer: (a) adhesive thickness ha=0.09 mm; (b) adhesive thickness ha=0.2 mm;
(c) adhesive thickness ha=0.4 mm; (d) adhesive thickness ha=0.6 mm; (e) adhesive thickness ha=0.8
mm; (f) adhesive thickness ha=1.0 mm.
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 199
The average interfacial strength σmax of the six groups (with error bar) are
plotted as a function of the adhesive thickness in Fig. (4.12). According to
the manufacturer, the tensile strength of the bulk adhesive material is
approximately 30.3 MPa (its yield strength is slightly lower than its tensile
strength). One can see that the interfacial strength was 88 MPa at the
thickness of 0.09 mm, which is approximately 3 times as large as the yield
strength of the bulk adhesive material. The authors expects that with further
decrease in the adhesive thickness, the measured equivalent interfacial
strengths should keep increasing. This tendency is correct because the
lower bound is when the adhesive thickness is reduced to the atomistic level,
where the traction-separation response of a discrete atom pairs shows a very
high strength but very low separation. On the contrary, with the increase in
adhesive thickness, the measured interfacial strengths asymptotically
approach a constant value. This asymptote seems being the yield strength of
bulk adhesive material. Another phenomenon is that the equivalent
interfacial strength seems increasing dramatically when the adhesive
thickness was thinner than a certain value (such as 0.2 mm).
Fig. (4.12). The effect of the adhesive thickness on the interfacial strength σmax.
200 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Based on the analytical results of Ouyang and Li [35], with the current
geometric configuration and the experimental data Table 3.2, it was found
that all adherends (low carbon steel) remained linear elastic during the
entire test process without any plastic deformation. This indicates that all
the energy dissipation during the crack initiation and propagation were
contributed by the intrinsic cohesive fracture and the possible plastic
dissipation in the adhesive layer. A typical force vs. displacement curve at
the loading point is shown in Fig. (5.1) (a specimen in group 1 with
thickness of 0.1 mm). The force linearly increases until the local damage
occurs. However, the load increases in a nonlinear pattern until the crack
initiation process is completed. After that, the crack tip starts propagating.
The crack continuously propagates with a continuous drop in the force until
the crack propagates to the end of the ENF specimen. With Eq. (2.44), the
experimental value of the strain energy release rate JII can be obtained by
combining experimental values of δ0 and QT.
Fig. (5.1). A typical relationship between loadline displacements Δ and loadline peel force P for
Group 1 (ha=0.1 mm).
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 201
The value of the shear slip δ between the two adherends at the location of
the initial crack tip is calculated based on the digital images, and they are
plotted as a function of the loadline displacement Δ (global displacement)
in Fig. (5.2) (a specimen in group 1 with thickness of 0.1 mm). By
combining Figs. (5.1 and 5.2), a typical experimental J-δ curve is obtained
as shown in Fig. (5.3) (a specimen in group 1 with thickness of 0.1 mm).
With Eq. (2.45), the experimental J-δ curves are used to determine the
equivalent interfacial traction-separation laws or τ=τ(δ) at different
adhesive thicknesses. A characteristic strain energy release rate J0 is
defined as the J value when the crack starts propagating immediately after
the maximum fracture load Pcr is reached. It can be seen that J0 is generally
smaller than the critical strain energy release rate or the fracture energy JIIC,
which reflects that the strain energy release rate JII keeps increasing after
the crack propagation. This is because the plastic dissipation (or plastic
zone) can be further increased after the crack initiation.
Fig. (5.2). A typical relationship between loadline displacement Δ and local crack tip slip δ for Group
1 (ha=0.1 mm).
202 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Fig. (5.3). A typical relationship between energy release rate J and local crack tip slip δ for Group 1
(ha=0.1 mm).
In this study, the crack initiation and propagation processes were recorded
by using the high resolution CCD camera. It was observed that the initial
crack tip started opening as the shear load increased. It was also observed
that the whiting zone appeared near the initial crack tip and was becoming
more visible as the load was increased, as shown in Fig. (5.4) (red circle
area).
10
9
8
7
6
τ (MPa)
5
4
3
2
1
0
0 100 200 300 400
δ (µm)
Fig. (5.6). Typical interface constitutive relationship (equivalent cohesive law) for Group 1 (ha=0.1
mm).
With the global experiment results for different groups, the maximum
peel loads are plotted as a function of the adhesive thicknesses (0.1 mm,
0.2 mm, 0.4 mm, 0.6 mm, and 0.8 mm) as shown in Fig. (5.7). It can be
observed that the average peak load is increased from 4912.24 N (Group
1) to 5836.33 N (Group 5) when the thickness of adhesive layer increases
(from 0.1 mm to 0.8 mm). However, the rate of increase in the peak load
reduces as the adhesive layer thickness increases.
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 205
Fig. (5.7). The average maximum shear load Pcr at different adhesive thicknesses (ha=0.1 mm).
The increase in the adhesive layer thickness might increase the bending
stiffness of the ENF specimen. However, this contribution is very small
since the adherends were made of low carbon steel with a much higher
thickness of 9.535 mm and much higher bending stiffness. The estimated
average fracture energy JC and the average characteristic J0 (at the
maximum peel load) were plotted as a function of the adhesive thickness
in Fig. (5.8) (0.1 mm, 0.2 mm, 0.4 mm, 0.6 mm and 0.8 mm). The
significant increase in J0 and JC are responsible for the increased load
capacity when the adhesive thickness becomes thicker. Meanwhile, one
may see that when the adhesive thickness is thin, J0 and JC are relatively
close to each other. When it is relatively thick, J0 and JC are relatively
departed. After data collecting and curve-fitting process of the
experimental J-δ curves, by applying Eq. (2.45), one can see that the
interfacial traction-slip laws are determined by numerically differentiating
the experimental J-δ curves.
206 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Fig. (5.8). The estimated fracture energy JC and the characteristic energy releaserate J0
(corresponding to Pcr) with different adhesive thicknesses.
Fig. (5.9). Typical shapes of the interfacial traction–separation laws at different adhesive layer
thicknesses ha.
Fig. (5.10). The effect of adhesive thickness on the maximum interface stress τf.
208 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Fig. (5.11). The shapes of all interfacial traction-slip laws with different thicknesses of bondline: (a)
bondline thickness ha=0.1 mm; (b) bondline thickness ha=0.2 mm; (c) bondline thickness ha=0.4 mm;
(d) bondline thickness ha=0.6 mm; (e) bondline thickness ha=0.8 mm.
The average interfacial strength τmax of the five groups (with error bar) are
plotted as a function of the adhesive thickness in Fig. (5.10). According to
the manufacturer, the shear strength of the adhesive material is
approximately 13.8 MPa (when the thickness is 0.75 mm with steel
adherends). One can see that the interfacial strength was 14.9 MPa at the
thickness of 0.8 mm, which is approximately the yield shear strength of the
bulk adhesive material. Finally, the complete fifteen interfacial
traction-separation laws of the bonded steel ENF joints (three specimens in
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 209
each of the five groups at various bondline thicknesses) are given in Fig.
(5.11). One may see the fairly consistent results of the three specimens in
each group at various bondline thicknesses.
The geometric parameters of each specimen have been given in Table 3.3. A
typical force vs. displacement curve (a specimen with ha=0.1mm and an
initial crack length a=80mm) at the loading point is shown in Fig. (6.1).
The force linearly increased until the local damage occurs. The load
nonlinearly increases until the crack initiation process is completed. After
that, the crack tip starts propagating. A continuous drop in the bending force
was seen as the crack continuously propagated until the crack reached the
specimen’s end.
Fig. (6.1). A typical relationship between loadline displacements Δ and loadline bending force P for
ha=0.1mm with initial crack length a=80mm.
210 Frontiers in Aerospace Science, Vol. 1 Ji et al.
The data of the rotation angel at the loading point were collected by the
inclinometer. With Eq. (2.50), the experimental energy release rate J is
determined by combining the measured θP and P. A typical relationship (a
specimen with ha=0.1mm and initial crack length of 20mm) between global
energy release rate J and loadline displacement Δ is given in Fig. (6.2). In a
real test, we found that it was fairly difficult to exactly define when the
initial crack tip was propagated. Therefore, a characteristic strain energy
release rate J0 was defined which represented the J value when the
maximum peel force P was reached. Note that with the growth of the crack,
the global strain energy release rate keeps increasing as shown in Fig. (6.2)
(J-Δ curve). This implies that the plastic dissipations beyond the crack tip
must keep increasing during this stable growth process. However, the
increase in J becomes slower and slower as the crack grows, which seems
nearly stable even by the end of the test. Therefore, an approximate
asymptote is added in Fig. (6.2) to estimate the interfacial toughness of the
bonded joints. And this estimated asymptotic value is denoted by JC or
fracture energy.
Fig. (6.2). A typical relationship between loadline displacements Δ and energy release rate J for
specimen with ha=0.1mm (a=80mm).
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 211
In this study, crack tip local deformations along the entire adhesive layer
were recorded using the high resolution CCD camera. The value of the
crack tip separation δ was measured as the relative normal displacement
between the two steel adherends at the location of initial crack tip through
the recorded digital images. By combining JI-δ curve and crack tip
separation δ, a typical experimental JI-δ curve was obtained as shown in Fig.
(6.3) (a specimen with ha=0.1mm and initial crack length of 80mm). With
Eq. (2.51), the experimental JI-δ curves were used to determine the
equivalent interfacial traction-separation laws or σ=σ(δ), as show in Fig.
(6.4).
Fig. (6.3). A typical relationship between energy release rate J and local crack tip separation δ for
specimen with ha=0.1mm (a=80mm).
212 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Fig. (6.4). A typical shape of the equivalent interfacial traction-separation laws with ha=0.1mm
(a=80mm) adhesive layer.
It was observed that the initial crack tip started opening as the shear load
increased. It was also observed that the whiting zone appeared near the
initial crack tip and was becoming more visible as the load was increased,
as shown in Fig. (6.5) (circle area).
Fig. (6.6). A typical relationship between energy release rate J and local crack tip slip δ for specimen
with ha=0.1mm (a=80 mm).
The value of the shear slip δ between the two adherends at the location of
the initial crack tip is calculated based on the digital images. By combining
P-Δ curve and local shear slip δ, according to Eq. (2.46), a typical
experimental JII-δ curve is obtained as shown in Fig. (6.6) (a specimen wiht
ha=0.1mm and initial crack length 80mm). With Eq. (2.49), the
experimental JII-δ curves are used to determine the equivalent interfacial
traction-separation laws or τ=τ(δ) at different adhesive thicknesses. A
214 Frontiers in Aerospace Science, Vol. 1 Ji et al.
characteristic strain energy release rate J0 is defined as the J value when the
crack starts propagating immediately after the maximum fracture load Pcr is
reached. It can be seen that J0 is generally smaller than the critical strain
energy release rate or the fracture energy JIIC, which reflects that the strain
energy release rate JII keeps increasing after the crack propagation. This is
because the plastic dissipation (or plastic zone) can be further increased
after the crack initiation.
Based on Eq. (2.47), these experimental JII-δ curves are used to determine
the equivalent interface cohesive laws or τ=τ(δ) at different adhesive
thicknesses. Fig. (6.7) gives a typical shear stress vs. local slip displacement
curve (a specimen with ha=0.1mm and initial crack length of 80mm).
Fig. (6.7). Typical interface constitutive relationship (equivalent cohesive law) for specimen with
ha=0.1 mm (a=80mm).
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 215
With the global experiment results for different groups, the maximum
bending loads are plotted as a function of the adhesive thicknesses (0.1
mm, 0.2 mm, 0.4 mm, 0.6 mm, and 0.8 mm) as shown in Fig. (6.8). It can
be observed that the average peak load is increased from 3417.67 N to
4791.10 N when the thickness of adhesive layer increases (from 0.1 mm
to 0.8 mm), as shown by the trend line in Fig. (6.8).
Fig. (6.8). The average maximum shear load Pcr at different adhesive thicknesses ha.
The increase in the adhesive layer thickness might increase the bending
stiffness of the specimen. However, this contribution is very small since the
adherends were made of low carbon steel with a much higher thickness of
6.35 mm and much higher bending stiffness. The estimated average fracture
energy JIC, JIIC and the average characteristic energy JI0, JII0 (at the
maximum peel load) were plotted as a function of the adhesive thickness in
Figs. (6.9 and 6.10) (0.1 mm, 0.2 mm, 0.4 mm, 0.6 mm and 0.8 mm). The
significant increase in J0 and JC are responsible for the increased load
capacity when the adhesive thickness becomes thicker. Meanwhile, one
may see that when the adhesive thickness is thin, J0 and JC are relatively
216 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Fig. (6.9). The estimated fracture energy JIC and the characteristic energy release rate JI0
(corresponding to Pcr) with different adhesive thicknesses (a=20mm).
Mode I Portion
Fig. (6.10). The estimated fracture energy JIIC and the characteristic energy release rate JII0
(corresponding to Pcr) with different adhesive thicknesses (a=20mm).
218 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Fig. (6.11). Typical shapes of the interfacial traction-separation laws of bonded steel with different
bondline thicknesses ha (a=20mm).
Mode II Portion
First, the interfacial strength increases for steel joint with the increase in
the bondline thickness. Second, the initial stiffness of the interfacial laws
(the initial slope of the τ-δ curve) is identical for steel joints with various
bondline thicknesses, which suggests that the shear deformation is
controlled by the adhesive layer. Finally, the fracture energy (the total area
under the δ-τ curve) increases with the bondline thickness. As shown in
Fig. (6.10), the fracture energy JIIC of the bonded joints is plotted as a
function of the adhesive thickness. The fracture energy JIIC at the
thickness of 0.1 mm is increased by approximately 143% when the
adhesive thickness becomes 0.8 mm. This significant increase in fracture
energy can explain the increase in load capacity when the adhesive
thickness is increased.
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 219
Fig. (6.12). Typical shapes of the interfacial traction-slip laws of bonded steel with different bondline
thicknesses ha (a=20mm).
Mode Mixities
The ratio of the estimated fracture energy JIIC and total fracture energy
(JIC +JIIC) with different adhesive thickness was plotted in Fig. (6.13).
The ratio of the shear strength and total strength with different adhesive
thickness was plotted in Fig. (6.14). It is observed that the ratio of the
estimated fracture energy JIIC and the ratio of the shear strength were
increased as the thickness of the adhesive layer increased. This indicates
that the ratio of mode I within the mixed mode I/II was decreased as the
thickness of the adhesive layer increased. The stiffness of the adhesive
layer is much lower than the stiffness of the adherends. Therefore, as the
thickness of adhesive layer increased, the plastic dissipation increased.
220 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Fig. (6.13). The ratio of the estimated fracture energy JIIC and total fracture energy (JIC + JIIC) with
different adhesive thicknesses (a=20mm).
Fig. (6.14). The ratio of the shear strength and total strength with different adhesive thicknesses
(a=20mm).
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 221
In this study, the effects of the initial crack tip length of the adhesive layer
were also considered with different thickness of the adhesive layer. Three
different initial crack lengths a=20mm, 50mm, and 80mm were selected in
this study.
The maximum bending load Pcr at each initial crack length with different
thickness of the adhesive layer was plotted in Fig. (6.15). It is observed that
the maximum load was decreased as the initial crack length increased; the
maximum load was increased as the thickness of the adhesive layer
increased at each initial crack length. As the initial crack length increased at
the same thickness of the adhesive layer, the bending moment was
increased at the initial crack tip. It caused the peeling force increase near the
crack tip area at the same time. In other words, the mode I portion was
increased as the initial crack length increased.
Fig. (6.15). The average maximum bending load Pcr at different initial crack length with different
adhesive thicknesses ha.
222 Frontiers in Aerospace Science, Vol. 1 Ji et al.
When the thickness of the adhesive layer was 0.1mm, as the initial crack
length increased, the mount of the adhesive material between the
adherends decreased and the moment around the initial crack tip area
increased. Additionally, the relative rotation angle between the two
adherends increased and as a result, the smallest peeling force can make
the crack propagate, as shown in Fig. (6.15). According to Eq. (2.50):
JI=P*θ/4, although the loading force increased from 2047N to 3417N as
the initial crack length decreased from 80mm to 20mm, the relative
rotation angle decreased from 3.23rad to 1.97rad. As the thickness of the
adhesive layer increased, the ratio of the increased relative rotation angle
increased. It can be explained that why the increased ratio of JIC with
initial crack length a=80mm is larger than the other two, as the thickness
of adhesive layer increased.
The estimated fracture energy JIC and JIIC at each initial crack length with
different adhesive thickness were plotted in Fig. (6.16 and 6.17). It is
observed that the JIC and JIIC were increased at each initial crack length as
the thickness of the adhesive layer increases.
In Fig. (6.16), the JIC was increased when the initial crack length was
increased from 20mm to 50mm. Further increasing the initial crack length
from 50mm to 80mm leads to reduction in JIC when the adhesive
thickness is 0.1mm. With the increase in the adhesive layer thickness, the
tread line of JIC with the initial crack length of 80mm line crossed the
other trending lines with initial crack length at 20mm and 50mm. The
phenomenon may be due to the combination effect of the thickness of the
adhesive layer and initial crack length. This is a very interesting point that
deserves in-depth investigation.
In Fig. (6.17), the JIIC was decreased as the initial crack length increased
with the same thickness of adhesive layer. It is understandable due to the
decreased loading force with increased initial crack length. Additionally,
as the thickness of the adhesive layer increased, the relative rotation angle
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 223
between the two adherends increased and caused the increase in shear
force.
Fig. (6.16). The estimated fracture energy JIC at each initial crack length with different adhesive
thicknesses.
Fig. (6.17). The estimated fracture energy JIIC at each initial crack length with different adhesive
thicknesses.
224 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Additionally, the interfacial strength τmax increases for steel joint with the
increase in the bondline thickness. The initial stiffness of the interfacial
laws (the initial slope of the τ-δ curve) is identical for steel joints with
various bondline thicknesses, which suggests that the shear deformation is
controlled by the adhesive layer. The fracture energy (the total area under
the δ-τ curve) increases with the bondline thickness.
Fig. (6.18). Typical shapes of the interfacial traction-separation laws of bonded steel with different
bondline thicknesses ha (a=20mm).
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 225
Fig. (6.19). Typical shapes of the interfacial traction-slip laws of bonded steel with different bondline
thicknesses ha (a=20mm).
Fig. (6.20). Typical shapes of the interfacial traction-separation laws of bonded steel with different
bondline thicknesses ha (a=20mm).
226 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Fig. (6.21). Typical shapes of the interfacial traction-slip laws of bonded steel with different bondline
thicknesses ha (a=20mm).
Fig. (6.22). The effect of the adhesive thickness on the interfacial strength σmax with different initial
crack length.
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 227
Fig. (6.23). The effect of the adhesive thickness on the interfacial strength τmax with different initial
crack length.
According to Figs. (6.22 and 6.23), as the initial crack tip length increased,
the interfacial shear strength was decreased with the increased thickness of
the adhesive layer. Additionally, as the initial crack tip length increased
from a=20mm to a=50mm, the interfacial peel strength decreased with
increased thickness of the adhesive layer. However, when the initial crack
length increased from 50mm to 80mm and the thickness increased from
0.1mm to 0.6mm, the peel strength increased due to the increased moment
and relative rotation angle which caused a compression zone near the crack
tip area and prevented the crack propagation. Therefore, the peel strength
increased. When the thickness increased from 0.6mm to 0.8mm, the
bending stiffness of the adhesive layer increased and caused the decreased
relative rotation angle. Therefore, the peel strength decreased without the
compression zone around the crack tip area.
228 Frontiers in Aerospace Science, Vol. 1 Ji et al.
Mode I Test
e. With the J-integral theory, the present study first presents the
experimental evidence that the cohesive strength is 88 MPa which
is about 3 times of the yield strength of the bulk adhesive when
the adhesive thickness is 90 μm (steel based specimen).
f. It is also noted that the significant plastic dissipation will not
necessarily be reflected in the globally measured value of the
fracture energy when the adhesive thickness is relatively thin
and even the intrinsic cohesive strength is much higher than the
bulk adhesive’s yield strength.
Mode II Test
In this study, the steel based adherends of adhesively bonded joints are
subjected to mode II loading using End Notched Flexure (ENF)
specimens. The objective is to measure the fracture toughness of the joint
corresponding to different thicknesses of the adhesive layer and to deduce
the nature of the cohesive (traction-separation) law that governs the crack
growth behavior. Within the boundaries of the used materials parameters
and based on the test results, the following conclusions are obtained:
a. The adhesive thickness has an influence on the critical loading
force. As the thickness of the adhesive layer increases, the critical
force is increased.
b. As the adhesive thickness increases, the fracture toughness JIIC
and characteristic energy JII0 (corresponding to critical load) are
increased.
c. The local crack tip slipping δ is also increased, as the thickness of
the adhesive layer increases.
d. The thickness of the adhesive layer also has a considerable effect
on the fracture strength (τ). As the adhesive thickness increases,
the fracture strength (τ) is increased.
230 Frontiers in Aerospace Science, Vol. 1 Ji et al.
In this study, the steel based adherends of adhesively bonded joints are
subjected to mixed mode I/II loading using Single Leg Bending (SLB)
specimens. The objective is to measure the fracture toughness (JIC & JIIC)
of the joint corresponding to different thicknesses of the adhesive layer
and different initial crack lengths and to deduce the nature of the cohesive
(traction-separation & traction-slipping) law that governs the crack
growth behavior. Within the boundaries of the used materials parameter
and based on the test results, the following conclusions are derived:
However, the fracture toughness JIC and fracture strength (σ) are not
always increased or decreased as the initial crack length increases. It is
very complicated due to the combination effect of the initial crack length
and adhesive thickness.
CONFLICT OF INTEREST
The authors confirm that they have no conflict of interest to declare for
this publication.
ACKNOWLEDGEMENTS
This study is based upon work supported by the NSF under grant number
(NSF/CMMI0900064) and by the NASA/EPSCoR under grant number
NASA/LEQSF (2007-10)-Phase3-01.
REFERENCE
[1] P. Mazza, "A new Palaeolithic discovery: tar-hafted stone tools in a European
Mid-Pleistocene bone- bearing bed", J. Archaeol. Sci., vol. 33, p. 1310, 2006.
[http://dx.doi.org/10.1016/j.jas.2006.01.006]
[2] Z. Ouyang, and G. Li, "Local damage evolution of DCB specimens during
crack initiation process: A natural boundary condition based method. ASME",
J. Appl. Mech., vol. 76, p. 051003, 2009.
[http://dx.doi.org/10.1115/1.3112742]
[3] A.J. Kinloch, and S.J. Shaw, "The fracture resistance of a toughened epoxy
adhesive", J. Adhes., vol.12, pp. 59-77, 1981.
[http://dx.doi.org/10.1080/00218468108071189]
232 Frontiers in Aerospace Science, Vol. 1 Ji et al.
[5] A.A. Griffith, "The phenomena of rupture and flow in solids", Philos.
Trans. R. Soc. Lond. A, vol. 221, pp. 163-198,
1921.[http://dx.doi.org/10.1098/rsta.1921.0006]
[6] G.R. Irwin, and J.A. Kies, "Critical energy rate analysis of fracture strength",
Weld. J., vol. 33, pp. 193-198, 1954.
[8] G.I. Barenblatt, "The formation of equilibrium cracks during brittle fracture.
General ideas and hypothesis. Axisymmetrical cracks", J. Appl. Math. Mech.,
vol. 23, pp. 622-636, 1959. [PMM].
[http://dx.doi.org/10.1016/0021-8928(59)90157-1]
[9] D.S. Dugdale, "Yielding of steel sheets containing slits", J. Mech. Phys.
Solids, vol. 8, pp. 100-104, 1960.
[http://dx.doi.org/10.1016/0022-5096(60)90013-2]
[10] J.R. Rice, "A path independent integral and the approximate analysis of strain
concentration by notches and cracks", ASME Journal of Applied Mechanics,
vol. 35, pp. 379-386, 1968.
[http://dx.doi.org/10.1115/1.3601206]
[11] Y. Wei, and J.W. Hutchinson, "Interface strength, work of adhesion and
plasticity in the peel test", Int.J. Fract., vol. 93, pp. 315-333, 1998.
[http://dx.doi.org/10.1023/A:1007545200315]
[12] J.G. Williams, and H. Hadavinia, "Analytical solutions for cohesive zone
models", J. Mech. Phys. Solids, vol. 50, pp. 809-825, 2002.
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 233
[http://dx.doi.org/10.1016/S0022-5096(01)00095-3]
[14] J.H. Rose, J.R. Smith, and J. Ferrante, "Universal features of bonding in
metals", Phys. Rev. B, vol. 28, pp. 1835-1845, 1983.
[http://dx.doi.org/10.1103/PhysRevB.28.1835]
[19] V. Tvergaard, and J.W. Hutchinson, "The relation between crack growth
resistance and fracture process parameters in elastic-plastic solids", J. Mech.
Phys. Solids, vol. 40, pp. 1377-1397, 1992.
[http://dx.doi.org/10.1016/0022-5096(92)90020-3]
[21] S.R. Chowdhury, and R. Narasimhan, "A finite element analysis of stationary
crack tip fields in a pressure sensitive constrained ductile layer", Int. J. Solids
Struct., vol. 37, pp. 3079-3100, 2000.
[http://dx.doi.org/10.1016/S0020-7683(99)00046-3]
[22] Q.D. Yang, M.D. Thouless, and S.M. Ward, "Elastic-plastic mode-II fracture
of adhesive joints", Int. J. Solids Struct., vol. 38, pp. 3251-3262, 2001.
[http://dx.doi.org/10.1016/S0020-7683(00)00221-3]
[23] Q.D. Yang, M.D. Thouless, and S.M. Ward, "Mixed mode fracture analysis
of plastically-deforming adhesive joints", Int. J. Fract., vol. 110, pp. 175-187,
2001. [http://dx.doi.org/10.1023/A:1010869706996]
[24] G. Alfano, and M.A. Crisfield, "Finite element interface models for the
delamination analysis of laminated composites: mechanical and computational
issues", Int. J. Numer. Methods Eng., vol. 50, pp. 1701-1736, 2001.
[http://dx.doi.org/10.1002/nme.93]
[25] R.H. Andruet, D.A. Dillard, and S.M. Holzer, "Two- and three-dimensional
geometrical nonlinear finite elements for analysis of adhesive joints", Int. J.
Adhes. Adhes., vol. 21, pp. 17-34, 2001.
[http://dx.doi.org/10.1016/S0143-7496(00)00024-5]
[27] J.L. Högberg, "Mixed mode cohesive law", Int. J. Fract., vol. 141, pp.
Effect of Bondline Thickness Frontiers in Aerospace Science, Vol. 1 235
[31] M.F. de Moura, R.D. Campilho, and J.P. Gonçalves, “Pure mode II
fracture characterization of composite bonded joints", Int. J. Solids Struct.,
vol. 46, pp. 1589-1595, 2009.
[http://dx.doi.org/10.1016/j.ijsolstr.2008.12.001]
[33] B.R. Blackman, H. Hadavinia, A.J. Kinloch, and J.G. Williams, "The use of a
cohesive Zone model to study the fracture of fibre composites and
adhesively-bonded joints", Int. J. Fract., vol. 119, pp. 25-46, 2003.
[http://dx.doi.org/10.1023/A:1023998013255]
[34] J. Pan, and C.K. Leung, "Debonding along the FRP-concrete interface under
combined pulling/peeling effects", Eng. Fract. Mech., vol. 74, pp. 132-150,
2007. [http://dx.doi.org/10.1016/j.engfracmech.2006.01.022]
[35] Z. Ouyang, and G. Li, "Nonlinear interface shear fracture of end notched
236 Frontiers in Aerospace Science, Vol. 1 Ji et al.
flexure specimens", Int. J. Solids Struct., vol. 46, pp. 2659-2668, 2009.
[http://dx.doi.org/10.1016/j.ijsolstr.2009.02.011]
[36] Z. Ouyang, and G. Li, "Cohesive zone model based analytical solutions for
adhesively bonded pipe joints under torsional loading", Int. J. Solids Struct.,
vol. 46, pp. 1205-1217, 2009.
[http://dx.doi.org/10.1016/j.ijsolstr.2008.10.021]
[37] Z. Ouyang, and G. Li, "Interfacial debonding of pipe joints under torsion
loads: A model for arbitrary nonlinear cohesive laws", Int. J. Fract., vol. 155,
pp. 19-31, 2009.
[http://dx.doi.org/10.1007/s10704-009-9318-7]
[39] C. Nguyen, and A.L. Levy, "An exact theory of interfacial debonding in
layered elastic composites", Int. J. Solids Struct., vol. 46, pp. 2712-2723,
2009.
[http://dx.doi.org/10.1016/j.ijsolstr.2009.03.005]
[40] H. Chai, "Shear fracture", Int. J. Fract., vol. 37, pp. 137-159, 1988.
[43] M.S. Kafkalidis, M.D. Thouless, Q.D. Yang, and S.M. Ward, "Deformation
and fracture of adhesive layers constrained by plastically-deforming
adherends", J. Adhes. Sci. Technol., vol. 14, pp. 1593- 1607, 2000.
[http://dx.doi.org/10.1163/156856100742401]
[44] C. Yan, Y. Mai, and L. Ye, "Effect of bond thickness on fracture behaviour in
adhesive joints", J. Adhes., vol. 75, pp. 27-44, 2001.
[http://dx.doi.org/10.1080/00218460108029592]
[46] D. Lee, T. Ikeda, N. Miyazaki, and N. Choi, "Effect of bond thickness on the
fracture toughness of adhesive joints", J. Eng. Mater. Technol., vol. 126, pp.
14-18, 2004.
[http://dx.doi.org/10.1115/1.1631433]
[47] B.F. Søensen, "Cohesive law and notch sensitivity of adhesive joints", Acta
Mater., vol. 50, pp. 1053- 1061, 2002.
[http://dx.doi.org/10.1016/S1359-6454(01)00404-9]
[49] T. Andersson, and A. Biel, "On the effective constitutive properties of a thin
adhesive layer loaded in peel", Int. J. Fract., vol. 141, pp. 227-246, 2006.
[http://dx.doi.org/10.1007/s10704-006-0075-6]
[52] L.A. Carlsson, J.W. Gillespie, and R.B. Pipes, "On the analysis and design of
the end notched flexure (ENF) specimen for mode II testing", J. Compos.
Mater.,vol. 20, pp. 594-604, 1986.
[http://dx.doi.org/10.1177/002199838602000606]
[53] A.J. Russell, and K.N. Street, Factors affecting the interlaminar fracture
energy of Graphite/Epoxy laminates. Progress in Science and Engineering of
Composites, T. Hayashi, K Kawata & S. Umekawa ed., ICCM-IV, Tokyo, p.
279-286, 1982.
[54] P. Davies, G.D. Sims, B.R. Blackman, A.J. Brunner, K. Kageyama, M. Hojo,
K. Tanaka, G. Murri, C. Rousseau, B. Gieseke, and R.H. Martin, "Comparison
of test configurations for determination of mode II interlaminar fracture
toughness results from international collaborative test programme", Plastics
Rubber Composite, vol. 28, pp. 432-437, 1999.
[http://dx.doi.org/10.1179/146580199101540600]
[55] B.D. Davidson, and X. Sun, "Effects of friction, geometry and fixture
compliance on the perceived toughness from three and four point bended
notched flexure tests", J. Reinf. Plast. Compos., vol. 4, pp. 1611-1628, 2005.
[http://dx.doi.org/10.1177/0731684405050402]
[58] A.J. Brunner, "Experimental aspects of mode I and mode II fracture toughness
testing of fibre- reinforced polymer–matrix composites", Comput. Methods
Appl. Mech. Eng., vol. 185, pp. 161-172, 2000.
[http://dx.doi.org/10.1016/S0045-7825(99)00257-1]
[59] A.J. Brunner, B.R. Blackman, and J.G. Williams, "Calculating a damage
parameter and bridging stress from GIC delamination tests on fibre
composites", Compos. Sci. Technol., vol. 66, pp. 785-795, 2006.
[http://dx.doi.org/10.1016/j.compscitech.2004.12.040]
[60] Q.D. Yang, M.D. Thouless, and S.M. Ward, "Numerical simulations of
adhesively-bonded beams failing with extensive plastic deformation", J.
Mech. Phys. Solids, vol. 47, pp. 1337-1353, 1999.
[http://dx.doi.org/10.1016/S0022-5096(98)00101-X]
[61] S. Li, M.D. Thouless, A.M. Waas, J.A. Schroeder, and P.D. Zavattieri, "Use
of a cohesive-zone model to analyze the fracture of a fiber-reinforced
polymer-matrix composite", Compos. Sci. Technol., vol. 65, pp. 537-549,
2005.
[http://dx.doi.org/10.1016/j.compscitech.2004.08.004]
240 Frontiers in Aerospace Science, 2016, Vol. 1, 240-268
CHAPTER 5
Abstract: With the continual development of the aircraft industry, aircraft engines
have provoked people’s attention more and more. The turbine blade plays a vital and
critical component of aircraft engines. In order to conform to the dimensional
tolerances of wax pattern die-profile for turbine blade in investment casting process,
this chapter provides an optimization method of geometric parameter for turbine
blades based on inverse adjustment. The geometric parameters for optimizing were
extracted, and the bending and torsional deformation can be compensation.
Therefore the nonlinear deformation compensation during solidification and cooling
procedure can be efficiently realized. This method set the theoretical foundation on
optimization method of die-cavity for turbine blade. The die-profile optimization
system which was developed in this paper proves better effect for the die-cavity
design. This chapter also offers a reverse design methodology for investment die
casting using ProCAST. In industry, the performance of the engine depends not only
on shape, but also on the dimensions of the components. This process is difficult as
super-alloy blade material cannot be easily machined. However investment casting
is an ideal process for such net - shape components, but it still requires an accurate
determination of the casting-die profile. In order to investigate and analyze the
methods that affect the shape and dimensions of the turbine blade most, similar
*Corresponding author Yangqing Dou: Department of Mechanical Engineering, Mississippi State University,
MS, 39762, USA; E-mail: yd120@msstate.edu.
NOMENCLATURE
INTRODUCTION
There are more than 20 kinds of parameters which are related to the
structure, strength, and the manufacturing process of turbine blade [19],
however, this paper mainly studies the method in determining die-profile
of turbine blade. Because the blade-design method can be based on
specifying a thickness distribution around the mean camber line, thus,
those parameters which can be used to represent and re-construct the
mean camber line are discussed.
2
The way of extraction blade exit angle (α1) is similar to the blade inlet
extraction method. Considering the angle between the line connecting the
first two points and the front of blade as the blade inlet angle, the blade
inlet angle can be established.
246 Frontiers in Aerospace Science, Vol. 1 Dou et al.
2 1
The stagger angle β is the parameter which can be used to represent the
torsional deformation of blade, hence, the compensation of the torsional
deformation is to compensate the deformation of β. As indicated in Fig. (2),
the solid line represents the schematic diagram of the blade section and the
dashed line represents the section after torsional deformation. β1 represents
the stagger angle of CAD model, and β2 represents the stagger angle of the
measuring model. The torsional deformation can be expressed as ∆β = β2 -
β1, the equation which can be used for torsion deformation compensation
can be rearranged as:
n 1
m (lim i )D c D c (1)
x i 1 1
The bending deformation can be denoted by the blade inlet angle α1 , the
blade exit angle α2, the maximum deflection position Pm, and the chord
248 Frontiers in Aerospace Science, Vol. 1 Dou et al.
y2 y1
Pxy ( x, y ) (1 )Wxy ( x, y ) P0 ( x0 , y0 ) (2)
x2 x1
Where P0(x0, y0) is the coordinate value of an arbitrary dot of CAD model,
Pxy (x, y) is the optimized coordinate value of this dot, Wxy (x, y) is the
distance between dot $A$ in CAD model and measuring model,
respectively.
The coordinate value of dot B, which is the last point of mean camber line
can be solved in the same way by the distance compensation method. (Eq.
(2).
M), the x-coordinate value of dot A, and the angle tan , the relations
between chord length L and cot α1 can be established as:
cot 1/ 2TA[ ( L M ) MT ( ( L M ) 2 ( L M ) MT
(3)
M 2T 2 4T 2 A2 4TA 4T 2 ( L M ) A) (1/ 2) )]
n
1
P lim i D P0 D P0 (4)
n
i 1 1
is the shape coefficient, for the form error is related to the structure of
the casting, it is noticed that each form error shows some similarities.
Hence, a shape coefficient noted is introduced.
Fig. (4) shows that the change of chord length is influenced by the height
of segment variation, the shape coefficient can be expressed as ∆Hi+1 /
∆Hi. Where H represents the difference between the height of segments
of measuring model and CAD model. ∆Hi represents the difference of
intervallic deformation between two iterative calculation steps.
Optimization of Geometric Parameters Frontiers in Aerospace Science, Vol. 1 251
The extraction of a type of certain turbine blade is shown in Fig. (5). After
the extraction of parameters, the corresponding data of CAD model and
measuring data are calculated. For the cross- section at height of 50mm,
the results are shown in Fig. (6), and the optimize cross-section is shown
in Fig. (7). The lines shown in Fig. (7) are the mean camber line, the
thicker line represents the optimized mean camber line, while the thinner
line represents the original one.
The displacement field of three directions is shown in Fig (8). Where Fig.
(8a) represents X direction, Fig. (8b) represents Y direction, and Fig. (8c)
represents Z direction.
If difference ∆Di is less than ∆Max (the threshold for maximum error),
Pi(xi, yi, zi) can be taken as the ideal deformed shape. Otherwise, the
deformed shape must be adjusted by -∆Di (as shown in Eq. 6) and
program iterates until the maximum error threshold is met. Fig. (11)
shows the program flow chart of such iteration. The final reverse
deformation function can be established as Eq. 7.
Pi Di 1 Pi 1 (6)
P( x, y, z ) Pi ( xi , yi , zi ) (7)
Fig. (11). Flow chart of the solution algorithm of the reverse deformation.
256 Frontiers in Aerospace Science, Vol. 1 Dou et al.
Volume
*_shell.mesh ViewCAST *_shellgd.dat
mesh
FEM PROCESSING
The FEM process using ProCAST can be divided into two aspects: A.
collect necessary node information by reading corresponding mesh files
(refer to Fig. 13); and B. generate plenty grid nodes for accuracy during
the simulation process.
During FEM process, sorting nodes has to be optimized for not only
reducing the CPU time but also shortening the node matching process.
Rearrangement of nodes first occurs after ProCAST automatically assigns
a contact node at the interface of casting and shell (those nodes are
generated in MeshCAST module) and put this new node at the end of the
file.
The sorted node information is then exported into *_shelld.out file, which
also contains an index table recording the corresponding relationship
among grid nodes.
Read node
Input mesh file
identification
N Y
characteristic word
code judgement End
Then the mesh files should be simplified and the necessary node
coordinates information is extracted. In current work, the demarcation
searching method is used to search the corresponding nodes. Demarcation
searching method is commonly used to compare the data sizes by
boundary values [18]. First, the data are sorted by initial boundary values
and the boundary value will be adjusted in accordance with demarcation
Optimization of Geometric Parameters Frontiers in Aerospace Science, Vol. 1 259
results. Repeat the comparison process until all data are divided into
orderly small units, as shown in Fig. (15).
Read *sm file Read *shelld.dat file Read *shelld.out file Read *shellgd.dat file
substract
Remove
Judgement
shell node
Index tableA--C
Corresponding node
elements which are less than the pivot come before the pivot and so that
all elements greater than the pivot come after it (equal values can go
either way). After this partitioning, the pivot is in its final position. This is
called the partition operation. Finally recursively sort the sub-list of lesser
elements and the sub-list of greater elements. The base cases of the
recursion are lists of size zero or one, which are always sorted.
EXAMPLE VERIFICAITON
DD6 alloy is chosen as the turbine blade material, and the mold material
is silica sand (thermo-physical properties of DD6 alloy and the mould
material can be found in Ref [22-23]). Fig. (16) shows a typical blade
mould design with a gating system. To simplify the geometry of the
turbine blade modeling and to improve the computational efficiency, only
one third blade module group is chosen (shown in Fig. 17). Further, we
adopt gravity filling method and ignore the influence of filling period. We
assume the initial casting temperature is 1550℃, the temperature of the
shell mould (pre-heat) is 1500℃, and the calculation is terminated at a
temperature of 600℃.
Based on UG, C++ language editor was used for developing "Investment
Casting Blades’ Mould Optimization System". One example of the
system interfaces is shown as Fig. (19).
This system module reads files generated from each simulation step,
orders and matches the grid nodes while updating the displacement field.
The output of the system is the new surface mesh file. Fig. (20) shows a
new surface mesh of die profile of a turbine blade.
| Q D |
i i
M i 1
(8)
n
The shape form error M reflects the coincidence degree of two surface
meshes. The smaller the form error is the more optimized mesh surface
from the original CAD one. Let X,Y,Z denote the average form error of
three directions, respectively, one can see from Table 1 that after 4th
iterations, the synthetic form error decreases from 0.516mm to 0.002mm.
Such convergence effect can meet the error requirement of the die cavity
design for turbine blades.
CONCLUSION
CONFLICT OF INTEREST
The authors confirm that they have no conflict of interest to declare for
this publication.
ACKNOWLEDGEMENTS
I hereby to acknowledge Dr. Kun Bu and her research team for their
contributions to my inspiration and knowledge and other help in creating
this chapter. This work was supported by National Key Technologies
R&D Program and was executed in the Key Laboratory of Contemporary
Design and Integrated Manufacturing Technology, Northwestern
Polytechnical University, China.
REFERENCES
digitized point data with discrete curvature", Comput. Aided Des., vol. 34,
no. 4, pp. 309-320, 2002.
[http://dx.doi.org/10.1016/S0010-4485(01)00091-4]
[16] S.C. Modukuru, N. Ramakrishnan, and A.M. Sriramamurthy, "Determination
of the die profile for the investment casting of aerofoil-shaped turbine blades
using the finite-element method", J. Mater. Process. Technol., vol. 58, no. 2-3,
pp. 223-226, 1996.
[http://dx.doi.org/10.1016/0924-0136(95)02127-2]
[17] M. Fackeldey, A. Ludwig, and P.R. Sahm, "Coupled modelling of the
solidification process predicting temperatures, stresses and microstructures",
Comput. Mater. Sci., vol. 7, no. 1-2, pp. 194-199, 1996.
[http://dx.doi.org/10.1016/S0927-0256(96)00080-8]
[18] Z. Dan, W. Zhang, and W. Min, "Reversing design methodology of the die
profile in investment casting based on the simulation of displacement field and
identification of featured parameters", Hangkong Xuebao, vol. 03, pp. 509-
514, 2006. [In Chinese].
[19] D.A. Reed, J. Wind Eng. Ind. Aerodyn., vol. 33, no. 3, pp. 487-494, 1990.
[http://dx.doi.org/10.1016/0167-6105(90)90003-U]
[20] Z. Bing, and L. Zhiqiang, "Three dimensional FEM simulation of Titanium
hollow blade forming process", Rare Metal Mat. Eng, vol. 39, no. 6, pp.
963-968, 2010.
[http://dx.doi.org/10.1016/S1875-5372(10)60106-3]
[21] A. Krimpenis, and P. Benardos, "Simulation-based selection of optimum
pressure die-casting process parameters using neural nets and genetic
algorithms", Int. J. Adv. Manuf. Technol., vol. 27, pp. 509- 517, 2006.
[http://dx.doi.org/10.1007/s00170-004-2218-0]
[22] X.H. Li, W. Mao, and W.L. Guo, "Transient liquid phase diffusion bonding of
a single crystal superalloy DD6", China Welding, vol. 458, pp. 281-289,
2005.
[23] W. Bonilla, S.H. Masood, and P. Iovenitti, "An investigation of wax patterns
for accuracy improvement in investment cast parts", Int. J. Adv. Manuf.
Technol., vol. 18, no. 5, pp. 348-356, 2001.
[http://dx.doi.org/10.1007/s001700170058]
Frontiers in Aerospace Science, 2016, Vol. 1, 269-301 269
CHAPTER 6
1
Department of Mechanical Engineering, SST Division-ASELSAN Inc.,
06172, TURKEY
2
Department of Mechanical Engineering, MGEO Division-ASELSAN Inc.,
06011, TURKEY
fatigue tests were compared to multi axial fatigue tests which were performed by
FEA simulations. Furthermore, the effects of various loading conditions and
geometries were investigated. The study showed that the assumption of multi axial
testing can be represented by uni-axial testing does not hold for various cases.
INTRODUCTION
This chapter will consist of two case studies which will focus on different
analysis and testing aspects of aerospace structures. The first section will
investigate an antenna integration process including flight tests and stress
analysis where the structure will not require fatigue analysis [4]. The
antenna is part of the MXF-484 V/UHF transmitter, which is used to
safely communicate without any enemy interference. Fig. (2) shows the
integration area on the helicopter.
Vibration Induced Fatigue Analysis Frontiers in Aerospace Science, Vol. 1 271
(a) (b)
(c) (d)
Fig. (1). (a) Comet fuselage failure (b) Stress concentrations at window corners [1] (c) Aloha
Airlines Boeing 737 fuselage failure due to multiple cracks at rivet holes (d) Crack growth on
fuselage [2].
(a) (b)
Fig. (2). (a) Antenna location on helicopter, (b) Measurement locations when antenna is not
installed.
272 Frontiers in Aerospace Science, Vol. 1 Aykan and Çelik
The edge numbered by "1" as shown in Fig. (4a) is a locked edge and the
edge numbered with "4" is far away from the location analyzed. Therefore,
the degrees of freedoms of these edges are fixed. Quarter turn fasteners
are used close to the edges "2" and "3" and these edges are not fixed but
simply supported. The materials used in the analysis consist of aluminum
2024-T3 and steel 4140.
Modal analysis of the cowling model is performed for with/without
antenna configurations. The mode shape for the first natural frequency of
the cowling changed from bending to rotation and the first natural
frequency of the antenna integration region has decreased from 154.3 Hz
to 67.9 Hz as shown in Fig. (5).
(a) (b)
Fig. (4). Cowling boundary conditions, (a) 1 and 4 fixed, 2 and 3 simply supported, (b) antenna
model (point mass element).
(a) (b)
Fig. (5). Mode shape of the antenna integration region for the first natural frequency, (a) without
antenna, (b) with antenna.
Vibration Induced Fatigue Analysis Frontiers in Aerospace Science, Vol. 1 275
Similarly, the stress distributions of the cowling with and without the
antenna are also needed to assess the safety of integration. A smaller
region of the cowling Fig. (6) is included in the vibration analysis as the
modal analysis results showed that the first 3 modes of the antenna region
are local modes. The analysis model has 19959 elements and 60935
nodes.
The random time history signals are usually defined in frequency domain
as Power Spectral Density (PSD) functions [7]. PSD presents this
information as a statistical spectrum where the area under the curve
represents the mean square amplitude of the wave other than the
amplitude. Since PSD values are needed for the vibration analysis, the
tri-axial vibration measurements are transformed to PSD and applied as
loading boundaries at the measurement locations in the FEA.
(a)
(b)
Fig. (7). Stress distributions of vibration (PSD) analysis, Max stress location at the antenna
integration area, (a) without antenna, (b) with antenna.
Numerical results showed that there is little risk for the installation of the
antenna. In order to validate the integration, flight tests with the antenna
are performed according to the same flight profile. A total of seven
tri-axial accelerometers are placed on the cowling portion on which the
antenna is mounted. Accelerometer locations for this configuration are
shown in Fig. (9).
The acceleration data collected during flight tests are compared with the
previous test where the antenna was not installed. According to the PSD
comparison, antenna integration has increased the overall vibration levels
at some of the measurement locations. A sample comparison is given for
location 5 in Fig. (10).
278 Frontiers in Aerospace Science, Vol. 1 Aykan and Çelik
Fig. (10). PSD comparison of location-5 on cowling, Y (lateral) and Z (vertical)-axes, with antenna
and without antenna.
Vibration Induced Fatigue Analysis Frontiers in Aerospace Science, Vol. 1 279
PSD graphs are also compared with the levels given in the military
standard MIL-STD-810F [8]. Fig. (11) shows the comparison of
measured lateral vibrations at location-5 with a sine-on-random graph
which is specified in MIL-STD-810F. Fig. (11) shows that the peaks
occur at the same frequencies as in the standards’ profile. In addition,
floor levels do not exceed the random portion of the standards’ profile.
This is valid for all the locations where vibration data is collected. This
comparison gives confidence that the integration process does not
significantly change the overall vibration characteristics of the helicopter
even at the locations where the vibration levels have increased.
Fig. (11). Comparison of experimental vibration level of location-5 at lateral axis with military
standard specification (MIL-STD-810F).
to complete the design with required safety factors. In this study, fatigue
analysis and testing of an aerospace structure exposed to broadband
loading is highlighted. Traditionally, fatigue analyses have been
performed in the time domain where the response is calculated by static
analysis. However, studies [9-12] show that when the loading excites the
natural frequencies of the structure, time domain approach lacked the
dynamics of the structure. Therefore, the analyses in this study are
performed in the frequency domain using random vibration theory. Fig.
(12) shows the basic flow for both time and frequency domain methods.
TIME DOMAIN
TIME RAINFLOW
Steady HISTORY COUNT PDF
state
or FATIGUE
LIFE
Transient
Analysis
FREQUENCY DOMAIN
PSD FATIGUE PDF
MODELLER
Transfer M0
M1
FATIGUE
Function M2
BLACK
BOX
LIFE
M
4
Given a measurement for a duration T of the random time series y(t), its
PSD, Gyy(f,T) can be defined [18] as Eq. 1 where all phase information is
lost in the process.
G yy ( f , T )
2
T
E Y ( f ,T )
2
(1)
282 Frontiers in Aerospace Science, Vol. 1 Aykan and Çelik
G yy ( f ) lim
T
2
T
E Y ( f ,T )
2
(2)
where f is the frequency, E the averaging operator and Y(f,T) is the Fourier
Transform of y(t).
Y ( f ) H ( f ). X ( f ) (3)
If the PSD of response is obtained by using Eq. 2 and Eq. 3 where the
input is written as a PSD function,
m m
G yy ( f ) H i ( f ).H j * ( f ).Gij ( f ) (5)
i j
Thus, using Eq. 5, the stress response PSD values can be calculated for a
specific location of the structure where the loading induced is partially
correlated.
Fig. (14). Schematic description of the elements of the loads’ counting process.
N S E[ P] T p (S d S S S d S )
(6)
where;
E[P]: the number of stress ranges of all levels per time unit, p(∆S-d∆S<∆S
284 Frontiers in Aerospace Science, Vol. 1 Aykan and Çelik
<∆S +d∆S ): the probability that the stress range ∆S will appear in interval
(∆S-d(∆S) : ∆S+ d(∆S)).
The key element of vibration fatigue analysis is the PDF and PDF can be
obtained from the PSD of stresses [20]. The characteristics of PSD that
are used to obtain this information are the spectral moments of the PSD
function. The nth spectral moment of the PSD is defined as shown in Fig.
(15) and expressed as in Eq. 7.
M
mn f n G yy ( f ) df f k G yy k ( f k ) f
n
(7)
0 k 1
which p(∆S)d∆S will have peak values in the stress range ∆S to ∆S +d∆S.
Thus, the damage at this stress level is calculated as
1
E P T p (S )d S .
N (S )
From this equation one can find the life of a component by setting the
E[D] equal to unity and obtain the life, T in seconds. For the integrations
given above a cut-off value for the upper limit is necessary. This value is
typically given in terms of Root Mean Square (RMS) values of stress. It is
common to set the cut-off value to 3 RMS in amplitude or 6 RMS in
range, but practice has shown that it should be at least set to 4.5 RMS in
amplitude in order not to miss fatigue damage.
There are many empirical solutions for the probability density function of
Rainflow stress ranges, p(∆S), like Tunna, Wirsching, Hancock,
Chaudhury and Dover [7] but the best correlation was obtained by Dirlik
[21] after computer simulations to model random signals using the Monte
Carlo method.
Zi Z 2 Zi 2
D1 D2 Z i 2 Ri2
e Q
2 e D3 Z i e 2
Q R
p (Si ) (9)
2 m0
where,
286 Frontiers in Aerospace Science, Vol. 1 Aykan and Çelik
Si E 0
, , E0 , EP
m2 m4 m1 m2
Zi xm
E P
,
2 m0 m0 m2 m0 m4
2( x m 2 ) 1 D1 D1
2
D1 , D2 , D3 1 D1 D2
1 2
1 R
1.25 ( D3 D2 R ) xm D1 2
Q ,R
1 D1 D1
2
D1
The damage equation (Eq. (8)) with the probability density function (Eq.
(9)) is used primarily in Aerospace, Military and in some industrial areas.
The ability to include broad band frequency loading environments to
fatigue damage calculations makes this method more powerful than time
domain methods when the dynamic characteristics of the structure cannot
be ignored.
In this study, due to the combined broad band and narrow band frequency
loading encountered which is typical for a helicopter; vibration fatigue
method is chosen as the analysis method. This section will include the
analysis of three different designs in order to investigate different
parameters effecting fatigue life and damage location.
The analysis method requires transfer functions of the structure and the
vibration loading. It is possible to measure or calculate the transfer
functions from prototypes or from numerical analysis of virtual
prototypes. In this study, they are obtained from a FEA software
(ANSYS). Verification of the numerical model was made by modal
testing and analysis. The modal tests and analysis showed that the first 3
modes of the structure had a maximum error of 2% between the
numerical and test model. The details of the modal testing and analysis
Vibration Induced Fatigue Analysis Frontiers in Aerospace Science, Vol. 1 287
The installation location of the bracket was chosen to have the maximum
dispersion of Chaff/Flare particles (Fig. 16). In order to find the load time
histories for this location, military standards (MIL-STD-810E/F,
GAM-EG-13 etc.) were investigated for the AH-1W helicopter but
unfortunately very coarse information was obtained. Therefore, the
loading time histories were obtained from flight tests performed on the
helicopter using two tri-axial accelerometers where the system will be
installed.
Fig. (16). Location of the Chaff/Flare dispenser bracket on the AH-1W Helicopter [23].
Frequency Response Case PSD matrix (i,1) PSD matrix (i,2) PSD matrix (i,3)
1 X axis PSD X-Y axis cross PSD X-Z axis cross PSD
2 X-Y axis cross PSD Y axis PSD Y-Z axis cross PSD
3 X-Z axis cross PSD Y-Z axis cross PSD Z axis PSD
(a) (b)
Fig. (17). PSD plots from flight tests, (a) X-axis, longitudinal axis of the helicopter, (b) Y-axis,
transverse axis of the helicopter.
(a) (b)
Fig. (18). PSD plots from flight tests, (a) Z-axis, vertical axis of the helicopter, (b) X-Y cross axis.
(a) (b)
Fig. (19). PSD plots from flight tests, (a) X-Z cross axis, (b) Y-Z cross axis.
During the fatigue analysis, the mean stress is not corrected due to the fact
that mean stresses cannot be directly defined in vibration fatigue analysis
method. In this method, the response always oscillates around zero stress
level. However, it is always possible to calculate the mean stress level for
the structure and then shift the S-N curve by applying any mean stress
correction theories (Goodman, Soderberg). Furthermore, a better and
faster solution is to calculate the life under oscillating stresses only and
then the life of the critical points can be recalculated with the added mean
290 Frontiers in Aerospace Science, Vol. 1 Aykan and Çelik
Analysis of Design-1
The fatigue life results of the first designed Chaff/Flare dispenser bracket
are shown in Fig. (20). This design is produced from aluminum blocks
and assembled using bolts. In order to validate the calculations, fatigue
tests had to be performed by testing the bracket three axes simultaneously.
However, the testing hardware was not suitable for the multi axial testing.
Thus, tests were performed on each axis separately as shown in Fig. (21).
The laboratory vibration tests were performed in 4 hours where the
operational loading was accelerated from 12000 flight hours for the
fatigue tests. The fatigue tests ended with no visual damage.
Fig. (20). Fatigue analysis results of the Chaff/Flare dispenser bracket showing the most critical
location under multi axis loading (Fatigue Life 8.59e-3 sec).
Vibration Induced Fatigue Analysis Frontiers in Aerospace Science, Vol. 1 291
However, this is not the same as applying all of the vibration profiles at
the same time as it is occurring in real life. Fatigue analyses were
recalculated for each axis separately which is encountered in accelerated
tests for test and analysis comparison. From the analysis results, it was
observed that x-axis (longitudinal axis of the helicopter) loading has
nearly no effect on the fatigue damage accumulation. Y-axis (transverse
axis of the helicopter) has the biggest effect and z-axis (vertical axis of the
helicopter) has a lower effect compared to the y-axis.
As for each axis a different critical location can exist, 445 locations (with
gradual damage distribution) are obtained from the first fatigue analysis,
in which all the three axes loading is applied at the same time, and then
the damage analyses are performed for these locations. In this analysis,
damage values from each axis are summed for these locations which gave
292 Frontiers in Aerospace Science, Vol. 1 Aykan and Çelik
the cumulative damage for axis by axis loading. The analysis results were
compared with the simulation results of multi axial loading. Furthermore,
the percent deviation values of uni-axial loading from multi axis loading
are calculated and a moving average algorithm is applied on the data. The
mean percent deviation of damage was obtained as ranging from 35% to
72%.
From this analysis it is evident that test time has to be extended in order to
fully obtain the multi axis loading effect when performing an axis by axis
test procedure. However, the increase in test time cannot be the same for
all locations due to having different percent deviations of damage. In
order to be safe, the testing duration for sequential testing can be
increased by 72%.
Analysis of Design-2
The new model where the stiffeners are removed is re-analyzed for 6061
HV T6 material at operational and MIL-STD-810F loading environment.
However, it must be noted that this profile is accelerated for 2600 flight
hours whereas the operational loading is used in the previous analyses.
The results of this analysis showed that the life distribution changed and
most damaged locations are at the bottom of the dispenser as shown in
Fig. (22). The mean percent deviations for these new locations are given
in Fig. (23).
In this case study, it is found out that when operational loading is applied
most of the damage is due to the z axis loading. However, for MIL STD
810F loading case, all of the axes have more or less equal amount of
contribution to damage. This phenomenon is dominant for locations
between "10" and "35" which causes a decrease in the mean percent
deviation of MIL STD 810F loading. The faster accumulation of damage
Vibration Induced Fatigue Analysis Frontiers in Aerospace Science, Vol. 1 293
in those locations for MIL STD 810F loading decreases the mean percent
deviation. However, for MIL STD 810F loading case different critical
locations are also formed. This clearly shows that the conservativeness of
military standards can change the critical location for some cases.
(a) (b)
Fig. (22). The new Chaff/Flare dispenser fatigue life distribution without the four stiffener
components (Fatigue Life 3.62e5 sec), (a) Isometric top view, (b) Isometric bottom view.
Fig. (23). Mean percent deviation values of sequential loading from multi axis loading for two case
studies.
294 Frontiers in Aerospace Science, Vol. 1 Aykan and Çelik
Analysis of Design-3
In the case study given above, it was shown that slight geometric change
affected the damage distributions considerably. In this case study, a
completely different design is analyzed for fatigue. In order to reduce the
weight of the dispenser bracket, the bracket was produced from sheet
metal where the connecting elements are mostly rivets as shown in Fig.
(24).
Fig. (25). The sheet metal Chaff/Flare dispenser fatigue life distribution (Fatigue Life 1.48e6 sec),
Isometric view [23].
Vibration Induced Fatigue Analysis Frontiers in Aerospace Science, Vol. 1 295
Fig. (26). Mean percent deviation values of sequential loading from multi axis loading for sheet
metal model [23].
The last analysis emphasizes the geometry dependence of the mean percent
deviation of damage for sequential loading from multi axis loading.
In order to compare the fatigue analysis results with tests for the sheet
metal model, fatigue tests were performed on each axis separately as
shown in Fig. (27). The operational loading is again accelerated to 4 hours
of laboratory test time which corresponds to 12000 flight hours for the
fatigue tests.
The failure criterion for these tests is defined as the formation of a visible
crack. The x and y axes concluded with no visual damage. However, at
the third hour in z axis, the dispenser loosened. As the x and y axis were
already completed, the corresponding flight hour where the dispenser
loosened cannot be accurately determined. A coarse approximation can be
that these deformations occurred at the 9000th flight hour for three axes
loading.
296 Frontiers in Aerospace Science, Vol. 1 Aykan and Çelik
-Y +Y
-X
+X
(a) (b)
+Z
-Z
(c)
Fig. (27). Sheet metal bracket fatigue test, (a) X-axis, (b) Y-axis, (c) Z-axis. [23].
At the end of the tests, the prototype is disassembled and each part is
examined. The following plastic deformations were observed Fig.
(28-29).
(a) (b)
Fig. (28). (a) Plastic deformation on the bolt hole of the magazine housing, (b) Zoomed view. [23].
Vibration Induced Fatigue Analysis Frontiers in Aerospace Science, Vol. 1 297
(a) (b)
Fig. (29). (a) Plastic deformation on the rivet/rivet hole of the side carrier legs, (b) Zoomed view
[23].
CONCLUSION
obtained from static analysis cannot be used. One more point to highlight
is that, although the loading is multi-axial, vibration fatigue method was
used as the dynamics of the structure dominate the response, instead of
multi axial fatigue methods. If stress data for the multi-axial loading can
be obtained, which for large and complex structures in FEM is impractical
(requires transient dynamic analysis) then multi-axial fatigue theories can
also be used. But this method, without using FEM, requires experimental
methods.
It should be noted that the accurate stress results obtained from the finite
elements analysis play a critical role for fatigue life of the structure.
Approximately 7% increase in stress value causes 50% reduction of life.
Therefore, it may not be enough that the finite element model simulates
the dynamic characteristics correctly. Stress concentrations occurring in
the finite element is the main reason for inaccurate stress results, which
can only be validated by tests. In the analyses performed, the damage
distribution of the structure was obtained which is a more valuable
information than the life. As a result, the most damaged areas can be
modified to be stronger. Furthermore, it should be noted that the critical
location found from a static analysis is different from the critical location
obtained from the dynamic analysis. This is due to the fact that when the
natural frequencies of the structure are excited, the response of the
structure is controlled by the mode shapes.
The fatigue tests performed also contain some assumptions. First of all,
the tests are sequential and cross correlations are not included in the
fatigue tests. Secondly, the moments are not applied. The parametric case
studies showed that the fatigue damage accumulating for a structures
varies with changes in loading, material and geometry and a formulated
factor that can take these parameter changes into account is not available.
This sensitive structure when combined with nonlinear response in the
frequency domain yields multi axis testing to be an industrial and military
standard. Also from (Fig. 23), it is evident that accelerating the vibration
loading can cause a further deviation from multi axis testing. The most
realistic case would be to use a six-DOF vibration testing machine where
Vibration Induced Fatigue Analysis Frontiers in Aerospace Science, Vol. 1 299
CONFLICT OF INTEREST
The authors confirm that they have no conflict of interest to declare for
this publication.
ACKNOWLEDGEMENTS
Declared none.
REFERENCES
[13] H. Zenner, A. Simbürger, and J. Liu, “On the fatigue limit of ductile
metals under complex multiaxial loading", Int. J. Fatigue, vol. 22, pp.
137-145, 2000.
[http://dx.doi.org/10.1016/S0142-1123(99)00107-3]
[14] Z. Xia, and F. Ellyin, "Multiaxial fatigue of an Alumina particle reinforced
aluminum alloy", Int. J. Fatigue, vol. 20, pp. 51-56, 1998.
[http://dx.doi.org/10.1016/S0142-1123(97)00114-X]
[15] A. Ahmadi, and H. Zenner, "Lifetime simulation under multiaxial random
loading with regard to the microcrack growth", Int. J. Fatigue, vol. 28, pp.
954-962, 2006. [http://dx.doi.org/10.1016/j.ijfatigue.2005.09.015]
[16] C.C. Chu, "Multiaxial fatigue life prediction method in the ground vehicle
industry", Int. J. Fatigue, vol. 19, pp. 325-330, 1997.
[http://dx.doi.org/10.1016/S0142-1123(97)00018-2]
[17] R.M. French, R. Handy, and H.L. Cooper, "A comparison of simultaneous and
sequential single- axis durability testing", Experimental Techniques,
September/October, pp. 32-37, 2006.
[http://dx.doi.org/10.1111/j.1747-1567.2006.00083.x]
[18] J.S. Bendat, and A.G. Piersol, Random Data: Analysis and Measurements
Procedures. John Wiley & Sons, 1971.
[19] Y.L. Lee, J. Pan, R. Hathaway, and M. Barkey, Fatigue Testing and Analysis:
Theory and Practice. Elsevier Butterworth-Heinemann, 2005.
[20] C. Lalanne, Mechanical Vibration & Shock, Fatigue Damage, Vol IV. Taylor
& Francis Books: London, 2002.
[21] T. Dirlik, “Application of computers in Fatigue Analysis”, PhD Thesis,
University of Warwick, 1985.
[22] MSC Fatigue Version 2003 User’s Manual. MSC Software Inc.: USA, 2003.
[23] M. Aykan, “Vibration Fatigue Analysis of Equipments Used in Aerospace”,
MSc. Thesis, Middle East Technical University, 2005.
302 Frontiers in Aerospace Science, 2016, Vol. 1, 302-325
CHAPTER 7
Abstract: Surface treatments, such as laser peening, can increase the life of the
component by generating compressive residual stresses on the surface. Laser peening
of an already peened component, termed as re-peening, can further increase the
fatigue life of the component. Re-peening has several applications in the aerospace
industry. The huge population of ageing aircraft components is one such application,
which can benefit significantly from the re-peening process. However, this process
is not optimized for maximum fatigue life due to the presence of many design
variables and the complex nature of the problem which requires a large number of
experimental testing to reach conclusions. Therefore, a computationally efficient
optimization strategy needs to be developed to conduct large-scale laser peening
simulations for problems related to fatigue life, such as aircraft lug failure, a problem
that requires consideration of component curvature and residual stress relaxation
effects. Deciding the time to peen an already peened component (re-peening time) is
another variable which makes the problem further complicated. The ultimate goal of
this research is to construct the framework to predict the optimum parameters for
maximum fatigue life on structural components. A two-step optimization strategy
is adopted for the fatigue life optimization of an aircraft lug component. The strategy
*Corresponding author Anoop Vasu: Department of Mechanical and Materials Engineering, Wright State
NOMENCLATURE
R Relaxation coefficient
N Number of load cycles
σa Alternating stress
σm initial mean stress
σ1 Mean stress after one cycle
σy Material yield strength
σu Ultimate strength
σr Residual stress
σNf Fatigue strength
Nf Number of cycles to failure
D Cumulative damage
P Peak pressure of laser pulse
O Percentage overlaps of sequential laser spots
t Mid-span duration of laser pulse
σCRS Compressive residual stress
σTRS Tensile residual stress
INTRODUCTION
lifetimes of testing. This region was repaired during the test and after 2.5
lifetimes of testing another crack emerged around the lug bore region
(Location 2). Replacing the lugs with new ones requires removing the
part from the entire assembly. This process is very cost prohibitive. Laser
peening serves the purpose of imparting favorable residual stresses in the
material surface thereby improving the fatigue life of such components.
Other aerospace applications include usage on engine parts, turbine blades,
fasteners, valves, transmission components, etc.
of many design variables and the complex nature of the problem which
requires a large number of experimental testing to reach conclusions.
Therefore, a computationally efficient optimization strategy needs to be
developed to conduct large-scale laser peening simulations for problems
related to fatigue life, such as aircraft lug failures, which is a problem that
requires consideration of curvature and relaxation effects.
Applied Load
Fig. (5). Loading for fatigue analysis (a) load spectrum (b) stress spectrum.
Many empirical and analytical models, which can be used to evaluate the
effect of relaxation during cyclic loading, have been formulated by
researchers over the years. Some of the most prominent empirical models
are given in this section. Morrow and Sinclair modeled residual stress
relaxation based on mean stress relaxation in axial fatigue tests [11].
Kodama proposed a linear logarithmic relationship based on surface
stresses on a shot peened material [12]. Their empirical models are given
by Eqs. 1, and 2 respectively.
5.7
R 1 a log N (1)
m
y
312 Frontiers in Aerospace Science, Vol. 1 Vasu and Grandhi
R C1 C2 log N (2)
C1, and C2 are constants which are determined by curve fitting the
experimental data performed under various loading conditions. More
sophisticated physics-based models have also been developed [13]. Any
of these models can be utilized in predicting the residual stress relaxation
of laser peened components based on the requirements of the application.
Fatigue Life
availability of material parameters and the nature of the problem [14]. Fig.
(7) shows the step-by-step procedure for life prediction of components
undergoing the laser peening process. The load on the aircraft component,
represented by the FALSTAFF spectrum, is used to create stress spectrum
on the component by conducting a unit load FEA. The alternating stress
(σa) and the mean stress (σm) due to the fatigue loading are obtained from
the stress spectrum. The residual stress (σr) from the laser peening
simulation is added to the mean stress. Using Modified Goodman
Equation, fatigue strength (σNf) is calculated. In the next step, number of
cycles to failure (Nf) is calculated for individual stress segments using
Basquin’s equation. The Cumulative Damage caused by the complete
stress spectrum is obtained by palmgen-miner linear damage rule. In the
last step, the fatigue life is calculated by inverting the cumulative damage.
Re-Peening
Fig (9). Critical location on the lug (a) experimental results [4] (b) FEA prediction.
OPTIMIZATION PROCEDURE
Application of laser peening on an aircraft lug component is used as an
example to illustrate the fatigue life optimization process. The critical
Fatigue Life Optimization Frontiers in Aerospace Science, Vol. 1 315
Design Variables
The design variables considered are re-peening time (R), peak pressure of
the laser pulse (P), mid-span duration of the laser pulse (t), and percentage
overlapping (O) of the laser spots. A brief description of the design
variables and the choice of design bounds are explained in the following
paragraphs:
Re-Peening Schedule
Fig. (10). Re-peening effect caused by maximum and weighted relaxation models.
Mid-Span Duration
Peak
Peak Pressure
Percentage Overlap
Optimization Strategy
The choice of cost function is based on the major output of the simulation,
which is to obtain the maximum fatigue life. A two-step optimization
strategy is followed in this chapter as shown in Fig (13).
(Table 1) contd.....
The optimized results are given by: P = 3.97 GPa, O = 80%, t = 30 ns,
σCRS = -542 MPa, and σTRS = 100 MPa. An extra simulation is run at the
optimized design point which yields σCRS = -511 MPa, σTRS = 103 MPa,
which is very similar to the predicted result. Therefore, this result is used
for the second phase of the optimization problem.
Fig. (14). Life optimization with re-peening (a) maximum relaxation model (b) weighted relaxation
model.
CONCLUSION
The optimum parameters to laser peen a structural component in order to
maximize the component’s fatigue life have to be determined. The fatigue
Fatigue Life Optimization Frontiers in Aerospace Science, Vol. 1 323
loading can cause the residual stress to relax; hence, the relaxation step
should be considered to give a good prediction of the fatigue life. By
conducting re-peening, we will be able to increase the fatigue life of the
component. A computational framework is created incorporating the key
constituents impacting the life such as load spectrum, residual stress
relaxation, and re-peening for the laser peening process to predict an
optimal re-peening schedule to improve the fatigue life. A two-phase
optimization strategy is employed to obtain prolonged fatigue life of an
aircraft lug. In the first phase, a three-variable RSM is constructed with
the SKA for both objective function and the constraint to conduct CRS
optimization. The predicted CRS matches the simulation results closely.
In the second phase, the optimum CRS is input into re-peening
optimization routine, which predicts the best time to re-peen the
component for maximum fatigue life. It is observed that re-peening the
component around 55% of its initial fatigue life yields the maximum
fatigue life for the structural component.
CONFLICT OF INTEREST
The authors confirm that they have no conflict of interest to declare for
this publication.
ACKNOWLEDGEMENTS
Declared none.
REFERENCES
CHAPTER 8
*Corresponding author Anoop Vasu: Department of Mechanical and Materials Engineering, Wright State
University, Dayton, OH 45435, USA; Tel/Fax: 937-775-5040; E-mail: anoop1984@gmail.com.
NOMENCLATURE
HEL Hugoniot elastic limit
dyn
y
Dynamic yield stress
v Poissons ratio
σvm Equivalent von mises stress
A Yield stress
B Strain hardening coefficient
n Strain hardening exponent
C Strain rate sensitivity constant
εp Equivalent plastic strain
*
Dimensionless strain rate
Strain rate from high strain rate experiments
0 Reference strain rate
dp Heat penetration depth
D Heat diffusion coefficient
Laser pulse width
C1 Kinematic hardening modulus
σij stress tensor
αij Backstress tensor
σ0 Equivalent stress defining size of yield surface
p Equivalent plastic strain rate
γ Material constant from cyclic test data
ij
p
Rate of plastic flow
R Stress ratio
328 Frontiers in Aerospace Science, Vol. 1 Vasu and Grandhi
INTRODUCTION
How to prevent failure in materials has been a great research interest all
over the world for centuries. It has been found out that cyclic loading is a
major cause of failure in many situations. There can be other contributing
factors, such as manufacturing defects and the presence of unfavorable
residual stresses. But residual stresses can also play a positive role and
can enhance the life of the component, if applied properly. There exist
methods named Surface enhancement techniques (SET), which can
induce favorable residual stresses (mechanical SET) on surface regions of
peened components to improve the fatigue life of the component [1]. Fig.
(1) depicts a pictorial representation of the impact of SET on fatigue life.
peened spring steel specimens [13]. Iida et al. found out that repeated
cyclic loading caused more relaxation when compared to reverse cyclic
loading [14]. Boyce et al. investigated relaxation effect due to foreign
object damage [15]. They found out that damage sites with high stress
concentrations which were initially under a state of strong compression
relaxed, and cracks tend to form in these regions. Vasu et al. developed a
computational methodology to incorporate the relaxation effect into
fatigue life prediction models [16]. However, very little information is
available to comprehend the relaxation behavior in laser peened
components subjected to low cycle fatigue loading conditions.
propagated into the material as a shock wave. When the stress created by
shockwave exceeds dynamic yield strength of the material, this results in
plastic deformation of the surface microstructure generating compressive
stresses inside the material. Black paint or tape is considered most of the
time as the opaque overlay while flowing water along the peening surface
is usually taken as the transparent overlay. Some special features observed
in the process of laser peening are discussed in this section. The features
are unique of laser peening among all surface enhancement methods.
pressure pulse generated by the shock wave are monitored and the
temporal profile representing them is shown in Fig. (4) [17]. This pressure
pulse has unique properties such as high peak pressure and very short
pulse duration.
Fig. (4). Temporal profile of the laser pulse and the generated pressure pulse.
The shock waves generate high strain rates inside the material during the
laser peening process on the order of 106/sec. These high strain rates can
play an important role in determining the material behavior. It has been
experimentally proven that many materials are sensitive to high strain rate
effects. It requires a higher stress to yield the material in case of high
strain rates when compared with a quasi-static process. The stress-strain
curve showing this high strain rate effect can be represented as shown in
Fig. (5) [18].
334 Frontiers in Aerospace Science, Vol. 1 Vasu and Grandhi
ydyn 1
HEL (1)
1 2
When the shock wave propagates into the material, the pressure pulse
amplitude drops (shock wave attenuation) as it goes deeper into the material.
When the stress wave magnitude goes below the HEL, no further plasticity
occurs and the surrounding material representing the elastically deformed
field compresses the plastically affected region to achieve static
equilibrium.
Plasticity experienced in a high strain rate process like laser peening can be
explained by means of the dependence of plastic strain on impact pressure
Prediction of Residual Stress Relaxation Frontiers in Aerospace Science, Vol. 1 335
with the use of the HEL as the measure, as illustrated in Fig. (6) [19]. This
model assumes a Von Mises yield criterion along with a uniform pressure
along the impact region. The first step involves the elastic loading until the
HEL is reached. The second step involves plastic loading where plastic
deformation occurs in the material. The third step is the elastic unloading
phase and the final step is the plastic unloading phase. When the impact
pressure is between 1 and 2 HEL, plastic strain can occur with a purely
elastic reverse strain. When the impact pressure is more than 2 HEL, elastic
reverse strain gets saturated and plastic reverse strain occurs. If the impact
pressures are of the order of 2.5 HEL or more, this can result in the
spallation on the material. Hence impact pressure should be optimized to
obtain best results. Theoretically, a load of 2 HEL should give the
maximum compressive residual stress according to Fig. (1.15). If the load is
beyond 1 HEL, permanent plastic deformation occurs. When the load is
between 1 and 2 HEL, as the plastic strain increases, the compressive
residual stress also increases.
2D Axisymmetric finite element models can predict the residual stresses for
one shot laser peening process. However, 3D models are required to predict
the residual stresses for a realistic application scenario. The details of the
2D and 3D FEA models have been described by the authors before [20, 21].
336 Frontiers in Aerospace Science, Vol. 1 Vasu and Grandhi
The temporal profile of the shock wave is constructed similar to the sharp
rise time pulse profile obtained from experimental testing as shown in Fig.
(7). Although non-homogeneity is observed along the spatial profile of the
laser shot impact, the effect on the peak pressure is fairly small [22]. This
research considers the initial condition of laser power density, which is
uniformly distributed along the laser spot with the same area dimensions for
all three geometries. A uniform spatial profile over the spot is used for the
current investigations.
Simulation Procedure
Laser peening simulation process can be divided into two stages. The first
stage involves the generation of pressure pulses by high-energy plasma,
which can be explained by laser physics. The second stage incorporates
Prediction of Residual Stress Relaxation Frontiers in Aerospace Science, Vol. 1 337
Constitutive Model
The strain rate for laser peening is in the order of 106 /second. In order to
give an accurate prediction of the material response, temperature effects
as well as the effect of strain rate on the flow stress should be included in
the constitutive model. The Johnson-Cook model has been an efficient
material model to deal with high strain rate processes and is used for all
the finite element simulations in the present work [24]. According to the
Johnson-Cook strength model, the equivalent Von Mises stress is given
by:
vm A B pn 1 C ln * (2)
Constant A is the yield stress at 0.2% offset strain; Constant B and
exponent n represent the strain hardening effect. The expression in the
second bracket represents the strain rate effect through constant C. Model
constants A, B, n and C are determined from experimental data in the
form of room temperature stress-strain curves at specific strain rates. It
can be observe that, the thermal term has been neglected from the
Johnson cook material model. The absorbent layer can protect the
material surface from thermal effects of laser irradiation. It not only
prevents the part surface from the direct laser ablation, but also prevents
the thermal effect propagating into the specimen; hence the thermal term
is dropped. More discussion will be given in the next section.
The heat effect is not considered in the simulation because the absorbent
layer can protect the material surface from thermal effect during laser
peening. Sources of heat in laser peening to consider include laser
irradiation of the material and temperature rise due to the plastic strain
caused by shock waves. The heat penetration depth dp shown in Fig. (9)
due to the nanosecond laser irradiation is given by:
Prediction of Residual Stress Relaxation Frontiers in Aerospace Science, Vol. 1 339
d p D
1/ 2
(3)
Taking black paint as an example of the absorbent layer, the heat diffusion
coefficient is D = 2.1*10-7 m2/s. If it is irradiated by a 20 ns laser beam,
the heat penetration depth would be on the order of 0.065 μm. The layer
thickness used in the laser peening process is usually tens of micrometers.
Therefore, the ablative layer not only prevents the part surface from the
direct laser ablation, but also prevents the thermal effect propagating into
the specimen.
Material A B 𝜀̇0 E ρ
n C ν
(MPa) (MPa) (/s) (GPa) (Kg/m3)
To see how the temperature rise affects the residual stress in the material,
a peak pressure of 5.5GPa (~2HEL) is applied to the material. The
residual stress imparted to the material on the surface as well as the depth
direction are shown in Fig. (11). The results indicate that the effect of
temperature due to laser peening in its operating conditions is small and
hence temperature effect needs not be considered to model the laser
peening process.
Prediction of Residual Stress Relaxation Frontiers in Aerospace Science, Vol. 1 341
Fig. (11). A comparison of mechanical and thermo-mechanical FEA in laser peening (a) Surface
residual stress (b) Residual stress along depth.
Curvature Effect
Fig. (12). Curvature effect on Ti-6Al-4V (a) Stress on the surface of the material (b) Stress in the
depth direction at 1.5 mm away from the center of the spot.
Prediction of Residual Stress Relaxation Frontiers in Aerospace Science, Vol. 1 343
The residual state just after peening should not be directly taken into
fatigue analysis. Residual stresses generated by laser peening or any other
surface enhancement method on structural components will have to deal
with complex loads during their life cycle. The cyclic fatigue loading, as
well as elevated temperature, can take off the beneficial effect of the
surface enhancement methods significantly. This phenomenon of stress
instability is called stress relaxation. There are three different ways to
impart relaxation to peened components. The first one is termed as
thermal stress relaxation, which occurs when the surface treated
components are exposed to elevated temperature [26]. The second
mechanism is called the static relaxation, which arises when the sum of
the residual stress and the applied load exceeds the yield criterion of the
material [27]. The third way to yield stress relaxation is through repeated
fatigue cycles and is the most prominent method among all three. This
process is called cyclic stress relaxation and can happen even when there
is no macroscopic plastic deformation. It should also be noted that
344 Frontiers in Aerospace Science, Vol. 1 Vasu and Grandhi
The main parameters which can influence stress relaxation are the number
of cycles, load amplitude, stress ratio, and degree of cold working [11].
Compared to shot peening, the effect of cold working is low in the case of
laser peening [28]. A framework has been constructed to determine the
relaxation behavior as shown in Fig. (14). It can be divided into two
Prediction of Residual Stress Relaxation Frontiers in Aerospace Science, Vol. 1 345
phases. The first phase is the peening phase where residual stresses are
generated in the model by the laser peening. An advanced
modified-explicit procedure is employed for simulating the multi-shot
laser peening process. The second phase involves the relaxation effect
where the residual stress is subjected to cyclic loading. Based on the
number of cycles applied onto the material, we can solve the problem
using FE models or available empirical/analytical models. The division is
based on two categories of fatigue failure, namely Low Cycle Fatigue
(LCF) and High Cycle Fatigue (HCF). LCF relaxation can be predicted
by finite element simulation models while HCF relaxation can be
predicted by already existing analytical models. The prediction of stress
relaxation with FE models has been detailed in this section.
Fig. (15). Strain controlled tests (a) Specimen geometry (b) Cyclic stress-strain behavior of
Ti-6Al-4V.
346 Frontiers in Aerospace Science, Vol. 1 Vasu and Grandhi
2 p p
where equivalent plastic strain rate is given by p ij ij . C1 is the
3
kinematic hardening modulus, γ is the material constant which determines
the rate at which the kinematic hardening modulus decrease with
increasing plastic deformation. The material hardening parameters are
calibrated from the cyclic testing data and are given by C1 = 1.38x1011 and
γ=635.
The numerical modeling can be broken down into two steps. The first step
is to laser peen the specimen to obtain the residual stress field. The
importance of curved geometry is taken into consideration while
modeling the residual stress imparted to the specimen. To make a good
representation of the real problem, multi-shot laser peening with
overlapping is conducted on a 3D convex model and the individual stress
components for each element are imported into the 2D axisymmetric FE
model as prestress. This task of modeling the initial residual stress field is
performed with the use of SIGNI.f, a user subroutine in abaqus. Step 1 is
shown in Fig. (16).
The second step is to apply fatigue load on the laser peened specimen.
Variable amplitude loading such as that obtained from FALSTAFF can be
used to describe the loading spectrum. But for the simplicity of the
procedure of calculating the cyclic response due to loading magnitudes,
constant amplitude loading is considered for this research. Step 2 is
Prediction of Residual Stress Relaxation Frontiers in Aerospace Science, Vol. 1 347
Fig. (17). Numerical prediction of stress relaxation due to applied cyclic loading.
Fig. (18). Cyclic response of residual stresses undergoing fatigue loading along (a) the longitudinal
direction (b) the transverse direction.
Fig. (19) shows the residual stress in the longitudinal and transverse
directions for a completely reversible loading condition with different
maximum stresses. We can observe that, as the maximum stress gets
higher, relaxation becomes significant in both directions. For a repeated
loading case scenario, the relaxation effect is minimal. This indicates that
compressive loading is the driving force for the relaxation of residual
stresses produced by methods like laser peening.
Prediction of Residual Stress Relaxation Frontiers in Aerospace Science, Vol. 1 349
CONCLUSION
CONFLICT OF INTEREST
The authors confirm that they have no conflict of interest to declare for
this publication.
ACKNOWLEDGEMENTS
Declared none.
REFERENCES
stresses and mechanical loading", Int. J. Fatigue, vol. 23, no. Issue.4, pp.
293-302, 2001.
[http://dx.doi.org/10.1016/S0142-1123(00)00104-3]
[13] G.H. Farrahi, J.L. Lebrun, and D. Couratin, "Effect of shot peening on residual
stress and on fatigue life of a spring steel", Fatigue Fract Eng. Mat. Struct.,
vol. 18, no. 2, pp. 211-220, 1995.
[http://dx.doi.org/10.1111/j.1460-2695.1995.tb00156.x]
[14] K. Iida, and M. Takanashi, "Relaxation of welding residual stresses by
reversed and repeated loadings", Welding in the World/Le Soudage dans le
Monde, vol. 14, no. 4, pp. 314-327.
[15] B.L. Boyce, X. Chen, J.O. Peters, J.W. Hutchinson, and R.O. Ritchie,
"Mechanical relaxation of localized residual stresses associated with foreign
object damage", Mater. Sci. Eng., vol. 349, no. Issues.1-2, pp. 48-58, 2003.
[http://dx.doi.org/10.1016/S0921-5093(02)00543-9]
[16] A. Vasu, K. Gobal, and R.V. Grandhi, "A computational methodology for
determining the optimum re-peening schedule for an increased fatigue life of
laser peened aircraft components", Int. J. Fatigue, vol. 70, no. January, pp.
395-405, 2015.
[http://dx.doi.org/10.1016/j.ijfatigue.2014.07.008]
[17] P. Peyre, R. Fabbro, P. Merrien, and H. Lieurade, “Laser shock
processing of aluminum alloys. Application to high cycle fatigue behavior",
Mater. Sci. Eng. A, vol. 210, no. 1-2, pp. 102-113, 1996.
[http://dx.doi.org/10.1016/0921-5093(95)10084-9]
[18] M.A. Meyers, Dynamic Behavior of Materials, J.. Wiley: New York, 1994.
[http://dx.doi.org/10.1002/9780470172278]
[19] P. Ballard, J. Fournier, R. Fabbro, and J. Frelat, “Residual stresses induced by
laser-shocks,” Le Journal de Physique IV, Vol. 1, No. C3, pp. 487-494.
[20] A. Vasu, Y. Hu, and R.V. Grandhi, "Differences in plasticity due to curvature
in laser peened components", J. Surf. Coat. Technol., vol. 235, no. November,
pp. 648-656, 2013.
[http://dx.doi.org/10.1016/j.surfcoat.2013.08.043]
Prediction of Residual Stress Relaxation Frontiers in Aerospace Science, Vol. 1 353
[21] A. Vasu, and R.V. Grandhi, "Effects of curved geometry on residual stress in
laser peening", J. Surf. Coat. Technol., vol. 218, no. March, pp. 71-79, 2013.
[http://dx.doi.org/10.1016/j.surfcoat.2012.12.029]
[22] Y. Hu, C. Gong, Z. Yao, and J. Hu, "Investigation on the non-homogeneity of
residual stress field induced by laser shock peening", J. Surf. Coat. Technol.,
vol. 203, no. 23, pp. 3503-3508, 2009.
[http://dx.doi.org/10.1016/j.surfcoat.2009.04.029]
[23] A. Vasu, and R.V. Grandhi, "Impact of plasticity generated by rayleigh waves
on the residual stress behavior of structural components subjected to laser
peening", Int. J. Struct. Integrity, vol. 6, no. Issue. 1, pp. 107-123, 2015.
[http://dx.doi.org/10.1108/IJSI-06-2014-0028]
[24] H.K. Amarchinta, R.V. Grandhi, A.H. Clauer, K. Langer, and D. Stargel,
"Simulation of residual stress induced by laser peening process through
inverse optimization of material models", J. Mater. Process. Technol., vol. 210,
no. 14, pp. 1997-2006, 2010.
[http://dx.doi.org/10.1016/j.jmatprotec.2010.07.015]
[25] Vasu, A., Influence of Curved Geometries on the Fatigue Life of Laser Peened
Components, Ph.D Dissertation, Wright State University, 2014.
[26] Z. Zhou, A.S. Gill, D. Qian, S.R. Mannava, K. Langer, Y. Wen, and V.K.
Vasudevan, "A finite element study of thermal relaxation of residual stress in
laser shock peened IN718 superalloy", Int. J. Impact Eng., vol. 38, no. Issue. 7,
pp. 590-596, 2011.
[http://dx.doi.org/10.1016/j.ijimpeng.2011.02.006]
[27] O.S. Zaroog, A. Ali, B.B. Sahari, and R. Zahari, "Relaxation of compressive
residual stress. Part 1: Relaxation of stage 1", J. Sci. Ind. Res. (India), vol. 68,
no. Issue. 12, pp. 1035-1037, 2009.
[28] P. Prevey, J. Telesman, T. Gabb, and P. Kantzos, "FOD resistance and fatigue
crack arrest in low plasticity burnished IN718".
354 Frontiers in Aerospace Science, 2016, Vol. 1, 354-392
CHAPTER 9
Abstract: The aerospace industry has utilized high strength aluminum alloys to
propel the production and manufacturing of advanced aerospace technology;
however, welding of Aluminum and its constituent alloys introduce challenges which
affect the structural integrity of the welded area if conducted indecorously. For this
reason, Non-destructive Evaluation (NDE) of welded areas is conducted to ascertain
defective regions to ensure structural integrity of the aerospace structure. NDE
techniques are noninvasive and can determine whether the object contains
irregularities, discontinuities, or flaws. Inspecting weld areas allows for cost
reduction by detecting discontinuities in the early stages of manufacturing;
consequently, reducing the time and money to rework the error and allows for the
validation of sound welds. A variety of NDE techniques are available depending on
the applications, each with its own advantages and disadvantages. Among the
number of NDE techniques, radiography and ultrasonics are the most widely utilized
for inspection of weld defects. In this study, a detailed analysis was thereby
conducted to ascertain the critical phased array ultrasonic testing (PAUT) parameters
for the detection of weld defects, more specifically with Friction-Stir-Welding (FSW).
Consequently a comparison to X-ray radiography is also included. It was observed
that both techniques produced similar detection results for defects in the range of
1.0mm; however, it was found that PAUT was the only technique able to discover
defects in the range of 0.15 mm.
INTRODUCTION
(Table 1) contd.....
Not applicable
Can detect
Magnetic Magnetic particles are Surface and to
flaws up to ¼ Cracks, Incomplete
Particle attracted to breaks in near subsurface non-magnetic
inch below penetration, overlap
(MT) magnetic lines of force defects metals or
surface
alloys
Economic
Penetrating rays (X-ray or limit to depth
Burn Through, Excessive
gamma rays) cast Defects penetration; Permits visual
Radio-gra /inadequate
shadows on the other side anywhere hazardous analysis of
phic reinforcement,
of “solid” objects; film within the operation; buried defects
Testing Incomplete penetration,
radiography records examined complex or components
(RT) Misalignment, Porosity,
shadow on photographic volume shapes are in assembly
Root concavity, undercut
film difficult to
analyze
Ultrasonic testing has become a widely used NDE technique with many
advancements and variations. Ultrasonic testing can be used to detect
358 Frontiers in Aerospace Science, Vol. 1 Dewan et al.
cracks, voids, and changes in geometric and material parameters such as:
thickness, stress concentration, and modulus [13]. In ultrasonic testing,
high frequency ultrasound vibrations are generated from piezoelectric
elements and thereby transmitted into test specimens. The transmitted high
frequency waves are reflected or scattered by discontinuities inside the
material as the waves propagate. The piezoelectric elements also act as
receivers which detect the ultrasonic reflections from the discontinuities.
In solids, sound waves can propagate in four principle methods that are
established by fundamental particle oscillation principles. Sound can pro-
pagate as longitudinal waves, shear waves, surface waves, and plate
waves. Longitudinal and shear waves are the two modes of sound wave
propagation most commonly used in ultrasonic testing. In longitudinal
waves, the oscillations occur in the longitudinal direction or the direction
of wave propagation. In the transverse or shear wave, the particles
oscillate at a right angle or transverse to the direction of wave propagation.
Three important properties of sound waves propagating in isotropic solid
materials are: wavelength (𝜆), frequency (𝑓), and velocity (𝑉). The
wavelength is directly proportional to the velocity of the wave and
inversely proportional to the frequency of the wave. This relationship is
shown below in Eq. (1). The wavelength is related to defect detection
capabilities, which vary with ultrasonic transducer capabilities. Generally
the measurable smallest defect size should be larger than one half of the
wavelength. In general, a defect size must be larger than one-half the
wavelength in order to be detected. Therefore, if a material’s velocity
remains constant, by increasing the frequency the size of λ will decrease
which will result in smaller defects that can be found; hence, low
frequency probes can overlook defects and can cause an operator to
believe that a defect-free weld has been conducted.
Non-Destructive Evaluation (NDE) Frontiers in Aerospace Science, Vol. 1 359
𝑉𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑉
𝑤𝑎𝑣𝑒𝑙𝑒𝑛𝑔𝑡, 𝜆 = (1)
𝐹𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 𝑓
𝐷2 𝐷2 𝑓
𝑁= = (3)
4𝜆 4𝑉
Fig. (1). Radiated fields from an ultrasonic transducer: near-field and far-field [17].
The energy in the ultrasonic beam does not remain in a cylinder, but
instead, spreads out as it propagates through the material. The
phenomenon is usually referred to as beam-spread but is sometimes also
Non-Destructive Evaluation (NDE) Frontiers in Aerospace Science, Vol. 1 361
𝑉
𝑆𝑖𝑛𝜃 = 1.2 (4)
𝐷𝑓
Fig. (3). Schematic of conventional UT, TOFD, and PAUT, three different ultrasonic set-up created
with ultrasonic simulation software (ESBeam Tool).
Non-Destructive Evaluation (NDE) Frontiers in Aerospace Science, Vol. 1 363
Fig. (4). Phased array ultrasonic scan pattern and different scan views [18].
Fig. (6). Typical A-scan signal indicating received signal voltage vs. time.
Non-Destructive Evaluation (NDE) Frontiers in Aerospace Science, Vol. 1 365
For A-Scan data, an ultrasonic echo will be found in between the front
and back wall echoes versus travel time (∆𝑡). For the particular material,
sound velocity ( 𝑉) is known. Accordingly, the sound path can be
calculated using Eq. (5).
Now, if the reflection angle (𝜃𝑅 ) and material thickness (𝑇) is known, the
location of the defect can be calculated using the following equations:
Fig. (7). Schematic of TOFD scanning process and typical TOFD A and B scan views showing
different echoes [9].
Non-Destructive Evaluation (NDE) Frontiers in Aerospace Science, Vol. 1 367
Type of
UT Signals Schematic and TOFD Scan View
Defect
Two signals (2, 3)
Incomplete
from top and
root
bottom of the
penetration
defect
More back wall
signal and
Lack of root
inverted as
penetration
compared to
lateral wave (LW)
One signals
Interphases
in-between LW
lack of
and back-wall
fusion
signal
Many small
signals
Cluster in-between LW
Porosity and back-wall
echoes, looks like
noise
Concave Distortion of
root back-wall echo
368 Frontiers in Aerospace Science, Vol. 1 Dewan et al.
PAUT has three distinct components. Firstly, the probe (transducer) which
functions as the transmitter and receiver of the high frequency sound
waves. Phased array ultrasonic probes utilize an array of piezoelectric
elements that in sequence generate high frequency sound waves.
Transducer frequencies are most commonly found to be in the range of 2
MHz to 10 MHz [24]. The probe is mounted on top of a wedge
component which couples the sound waves generated by the probe to the
work piece. Wedges are also used to steer the sound waves at a certain
angle as they enter the work-piece according to Snell’s Law. The couplant,
typically a moderately viscous, nontoxic liquid or gel, is used between the
wedge and the work-piece to facilitate the transmission of sound energy
between the wedge and the test piece. Lastly, a data acquisition unit is
used to generate images that can utilize the sound waves to determine
defects of a material. These data acquisition units are very sophisticated
and have many input variables that will affect the images produced. In Fig.
(8), an illustration is given of a probe/wedge unit that is situated on top of
Non-Destructive Evaluation (NDE) Frontiers in Aerospace Science, Vol. 1 369
Fig. (8). A typical probe and wedge configuration with illustration of wave propagation inside a
friction stir (FS) welded specimen.
For PAUT inspection, multiple input parameters must be entered into the
data acquisition unit to accurately determine defect sizes and locations.
These variables include the characteristics of the transducer, wedge, and
welded area. Setup variables also include parameters that assist in
generating an accurate focal law for the high frequency sound waves.
370 Frontiers in Aerospace Science, Vol. 1 Dewan et al.
The sound waves are predictable due to wave physics principles. These
wave principles are classical physics phenomena and have been well
documented as they pertain to ultrasonics, as seen in [27-34]. Work has
also been conducted to determine the effect of element size in linear
phased arrays [35]. Optimization of PAUT probe design has likewise been
discussed in [36, 37]. Simulations of PAUT systems have been conducted
to predict phased array ultrasonic wave interactions and behavior inside
materials in [38-40]. With this new technology, simulation software has
been developed to aid in the scanning process to create the required
PAUT software to accurately detect defects. A requirement is to describe
the inspection task and enter fundamental information such as site
location, number of welds, pipe diameter, thickness, material properties,
weld preparation, procedure, and ultrasonic method to be used. The
software will then calculate and generate all the UT set-up parameters
required to perform Phased Array inspection of the welds.
Non-Destructive Evaluation (NDE) Frontiers in Aerospace Science, Vol. 1 371
As mentioned earlier the most common scan views (images) that a PAUT
system utilizes are A, B, C, and S scans. If there is a defect present in the
scanned samples, it can be visualized with 6-dB color change [25, 41, 42].
Fig. (9) shows typical A, S, and C-scan views obtained from phased array
ultrasonic testing. The S-scan is perhaps the most useful asset to current
PAUT post-weld evaluation due to the ability to steer the sound waves in
a range of angles which allows for easy visualization of a specimen [20].
The horizontal axis corresponds to location of the defect from probe
position, and the vertical axis to depth of the test piece. During inspection,
the inspection data for each weld is stored into the memory. The location
can be identified by incorporating an encoder or scanner. After inspection,
a sophisticated software package is employed allowing a qualified
ultrasonic inspector to review and analyze the inspection data utilizing
advanced analysis tools such as: real time volume corrected imaging, as
well as, conventional image analysis, image enhancement, and
measurement determination.
Fig. (9). PAUT A, S, and C-scan view acquired using Olympus Omniscan MX2.0 data acquisition
unit [18].
372 Frontiers in Aerospace Science, Vol. 1 Dewan et al.
Defect sizing capabilities of the PAUT unit was examined using the
Olympus OmniScan MX2.0 unit and TomoView software version 2.10. For
defect sizing, two important parameters need to be considered: (i) gain
and (ii) index offset. In this study, defects were measured with (-2 dB), (-3
dB), and (-6 dB) color drop techniques and was compared with actual
defect sizes. Higher gain value results in higher A-scan amplitude; and
higher A-scan amplitude leads to oversize estimation. It is also observed
that the higher color drop (-6 dB) technique implies larger defect sizing
compared to smaller color drop (-2 dB) techniques. Hence, during
post-weld inspection the defect size is dependent on the peak A-scan
amplitude and color drop technique (Fig. 10). The index offset is the
distance of the tip of the wedge from the weld-centerline. For constant
A-scan amplitude, a higher index offset results in higher defect size
estimation (Fig. 11).
Fig. (10). Effect of A-scan amplitude and color drop on defect size estimation.
Non-Destructive Evaluation (NDE) Frontiers in Aerospace Science, Vol. 1 373
Fig. (11). Effect of index offset and color drop on defect size estimation (fixed gain value).
Fig. (12). Variations of A-scan amplitude with index offset to illustrate the effect of TCG calibration.
374 Frontiers in Aerospace Science, Vol. 1 Dewan et al.
Fig. (13): Aluminum alloy plate with seven varying-hole sizes with associated C-scan and
eco-dynamic A-Scan images.
Non-Destructive Evaluation (NDE) Frontiers in Aerospace Science, Vol. 1 375
This results in the elongation of the defect image; alternatively, for defects
at smaller depths, the wave has not reached its focal length and thereby
underestimates the defect. In other words the defect image depends on the
near and far fields focusing distances of the wave. These results indicate
that PAUT has the ability to accurately detect defect sizes and locations.
-ness Dia. Dia. Dia. Depth Depth Depth Dist. Dist. Dist.
(mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm)
25.5 2.8 3.1 0.3 8.1 8.25 0.15 16.1 16.6 0.5
25.5 1.6 2.2 0.6 11.8 11.9 0.1 17.9 18.3 0.4
25.5 3.5 3.6 0.1 12.5 11.9 -0.6 18.5 19.0 0.5
8.35 1.6 1.5 0.1 4.2 3.9 -0.3 4.1 3.6 -0.5
8.35 2.4 2.0 -0.4 4.2 3.8 -0.4 3.6 3.0 -0.6
Fig. (14). Welding defect sizes measured with optical microscope and PAUT [43].
Fig. (15). PAUT’s S and A-scan views, defect location, and cross section optical micrograph of a
FS welded panel illustrating a surface-defect.
Non-Destructive Evaluation (NDE) Frontiers in Aerospace Science, Vol. 1 377
Fig. (16). PAUT’s S-Scan view, A-Scan View, defect location, and cross sectional optical
micrograph of a FS welded panel illustrating an internal defect.
Fig. (17). Schematic of basic set-up for film radiography [9, 44].
𝑐
𝐸 = . 𝜗 = (8)
𝜆
Non-Destructive Evaluation (NDE) Frontiers in Aerospace Science, Vol. 1 379
The energy of the radiation is responsible for its ability to penetrate matter.
Higher energy radiation can penetrate more and higher density matter
than low energy radiation. Radiation intensity is the amount of energy
passing through a given area that is perpendicular to the direction of
radiation travel in a given unit of time. The penetrating intensity can be
expressed as Eq. (9).
𝐼 = 𝐼0 . 𝑒 −𝜇𝑡 (9)
𝐼1 𝑑1 2 = 𝐼2 𝑑2 2 (10)
𝑡(𝑈𝑡 +4𝑠)
𝐹= (11)
𝑈𝑡
𝐼
𝑅𝑎𝑑𝑖𝑜𝑔𝑟𝑎𝑝𝑖𝑐 𝑑𝑒𝑛𝑠𝑖𝑡𝑦 = log( 0 ) (13)
𝐼𝑡
(Table 4) contd.....
Incomplete
A darker density band, with
or Lack of
very straight parallel edges, in
Penetration
the center of the width of the
(LOP)
weld image
Elongated
Elongated parallel or single
slag lines
darker density lines, irregular
(wagon
in width and slightly winding
tracks)
lengthwise
(Table 4) contd.....
(Table 4) contd.....
Fig. (18). Digital Radiographic images of FS welded AA 2219-T87 plates showing (a) internal
defect and (b) surface-defect.
Non-Destructive Evaluation (NDE) Frontiers in Aerospace Science, Vol. 1 385
Fig. (19). Radiographic and PAUT’s B-Scan and C-Scan images illustrating a surface-defect in a
FS welded AA 2219-T87 panel.
386 Frontiers in Aerospace Science, Vol. 1 Dewan et al.
Fig. (20). Radiographic and PAUT’s B-Scan and C-Scan images of a FS welded AA 2219-T87
panel.
CONCLUSION
CONFLICT OF INTEREST
The authors confirm that they have no conflict of interest to declare for
this publication.
ACKNOWLEDGEMENTS
Declared none.
REFERENCES
[1] T.M. Morton, "Defects in 2021 aluminum alloy welds", Weld. J., vol. 7, pp.
304-312, 1971.
[3] F.V. Lawrence Jr, W.H. Munse, and J.D. Burk, "Effects of Porosity on the
Fatigue Properties of 5083 Aluminum Alloy Weldments", Welding Research
Council Bulletin, vol. 206, pp. 95s-98s, 1975.
correlated with ultrasonic and radiographic evaluation", Weld. J., vol. 4, pp.
198-206, 1972.
[9] "ASME Boiler and Pressure Vessel Code (BPVC)," In: Section V-
Nondestructive Examination, ASME International, New York, NY, 2013.
[12] "Specification for Friction Stir Welding of Aluminum Alloys for Aerospace
Applications", In: AWS D17.3/D17.3M:200X. American Welding Society:
Miami, FL.
[14] G.S. Kino, Acoustic Waves: Devices, Imaging, and Analog Signal Processing.
Prentice Hall PTR: New Jersey, 1987.
[15] K.B. Ocheltree, and L.A. Frizzel, “Sound field calculation for rectangular
Non-Destructive Evaluation (NDE) Frontiers in Aerospace Science, Vol. 1 389
sources", IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 36, no. 2, pp.
242-248, 1989.
[http://dx.doi.org/10.1109/58.19157] [PMID: 18284974]
[18] “Phased array testing: Basic theory for industrial applications,” Tech. Report.
Olympus NDT: Waltham, MA, 2010.
[21] M. Moles, N. Dube, and F. Jacques, "Ultrasonic Phased Arrays for Thick
Section Weld Inspections," In: Joining of Advanced and Specialty Materials
VI: Proceedings from Materials Solutions 2003 on Joining of Advanced and
Specialty Materials, 2004, pp. 112-118.
[22] A.S. Birks, R.E. Green, and P. McIntire, Nondestructive Testing Handbook. 2nd
ed American Society for Nondestructive Testing: Columbus, OH, 1991.
390 Frontiers in Aerospace Science, Vol. 1 Dewan et al.
[23] B. Li, Y. Shen, and W. Hu, "The study on defects in aluminum 2219-T6 thick
butt friction stir welds with the application of multiple non-destructive testing
methods", Mater. Des. vol. 32, pp. 2073-2084, 2011.
[http://dx.doi.org/10.1016/j.matdes.2010.11.054]
[29] J. L. Rose, "A baseline and vision of ultrasonic guided wave inspection
potential," J. Press. Vess. Tech., vol. 124, August 2002.
[http://dx.doi.org/10.1115/1.1491272]
[30] J.L. Rose, Ultrasonic Waves in Solid Media. Cambridge University Press:
New York, 2004.
[31] J.L. Rose, "Guided wave nuances for ultrasonic nondestructive evaluation",
IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 47, no. 3, pp. 575-583,
2000.
[http://dx.doi.org/10.1109/58.842044] [PMID: 18238584]
[32] L. Azar, Y. Shi, and S. C. Wooh, "Beam Focusing Behavior of Linear Phased
Non-Destructive Evaluation (NDE) Frontiers in Aerospace Science, Vol. 1 391
[34] B.A. Auld, Acoustic Fields and Waves in Solids, vol. Vol. 2. 2nd ed Kreiger
Publishing Co.: FL, 1990.
[35] S. Wooh and Y. Shi, "Influence of phased array element size on beam steering
behavior," Ultrasonics, vol. 36, April 1998.
[http://dx.doi.org/10.1016/S0041-624X(97)00164-9]
[36] V. Lupien, "Phased array ultrasonic probe design: From art to science through
optimization", AIP Conf. Proc., vol. 700, pp. 761-768, February 2004.
[http://dx.doi.org/10.1063/1.1711697]
[37] R.A. Roberts, "Efficient design of ultrasonic phased array transducer surface
geometry", AIP Conf. Proc., vol. 820, pp. 797-804, March 2006.
[http://dx.doi.org/10.1063/1.2184608]
[40] X. Na, and Z. Zhenggan, "Numerical simulation and experiment for inspection
of corner-shaped components using ultrasonic phased array", NDT Int., vol. 63,
392 Frontiers in Aerospace Science, Vol. 1 Dewan et al.
2014.
[41] M.W. Dewan, J. Liang, M.A. Wahab, and A.M. Okeil, "Effect of post-weld
heat treatment and electrolytic plasma processing on tungsten inert gas welded
AISI 4140 alloy steel", Mater. Des. vol. 54, pp. 6-13, 2014.
[http://dx.doi.org/10.1016/j.matdes.2013.08.035]
CHAPTER 10
1
Department of Engineering Technology and Construction Management,
University of North Carolina at Charlotte, NC, USA
2
School of Mechanical, Industrial, and Manufacturing Engineering,
Oregon State University, OR, USA
3
Ecole Polytechnique de Montreal, Quebec, Canada
4
Sustainable Environment and Energy Systems Department, Middle East
Technical University, Northern Cyprus Campus, Turkey
5
Aerospace Engineering Department, Amirkabir University of Technology,
Tehran, Iran
6
Aerospace Engineering Department, University of Tehran, Tehran, Iran
NOMENCLATURE
Symbol Description
Acronyms
FMEA Failure Modes and Effect Analysis
FMECA Failure Modes, Effects, and Criticality
Analysis
MTBF Mean Time between Failures - s
MTTF Mean Time to Failure - s
PDF Probability Density Function
RPH Remotely Piloted Helicopter
UAV Unmanned Aerial Vehicle
Greek Symbols
βc,CR Control rotor longitudinal flapping - rad
βs,CR Control rotor lateral flapping - rad
λs(t) Failure rate - s
Ω Rotational speed - rad/s
ρ Air density - kg/m3
σ Solidity factor
ΘOM Main rotor blade pitching angle - rad
Development of a Remotely Piloted Helicopter Frontiers in Aerospace Science, Vol. 1 395
INTRODUCTION
1. Conceptual design
2. Detailed design
3. Manufacturing and assembly
4. Tests and reliability analysis.
CONCEPTUAL DESIGN
The main goal in the conceptual design phase is to specify the desired
application of the RPH and determine the proper range of weights,
dimensions, and stability and performance characteristics at different
flight modes including hover, climb, and forward flights. There are
different methods to obtain these characteristics. Two different methods
can be implemented: 1) deriving theoretical equations based on
aerodynamic and dynamic principles of helicopter performance, such as
blade element theory and momentum theory [11-14], and 2) deriving
empirical equations based on similar helicopters characteristics. Here, we
employ a combination of both methods to obtain the desired RPH
characteristics for the defined mission. A strong database that includes
dimensions and performance characteristics of similar operational
unmanned helicopters is needed. Table 1 summarizes the studied parameters
of more than 100 RPHs all around the world. These RPHs are classified in
four major groups of power unit, weight, performance, and dimension [1]. In
order to have a complete database the performance characteristics of the
398 Frontiers in Aerospace Science, Vol. 1 Goudarzi et al.
Max available
Max combat load Max cruise speed @ SL Overall height
power
Using the obtained trends [1], one can estimate the empty weight (WEmpty),
payload weight (WPayload ), fuel weight (WFuel ), and overall weight (WOverall)
of an RPH for a desired mission. Mission requirements affect the RPH
performance (altitude, range, endurance, and climate), design,
monitoring requirements (real time or post flight, bandwidth for real
Development of a Remotely Piloted Helicopter Frontiers in Aerospace Science, Vol. 1 399
Remote
5 1.2 80 18 140 43 2100
sensing
Forest fire
7 1.4 300 5 150 15 1500
surveillanc
e
Rescue
14.3 2 300 3.9 120 100 1650
and relief
Parvan
8.5 2 50 10 150 80 1500
(traffic
monitoring
)
2.2. Configuration Selection
The next step is to characterize the main and tail rotor geometric
dimensions. Interpolating in the trends of different RPH design geometric
characteristics in the created database can be performed through defining
a set of dimensionless parameters [1]. One of the well-used
non-dimensional parameters is the solidity factor σ, the ratio of blades
area to rotor area which has a significant role in rotor sizing. Using σ
helps the designer to interpolate between three main design parameters:
number of blades b, blade chord c, and rotor radius R. The interpolation
of c and R with two blades for the Parvan gave the initial σ value of
0.081, the blade chord of 0.093 m and the blade radius of 0.73 m.
Besides defining dimensionless parameters that reduce the number of
design variables and simplify the interpolation of geometric
characteristics, there are certain relationships between some design
parameters. Fig. (2) illustrates the relationship between main and tail
rotor diameters and Fig. (3) illustrates the relationship between RPH
body height and main rotor diameter using the database [1]. These
Development of a Remotely Piloted Helicopter Frontiers in Aerospace Science, Vol. 1 401
1.4
1.2
0.8
0.6
0.4
0.2
0
0 2 4 6 8 10
Main rotor diameter (m)
Fig. (2). The trend line of tail rotor and main rotor diameter relationship: the data collected for several
RPHs ( ), and fitted curve (-).
3.5
2.5
RPH Body Height (m)
1.5
0.5
0
0 1 2 3 4 5 6 7 8 9
Fig. (3). The trend line of RPH body height with respect to the main rotor diameter: the data collected
for several RPHs ( ), and fitted curve (-).
402 Frontiers in Aerospace Science, Vol. 1 Goudarzi et al.
The blade element theory together with the created database are the
basis for aerodynamic sizing, thrust sizing, and rotational speed (in
RPM) selection. Due to the importance of this step, three different
methods are performed and compared to each other to obtain the final
values:
1. Using the database to estimate the required power, select the RPM,
and obtain the lift/drag coefficients based on the required thrust;
2. Using the database to estimate the lift/drag coefficients and RPM
through the power and thrust relations;
3. Using conventional airfoil databases to select an airfoil, determine
the thrust, required power, and RPM.
All the generated and lost powers should be calculated in order to estimate
the required power and choosing the right engine type. Through the Parvan
project, it was found out that the most important parts of the power
contributions are the induced power (Pi) and the profile power (Po) which
are defined by [13]:
Pi C pi A R
3
(1)
Development of a Remotely Piloted Helicopter Frontiers in Aerospace Science, Vol. 1 403
CT3 2
C pi (2)
2
CD A R
3
Po (3)
8
where Cpi is the induced power coefficient, ρ is the air density, A is the rotor
area, R is the rotor radius, ω is the rotor blade angular velocity (in RPM), κ
is an empirical factor equal to 1.15, CT is the thrust coefficient, and CD is the
drag coefficient. The total power at different altitudes, rate of climbs,
speeds, and different flight modes including hover, climb, and forward
flights should be calculated [1]. Different software packages or self-coding
can be employed for performance analysis and calculating performance
parameters such as maximum sustainable speed, maximum ROC, hover
ceiling, service ceiling, and power matching. These parameters were
calculated through a novel coding developed by the Parvan group. The
changes in the main rotor power, total power, and the maximum ROC at
different forward speeds have been shown in an earlier study [1]. It was
seen that at a specific altitude, the main rotor power decreases with
increasing the forward speed. Also, at a specific altitude, the total power
(the summation of induced power, profile power, parasite power, and main
rotor power) has its minimum for maximum forward speed in the range of
32-72 km/hr and it increases as the maximum forward speed increase.
Finally, the maximum ROC decreases with altitude.
DETAILED DESIGN
The very first choice for an RPH would be an electric engine. These
engines are the most convenient ones to setup and obtain a moderate
power to weight ratio compared to internal combustion engines [19].
The process of tuning carburetors, fuel systems and exhaust does not
need to be tuned. They are more efficient than piston engines and they
Development of a Remotely Piloted Helicopter Frontiers in Aerospace Science, Vol. 1 405
are less likely to crash during the flight as a result of the fuel/air mixture
or ignition problems. However, these engines are not economically as
efficient as nitro-gas powered engines due to the price of large-capacity
lithium batteries. The batteries are not durable in the severe maneuvers
due to the power required for these regimes. An advantage of these
types of engines would be the quick start up compared to other engine
types such as the gas powered engines and turbine powered engines.
One of the most profitable factors of using electric engines would be the
constant center of gravity during the flight compared to the other types
which use the chemical fuels. Moreover, the noise created from these
engines would be quite negligible and their operation at high frequencies
results in low vibration levels. On the other side, to name a few of
their main disadvantages: 1) the flight time can be short, 2) the cooling
time for the batteries between flights would be essential, and 3) large
lithium batteries can be dangerous and easily ruined if not cared
properly.
Gas engines are very powerful and run at lower RPMs. Their operational
procedure is easier than the nitro engines, whilst they provide higher
flight endurance, and are less expensive [24]. The main disadvantages of
such engines are: 1) lower power to weight ratio, 2) higher price
than electric engines, and 3) high voltage igniters which requires a
powerful battery [19].
Turbine engines are one of the most reliable ones to be used in the
unmanned helicopters. There is no requirement for tuning the system
manually since they are controlled using their built-in computers. These
engines are very noisy and are the heaviest type of helicopter engines.
Moreover, they are more expensive than the electric ones and possess a
higher operating cost and maintenance than any other type. However, the
startup and stop processes of turbine engines are much easier than the
other types. Although these engines are powerful, the power to weight
ratio is not as efficient as other types such as electric ones.
1.2
1200
1
1000
0.8
Power (Watt)
Torque (N.m)
800
0.6
600
400 0.4
200 0.2
0 0
7000 8000 9000 10000 7000 7500 8000 8500 9000 9500 10000
(rpm) (rpm)
(a) (b)
Fig. (4). The Parvan performance charts derived for the OS-91 SZH Engine; (a) estimated power using
the experimental data obtained from P = 4.47Ω3 − 1.3Ω2 + 1.28Ω − 4.15, and (b) estimated torque
using the same data form and the torque equation of Q = 6.81Ω3 − 1.96Ω2 + 1.89Ω − 5.98.
Since the defined mission for Parvan, requires a high endurance in hovering
regime, as opposed to other modes of flight, the design team decided to
choose the main rotor rpm according to the thrust required for hovering
flight. Nevertheless, this number might be changed for different flight
missions. Hence, the gear ratio should be selected to maintain the hovering
flight (based on the defined mission), while other flight regimes are
considered as well. The best way to choose the optimal gear ratio is based
on measuring different torque and angular velocity of the engine while it is
running the power transmission system, main rotor, and tail rotor. A thrust
test stand can be designed and manufactured to measure the main rotor
thrust and power to find the gear ratio. Due to the complexities of using the
OS-91 SZH engine on the test stand, an electric engine was used in place of
the piston engine with a higher gear ratio than the one used on Parvan
(Parvan’s power transmitting system is explained in section 3.3). The gear
train was designed to rotate the main rotor with an angular velocity of 1500
rpm. Fig. (5) illustrates the required power and thrust of main rotor which
are obtained experimentally.
408 Frontiers in Aerospace Science, Vol. 1 Goudarzi et al.
3000
rpm = 1000 rpm = 1000
rpm = 1100 rpm = 1100
2500 rpm = 1200 200 rpm = 1200
rpm = 1300 rpm = 1300
rpm = 1400 rpm = 1400
rpm = 1500 rpm = 1500
2000
Power (watts)
rpm = 1600 rpm = 1600
150
Thrust (N)
1500
100
1000
50
500
0 0
6 7 8 9 10 11 12 13 6 7 8 9 10 11 12 13
(degrees) (degrees)
(a) (b)
Fig. (5). Main rotor power and thrust for evaluating the optimum gear ratio (a) the main rotor power,
and (b) the main rotor thrust measured using the experimental test stand for different angular velocity
of the main rotor.
The required thrust for stabilizing the RPH in hovering flight can be
calculated using the existing models [1], hence, a total collective pitch is
picked for neutralizing the weight; and the power associated with the
pitching angle gives us the required rpm for the main rotor. This angular
velocity is found to be 1500 rpm which gives an optimum power for the
Parvan engine based on the power equation shown in Fig. (4). Hence, the
total gear ratio is determined through the aforementioned procedure.
Small engines can drive a broad range of desired rotors by adjusting the
optimal RPM of rotors through a proper transmission design. Indeed, a
high-speed low-torque engine output can be transferred to low-speed
410 Frontiers in Aerospace Science, Vol. 1 Goudarzi et al.
high-torque values via gearbox in order to drive a large and heavy main
rotor. The best engine performance, main, and tail rotors are achieved at
certain rotational speeds. While obtaining this optimal rotational speed is
achieved through a proper gearbox design, a careless design of torque and
RPM setting might result in serious engine damages.
Gear ratio
& Vibration
Gearbox analysis
design
Frequency
separation
adequacy
End
Fig. (6). The schematic of the transmission system design in overall helicopter design process.
After gear ratio selection for gearbox that is done by analyzing the main and
tail rotors and engine thrust and power values, in order to prove the
adequacy of frequency separation between natural frequency of the power
transmitting system and the rotor RPM for a safe flight, vibration behavior
of the transmission system were studied. Holzer method [27] is used for
calculating the natural frequencies of the power system. According to Fig.
(6), if frequency separation to avoid coupling with the operating modes of
the vehicle is not enough, then gear box should be redesigned in order to
meet the requirements.
4.1.1. Methodology
(a) (b)
Fig. (8). Rotor-body interaction in (a) climb or hovering flight, and (b) forward flight.
Fig. (9). The control rotor stabilizing the main rotor inputs in Parvan.
Clockwise
Rotation
Main Rotor
Blade
Stabilizer Bar
Pitch Input
Roll Input
Fig. (10). Schematic of the Bell-Hiller stabilizer bar used in Parvan [28].
Newton’s second law is the key to study the physics of the helicopter
motion. Assuming that the reference frame is the stationary earth, one
can develop the kinetics and kinematics of the rigid body motion to be
expressed in the following format:
where Ib is the moment of inertia tensor, and wbf is the body frame matrix
Development of a Remotely Piloted Helicopter Frontiers in Aerospace Science, Vol. 1 419
X M X T X H XV X F mg sin m( x yr zq )
YaM YaT YaV YaF mg sin m( y xr zq)
Z M ZT Z H ZV Z F mg cos m( z xq zp )
RoM YaT hM Z M yM YaT hT yV hV Ya Fh F RoF I xx p qr ( I yy I xx )
RoM YaT hM Z M lM M T X T hT ZT lT XHhH Z H lH XV hV M F Z F lF
X F hF I yy q pq ( I xx I yy )
N M YaM lM YaT lT Ya Vl V N F YaF lF I zz r pq ( I xx I yy )
(8)
where X, Y, and Z are corresponding to the forces in the x, y, and z axes,
respectively, whereas, the pitching, rolling and yawing moments are
denoted by M, R, and N, respectively. The variable h defines the height
from body axis to the point that the force is exerted and so forth for l.
Those variables described by the index F are in association with
fuselage, while H refers to the horizontal stabilizer mounted on the
helicopter tail, and V refers to the vertical stabilizer.
Studying the cyclic motion of the main rotor disk due to the aerodynamic
forces and moments associated with the gyroscopic behavior of the rotor
which is also affected by the type of control input whether to be
longitudinal or lateral. Hence, the lateral and longitudinal cyclic pitch
of the main rotor is dependent to the flapping dynamics of the control
rotor in the systems that are working with this analogue controller to
dampen the pilot’s control inputs for two cyclic angles in lateral and
longitudinal directions. These two angles create two more degrees of
freedom for the rigid body of the helicopter, which will be added to the
two more degrees of freedom for the body frame rotations respect to the
earth reference frame. The latter 2-DOFs are created based on the
pitching and rolling Euler angles. Hence, the dynamical system has
10-DOF which is investigated in this study.
X X X X X X X
X x y z q p r c,CR
x y z q p r c,CR
X X X X X (9)
s,CR 0M 0T A1 B1
s,CR 0 0 A1 B1
M T
where, βc,CR and βs,CR are the longitudinal and lateral flapping angles for
the stabilizer bar, respectively. According to the Eq. 9, there are
nonlinear terms on the right hand side which can be linearized using the
Taylor expansion over the trim point of the system. The requirement to
use the Taylor expansion in linearizing the equations is based upon the
assumption of analytic functions of aerodynamic forces and moments to
be valid over the trim point [11]. Having spikes without smoothness in
the functions with discontinuities breaks this assumption.
4.2.2. Stability
The main problem in analyzing the dynamics of flight and the system
response to the control inputs are the number of degrees of freedom in
the model. The best way to analyze the nonlinear systems linearly is to
divide the longitudinal and lateral modes of flight with the certain
assumption suggesting the small coupling between these modes as it has
been asserted for Parvan in [31]. According to this study, the stability
and aerodynamic derivatives for the case study (Parvan) delineating the
coupling between the longitudinal and lateral modes, as well as roll and
yaw axes [11] are sufficiently negligible compared to the pure
derivatives in each aforementioned mode. Hence, the process of
isolating these modes in the analysis may have only a weak drawback in
maneuvering phases in which the coupled dynamic forces play key role
in the analysis system; a case where the small perturbation assumption is
no longer valid and the system cannot be modeled linearly. The 10-DOF
flight model studied in [31] was tested by the linear method of analyzing
422 Frontiers in Aerospace Science, Vol. 1 Goudarzi et al.
stability proposed by Prouty and Padfield in [11, 29]. In this section, the
stability is studied in hover and forward flight.
where the roots are almost stable, while there are two unstable roots that
belongs to the lateral motion of the aircraft as it is illustrated by Eq. 11:
where the oscillation is stable but there are two unstable roots that is
illustrated by Eq. 13:
As the details of this mode are discussed in [31], the lateral motion is a
little unstable in Parvan unmanned helicopter.
424 Frontiers in Aerospace Science, Vol. 1 Goudarzi et al.
Fig. (11). The block diagram of the 3-DOF simplified longitudinal mode in hover for Parvan [31].
s1 0.951
5-DOF s2,3 0.003 0.265i 23.7 217 -
s4,5 1.954 4.058i 1.5 - 0.35
2-DOF - 23.5 - -
Development of a Remotely Piloted Helicopter Frontiers in Aerospace Science, Vol. 1 425
where the roots are almost stable, while there are two unstable roots that
belongs to the lateral motion of the aircraft as it is illustrated by Eq. 15:
system to the cyclic pitch consists of two modes of Phugoid and short
period. According to this, one degree in cyclic pitch will cause
approximately 10 degrees of response which is damped after
approximately 5 seconds.
Fig. (12). The step response of the 5-DOF simplified longitudinal mode in forward flight to
cyclic pitch of B1 [31].
The dynamics of the helicopter in lateral flight is based on the lateral
wind, yawing rate, rolling rate, lateral flapping of control rotor, and the
rolling angle. Basically, the lateral dynamics of helicopters is not quite
stable inherently and is controlled with the thrust of the main rotor and
tail rotor to compensate for the discrepancies between their available
torques in the case of the lateral instability such as a lateral wind. One of
the main contributions of the instabilities is emanated from the tail rotor
thrust. The increase in forward speed results in more instabilities since
the torque of main rotor is increased. The dynamics of pure lateral
motion in forward flight is described using Eq. 18.
CATIA MODELING
There is a transition phase from the conceptual to detailed design, we
called it the “preliminary design”, where the overall configuration of the
helicopter is specified according to the main subsystems. Computer-
aided design (CAD) software would be the best option to create this
preliminary design and also for the following modification, analysis, or
optimization studies. One of the most suitable CAD software packages is
CATIA that has lots of powerful modules (Part Design, Product Design,
DMU Kinematics, Generative Shape Design and Drafting and so on) for
designing mechanical parts and products together with defining the
physical properties of different parts. Fig. (13) illustrates the 3-D
428 Frontiers in Aerospace Science, Vol. 1 Goudarzi et al.
The final technical drawings from the detailed design phase should be
sent for manufacturing. Before considering the required accuracy,
precision, and optimum machining techniques, the best material should
be selected. It should meet the design and mission requirements, and it
meets an optimum value from a trade-off between cost, weight, life
cycle and reliability considerations. The product tree is found the best
tool to show relations of components and subsystems. Concerning a
complex system like a helicopter, for providing a detailed product tree,
Parvan was divided into eight distinct groups from the assembly
viewpoint including engine, fly-bar, hub, power train, tail, and avionics.
Therefore, each group was assembled separately to montage the whole
system. For instance, Fig. (14) illustrates the product tree of the engine
group with its subsystems. It is important to make an accurate
documentation at this step in order to assure the procedure of
assembling and disassembling is followed correctly each time.
Engine Group
(a) (b)
Fig. (15). Manufactured parts, (a) hub, and (b) swashplate.
The main and tail rotor hub and swashplates are made from Al 7075 and
steel is used for the main shaft of Parvan. Al-7075 is a strong aluminum
alloy with high strength, light weight, and less resistance to corrosion
than many other aluminum alloys. Among different manufacturing
techniques, CNC was selected for small and fine aluminum based
components and wire cut technique was used for fuselage and fly-bar
frames. Some parts such as hub Fig. (15a) and swashplate Fig. (15b) are
manufactured for this specific design, and some others are bought from
existing off-the-shelf market such as the OS 91SZ-H ring engine.
Fig. (16) shows major subsystems of Parvan. Fig. (17) shows the final
assembly.
430 Frontiers in Aerospace Science, Vol. 1 Goudarzi et al.
Parvan
As explained in the detailed design section and in Fig. (6), the engine
torque/power rating is the input for designing the gearbox system. Fig.
(18) shows the designed engine torque/power test-stand for the Parvan
project. The engine torque/power values at different rotational speeds
and various blade sizes were calculated using Eq. 20 [13, 29]:
R
R4
P r 33 dr 3 (20)
0 4
Having the final gearbox design as shown in Fig. (19), the RPH main
rotor thrust values at different rotational speeds and input powers are
obtained using the designed thrust test stand shown in Fig. (20).
432 Frontiers in Aerospace Science, Vol. 1 Goudarzi et al.
(a) (b)
Fig. (19). The final gearbox system for main rotor thrust test stand: (a) side view of the main frame,
(b) transmission system on the Parvan.
Development of a Remotely Piloted Helicopter Frontiers in Aerospace Science, Vol. 1 433
t
R(t ) et e MTBF (21)
Failure Rate
Lifetime of component
Table 6 summarizes the criticality numbers for all the subsystems in the
Parvan project. The control subsystem has the most average critically
number; hence, more detailed reliability calculations are performed in [33].
Finally, an improved reliability of the control system is achieved through
analyzing and modifying the control components and the fabrication
methods.
436 Frontiers in Aerospace Science, Vol. 1 Goudarzi et al.
Table 5. A sample of FMECA table developed for one component of Parvan, Hub.
Failure Preventive
Part Failure Failure Detection and
effects P PN n d C
Name mode cause technique recovery
precautions
M-5 screw
(on the Extra Blade
Preflight
horizontal Undesirable flapping and unbalancing
check (every 1/3000 1 2 2 4
shaft) might vibration Unstable and rotor
get 10hr)
flight vibration
loosened
Hub Bearing of
Delay in
hub Loss of
Hub control Preflight
teetering lubricant;
mechanisms commands; check (every 1/1000 2 2 2 8
damage Undesirable
loose Rotor 10hr)
vibration
vibration
Table 6. Final Parvan subsystems criticality numbers after including design and maintenance
considerations.
Engine 5.78
Avionics 7.10
Gear Box 4.00
Control 7.38
Power Train 7.20
Tail 5.33
Development of a Remotely Piloted Helicopter Frontiers in Aerospace Science, Vol. 1 437
CONCLUSION
CONFLICT OF INTEREST
The authors confirm that they have no conflict of interest to declare for this
publication.
438 Frontiers in Aerospace Science, Vol. 1 Goudarzi et al.
ACKNOWLEDGEMENTS
The authors wish to thank the other members of the Parvan group
throughout the course of this project; in particular Dr. Mortazavi for
supervising the project, and Hoofar Pourzand for technical assistance.
REFERENCES
[31] M.S. Sajedi, and R.M. Ziazi, Development of a software for the stability and
performance of unmanned helicopter, B.S. thesis, Aerospace Eng. Dept.,
Amirkabir University of Tech., Tehran, Iran, 2009.
[32] D.C. Lombardo, and K.F. Fraser, Importance of reliability assessment to
helicopter structural component fatigue life prediction. DSTO Aeronautical and
Maritime Research Laboratory, 2002.
[33] F. Kermanshahi, M. Mohagheghi, and S. Sadati, "Reliability analysis, Design
improvements, and maintenance procedure compilation of a novel remotely
piloted helicopter", AIAA Info Tech Conference, 2011 St. Louis, MI, USA.
[http://dx.doi.org/10.2514/6.2011-1545]
[34] Engineering design handbook helicopter engineering. Part one. Preliminary
design. Storming Media, 1997.
[35] H. Pham, Handbook of reliability engineering. Springer Verlag, 2006.
442 Frontiers in Aerospace Science, 2016, Vol. 1, 442-467
CHAPTER 11
Abstract: This chapter reviews the advances in the materials for applications in
structures of both subsonicas well as supersonic aircrafts. An account of the
operating and ambient environmental conditions during flight is first given and the
resulting material requirements have been discussed. Design relationships have been
established taking into consideration the loading conditions and the strength
requirements. In particular, the aircraft skin temperatures at various mach numbers
have been taken into account for selecting appropriate structural materials for both
subsonic and supersonic aircrafts; and consequently various aerospace aluminum
alloys, titanium alloys, superalloys, and composites have been suggested. Finally, a
new materials-selection chart is presented which would help aerospace designers to
select appropriate materials for structural application in subsonic and supersonic
aircrafts.
*Corresponding author Zainul Huda: Department of Mechanical Engineering, King Abdulaziz University,
Jeddah 21589, Saudi Arabia; Tel/Fax: +966537701246; E-mail: drzainulhuda@hotmail.com.
NOMENCLATURE
M Mach number
CT Civil transport
SST Supersonic transport
CFRP Carbon fiber reinforced polymer
SCC Stress corrosion cracking
W(a) Weight of a structural member using material (a)
W(b) Weight of a structural member using material (b)
ρ(a) Density of material (a),
ρ(b) Density of material (b),
σy(a) Yield strength of material (a)
σy(b) Yield strength of material (b)
τ(a) Torsional shear stresses acting on materials (a)
τ(b) Torsional shear stresses acting on materials (b)
E Young’s modulus
K Stress intensity factor
KIC Plane strain fracture toughness
σ Stress applied normal to a crack
RMAF Royal Malaysian Air Force
Al-Cu Aluminum-copper alloy
P/M Powder metallurgy
Ti-Al-V Titanium-aluminum-vanadium alloy
UTS Ultimate tensile strength
BMIs Bismaleimides
CEs Cyanate esters
FGM Functionally graded materials
INTRODUCTION
Fig. (1). Fuselage lap joint showing the localized pillowing caused by crevice corrosion occurring
between the two layers.
Materials Selection in Design of Structures Frontiers in Aerospace Science, Vol. 1 447
During its flight, an aircraft may be subjected to four types of loading: (1)
tension, (2) compression, (3) bending, and (4) torsion. Minimization of
aircraft weight while still satisfying strength requirements is an important
objective in material selection for both subsonic and supersonic aircrafts.
For subsonic applications, lightweight structure results in cost-saving on
fuel; whereas high speeds are achieved by lightweight structures for
supersonic aircrafts. The relationship of each of the four loading types and
the overall weight of the aircraft are discussed as follows [16].
For a tensile loading below the yield limit, the applied stress on a structural
component (e.g. fuselage) should be considered regarding its weight (W). In
comparing specific strengths of two materials: (a) and (b), the design
relationship may be expressed as:
where W(a) is weight of a part using material (a), W(b) weight of the part with
material (b); density of materials (a) and (b) are denoted as ρ(a) and ρ(b) ,
respectively; σy(a) and σy(b) are yield strengths of (a) and (b), respectively.
448 Frontiers in Aerospace Science, Vol. 1 Zainul Huda
Under compression, an aircraft part may fail due to buckling; this loading
condition requires the selected material has a high stiffness (Young’s
modulus, E). In comparing two materials (a) and (b), the following
compression relationship is recommended for the aircraft designer:
where the symbols have the same meanings as mentioned in Eq. (1).
where τ(a) and τ(a) are torsional shear stresses acting on materials (a) and (b),
respectively.
Eqs. (1-4) provide useful tools for an aerospace designer to select the best
material for a specific aircraft structural member with respect to the type of
loading applied on it. Although Eqs. (1–4) provide useful design
relationships in selection of aircraft materials, they do not completely
Materials Selection in Design of Structures Frontiers in Aerospace Science, Vol. 1 449
where K is stress intensity factor, KIC the plane strain fracture toughness, σ
is the stress which acts along a normal direction to the major axis of a
central crack of length 2a (or an edge crack of length a) in the component,
and Y denotes geometric compliance function that describes the geometry
of the structural component. In order to achieve a safe design, material’s
fracture toughness has to be selected to guarantee that KIC> K in order to
prevent crack propagation.
smaller than the diameter of the rivet head; and hence it was impossible
for the inspection people to detect the crack which grew to fracture. In
fact, the designers should have used a different approach. They should
have first fixed the two variables namely high-level stress, σ, and the
critical flaw size, ac; and then should have applied Eq. (5) to define the
required fracture toughness KIC. This approach would have resulted in the
selection of material of appropriate fracture toughness to avoid the failure.
Because of the stringent operating conditions that supersonic flight, it will
be helpful to combine the application of high-performance aluminum alloys,
titanium alloys and polymer matrix composites in the aircraft structure.
Section 3 discusses the performance characteristics of each class of the
three aerospace materials.
[23-25].
The skin temperature during a subsonic flight is usually below 70 oC. This
non-stringent thermal condition enables aerospace designers to use
non-expensive age-hardened aluminum-copper alloys or aluminum lithium
alloys for subsonic aircraft’s structures. Aluminum (alloy) is the primary
aircraft material, comprising about 80 per cent of an aircraft's un-laden
weight. Because the metal resists corrosion, some airlines do not paint
their aero-planes, saving several hundred of kilograms in weight. In
particular, series 2xxx, 6xxx, and 7xxx Al alloys find wide applications in
subsonic aerospace structures. For instance, the ALCLAD 2024-T3 alloy is
being used in C-130 Hercules (a subsonic aircraft for military transport
applications) by Royal Malaysian Air Force (RMAF) [2]. Additionally, the
age-hardened 2024-T3 aluminum-copper alloy is used in fuselage, wing
tension members, shear webs, and structural areas where stiffness, fatigue
performance, and good strength are required. The selection of heat
treatment approaches for the aerospace Al alloys play an important role in
strengthening the alloys for supersonic applications [26]. The
age-hardening heat treatment for the alloy involves solution treatment
(heating) in the range of 450-550 oC and holding for about 1 hour followed
by quenching. The solution treated alloy is then either naturally or
artificially aged [17]. Fig. (2) illustrates the steps in age hardening of Al-Cu
alloys.
SOLUTION TREATED
K K
K
Liquid
K
K
600 K +L
K 5,6%
K
QUENCHED
Temperature (oC)
400
K +0 AGED
200 K + submicroscopic 0
OVERAGED
0
0 2 4 6 8 10
K +0
Wt. % Copper
Fig. (4). Airbus A330 subsonic aircraft; composites contribute to 25% structure [29].
Table 1. Mechanical properties of selected aluminum alloys for application in SST aircrafts [30,
31].
Al Alloy Density Yield Strength UTS (MPa) Elongation at
(g/cm3) (MPa) Break
2024-T81 2.78 450 485 6%
Materials Selection in Design of Structures Frontiers in Aerospace Science, Vol. 1 455
(Table 1) contd.....
It is evident from the data in Table 1 that although the 2219-T87 aluminum
alloy is easy to manufacture (10% elongation at break), its applicability in
modern supersonic aircrafts is limited by its low yield strength.
Comparatively, the easy-to-fabricate 7075-T6 alloy and the new
(heat-treated) type 2-mm Al-Zn-Mg-Sc-Zr alloy (with the highest
elongation at break of 12%) meet strength requirements for the modern
SCT aircrafts [27, 31]. In fact, 7075-T6 is the primary material for making
honeycomb panel frames, which are machined from plate to eliminate
corner joints. The aluminum honeycomb can also be used in the beaded
areas of skin doublers to stiffen the skin of fuselage. At elevated
temperatures, 2024-T81 foil provides higher strength than the work
hardened alloys, such as 5052-H39 and 5056-H39. Since lithium is the least
dense metallic element, the alloying of Li with Al can substantially reduce
the weight of aerospace alloys. For instance, 2090-T651 can be used in
fuselage bulkhead webs and internal framework parts. Another lightweight
Al-Li alloy is 8090-T651 has a high melting range of 600-655 oC, which
places it the best option for building the extendable nose.
The skeleton of the supersonic aircraft needs to be light but strong enough
so as to withstand high pressure applied on it. Titanium alloy Ti-6Al-4V has
excellent corrosion resistance, machinability, and the strength-to-weight
ratio and is an ideal material for the skeleton [32, 33]. A skeleton made of
Ti-6Al-4V will not be affected by the heat due to the friction because of the
456 Frontiers in Aerospace Science, Vol. 1 Zainul Huda
high melting point of titanium (1668 °C). In addition, the very low thermal
expansion (8.6 μm∙m−1·K−1) at 25 °C of titanium provides excellent thermal
stability of the aircraft during supersonic operation.
Materials Selection for Structure for Speeds Between 2.0 and 4.0 Mach
The temperature of skin of the supersonic aircraft is a function of its speed
and also depends on its environment. The skin temperature at first drops
when the aircraft climbs and goes to contact with the decreasing ambient air
temperatures. As the speed exceeds Mach 1, the ambient air temperature
begins to increase, and the skin temperature reaches a maximum of 120 °C
(248 °F) when the aircraft is cruising at Mach 2.2. The skin temperature
stays below 100 °C (212 °F) at a speed of Mach 2.0 and reaches 150 °C
(302 °F) when the speed reaches Mach 2.4. As shown in Table 2, for a
hypersonic aircraft whose speed higher than 4.0 Mach, the skin temperature
may exceed 370 oC.
The data in Table 2 clearly suggests that high strength and high
heat-resistant composites should be applied for current supersonic aircrafts
that fly at speeds in the range of 2.5-4.0 Mach. For example, the SR-71
Blackbird jet could fly continuously at Mach 3.1 with the temperature of
some parts reaching or even exceeding 315 °C (600°F). Most recent
high-temperature carbon fiber reinforced polymers (CFRP), such as
graphite fiber/PMR-15 and graphite fiber/PMR-11-55, can properly
function for thousands of hours under temperatures between 290 and
Materials Selection in Design of Structures Frontiers in Aerospace Science, Vol. 1 457
345 °C. Due to this reason, both SCTAs and military aircrafts are made of
CFRP to achieve a higher fuel efficiency.
The data in Table 3 provides a guideline for designers to select best CFRP
for supersonic aircrafts under different operating conditions. For example,
the CFRP with high elastic modulus should be selected for high Mach
speeds.
Table 3. Mechanical properties of three classes of CFRP composites [30].
Specific materials for each part of the SST aircraft are now selected based
on the operating conditions of the parts in supersonic aircrafts and the
temperature resistance of the materials. Fig. (5) indicates that 2650 Al alloy,
a cost effective material which can resist long-term creep at temperatures
ranging from 100 to 130 °C, is the ideal candidate for the fuselage skin of
the SST aircraft. For example, the application of 2650-T8 alloy showed that
it did not deform by more than 0.1% under a 150 MPa stress during a period
of its 60,000 h service life [34].
As shown in Fig. (5), different parts of the aircraft are heated up to different
temperatures during flight, therefore they have to be made of different
materials. Specifically, the temperature at the nose and around wings are
significantly higher than that along the fuselage. It is then decided that
high-temperature CFRP (such as graphite fiber/PMR-15, graphite fiber/
458 Frontiers in Aerospace Science, Vol. 1 Zainul Huda
PMR-11-55) should be used for making the fuselage and advanced CFRP
with better heat resistance should be used for the nose and around the wings.
A good candidate is carbon fiber reinforced epoxy, which possesses
excellent specific strength (450 GPa/kg/m3) (see Fig. 6) [7], and is
recommended for speeds in the range of 2.5-3.5 Mach.
Fig. (5). Variation and distribution of skin temperatures for Mach 2.2 and 2.4 SST aircrafts.
Fig. (6). Density normalized stiffness – strength curve for lightweight aerospace materials [7].
460 Frontiers in Aerospace Science, Vol. 1 Zainul Huda
The analysis of data in Table 4 indicates that stainless steel should be the
optimum choice for the specific application. However, with its outstanding
corrosion-resistance capability and low density, titanium alloy should be
preferred if the supersonic aircraft speed is below 3.0 Mach.
Materials Selection in Design of Structures Frontiers in Aerospace Science, Vol. 1 461
Material σy/ρ Abs Rel. (KIC/σy)2 mm E1/3/ρ Abs. Temp limit Cost (£/ton) Abs Overall
o
=A Abs Rel. = B Rel. = C ( C) Abs Rel Rel =E Rating
=D [10A+10
B+10
C+20
D+(1E)]/51
Al alloy 1 130 0.64 16.5 1.00 1.5 1.0 150 0.38 13,100 0.134 0.683
Al alloy 2 204 1.00 2.1 0.13 1.5 1.0 15 0.38 14,000 0.142 0.583
Ti alloy 196 0.96 4.6 0.27 1.0 0.7 30 0.75 98,000 1.000 0.674
Stainless 115 0.56 12.3 0.75 0.7 0.5 40 1.00 10,900 0.111 0.764
steel
Fig 8. Temperature variation within the military aircraft wings during a supersonic flight [2].
462 Frontiers in Aerospace Science, Vol. 1 Zainul Huda
In conclusion, the best scheme of selecting materials for the military aircraft
wings should be a combination of stainless steel (or composite), titanium
alloy, and high-temperature aluminum alloy (2650-T8). This
materials-selection recommendation is made based on the wide
temperature range to which the wings may be heated up during the
supersonic flight (Fig. 8).
Table 5. A new materials selection guide for subsonic and supersonic aircraft structural
components.
Current Subsonic with Current Supersonic Current Supersonic Future Hypersonic
Cruise Speed up to 1.0 with Cruise Speed up with Cruise Speed: with Cruise Speed
Mach to 2.0 Mach 2.0-4.0 Mach above 4.0 Mach
Possible Material(s) to Possible Material(s) to Possible Material(s) to Possible
be Selected be Selected be Selected Material(s) to be
Selected
Fuselage: Fuselage: Fuselage: Fuselage:
Al: 2024-T4; 7075-T6; Al: 2090-T651; CFRP: PMR-15, PMR- Polyimides, BMIs,
6061-T6, Al-Li alloys 7075-T6; 8090-T651 11-55, or similar high- CEs, Graphite fiber-
(nose) modulus composites; or phthalonitrile, or
New type 2-mm Al-Zn- FGM (ceramic-
Mg-Sc-Zr alloy metal composite)
Materials Selection in Design of Structures Frontiers in Aerospace Science, Vol. 1 463
(Table 5) contd.....
CONFLICT OF INTEREST
The author confirms that author have no conflict of interest to declare for
this publication.
464 Frontiers in Aerospace Science, Vol. 1 Zainul Huda
ACKNOWLEDGEMENTS
Declared none.
REFERENCES
[19] R.W. Hertzberg, R.P. Vinci, and J.L. Hertzberg, Deformation and Fracture
Mechanics of Engineering Materials. 5th ed John Wiley & Sons Inc: New
York, 2012.
[20] K.G. Budinski, and M.K. Budinski, Engineering Materials: Properties and
Selection. 8th ed Pearson Prentice Hall: New Jersey, 2005.
[21] G. Design, "Materials Decision Support (software)", Cambridge. Available
from: http://www.grantadesign.com, 2012.
[22] M.F. Ashby, Materials Selection in Mechanical Design. 4th ed
Butterworth-Heinemann: Burlington, 2011.
[23] L. Yuan-pei, M. Qing-chun, L. Feng, X. Guang-xing, and G. Zhi-dong,
"Aircraft design material- selection method based on MAUT theory", J.
Aeronaut. Mater., vol. 30, no. 3, pp. 88- 94, 2010.
[24] E. Norman, "Expert systems in the Design Process", Studies in Design
Education Craftand Technology, vol. 20, no. 1, 1987.
[25] Y. LAN, Z. Guan, Q. Jiao, and G. Xu, "A web-based computer-aided
material-selection system for aircraft design", J. Comput. (Taipei), vol. 6, no. 5,
pp. 976-983, 2011.
[26] J. Beddoes, "Design of solution heat treatments for aerospace alloys",
Canadian Metallurgical Quarterly, vol. 20, no. 3, pp. 215-221, 2011.
[27] M. Tajally, Z. Huda, and H.H. Masjuki, "A comparative analysis of tensile and
impact-toughness behavior of cold-worked and annealed 7075 aluminum
alloy", Int. J. Impact Eng., vol. 37, pp. 425- 432, 2010.
[http://dx.doi.org/10.1016/j.ijimpeng.2009.08.009]
[28] R.F. Simenz, and M.K. Guess, "Structural Aluminum Materials for 1980's", J.
Aircr., vol. 17, no. 7, pp. 514-520, 1980.
[http://dx.doi.org/10.2514/3.57933]
[29] AIRBUS, "Innovative Materials: Setting the Innovation Standards", Available
from: http://www.airbus.com/innovation/proven-concepts/in-design/
innovation-materials/, 2014.
Materials Selection in Design of Structures Frontiers in Aerospace Science, Vol. 1 467
[30] W.D. Calister, and D.G. Rethwisch, Materials Science and Engineering. 8
John Wiley & Sons Inc: Singapore, 2011.
[31] Y. Deng, Z. Yin, J. Duan, K. Zhao, B. Tang, and Z. He, "Evolution of
microstructure and properties in a new type 2 mm Al–Zn–Mg–Sc–Zr alloy
sheet", J. Alloys Comp., pp. 118-126, 2012.
[32] M.J. Donachie Jr, TITANIUM: A Technical Guide. ASM International: Metals
Park, OH, 1988.
[33] S.I. Jaffery, and P.T. Mativenga, "Assessment of the machinability of
Ti-6Al-4V alloy using the wear map approach", Int. J. Adv. Manuf. Tech., vol.
40, no. 7-8, pp. 687-696, 2009.
[34] J. Majimel, M.J. Casanove, and G. MolA(c)nat, "A 2xxx aluminum alloy crept
at medium temperature: role of thermal activation on dislocation mechanisms",
Mater. Sci. Eng. A, vol. 380, no. 1-2, pp. 110-116, 2004.
[http://dx.doi.org/10.1016/j.msea.2004.03.050]
[35] V.P. McConnell, "Resins for the Hot Zone, Part II: BMIs, CEs, benzoxazines
and phthalonitriles", Available from: http://www.compositesworld.com/articles
/resins-for-the-hot-zone-part-ii--mis-ces-benzoxazines-and-phthalonitriles,
2009.
[36] H.N. Jones, and T.M. Keller,o High-Temperature Tensile Properties of
Graphite Fiber- Phthalonitrile Composites ??, Tech. Report. United States
Naval Research Laboratory: Washington, DC, 2002.
468 Frontiers in Aerospace Science, 2016, Vol. 1, 468-496
CHAPTER 12
*Corresponding author Omid Gohardani: Springs of Dreams Corporation, 340 East 1st Street, No. 8,
Tustin, California 92781, U.S.A; E-mail: omid.gohardani@springsofdreams.org.
NOMENCLATURE
ρ Density
ς Electrical resistivity
σmax Tensile strength
dt Nanotube diameter
k Thermal conductivity
m, n Integers
AR Aspect ratio
Ch Chiral vector
E Young’s modulus
INTRODUCTION
Carbon nanotubes (CNTs) have since their relatively recent discovery, been
employed in a multitude of different scientific applications such as sensing
[1], mechanical systems [2], energy storage [3, 4], biological applications
[5, 6], and field emission [7, 8]. The expanding usage of composite
materials on current commercial and military aircraft has in this respect
highlighted these materials as viable candidates that can sustain different
operational requirements in aeronautics and astronautics.
1.80%
and "aerospace" from total CNT publications
1.60%
1.40%
1.20%
CNT
1.00% CNT and aerospace
CNT and aircraft
0.80%
0.60%
0.40%
0.20%
0.00%
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Year
Fig. (1). CNT publications with the keywords “aerospace” and “aircraft”, between the years 2000 and
2013. (Source: Adapted from [12]).
Applications of Carbon Nanotubes Frontiers in Aerospace Science, Vol. 1 473
Fig. (2). Selected milestones of CNTs and their implementations in aerospace sciences. (Source:
Adapted from [12]).
474 Frontiers in Aerospace Science, Vol. 1 Omid Gohardani
The large surface area for a given volume and reinforcement effects caused
by a high aspect ratio [32] therefore, render nanocomposites suitable for
aeronautical applications. The influence of CNTs on the weight, range and
fuel consumption of commercial aircraft, was in a pioneering study
examined by O’Donnell [33] and colleagues [34, 35]. In these simulations,
Applications of Carbon Nanotubes Frontiers in Aerospace Science, Vol. 1 475
Fig. (3). Aircraft lightning zones as designated by SAE aerospace recommended practices 5414.
(Source: Adapted from [12]).
For military aircraft, stealth technology is crucial for survival of the system.
Stealth performance is often represented by the radar cross section (RCS),
which is described as the strength of the radar signal backscattered from an
object for an incident electromagnetic wave and equally a measure of the
amount of reflected electromagnetic wave from an incident electromagnetic
wave [54]. The RCS can be minimized by modification of the external
features of the target, utilizing radar absorbing materials [55] and radar
absorbing structures. In one such study, Kim and Lee [54] have utilized
CNTs to achieve absorption of incident electromagnetic waves up to 90% in
the X-band frequency.
Fig. (4). NASA’s Morphing Airplane. Source: NASA Dryden Flight Research Center Photo
Collection.
Applications of Carbon Nanotubes Frontiers in Aerospace Science, Vol. 1 479
Future rotorcraft and helicopters may benefit from CNTs, for structural
damping applications. Studies carried out by Liu et al. [57] and Bakis and
Wang [58] on CNTs embedded in polymer resins through
micromechanics/molecular dynamics models exhibit that the interfacial
shear strength is ~25 times the value of non-functionalized CNTs, when
approximately 1% of the carbon atoms in CNTs are covalently bonded to
the epoxy resin.
Recent Studies
An extensive research effort has been carried out by Gohardani [41] since
2011, on the prospects of CNTs in aeronautics. A review study conducted
on erosion modeling in particular with materials reinforced with CNTs,
exhibited such modeling to be highly complex in nature [59]. In a different
study, empirical liquid erosion experiments conducted on polymeric matrix
composites reinforced with CNTs, exhibited nominally similar damage
threshold velocities as the non-reinforced composites. The liquid erosion
experiments were conducted upon utilizing Cavendish Laboratory's
Multiple Impact Jet Apparatus (MIJA) [60]. Based on the aforementioned
study, the CNT dispersion level did not appear to result in significantly
higher damage threshold velocities (DTVs), confining the aerospace
applications of the candidate materials to UAVs, with a cruise speed less
than Mach 0.4, such as the generic unmanned air vehicle (GUAV) [61].
Given the numerous future application areas of UAVs, the need for icing
protection systems has recently been more emphasized. Granted that
UAVs exhibit a lower climb rate than other military, rotary- and
ifxed-wing aircraft, ice accretion is more prone to ensue in destabilization
of the aircraft [66]. For this purpose, many different methods for icing
mitigation for UAVs have been undertaken by researchers in recent years
[67, 68].
densities in line with the limited available power for UAVs and traditional
coating application methods, enabling the retrofitting of this system onto
existing airframes.
Recent Studies
The Space Elevator concept has been one of the envisioned ideas for space
transportation that has attracted a considerable amount of attention, since
the discovery of CNTs [12, 71-81]. A Space Elevator is essentially a
physical connection from Earth’s surface to Geostationary Earth Orbit
(GEO) located approximately at an altitude of 36,000 km, by means of a
proposed cable made of CNTs, on which magnetic levitating vehicles travel
[12].
There are many benefits associated with the realization of the Space
Elevator. Upon its establishment, the Space Elevator may reduce the
current high costs associated with placement of small payloads into Low
Earth Orbit (LEO), enabling equipment to be brought back to Earth and a
rapid cargo launch [82]. Equally, numerous challenges have been identified
with this concept consisting of susceptibility to lightning strikes, potential
meteorite damages, LEO cable impacts, atomic oxygen damage, and micro
scale defects [12, 74, 77].
482 Frontiers in Aerospace Science, Vol. 1 Omid Gohardani
Recent Studies
In a recent study [86], polyimide (PI) infiltration into entangled cup-stacked
CNT (CSCNT) sheets preserved the beneficial properties of CNTs, while
preventing agglomeration and enabling CNT patterning. The obtained
CNI-PI films exhibited low sheet resistivity values in both lateral and
vertical direction, in line with as-grown CNT sheets, while preserving a
high conductivity during thermal and mechanical manipulations. The
developed CNT-PI films are hence rendered suitable as an alternative for
electrostatic discharge (ESD) protection layer in spacecraft thermal blanket,
given their durable exhibited character.
Fig. (5). An overview of the benefits of using CNTs in commercial aircraft, military aircraft,
rotorcraft, and space applications. (Source: Adapted from [12]).
484 Frontiers in Aerospace Science, Vol. 1 Omid Gohardani
(Table 3) contd.....
CONFLICT OF INTEREST
The author confirms that author have no conflict of interest to declare for
this publication.
ACKNOWLEDGEMENTS
Any opinions, findings and conclusions expressed in this article are those of
the author only and do not necessarily reflect the corporate views of any
organization(s) affiliated with the author. Furthermore, the author is not
aware of any affiliations, memberships, funding, or financial holdings that
might be perceived as affecting the objectivity of this study.
REFERENCES
[1] J. Gooding, "Nanostructuring electrodes with carbon nanotubes: A review on
electrochemistry and applications for sensing", Electrochim. Acta, vol. 50, no.
15, pp. 3049-3060, 2005.
[http://dx.doi.org/10.1016/j.electacta.2004.08.052]
[http://dx.doi.org/10.1002/adma.200501851]
[6] N.W. Kam, M. O’Connell, J.A. Wisdom, H. Dai, and H. Dai, "Carbon
nanotubes as multifunctional biological transporters and near-infrared agents
for selective cancer cell destruction", Proc. Natl. Acad. Sci. USA, vol. 102, no.
33, pp. 11600-11605, 2005.
[http://dx.doi.org/10.1073/pnas.0502680102] [PMID: 16087878]
[7] S. Fan, M.G. Chapline, N.R. Franklin, T.W. Tombler, A.M. Cassell, and H.
Dai, "Self-oriented regular arrays of carbon nanotubes and their field emission
properties", Science, vol. 283, no. 5401, pp. 512- 514, 1999.
[http://dx.doi.org/10.1126/science.283.5401.512] [PMID: 9915692]
[9] V. Popov, "Carbon nanotubes: properties and application", Mater. Sci. Eng.
Rep., vol. 43, no. 3, pp. 61-102, 2004.
[http://dx.doi.org/10.1016/j.mser.2003.10.001]
[10] R.H. Baughman, A.A. Zakhidov, and W.A. de Heer, "Carbon nanotubes--the
488 Frontiers in Aerospace Science, Vol. 1 Omid Gohardani
route toward applications", Science, vol. 297, no. 5582, pp. 787-792,
2002.
[http://dx.doi.org/10.1126/science.1060928] [PMID: 12161643]
[14] M. Al-Saleh, and U. Sundararaj, "A review of vapor grown carbon nanofiber/
polymer conductive composites", Carbon, vol. 47, pp. 2-22, 2009.
[http://dx.doi.org/10.1016/j.carbon.2008.09.039]
[16] P.M. Ajayan, "Nanotubes from Carbon", Chem. Rev., vol. 99, no. 7, pp.
1787-1800, 1999.
|[http://dx.doi.org/10.1021/cr970102g] [PMID: 11849010]
[17] E. Thostenson, Z. Ren, and T. Chou, "Advances in the science and technology
of carbon nanotubes and their composites: A review", Compos. Sci. Technol.,
vol. 61, pp. 1899-1912, 2001.
[http://dx.doi.org/10.1016/S0266-3538(01)00094-X]
[18] P. Ehrburger, "Carbon nanotubes: the present state", Carbon, vol. 40, no. 10,
Applications of Carbon Nanotubes Frontiers in Aerospace Science, Vol. 1 489
[19] F. Komarov, and A. Mironov, "Carbon nanotubes: present and future", Phys
Chem Solid State, vol. 5, no. 3, pp. 411-429, 2004.
[23] "N. Lange, “The structure of graphite filaments", Z. Kristallogr., vol. 111, pp.
24-34, 1958.
[26] S. Iijima, "Helical microtubules of graphitic carbon", Nature, vol. 354, pp.
56-58, 1991. [http://dx.doi.org/10.1038/354056a0]
[27] M. Monthioux, and V. Kuznetsov, "Who should be given the credit for the
discovery of carbon nanotubes?", Carbon, vol. 44, pp. 1621-1623, 2006.
[http://dx.doi.org/10.1016/j.carbon.2006.03.019]
[37] R. Van Noorden, "Chemistry: The trials of new carbon", Nature, vol. 469, no.
7328, pp. 14-16, 2011.
[http://dx.doi.org/10.1038/469014a] [PMID: 21209637]
[38] T. Kukowski, "Lightweight CNT cables for aerospace", In: NESC NDE
technology assessment NASA NDI workshop. Johnson Space Center Houston:
TX, 2012.
[41] O. Gohardani, The influence of erosion and wear on the accretion and
adhesion of ice for nano reinforced polymeric composites used in aeronautics,
Ph.D. thesis, Cranifeld University, United Kingdom, 2011�
[42] A. Amendola, and G. Mingione, "On the problem of icing for modern civil
aircraft", Air Space Eur.,vol. 3, no. 3–4, pp. 214-217, 2001.
[http://dx.doi.org/10.1016/S1290-0958(01)90098-X]
[43] H. Chu, Z. Zhang, Y. Liu, and J. Leng, "Self-heating fiber reinforced polymer
composite using meso/macropore carbon nanotube paper and its application in
deicing", Carbon, vol. 66, pp. 154-163, 2014.
[http://dx.doi.org/10.1016/j.carbon.2013.08.053]
[44] W. Zhao, M. Li, Z. Zhang, and H. Peng, "Carbon nanotube based composites
film heater for de-icing application", In: 14th European conference on
composite materials, Budapest, Hungary, Paper ID: 545-ECCM14, 2010.
492 Frontiers in Aerospace Science, Vol. 1 Omid Gohardani
[50] C. Wu, H. Lu, Y. Liu, and J. Leng, "Study of carbon nanotubes/short carbon
fiber nanocomposites for lightning strike protection", Proceedings of the SPIE
7644, behavior and mechanics of multifunctional materials and composites,
2010
[53] A. Sehra, and J. Shin, "Revolutionary propulsion systems for 21st century
aviation", Technical report. National Aeronautics and Space Administration
(NASA), NASA/TM-2003-212615, 2003.
Applications of Carbon Nanotubes Frontiers in Aerospace Science, Vol. 1 493
[54] P. Kim, and D. Lee, "Composite sandwich constructions for absorbing the
electro- magnetic waves", Compos. Struct., vol. 87, no. 2, pp. 161-167, 2008.
[http://dx.doi.org/10.1016/j.compstruct.2008.05.015]
[55] K. Vinoy, and R. Jha, Radar absorbing materials: from theory to design and
characterization. Vol. 1. Kluwer Academic Publishers London: United
Kingdom, 1996.
[http://dx.doi.org/10.1007/978-1-4613-0473-9]
[56] B. Rosenberg, "Product focus: wire and cable", Avionics Magazine. Available
from: http://www.aviationtoday.com/av/military/Product-Focus-Wire-and-
Cable-79536.html, 2013.
[57] A. Liu, C. Bakis, and K. Wang, "Multiscale damping model for polymeric
composites containing carbon nanotube ropes", J. Compos. Mater., vol. 44, pp.
2301-2323, 2010.
[http://dx.doi.org/10.1177/0021998310365176]
[61] A.S. Gohardani, and O. Gohardani, "Ceramic engine considerations for future
aerospace propulsion", Aircr. Eng. Aerosp. Technol., vol. 84, no. 2, pp. 75-86,
2012.
[http://dx.doi.org/10.1108/00022661211207884]
[63] O. Gohardani, and D.W. Hammond, "Ice adhesion to pristine and eroded
polymer matrix composites reinforced with carbon nanotubes for potential
usage on future aircraft", Cold Reg. Sci. Technol., vol. 96, pp. 8-16, 2013.
[http://dx.doi.org/10.1016/j.coldregions.2013.09.005]
[67] "Unmanned aerial vehicle reliability study", Technical report. Ofifce of the.
Secretary of Defense, 2003.
[69] "Battelle’s anti-icing system for remotely piloted aircraft (RPAs)", Available
from: http://www.battelle.org, [accessed 15.09.13].
[70] K.L. Sørensen, A.S. Helland, and T.A. Johansen, "Carbon nanomaterial- based
wing temperature control system for in-flight anti-icing and de-icing of
unmanned aerial vehicles", IEEE Aerospace Conference, Big Sky, 2015.
[http://dx.doi.org/10.1109/AERO.2015.7119206]
[71] K. Tsiolkovsky, Grezy o Zemle i Nebe (i) Na Veste (Speculations about Earth
and Sky and on Vesta). Izd-vo AN SSSR: Moscow, 1895.
[74] B. Edwards, "Design and deployment of a space elevator", Acta Astronaut., vol.
47, no. 10, pp. 735- 744, 2000.
[http://dx.doi.org/10.1016/S0094-5765(00)00111-9]
[76] A. Bolonkin, "Non-rocket earth-moon transport system", Adv. Space Res., vol.
31, no. 11, pp. 2485- 2490, 2003.
[http://dx.doi.org/10.1016/S0273-1177(03)00565-9]
[77] N. Pugno, "On the strength of the carbon nanotube-based space elevator cable:
from nanomechanics to megamechanics", J. Phys. Condens. Matter, vol. 18,
no. 33, pp. S1971-S1990, 2006.
[http://dx.doi.org/10.1088/0953-8984/18/33/S14]
[78] J.M. Wörle-Knirsch, K. Pulskamp, and H.F. Krug, "Oops they did it again!
Carbon nanotubes hoax scientists in viability assays", Nano Lett., vol. 6, no. 6,
pp. 1261-1268, 2006.
[http://dx.doi.org/10.1021/nl060177c] [PMID: 16771591]
[79] B. Iannotta, "Nanotubes lift hope for space elevator", Aerosp. Am., vol. 44, no.
3, pp. 30-35, 2006.
[80] M. Avnet, "The space elevator in the context of current space exploration
policy", Space Policy, vol. 22, no. 2, pp. 133-139, 2006.
[http://dx.doi.org/10.1016/j.spacepol.2006.02.005]
[81] J. Dempsey, System and method for space elevator, US patent no. US
6,981,674 B1, 2006.
[84] S. Keller, and A. Zaghloul, "Carbon nanotube yarn and sheet antennas", In:
Nanotube superfiber materials – changing engineering design.. Elsevier,
William Andrew: Boston, 2014, pp. 749-787.
[http://dx.doi.org/10.1016/B978-1-4557-7863-8.00026-8]
[87] H. Tennent, J. Barber, and R. Hoch, Carbon fibrils, method for producing same
and adhesive compositions containing same, US patent no. 5,578,543, 1996.
Frontiers in Aerospace Science, 2016, Vol. 1, 497-514 497
CHAPTER 13
1
Institute of Production Engineering, Helmut Schmidt University, 22043
Hamburg, Germany
2
Polynt Composites Germany GmbH, Kieselstraße 2, 56357 Miehlen,
Germany
3
Composite Technology Center GmbH (An Airbus Company),
Airbusstraße 1, 21648 Stade, Germany
4
Institute of Polymer Materials and Plastics Engineering, TU Clausthal,
Agricolastrasse 6, 38678 Clausthal, Germany
Abstract: The current use of fuel efficient and environmentally friendly aircraft is
only possible by the development of innovative lightweight constructions and the use
of lightweight materials, such as carbon fiber reinforced plastics. With the rising
demand on fiber reinforced components in the aerospace industry new production
processes have been built up. However, current production technologies for
composites cause higher costs and obtain longer process cycle times in comparison
to the manufacturing processes of metals. Moreover raw materials, such as carbon
fibers and resin, and semi-finished products are very expensive. In contrast to this
and compared with other manufacturing technologies for fiber reinforced plastics
*Corresponding author Marc Fette: Institute of Production Engineering, Helmut Schmidt University,
Hamburg, Germany; E-mail: marc.fette@hsu-hh.de.
NOMENCLATURE
INTRODUCTION
In the recent decades, the demand for efficient and fuel-saving aircraft for
the commercial air traffic has led to the development of innovative
lightweight solutions and to an increasing use of modern lightweight
Tailored Carbon Fiber Reinforcements Frontiers in Aerospace Science, Vol. 1 499
The main objectives of this article are researches and analysis of the
potentials of this technology in reference to aerospace applications with
the requirements for secondary structures of commercial aircraft.
500 Frontiers in Aerospace Science, Vol. 1 Fette et al.
Fig. (1). Process cycle of the combination of pre-impregnated carbon fiber fabrics and carbon fiber
SMC.
Possible materials for reinforcements are carbon fibers, glass fibers and
natural fibers which can be combined with different thermoset matrix
materials, for example unsaturated polyester resin, vinyl ester resin or
epoxy resin. Besides it is possible to use carbon fiber recyclates,
especially as long fiber reinforcements in SMC formulations. Using
recycled carbon fibers made out of end-of-life cycle aircraft parts or
cutting scrap of the current aerospace production can close the ecological
value chain referring to future aircraft. These fibers can be final products
from recycling processes like pyrolysis. These carbon fibers keep up to 95
percent of the original mechanical properties and obtain a lower price
level than new carbon fibers. In this context, this technology realizes the
reuse of expensive and energy-intensive resources as reinforcement
materials for aircraft parts again. Possible areas of application can be
cabin and cargo structures. As a result the ecological sustainability and the
resource efficiency of aerospace production for composite aircraft
components will be improved.
In a first feasibility study the substitution of a cargo foot step for a current
commercial short and middle range aircraft, which is currently made of
titanium, has been analyzed. Lead time reduction, cost savings and weight
reduction as results of increased mechanical properties, especially tensile
strength, flexural strength, stiffness and impact strength, are the key
aspects for the planned optimization. Subsequently, the industrial
introduction of this manufacturing process and the substitution of other
aircraft components will follow. Therefore completely new possibilities
for aircraft production and a new generation of hybrid composite
structures for future aircraft will be created.
Experimental Procedures
Fiber content 25 %
Glass fiber reinforcement
Fiber length 25 mm
Flame-retardant components 66 %
Thermoset matrix components Curing system and process
6 %
additives
Thickening agents 1 %
Shrinkage 0.08 %
HUP 27- HUP 27- HUP 27- HUP 27- HUP 27-
GF, fiber GF, fiber CF, fiber CF, woven CF,
Inner layer content content content fabric unidirect-i
25%, 50%, 50%, onal fiber
Outer layer chopped chopped chopped reinforce-
fibers fibers fibers ment
7DEOH2 FRQWG
The materials with chopped fiber reinforcements are used as inert systems
for mechanical data base and for the outer systems. The materials with
continuous fiber fabrics or unidirectional fibers are used for the inner
layer, illustrated in Fig. (2). The mechanical properties determined for flat
Sheet Molding Compound plates are molded according to DIN EN 14598.
These plates have a thickness between 2.0 mm and 3.0 mm. The test
specimens are cut from the non-flow area of a sample plate with the
dimension of 250 mm x 120 mm and with molding conditions of 180
seconds at 145 °C. Afterwards the cut specimen is milled to the accurate
dimension according to the standard specification.
Fig. (2). Schematic illustration of outer and inner layers of the material samples.
DISCUSSION ON RESULTS
The results generated by the experimental investigations on the different
material combinations are shown in Figs. (3 to 5). The parameters
investigated are flexural modulus, flexural strength, tensile modulus,
tensile strength and impact strength. There are significant differences
between the measurements, especially in reference to the used type of
fiber and depending on the combination of the different fiber
reinforcements.
In general, the measurements indicate that higher fiber contents and the
use of chopped carbon fibers, instead of glass fibers, for Sheet Molding
Compound formulations can realize an improvement on flexural and
tensile modulus, illustrated in Fig. (3). Furthermore woven fabrics and
unidirectional layers of carbon fiber combined with Sheet Molding
Compounds tend to result in an increase of the mentioned mechanical
properties. Especially, HUP 27 with chopped carbon fibers (50 percent
fiber content) combined with an unidirectional carbon fiber layer or a
carbon fiber woven fabric can realize a significant improvement of
flexural and tensile modulus. In contrast, Fig. (4) shows that the flexural
Tailored Carbon Fiber Reinforcements Frontiers in Aerospace Science, Vol. 1 509
Fig. (3). Flexural and tensile modulus for different material combinations.
Fig. (4). Flexural and tensile strength for different material combinations.
Fig. (6). SEM pictures of the combination of carbon fiber SMC and unidirectional carbon fiber
fabrics.
512 Frontiers in Aerospace Science, Vol. 1 Fette et al.
CONCLUSION
CONFLICT OF INTEREST
The authors confirm that they have no conflict of interest to declare for
this publication.
Tailored Carbon Fiber Reinforcements Frontiers in Aerospace Science, Vol. 1 513
ACKNOWLEDGEMENTS
Declared none.
REFERENCES
[1] J.E. Composites, "Over-moulded organo sheets", JEC Composites, vol. 85, pp.
22-27, 2013.
[2] M. Cabrera-Rios, and J.M. Castro, An economical way of using carbon fibers
in sheet molding compound compression molding for automotive applications.
Polymer Composites 27, vol. 6, pp. 718-722, 2006.
[3] H.G. Kia, Sheet moulding compounds - science and technology. Hanser /
Gardener Publications, Inc., Ohio, USA.
[4] P. Stachel, "Carbon fibre reinforced SMC for automotive applications",
Proceedings of the 5th Automotive Seminar, Topic: SMC/BMC - New
challenges in Automotive, Landshut, 2006.
[6] C. Lauter, T. Tröster, B. Sköck-Hartmann, T. Gries, and M. Linke, Höchst
feste Multimaterialsysteme aus Stahl und Faserverbundkunststoffen. VDI
Konstruktion: Ingenieur-Werkstoffe, Springer-VDI-Verlag, Düsseldorf, vol.
11-12/2010, pp. IW 8 - 9, 2010.
[7] M. Oldenbo, S.P. Fernberg, and L.A. Berglund, "Mechanical behaviour of
SMC composites with toughening and low density additives", Compos., Part
A Appl. Sci. Manuf., vol. 34, pp. 875-885, 2003.
[http://dx.doi.org/10.1016/S1359-835X(03)00155-6]
[8] S. Pimenta, and S.T. Pinho, "Recycling carbon fibre reinforced polymers for
structural applications: technology review and market outlook", Waste Manag.,
vol. 31, no. 2, pp. 378-392, 2011.
[http://dx.doi.org/10.1016/j.wasman.2010.09.019] [PMID: 20980138]
[9] E. Reuther, "Kohlefaser SMC für Strukturteile", In: Proceedings of the 7th
Internationale AVK-TV Tagung, A6-1 – A6-6, 2004.
[10] K. Steinbach, G.P. Ehnert, and K. Bieniek, "Neue Entwicklungen zur
514 Frontiers in Aerospace Science, Vol. 1 Fette et al.
CHAPTER 14
*Corresponding author Mohammad Rouhi: Concordia Center for Composites, Department of Mechanical and
Industrial Engineering, Concordia University, Montreal, Quebec, H3G 1M8 Canada; Tel/Fax: 514 - 848 4596;
E-mail: m.rouhi@gmail.com
INTRODUCTION
Hyer and Charette [25] have studied the structural improvement by using
curvilinear fibers in a flat plate with central hole subjected to tensile and
compressive loads. They concluded that in tension the curvilinear designs
lead to improved performance. In compression, however, the buckling
loads were not as high as for quasi-isotropic (QI) laminates. In a follow-up
work, Hyer and Lee [26] constrained the variation of the orientation angles
to increase buckling resistance while producing designs that could be
fabricated. As a result, and by combining both straight and curvilinear
fibers, the best buckling and failure results were achieved. Afterwards,
Nagendra et al. [27] explored that VS laminate made bu curved fibers
could improve the buckling load of the plate-with-hole problem addressed
by Hyer and Lee [26] up to 85% compared with the initial constant
stiffness (CS) design. The research on VS laminates was continued by
Gu¨rdal and Olmedo [28] in the early 1990’s at Virginia Tech (USA).
Followed by achieving promising results in terms of stiffness and buckling
generated by analytical and numerical methods [28-30], several VS
structures were fabricated [31] to validate the manufacturability of
tow-steered laminates. Subsequent testing [32] confirmed the improved
load-carrying capability of the VS panels over their CS counterparts. The
reason for the substantial improvements was attributed to the redistribution
of the load from the centre of the panels to the higher-stiffened edge
regions. The research was followed by a number of Prof. Gürdal’s research
group [33-38] at Delft University of Technology.
Besides plates and panels, VS cylinders were also investigated for their
potential improvement in structural performance. They have been shown
to be very promising in terms of the structural performance improvements
they can offer compared with their constant stiffness (CS) counterparts
Variable Stiffness Composite Structures Frontiers in Aerospace Science, Vol. 1 519
[3-5, 22, 23, 26]. Circular and conical cylinders have been extensively
studied for structural improvement via stiffness tailoring. Blom et al. [4]
could improve the bending-induced buckling load of a VS circular
cylinder up to 17% with respect to a baseline laminate at the same weight.
Khani et al. [22] could improve the buckling load of a VS cylinder about
24% over its best constant stiffness (CS) counterpart. The research works
on VS cylinders were not limited to those with circular cross sections.
Elliptical cylinders were also studied for their potential improvement in
axial buckling performance [5, 22, 39]. The axial buckling capacities for
elliptical cylinders were reported up to 30% for the cross sectional aspect
ratio of 0.7 [5]. As an objective different from the buckling performance,
White and Weaver [23] have studied the potential of stiffness tailoring to
achieve bend-free states in pressure vessels with elliptical cross sections.
path for VS plies so that the buckling capacity is maximum?”. To limit the
infinite number of variations of the fiber orientation angle for VS plies, a
piece-wise constant model is used where the cylinder is divided into a
number of narrow bands in which the fiber orientation angle is assumed to
be constant but different from its neighboring regions. This way, the
stiffness tailoring problem is reduced to finding the fiber orientation angle
in each narrow band shown in Fig. (1b). To further reduce the number of
design variables, the orientation angles in a number of equally spaced
narrow bands are assumed to be the design variables (Ti’s) as illustrated in
Fig. (1c). The orientation angles of the narrow bands in between these
regions are calculated by a linear interpolation between the values of the
orientation angles in these design points. Therefore, the orientation angles
in the equidistant narrow bands on the surface of the cylinder are
considered as design variables to be determined for maximum buckling
capacity of the cylinder:
where Fcr is the buckling capacity of the cylinder in terms of the design
variables (Ti’s) and NDV is the number of design variables. It should also
be noted that in case of a different loading scenario, the piece-wise
constant model for the continuous variation of the orientation angles may
need to be arranged in a different direction.
For any set of design variables, using the commercial software ABAQUS,
a finite element model is constructed and the structure is analyzed for
buckling. Table 1 shows the mechanical properties of the constituent
materials used in the composite laminate (AS4D/9310 Carbon/Epoxy).
Followed by a mesh convergence study, the cylinder is discretized into an
appropriate number of segments on the circumference and the length. The
S8R5 shell elements of ABAQUS were used in this study that have 8
Variable Stiffness Composite Structures Frontiers in Aerospace Science, Vol. 1 521
Fig. (1). (a) The composite cylinder subjected to bending-induced buckling, (b) the
piece-wise constant model to approximate the continuous variation of the orientation
angle in a VS ply, and (c) reducing the number of design variables to the values of the
orientation angles in some particular narrow bands.
Property AS4D/9310
EL (GPa) 134
ET (GPa) 7.71
GT (GPa) 2.76
νLT 0.301
νT 0.396
Vf 0.55
DESIGN OPTIMIZATION
by using both metamodel function and FEA. The average error was
calculated as a representation of the metamodel’s error.
RESULTS
Fig. (3) shows the buckling mode shapes of a composite cylinder with
quasi-isotropic (QI) laminate along with its optimized VS counterpart. The
stacking sequence for the cylinders is assumed to be [0/θ/90/-θ/-θ/90/θ/0]s
in which θ = 45° for QI cylinder and fiber steered (variable in
circumferential direction) for VS cylinder. For a cylinder with
length=diameter=18 in, the buckling capacity of VS cylinder (VS: Fcr =
11.62e5 Nm) is about 25% larger than its QI counterpart (QI: Fcr = 9.31e5
Nm). As Fig. (3) reveals, the VS cylinder exploits more area in carrying
the compressive load compared with the QI one. The more efficient use of
Variable Stiffness Composite Structures Frontiers in Aerospace Science, Vol. 1 525
QI VS
(a) (b)
Fig. (3). Bending-induced buckling mode shape of cylinders with (a) QI and (b) VS
design.
526 Frontiers in Aerospace Science, Vol. 1 Rouhi et al.
Fig. (4). Orientation angle distribution over the circumference of the QI and VS cylinders.
Keeping the thickness and radius of the cylinder unchanged, one may
change the length of the composite cylinder and design the optimum VS
laminate for each length to study the effect of the aspect ratio (L/R). Fig.
(7) shows the result of such a study in which the potential improvement of
the bending-induced buckling capacity of a VS cylinder increases with the
decrease of its aspect ratio (L/R). In other words, there is more room for
improvement in the buckling performance for shorter cylinders than the
long ones. The baseline to calculate the improvements in both cases are
the composite cylinders with QI laminate.
Variable Stiffness Composite Structures Frontiers in Aerospace Science, Vol. 1 527
Fig. (5). Normalized axial section force on the normalized circumferential distance for
QI and VS designs.
Fig. (6). The curved fiber paths on the developed surface of a VS ply.
528 Frontiers in Aerospace Science, Vol. 1 Rouhi et al.
Fig. (8). Bending-induced buckling improvement of VS over QI designs for different radii
(R).
Fig. (9). Maximum buckling load improvement of VS over QI designs in terms of the
percentage of the steered plies.
25% improvement of the buckling load for one direction, but at the
expense of 57% decrease in the buckling capacity if the bending load is
applied in the opposite direction. To simultaneously improve the buckling
capacity in both directions a multi-objective design optimization is needed.
The general multi-objective utility function that combines all individual
objectives can be expressed as the weighted exponential sum:
(3)
Fig. (10). Buckling performance of VS (black line) compared with QI (cyan dashed line)
composite cylinder in terms of the direction of applied bending moment (M). The VS
cylinders are optimized for bending-induced buckling performance in (a) only one
direction: 𝜔1 = 1, 𝜔2 = 0, and (b) two opposite directions: 𝜔1 = 𝜔2 = 0.5.
Fig. (13). The buckling mode shapes of (a) QI, (b) CS, and (c) VS elliptical composite
cylinders with b/a=0.4.
VS CS VS CS
90 90
θ1 60
θ2 60
30 30
0 0
1 2 3 4 1 2 3 4
Circumferential Location (S) Circumferential Location (S)
Fig. (14). Optimum distribution of the orientation angles (θ 1 and θ 2 ) over a quarter
circumference of an elliptical cylinder with the stacking sequence of [±θ 1 /±θ 2 ] s and
b/a=0.4.
Variable Stiffness Composite Structures Frontiers in Aerospace Science, Vol. 1 535
CONCLUSION
CONFLICT OF INTEREST
The authors confirm that they have no conflict of interest to declare for
this publication.
ACKNOWLEDGEMENTS
REFERENCES
[1] S.T. IJsselmuiden, M.M. Abdalla, and Z. Gu¨rdal, "Optimization of variable-
536 Frontiers in Aerospace Science, Vol. 1 Rouhi et al.
[5] M. Sun, and M. Hyer, "Use of material tailoring to improve buckling capacity
of elliptical composite cylinders", AIAA J., vol. 46, no. 3, pp. 770-782, 2008.
[http://dx.doi.org/10.2514/1.32495]
[6] D.O. Evans, M.M. Vaniglia, and P.C. Hopkins, "Fiber placement process
study", International SAMPE Symposium and Exhibition (Proceedings), vol.
Vol. 34, 1989pp. 1822-1833
CA, USA
[10] R.L. Pinckney, "Fabrication of the v-22 composite aft fuselage using
automated fiber placement", In: Proceedings of First NASA Advanced
Composites Technology Conference, 1991, pp. 385-397.
[11] R. Measom, and K. Sewell, "Fiber placement low-cost production for complex
composite structures", Annual Forum Proceedings of American Helicopter
Society, 1996 pp. 611-622.
[12] J.P. Martin, R.J. Langone, M.J. Pasanen, and J.A. Mondo, "Cost-effective,
automated equipment for advanced composite structure development and
production", International SAMPE Symposium and Exhibition (Proceedings),
vol. Vol. 42, 1997pp. 48-55
[13] M.J. Pasanen, J.P. Martin, R.J. Langone, and J.A. Mondo, "Advanced
composite fiber placement: process to application", In: Proceedings of the
Japan International Sampe Symposium. Japan Chapter of Sampe, 1997, pp.
1055-1060.
[15] D.O. Evans, "Fiber placement", In: D.B. Miracle, S.L. Donaldson, Eds., ASM
Int., vol. 21. 2001, pp. 477-479.ASM Handbook
[16] D.A. McCarville, J.C. Guzman, and D.M Rotter, "Automated material
placement: 2008 industry overview", Proceedings of International SAMPE
Symposium and Exhibition, 2008
[17] B. Morey, "Automating composites fabrication", Manuf. Eng., vol. 140, no. 4.
[19] S.B. Biggers, and S.S. Pageau, "Shear buckling response of tailored composite
plates", AIAA J., vol. 32, no. 5, pp. 1100-1103, 1994.
[http://dx.doi.org/10.2514/3.12107]
[21] G. Raju, Z. Wu, and P.M. Weaver, "Buckling and postbuckling of variable
angle tow composite plates under in-plane shear loading", Int. J. Solids Struct.,
vol. 58, pp. 270-287, 2015.
[http://dx.doi.org/10.1016/j.ijsolstr.2015.01.011]
[23] S. White, and P. Weaver, "Bend-free shells under uniform pressure with
variable-angle tow derived anisotropy", Compos. Struct., vol. 94, no. 11, pp.
3207-3214, 2012. [http://dx.doi.org/10.1016/j.compstruct.2012.05.009]
[26] M.W. Hyer, and R.F. Charette, "Use of curvilinear fiber format in composite
structure design", AIAA J., vol. 29, no. 6, pp. 1011-1015, 1991.
[http://dx.doi.org/10.2514/3.10697]
[30] B.F. Tatting, Analysis and design of variable stiffness composite cylinders,
Ph.D. thesis, Virg. Tech, Blacksburg, Virginia, 1998.
[31] K.C. Wu, and Z. Gu¨rdal, "Thermal testing of tow-placed variable stiffness
panels", In: Proceedings of the 42nd AIAA/ASME/ASCE/AHS/ASC structures,
structural dynamics and materials (SDM) conference, Seattle, WA, 2001.
[http://dx.doi.org/10.2514/6.2001-1190]
[38] A. Khani, Optimum design of steered fibre composite cylinders with arbitrary
cross-sections, Ph.D. thesis, Delft University of Technology (2013).
[40] Pawtucket, RI, USA, ABAQUS Version 6.11 User’s Manual. 2011.
[44] N.V. Queipo, R.T. Haftka, W. Shyy, T. Goel, R. Vaidyanathan, and K.P. Tucker,
"Surrogate-based analysis and optimization", Prog. Aerosp. Sci., vol. 41, no. 1,
pp. 1-28, 2005.
[http://dx.doi.org/10.1016/j.paerosci.2005.02.001]
[51] M. Rouhi, H. Ghayoor, S.V. Hoa, and M. Hojjati, "The effect of the percentage
of steered plies on the bending-induced buckling performance of a variable
stiffness composite cylinder", Science and Engineering of Composite Materials,
vol. 22, no. 2, pp. 149-156, 2015.
[http://dx.doi.org/10.1515/secm-2014-0258]
CHAPTER 15
*Corresponding author Tracie Prater: NASA Marshall Space Flight Center, Huntsville, AL, 35811; E-mail:
tracie.prater@nasa.gov.
FSW is employed to join structural components in Delta IV, Atlas V, and Falcon IX
rockets as well as NASA’s Orion Crew Exploration Vehicle and Space Launch
System. Currently, FSW researchers are interested in extending the application of
the process to new materials which are difficult to weld using conventional fusion
techniques, such as MMCs. Rapid wear of the welding tool in FSW of MMCs is a
consequence of the large discrepancy in hardness between the steel tool and the
reinforcement material. This chapter summarizes the challenges encountered when
joining MMCs to themselves or to other materials in structures. Specific attention is
paid to the influence of the process variables for FSW on the wear process. A
phenomenological model of the wear process was established based on the rotating
plug model of FSW. The effectiveness of tool materials with high hardness (e.g.
Tungsten Carbide, high speed steel, and tools with diamond coatings) in resisting
abrasive wear is also considered. In-process force, torque, and vibration signals are
analyzed to determine the feasibility of in situ monitoring of tool shape changes as
a result of wear. One advantage of this model is that its successful implementation
would eliminate the need for off-line evaluation of tool condition during joining.
Monitoring, controlling, and reducing tool wear in FSW of MMCs are critical to
full application of these materials in aerospace structures where they would be of
most benefit. The work presented in this chapter can be further extended for
machining of MMCs, where the wear of the tool materials is also a limiting factor.
NOMENCLATURE
FSW Friction Stir Welding
MMC Metal Matrix Composite
RPM Rotations per minute
SiC Silicon Carbide
ℓ Length of joint (inches)
𝜈 Traverse rate (inches per minute)
𝜔 Rotation rate (rotations per minute)
𝑊 Wear experienced by the tool during welding
544 Frontiers in Aerospace Science, Vol. 1 Tracie Prater
INTRODUCTION
The reliance on ballistic techniques in launch architectures makes mini-
mization of weight one of the most preeminent issues in spacecraft design.
For any launch vehicle, cargo weights only comprise a small portion of
the vehicle’s weight at launch. Structural components and fuel account for
the vast majority of the launch weight of a rocket. In order to improve the
structural efficiency of a vehicle, the vehicle’s dry weight can be reduced
and the weight reduction represents a commensurate gain in cargo
capacity.
Fig. (1). The FSW process. The tool consists of a cylindrical pin that penetrates the workpiece
material and a larger diameter shoulder which rests on the surface of the material. In the FSW
process, the tool rotates at an angular speed 𝜔 while advancing through the material at a traverse
speed v. During its traversal, plasticized material is picked up on its advancing side and is
deposited on its retreating side. Materials cools and consolidates to form a welded region.
546 Frontiers in Aerospace Science, Vol. 1 Tracie Prater
PROBLEM DESCRIPTION
Large defects present in the weld result in reduced mechanical properties.
For MMCs, it is therefore critical to reduce wear and preserve the tool shape
that will facilitate material stirring and prevent the formation of such
defects. To avoid weld-related structural failures, significant investments
have been made by NASA and its commercial partners on post-process
inspection and non-destructive testing techniques (such as ultrasonics and
X-rays) to inspect welds of flight hardware. The non-destructive evaluation
results and mechanical tests require parameterization studies. As part of
weld development, a set of experiments is designed to determine a window
of welding parameters that can produce joints satisfying design and safety
criteria. Even though many parameterization studies about FSW of MMCs
have demonstrated that it is possible to generate an acceptable operating
window of parameters for virtually any tool/workpiece combination, these
studies did not specifically consider defects resulting from the wear
mechanisms that affect the system [5, 6]. Even if parameters used to
produce the MMC joint fall in the operating window, there is still a
possibility that a defect may be created over time as the tool loses volume
and the workpiece undergoes a corresponding reduction in the flow of
plasticized material. Therefore a better control of the wear process during
joining is required to fully guarantee the successful integration of MMCs
into aerospace structures. Measures of mitigating wear may include careful
selection of process parameters (rotation speed, traverse rate, length of weld,
tilt angle, etc.) and material properties (percentage reinforcement, type of
reinforcement, particle size, and tool material). In cases of longer welds or
scenarios in which coatings fail to guard against wear, wear maybe
inevitable and in-process monitoring becomes essential. As emphasized in
the previous discussion, in process wear detection is often synonymous
with in-process fault detection due to the loss of tool features which
facilitate material stirring. Techniques used to detect wear in this scenario
548 Frontiers in Aerospace Science, Vol. 1 Tracie Prater
The first key to unraveling the tool wear problem in FSW of MMCs is to
understand the fundamental physics which underlie the wear process. A
study of the variation of wear with FSW process parameters can provide
some initial insight. The experiments for the preliminary wear study
conducted in this work comprise an L27 orthogonal Taguchi array with
three factors (rotation rate 𝜔, traverse speed v, and distance welded l) at
three levels (Table 1). The deterioration in the cross-sectional area of the
tool pre and post-experiment was quantified using imaging software. Tool
wear in FSW of MMCs displays a circumferentially symmetric pattern,
which means that the degradation measured in the cross-section of the
probe is representative of the volume lost by the entire probe. Fig. (2) shows
an overlay of close-up images of the tool probe taken after successive welds
of Al 359 reinforced with 20% Silicon Carbide particles. The characteristic
rounding of the tool that accompanies wear is undesirable because it limits
vertical stirring of material and frequently results in root flaw defects [4].
Fig. (2). Successive overlay of probe cross-sections for welds of Al 359/SiC/20p at 1500 RPM and
7 IPM traverse speed for 0, 8, 16, and 24 inches welded (from outside to inside) [7].
Metal Matrix Composites in Aerospace Structures Frontiers in Aerospace Science, Vol. 1 549
The wear data obtained from these experiments was used to construct a
multiple regression model (MRM) to predict the volume loss of the tool in
welding based on the values of the three major process parameters [7]. An
analog can be drawn between the MRM model and the machinability maps
used to determine optimal parameters for machining abrasive materials. In
both instances, we seek to develop parameters which minimize wear but
still produce a cut, hole or other feature (ex. weld) with acceptable quality.
The relationship derived from the regression analysis (Eq. 1) is strongly
correlated with the experimental data. The empirical predictive model was
tested on a validation set that consists of cases separate from those in the
original data set used to construct the regression model. The calculated and
observed wear values for the validation set closely agree with each other. A
direct comparison of observed and calculated values is presented in Table 2.
The model should thus be applicable to parameter sets different from those
used in its construction.
Table 2. Comparison of observed and predicted wear for model validation data set.
From the multivariate regression analysis results it can be seen that tool
wear in FSW of MMCs is directly proportional to the rotation speed and
welding distance but inversely proportional to the traverse speed. A
dimensionless group (𝜔𝑙/𝑣) is defined based on these parameters, which is
linearly correlated with wear as expressed in Eq. 2. Physically, Eq. (2)
means that each unit increase in the group 𝜔ℓ/𝜈 is accompanied by
a .0004% increase in wear [8]. This constant, like the constants in the
multiple regression model (Eq. 1), can be used to predict relative changes in
wear based on changes in process parameters. For instance, a 10,000-fold
increase in 𝜔ℓ/𝜈 would be associated with a 4% increase in wear
percentage.
𝜔ℓ
𝑊 = 0.0004 (2)
𝜈
the probe tip while the material loss at the interface between the probe and
the shoulder is minimal. The rotating plug model for material flow in FSW,
originally developed by Dr. Arthur Nunes of NASA’s Marshall Space Flight
Center, provides a framework for understanding the physical phenomena
which govern the relationships observed experimentally in FSW of MMCs
[9]. In this formulation, it is assumed that the probe is surrounded by a plug
of plasticized metal, also referred to as the “rotating plug”. The width of this
plug, δ, can be considered as a function of the process parameters ω and ν,
and the angular position θ in the x-y plane (Eq. 3). As shown in Figs. (3 and
4), the plug is symmetric about the y-axis and changes from δ= 𝛿0 at
θ = -π/2 to δ 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 at θ = π/2.
𝑣
𝛿 = 𝛿0 + (1 + 𝑠𝑖𝑛𝜃) (3)
𝜔
HBULGE
w
Exit Hole
5 mm
The analytical model can be slightly adjusted to account for the presence of
abrasive particles like the inclusions found in Al-MMCs. The model as it
applies to FSW of MMCs is shown in Fig. (4). For a given particle diameter
Metal Matrix Composites in Aerospace Structures Frontiers in Aerospace Science, Vol. 1 553
D, wear occurs only when the particle is able to span the width of the shear
zone and impinge on the tool surface, removing material as the tool rotates
past it. The analytical results calculated based this model are consistent with
the experimental data. Greater wear is produced by parameters which
correspond to thinner shear zones, while parameter sets associated with
larger δ values lead to less wear [9]. A piecewise condition for abrasion is
also suggested by the rotating plug model: wear only occurs at locations
where the radius of the particle exceeds the width of the shear zone
(𝐷/2 > 𝛿). This corollary implies that the wear is also influenced by
material properties, such as particle size and percentage of reinforcement
(number density of particles present in the material).
Fig. (4). Rotating plug model for FSW of MMCs. Pink shaded region is the rotating plug (with a
thickness 𝛿). The solid black circle depicts an abrasive particle in the workpiece material.
The rotating plug model presented in this chapter can be used to define the
region where scoring of the tool by abrasive is possible. The original plug
model predicts that the shear surface widens along the length of the tool
(Fig. 5), making the region near the tip of the probe the most susceptible to
554 Frontiers in Aerospace Science, Vol. 1 Tracie Prater
abrasion. The percent wear predicted by the model can be calculated with
equation 4, where D is the diameter of the reinforcement particle, P denotes
the proportion of reinforcement relative to the volume of the entire
workpiece, 𝜔 is the rotation speed, 𝑣 is the traverse speed, l is the length of
weld, R is the radius of the pin, and ∆𝐶𝑚𝑎𝑥 represents the maximum value of
the cutting arc (an angular area where the abrasion can occur). For a
detailed derivation of this equation please refer to references [9] and [10]. It
is assumed that the depth to which the particle impinges on the tool surface
is equal to the radius of the particle. The volume of material removed in a
single rotation is integrated along the length of the tool where abrasion may
occur. That length is estimated at about half of the probe length. The
deterioration in the width of the cutting arc ∆𝐶 as the distance from the
probe increases is also considered. Analytical results calculated using this
model are close (generally within 10 percent) to values measured in
experiments, with differences between predicted and observed values
[9, 10].
5𝐷∆𝐶𝑚𝑎𝑥 𝑃𝜔𝑙
% 𝑤𝑒𝑎𝑟 = (4)
24𝑅𝑣
Fig. (5). Growth of plug in the axial direction along length of the tool probe. Black circles represent
abrasive particles.
Metal Matrix Composites in Aerospace Structures Frontiers in Aerospace Science, Vol. 1 555
Fig. (6). Plot of percent weight reduction for FSW probe vs particle size. Each data series represents
a separate set of parameters [10]. X-axis is particle diameter in inches and y-axis is percent volume
loss.
Surface metrology of FSW tool probes used for welding joints containing
particles of three different sizes (FEPA grade F150, F60, and F14) were
compared using contact profilometry. Surface texture parameters such as
the average roughness Ra and peak to valley roughness Rt dramatically
556 Frontiers in Aerospace Science, Vol. 1 Tracie Prater
change with the wear. The linear increase in roughness (relative to the
unworn probe) directly varies with the particle size [10]. It is generally
expected that the larger the particles, the greater the observed wear. The
experimental results of the particle size experiments suggest the following:
1. They provide further evidence for the rotating plug model of wear,
suggesting that wear in this process is strongly dependent on
particle size. Within the context of the model, larger particles are
more likely to span the clearance 𝛿 between the probe and the
rotating plug and impinge on the tool surface.
2. From the perspective of material design, the experimental results
demonstrate that one technique to minimize wear in FSW of MMCs
is to select materials with reinforcement particles lying at the upper
end of the FEPA scale.
3. The experimental method adapted based on Colligan’s approach
represents an economical means to study the effect of materials with
specific particle sizes for research purposes that does not require the
production of custom material.
Fig. (7). SEM image (100X) of worn FSW probe. The circumferential, parallel grooves exhibited on
the surface are characteristic of abrasive wear processes.
Metal Matrix Composites in Aerospace Structures Frontiers in Aerospace Science, Vol. 1 557
The tools used in the particle tracer experiments were further inspected
using SEM to evaluate the abrasive wear mechanism (Fig. 7). The best way
to prevent wear in this instance is to select a tool material that is harder than
the reinforcement particles. According to classical wear theory, when the
hardness ratio between the tool and the abrasive is greater than one,
abrasive wear cannot occur. The best candidate for a wear-resistant tool
material is diamond. However, monolithic diamond is typically only used
in critical manufacturing applications because of its high cost. Diamond
coatings represent a cost effective alternative. The major barrier to
application of coated tools in FSW is the coating’s tendency to delaminate
under the stresses imposed by the joining process. When delamination
occurs, the wear behavior of the coated tool regresses to levels associated
with its substrate. Hence a strong bond between substrate and coating is
necessary to prevent delamination and preserve wear-resistant
characteristics of the coated tool. The effect of the hardness ratio on wear in
FSW of MMCs was evaluated using four different tool materials (O1 tool
steel, WC, WC micrograin, and WC coated with diamond) in two substrate
materials (Al 359/SiC/20p and Al 359/SiC/30p) at the previously optimized
parameter set 𝜔 = 1000 RPM and 𝑣 = 3 IPM (the brittle behavior of ceramic
and refractory metals significantly narrows the process window for FSW).
A higher rotation rate coupled with a comparatively slower traverse speed
tends to provide sufficient heating to reduce the of tool failure. Fig. (8)
compares the cumulative wear of the candidate materials over a 4 feet of
length of weld. The wear resistance of the coated tool is the best of the
materials considered but the margin between the performance of the
diamond and WC-Co probes is small [12].
558 Frontiers in Aerospace Science, Vol. 1 Tracie Prater
WC/Co micrograin
WC/Co Submicrograin
30% reinforcement
20% reinforcement
O1 Steel
Diamond*
0 10 20 30 40
Fig. (8). Relationship between the wear percentage and the tool material, where each cylinder
represents the wear percentage measured for the tool material after three 14” long welds (at 1000
RPM/3 IPM) in an Al 359 MMC with either 20% or 30% SiC reinforcement [12].
An inspection of the torque equation derived from the rotating plug model
(equation 5) indicates that radial deterioration of the probe during joining of
MMCs by FSW should coincide with a decrease in the magnitude of the
torque the tool experiences during welding. Simulations carried out by
Gibson, Prater, et al. showed that in situ estimates of wear could be made
with an adaptive torque controller [13]. According to Nunes’s torque
equation developed based on the rotating plug model, the torque signal is
sensitive to flow stress 𝜎, temperature, plunge depth t, as well as geometric
parameters (shoulder and pin radii R and r, respectively). These additional
(and sometimes coupled) relationships make it more difficult to isolate
changes in the torque signal that can be attributed solely to radial loss of
tool material. Nonetheless, in-process sensing merits further investigation
since such a capability would minimize disruption of the manufacturing
process for wear evaluation (currently tools must be taken off-line to assess
Metal Matrix Composites in Aerospace Structures Frontiers in Aerospace Science, Vol. 1 559
wear) and in some cases be tantamount to flaw detection. Fig. (9) depicts
the evolution of the torque signal (measured from motor current) over a 6ft
weld of Al 359/SiC/20p for a conventional tool material (O1 steel hardened
to RC 50). The steady-state torque signal for each pass was related to the
amount of wear. Over the course of several experiments where longer
panels are welded, the results demonstrated the change in torque signal due
to wear is both detectable and a viable technique for wear prediction. One
primary application of such wear detecting technology would be longer
length welds of reinforced Al alloys where an off-line in media res
evaluation of wear is not feasible.
2𝜋𝑅3 𝑟2𝑡
𝑇𝑡𝑜𝑡𝑎𝑙 = (1 + 3 )𝜎 (5)
3 𝑅3
Fig. (9). Motor current (Amperes) versus time (sec) for 6ft long weld of Al 359/SiC/20p at 1400
RPM and 9 IPM. Purple shaded regions represent successive 12” weld passes along the panel.
CONCLUSION
The ultimate goal of the present study is to extend the application of metal
composites to aerospace structures by developing process models and
560 Frontiers in Aerospace Science, Vol. 1 Tracie Prater
CONFLICT OF INTEREST
The author confirms that author have no conflict of interest to declare for
this publication.
ACKNOWLEDGEMENTS
Declared none.
REFERENCES
[1] "Space Shuttle Technology Summary: Friction Stir Welding", Available from:
www.nasa.gov/centers/marshall/pdf/104835main_friction.pdf, 2009.
[2] D. Storjohann, O.M. Barabash, S.S. Babu, and S.A. David, "Fusion and
friction stir welding of aluminum metal matrix composites", Metall. Mater.
Trans., A Phys. Metall. Mater. Sci., vol. 36A, pp. 3237-3247, 2005.
[http://dx.doi.org/10.1007/s11661-005-0093-4]
[3] T. Prater, "Solid state joining of metal matrix composites: a survey of
challenges and potential solutions", Mater. Manuf. Process., vol. 26, pp.
636-648, 2011.
[http://dx.doi.org/10.1080/10426914.2010.492055]
[4] T. Prater, G.E. Cook, A.M. Strauss, J. Davidson, and M. Howell,
"Parameterization of friction stir welding of Al 6061/SiC/17", In: Proceedings
of the 8th International Conference on Trends in Welding Research, , 2008, pp.
102-108.
[5] L.M. Marzoli, A.V. Strombeck, J.F. Dos Santos, C. Gambaro, and L.M.
Volpone, "Friction stir welding of an AA6061/ Al2O3preinforced alloy",
Compos. Sci. Technol., vol. 66, pp. 363-371, 2006.
[http://dx.doi.org/10.1016/j.compscitech.2005.04.048]
[6] P. Cavaliere, E. Cerri, L. Marzoli, and J. Dos Santos, "Friction stir welding of
562 Frontiers in Aerospace Science, Vol. 1 Tracie Prater
[14] T. Prater, "Friction stir welding of metal matrix composites for use in
aerospace structures", Acta Astronaut., vol. 93, pp. 366-373, 2014.
[http://dx.doi.org/10.1016/j.actaastro.2013.07.023]
56 Frontiers in Aerospace Science, 2016, Vol. 1, 56-57
SUBJECT INDEX
A unsteady 3, 6, 9
Aerodynamic models 7, 23
Abrasive particles 553, 554, 555 Aerodynamic nonlinearities 7, 70
Absorbed energy 114, 125, 127, 130, Aerodynamics 7, 68, 70, 298, 398, 414,
133, 135, 144, 148 415, 416, 417, 420, 424
Absorbent material vaporizes 305, 332 Aerodynamic study 67
Acoustic emission (AE) 357, 477 Aeroelastic analysis 6, 7, 25, 26, 29, 32
Adhesive joints 154, 155, 160, 161, Aeroelastic behavior 3, 7, 29, 36, 41, 56
165, 170 Aeroelasticity 4, 7
Adhesive layer 159, 160, 163, 164, 166, Aeroelastic model 3, 26, 34
170, 173, 174, 175, 177, 182, 183, Aeroelastic problem 5, 7, 9
184, 188, 190, 191, 193, 195, 198, Aeroelastic stability 8, 29, 31, 52
199, 201, 205, 207, 212, 213, 216, Aeroelastic stability boundaries 3
219, 220, 222, 223, 225, 228, 229, Aerospace designers 443, 449, 451,
230, 231, 232 452, 455, 463
thin 166, 175, 177, 191 Aerospace hydraulic systems 116
Adhesive layer thicknesses 155, 160, Aerospace industry 303, 307, 355, 356,
164, 166, 170, 198, 199, 205, 206, 387, 445, 446, 469, 470, 475, 483,
208, 216, 220, 223, 229, 231 484, 486, 498, 500, 501, 502, 504
Adhesive material 154, 155, 160, 163, Aerospace materials 451, 463, 464, 469,
164, 165, 170, 181, 196, 200, 208, 486, 505
209, 223 Aerospace nanotechnology 469
Adhesive thicknesses 155, 160, 164, Aerothermoelastic problems 6
173, 183, 185, 190, 194, 195, 196, AFP technology 518
197, 198, 199, 200, 202, 204, 205, AH-1W Helicopter 288
206, 207, 209, 214, 215, 216, 217, Aircraft 66, 67, 70, 76, 307, 311, 397,
218, 219, 220, 221, 222, 223, 224, 406, 414, 415, 416, 417, 418, 421,
225, 227, 228, 229, 230, 231, 232 423, 426, 443, 444, 445, 446, 447,
effects of 154, 160, 163, 164, 165, 448, 449, 450, 451, 452, 455, 457,
196, 208 458, 459, 461, 469, 470, 473, 476,
Aerodynamic coefficients 25 477, 479, 481, 484, 485, 486, 499,
Aerodynamic derivatives 26, 27, 395, 502, 503, 504, 505, 543
422 military 446, 458, 459, 470, 479,
Aerodynamic forces 13, 66, 67, 72, 74, 484, 485
77, 92, 107, 110, 421, 422 Aircraft design, advanced 451
Aerodynamic lift 9, 13, 24 Aircraft designer 449
Aerodynamic loads 3, 5, 6, 7, 9, 41 Aircraft engines 241
Carbon fibers 459, 498, 500, 504, 507, Composite cylinders 516, 517, 520,
509, 510, 511 522, 523, 524, 525, 527, 530, 532,
Carbon nanotube paper 476 533, 534, 535, 536
Carbon nanotubes 469, 470, 471, 472, Composite materials 446, 454, 470,
475, 481, 482 474, 477, 485, 499, 501, 516, 518,
Card, internal pressure curve 120 534, 546
Casting-die profile 241 Composite structures 517, 518
Catastrophic failures 157 Compression process 503, 507
Chaff/Flare dispenser bracket 280, 288, Compressive residual stress (CRS) 303,
291, 292 304, 309, 311, 315, 320, 322, 327,
Chaotic response 8, 39, 41, 42, 52 330, 331, 336, 342, 343, 348, 351
Chaotic systems 22 Computed radiography (CR) 357, 422
Chaotic vibrations 8, 22 Computer-aided design (CAD) 265, 429
Chord length 72, 88, 250, 251 Computer modeling and simulation
Circumferential Location 535 115, 271
CNT implementation 483, 484, 485 Constant model, piece-wise 521, 522
CNTs in aerospace applications 474, Constant stiffness (CS) 12, 517, 519,
487 520, 534, 535, 536
Cohesive fracture 157, 190, 191 Constitutive models 327, 332, 339
Cohesive laws 154, 155, 160, 165, 171, Control command system 405
174, 198, 204, 215, 218, 229 Controllers, designing 415, 416
Cohesive strength 155, 160, 230 Conventional ultrasonic 357, 362, 364,
Cohesive zone models (CZMs) 154, 365
155, 159, 160, 161, 165, 173 Coordinate value, optimized 249
Collapse mechanism 115 Cowper-Symonds relation 118
Commercial aircraft 475, 484, 485, 500, Crack growth 157, 161
504, 505 Crack initiation 201, 202, 204, 215,
Components 155, 241, 243, 271, 286, 304, 308, 330
303, 305, 307, 308, 310, 314, 315, Crack propagation 160, 163, 198, 202,
316, 320, 322, 324, 329, 330, 344, 215, 218, 225, 228, 229, 231, 315,
345, 358, 369, 411, 416, 421, 422, 445, 450
429, 432, 435, 436, 437, 450, 478, Crack tip 157, 158, 159, 162, 163, 172,
501, 502, 504, 512, 513, 546 173, 175, 176, 177, 178, 179, 180,
metal 501, 502, 513 181, 188, 190, 192, 193, 201, 210,
treated 344, 345 211, 212
Subject Index Frontiers in Aerospace Science 56
End notched flexure (ENF) 161, 163, Fatigue life prediction 308, 309, 310,
164, 175, 177, 182, 229, 230 313
Energy release rate (ERR) 157, 158, Fatigue loading 309, 314, 315, 349
193, 195, 203, 211, 212, 214 Fatigue prediction methods 313
Energy release rate component JI 180, Fatigue strength 278, 304, 314, 445,
181 446
ENF specimen 178, 186, 189, 201, 206 Fatigue tests 270, 271, 281, 291, 296,
Engine and fuel system 395, 405 299, 432
Engine gears 411, 412 FEA models 116, 118, 119, 120, 121,
Engine group 430 122, 125, 149, 316, 336
Engines 241, 395, 405, 406, 407, 408, FEA software 287
411, 414, 429, 431, 438, 446 Fiber, glass 169, 504, 507, 509, 510,
electric 405, 406, 407, 408 511
piston 405, 406, 407, 408 Fiber content 502, 506, 507, 508, 509,
turbine 405, 407 510
Equations 3, 10, 21, 89, 169, 170, 423, Fiber orientation angle 517, 521
424, 426 Fiber-reinforced polymer composite
characteristic 21, 423, 424, 426 (FRPC) 517
governing 3, 10, 89, 169, 170 Fiber reinforcements 500, 501, 502,
Estimated forces 78, 82, 92, 93, 94, 95, 509, 512
96, 97, 98, 110 continuous 509, 512
Estimated forces and moments 78, 82, continuous carbon 500, 501, 502
92, 96, 110 Fighter aircraft loading standard for
Excitation system 71 fatigue (FALSTAFF) 310
Expansion waves 15, 16, 17 Finite element analysis (FEA) 114, 122,
Explicit FEA solver 116 124, 135, 138, 139, 145, 150, 160,
244, 270, 276, 311, 327, 331, 523,
F 524
Finite element model 34, 66, 68, 70,
Failure mode effect analysis (FMEA) 299, 336, 338, 521
435 Flap and plunge direction 39, 40, 41,
Fatigue analysis 271, 280, 281, 290, 43, 44, 46, 47, 50, 51, 54, 55
291, 292, 296, 298, 300, 311, 322, Flapping motions 9, 418
344 Flexural strength 502, 505, 506, 507,
Fatigue failures 271, 298, 304, 309, 508, 509
327, 346 Flight missions 406, 407, 408
Flight regimes 408, 418
Subject Index Frontiers in Aerospace Science 5
Flight tests 271, 274, 278, 288, 289, Frequency domain methods 281, 282
290, 432, 434 Frequency response functions (FRFs)
Flow, supersonic 11, 15, 16, 27 66, 77, 78, 82, 92, 99, 101, 104,
Flow properties 6, 15, 18 107, 108
Flutter divergence 30, 31, 32 Frequency response matrix 74, 75, 76,
Force determination error 93, 94, 95, 77, 99
97, 98, 99 Frequency separation 413
Force determination methods 66, 68, Friction stir (FS) 370, 376, 377, 378,
75, 77, 78, 86, 92, 110 385, 386, 387, 556
Forces 4, 10, 13, 14, 66, 72, 74, 76, 77, Friction-stir-welding (FSW) 355, 376,
78, 82, 83, 92, 94, 96, 99, 101, 543, 544, 546, 547, 549, 553, 556,
107, 114, 126, 139, 190, 201, 210, 558, 559, 561
229, 230, 231, 349, 358, 396, 417, FRPC materials 517
420, 422, 424 FSW, application of 547
actual 66, 74, 107 FSW of MMCs 548, 549, 552, 553,
critical 229, 230, 231 554, 557, 558, 561
excitation 72, 74, 78, 101, 107 FSW process 546
generalized 4, 10 Fuel consumption 407, 475, 476
Forward flight 398, 404, 405, 416, 417, Fuel systems 395, 405
418, 423, 426, 427, 428 Functional composite components 500
Fracture, basic modes of 156
Fracture energy 159, 160, 163, 164, G
173, 192, 207, 211, 219, 225, 230
Fracture energy JIIC 202, 215, 218, Gear ratio 408, 410, 411, 412, 413
219, 20, 221, 224 Geometrical parameters 3, 29, 36
estimated 218, 220, 221, 224 Geometric characteristics 401, 402, 438
Fracture process 158, 160, 165 Geometries 11, 15, 27, 29, 33, 160, 163,
Fracture strength 229, 230, 231, 232 184, 185, 196, 261, 271, 293, 299,
Fracture tests 155, 173, 174, 187 308, 311, 316, 337, 342, 347, 358,
Fracture toughness JIC 229, 231, 232 450, 513
Free-play 3, 4, 5, 8, 9, 11, 13, 14, 36, concave 316, 342
37, 56, 57 curved 308, 347
Free-stream surface 18 Ground vibration test (GVT) 71, 75, 76,
Frequency domain 69, 70, 73, 89, 108, 101, 102, 108, 110
276, 281, 299
57 Frontiers in Aerospace Science Yucheng Liu
Group 184, 185, 190, 191, 192, 193, Impact response 116, 133, 135, , 142,
164, 195, 196, 198, 200, 201, 202, 145, 150
203, 204, 205, 207, 210, 216, 217, lateral 116, 142, 145, 150
218, 429, 499, 547, 552 Impact simulations 119, 120, 122
specimen in 190, 191, 192, 194, 201, Impact strength 505, 506, 507, 509, 511
202, 204 Impulse response functions (IRFs) 104
Inclinometer 188, 191, 211
H Index offset 373, 374, 388
Industry, modern 135, 154, 155
Helicoids hole 100 Initial angle of attack 3, 24, 25, 26, 27,
Helicopter components 271, 272 29, 35, 45
High-temperature CFRP 458 Initial crack 186, 194, 204
Hinge frictional torque 3, 56 Initial crack length 184, 185, 210, 211,
Hovering flight 408, 409, 417, 418, 212, 214, 215, 217, 222, 223, 224,
423, 424, 426 225, 227, 228, 231, 232
H Plunging displacement 4 Initial crack tip 157, 192, 193, 194, 202,
Hugoniot elastic limit (HEL) 318, 328, 204, 211, 212, 214, 222, 229
335, 336, 341 Initial crack tip length 222, 228
Hydraulic system of aerospace 115, 150 Integration process 270, 274, 280
Hydrophobicity 469 Integrity, structural 355, 356, 475
Interfacial strength 155, 173, 198, 200,
I 207, 209, 218, 219, 225, 229, 231
Interfacial strength σmax 198, 200, 218,
Impact 118, 119, 133, 134, 145, 150, 225, 227
329, 543, 556, 561 Interfacial traction-separation laws 154,
mid-position 133, 134, 150 160, 161, 164, 173, 198, 209, 225
quarter-position 133, 134, 150 Interfacial traction-separation laws of
Impact behavior 116, 117, 149, 511 bonded steel 219, 225, 226
Impact energy 117, 134, 135, 145 Interlayer, adhesive 158, 163, 164
Impact force 125, 126, 135 Internal defect 377, 378, 385, 386
Impact forces, higher 134, 145 Internal pressure 114, 115, 116, 117,
Impact hammer 101, 102, 103 119, 120, 125, 127, 131, 133, 134,
Impact positions 114, 116, 125, 127, 135, 139, 142, 145, 149, 150
128, 129, 130, 133, 145, 150 effects of 114, 116, 131, 142
Impact resistance 134, 145, 149, 181 influence of 116, 117, 133, 142
Inverse deformation 242
Subject Index Frontiers in Aerospace Science 57
Maximum fatigue life 303, 308, 309, empirical 312, 331, 561
318, 319, 324 Modeling 5, 9, 137, 160, 274, 318, 327,
Maximum impact force 127, 128, 133, 337, 347, 405, 415, 480, 520
134, 142, 145, 146, 149, 150 Modern toughened adhesives 157
Maximum Lyapunov exponent 22, 23 Mode shapes 4, 5, 10, 34, 66, 89, 90,
Maximum transverse displacement 134, 107, 110, 275, 299
150 natural 4, 5, 10, 34
Mean time between failures (MTBF) second 89, 90
395, 435 Molten aluminum 543, 547
Mean time to failure (MTTF) 395, 435 Monte carlo method 286
Measurement locations 273, 276, 278, Morozov’s discrepancy principle 69
279 Multi-axial testing 271, 281
Metal fatigue 270 Multi-degree-of-freedom model 70, 88,
Metal matrix composites (MMCs) 543, 92, 93, 94, 95, 96, 97, 98, 99, 110
544, 547, 548, 549, 552, 553, 554, Multiple-degree-of-freedom system 66
557, 558, 559, 561 Multiple regression model (MRM) 551,
Metamodeling based design 552
optimization (MBDO) 516, 520,
536 N
Method of die-cavity 241
Microchips to large aircrafts 155 Nanomaterials 474, 485
Mid-span duration 316, 317, 320 Nanotube diameter 471, 472
Mid-span impact 138, 140, 142, 143, National aeronautics and space
144, 145, 146, 147, 148, 149 administration (NASA) 474, 476,
Mild-steel pipelines 135 543, 546, 548
Military aircraft wings 461, 462, 463 NDE techniques 355, 356, 357, 358
Military standards 270, 288, 294 Non-destructive evaluations 356
Model 3, 7, 11, 28, 29, 32, 66, 70, 71, Non-linear aeroelastic behavior 3
87, 99, 100, 101, 103, 104, 110, Nonlinear dynamical systems 20
118, 120, 126, 127, 160, 161, 242, Non-periodic responses 22
255, 274, 275, 286, 293, 312, 313, Numerical models 155, 160, 161, 287,
316, 327, 336, 338, 341, 346, 409, 327, 350
414, 415, 416, 417, 421, 422, 431, Numerical simulations 114, 149, 242,
544, 551, 553, 554, 555, 557, 562 252, 255, 263, 266, 346
cowling 274, 275
Subject Index Frontiers in Aerospace Science 57
Sound waves 359, 360, 361, 369, 370, Stress relaxation model 316, 327
371, 372, 375 Stress spectrum 311, 314, 316
high frequency 369, 370 Stress-strain behavior 332, 346, 347
Space elevator 469, 482 Stress values 277, 278, 286, 299, 300
Specimens 160, 162, 172, 175, 177, Structural components 303, 304, 308,
178, 182, 183, 184, 185, 190, 195, 309, 316, 323, 324, 330, 344, 448,
196, 198, 209, 210, 211, 212, 214, 450, 451, 463, 464, 503, 517, 518,
215, 216, 218, 229, 230, 231, 281, 544
331, 339, 340, 347, 372, 378, 381, Structural improvement 519, 520, 530
508 Structural materials 445, 447, 454, 464,
effective 185, 190 545, 548
five 195, 196, 198 Structural member 444, 449
five groups of 184, 185 Structural model 3, 7, 9, 29, 71
groups of 183, 184, 185, 190 Structural nonlinearities 7, 8, 70, 300
SST aircrafts 446, 455, 458, 459 Structural performance 516, 517, 519
Stability derivatives 416, 420, 424 Structural responses 66, 67, 74, 75, 270,
Stacking sequence 410, 525, 530, 533, 523
535 Structures, bonded 154, 156, 165
Stagger angle of CAD model 248 Subsonic aircrafts 445, 452, 453, 454
Static mass moment 5, 12 Subsonic aircraft’s structures 452, 453,
Steel adherends 164, 209, 212 454
3-D simulation 395 Subsonic structures 445, 452
Strain rate effects 118, 335, 339 Supersonic aerodynamic characteristics
Strain rates 118, 328, 339 6
Stress115, 156, 176, 177, 179, 181, 261, Supersonic cruise conditions 447
304, 318, 320, 330, 331, 348 Supersonic fins 7, 29, 31, 52
normal 176, 177, 179, 181 Supersonic flight 451, 462, 463
tensile 115, 156, 261, 304, 318, 320, Supersonic/hypersonic flow 5
330, 331, 348 Supersonic regime 6
Stress corrosion cracking (SCC) 444, Surface enhancement methods 333, 344
448 Surface enhancement techniques 304,
Stress field, residual 316, 338, 345, 347 308, 329, 330
Stress intensity factor (SIF) 157, 444,
450 T
Stress range 285, 286
Stress relaxation 308, 344, 345, 346, Tail rotor diameter 399, 401, 403
347, 348, 350 TCG calibration 374, 388
576 Frontiers in Aerospace Science Yucheng Liu
Techniques 6, 115, 136, 162, 356, 357, Traction-separation laws 154, 161, 163,
373, 381, 386, 387, 416, 431, 549, 164, 194, 198, 199, 202, 212, 213,
556, 557, 560, 561 214
traditional inverse 79, 93 equivalent interfacial 163, 194, 198,
Temperature effects 306, 339, 341 199, 202, 212, 213, 214
Temperature resistance 458, 459, 461 local interfacial 154, 161, 163, 164
Tensile modulus 507, 509, 510 Traditional pseudo-inverse technique
Tensile strength 181, 200, 458, 460, 77, 92
472, 502, 505, 506, 507, 508, 509, Traffic monitoring 397, 398, 400, 401,
510, 511 438
Tests, wind tunnel 26, 68, 71, 99, 107, Translational force 80, 82, 84, 85, 94,
108, 110, 432 95, 97, 98
Test specimens 161, 187, 188, 359, 508 Transmission system design 411, 412,
Theoretical model 8, 134, 165 413
Theory, classical beam 165, 166, 167 Transmitting system 405, 408, 411, 413
Thermal conductivity 470, 471, 472 Transverse deformations 18
Thick fins 3, 6, 15, 16, 32 Transverse displacements 133, 150
Thick supersonic fin 4, 6, 9, 36, 57 Truncated singular value decomposition
Ti-6Al-4V material 327, 340 (TSVD) 69
Time-corrected-gain (TCG) 374 Turbine blade 241, 242, 243, 245, 249,
Time derivatives 37, 56 252, 254, 255, 261, 262, 264, 265,
Time-of-flight-diffraction 362, 366, 368 308
Time of flight diffraction (TOFD) 357, Two degree-of-freedoms pitch-plunge
362, 363, 366, 368 system 72
Time series, torsional angle 37, 41 Two-step optimization strategy 303,
Titanium 445, 446, 457, 471, 505, 547 309, 319
Tool materials 544, 548, 558, 559, 561
Tool probe 549, 555, 556 U
Tool surface 554, 555, 557
Tool wear 544, 552, 556, 561 Ultimate tensile strength (UTS) 444,
Torque 396, 408, 410, 411, 428, 544, 455, 460
559, 561 Ultrasonic testing (UT) 357, 358, 362,
Torsional control parameters 245, 248 364
Torsional shear stresses 444, 449 Uni-axial fatigue testing error 271
Traction-separation 229, 230, 231 Unmanned aerial vehicles (UAVs) 395,
397, 405, 414, 469, 470, 480, 481,
482
Subject Index Frontiers in Aerospace Science 577