Reduced Order Modeling of

Aircraft Structural Assembly using

Spectral Finite Element Method

A Dissertation

Submitted in the Partial Fulfillment of the

Requirements for the Degree of Master of Engineering

in Aerospace Engineering

By
K. Sainath

June 2012

Department of Aerospace Engineering

Indian Institute of Science
Bangalore 560012
 K. Sainath

June 2012

All rights reserved

Dedicated

To

My Parents
Acknowledgment
I now take the opportunity to express my gratitude to all those who supported me
to complete this thesis work.

I express my sincere gratitude to Dr. Debasish Ghose, Chairman of Aerospace

Department, Indian Institute of Science, Bangalore, for providing me the
opportunity and all academic facilities with lots of support and encourageme nt
and valuable guidance to successfully complete M.E in Aerospace Engineering.

I express my profound thanks to my supervisors Dr. D. Roy.Mahapatra and Dr.

S. N. Omkar, Department of Aerospace Engineering, Indian Institute of Science,
Bangalore, for there excellent and valuable guidance and encouragement in
successful completion of thesis.

I thank the faculty members and of the Aerospace Department staff for their co-
operation. I thank all my friends and my batch mates in IISc for their constant
support and encouragement.

I am very much thankful to my beloved parents for their encouragement and

support in all my efforts. Finally I thank almighty for giving me sound health and
mind during my time at IISc.

K. Sainath

i
Table of contents
List of Figures .................................................................................................................. iv
Abstract .............................................................................................................................vii
Chapter 1 Introduction .................................................................................................... 1
1.1 Introduction ......................................................................................................... 1
1.2 Research Objectives ......................................................................................... 4
1.3 Outline of thesis ................................................................................................. 4
Chapter 2 Review of Aircraft Structural Components in Context of SHM............. 5
2.1 Aircraft structural design and its components ............................................... 5
2.2 Aircraft wings ...................................................................................................... 7
2.2.1 Components of the aircraft wing .............................................................. 8
2.3 Aircraft fuselage ................................................................................................. 9
2.3.1 Components of the aircraft fuselage ..................................................... 10
2.4 Aircraft Empennage Section and primary flight control surfaces ............. 10
2.4.1 Aircraft empennage components and flight control surfaces ............ 10
2.5 Load transfer .................................................................................................... 12
2.5.1 Wing structure ........................................................................................... 12
2.5.2 Fuselage .................................................................................................... 13
2.6 Summary ........................................................................................................... 13
Chapter 3 Wave Propagation and Spectral Finite Element Method ..................... 14
3.1 Wave Propagation ........................................................................................... 14
3.2 Spectral Analysis ............................................................................................. 15
3.2.1 Waveguide................................................................................................. 15
3.2.2 Wave number and Spectrum relation ................................................... 15
3.2.3 Phase speed, Group and Dispersion relations .................................... 16
3.2.4 Relation between phase speed and group speed............................... 16
3.2.5 Cut-off frequency ...................................................................................... 17
3.3 Spectral finite Element Method (SFEM) ...................................................... 17
3.4 Spectral Element for Rods ............................................................................. 18
3.4.1 Shape functions ........................................................................................ 20
3.4.2 Dynamic stiffness for rods....................................................................... 21
ii
 3.5 Spectral elements for beams ......................................................................... 22
3.5.1 Shape functions ........................................................................................ 22
3.5.2 Dynamic stiffness for beams .................................................................. 25
3.6 Spectral element for 2D frames .................................................................... 28
3.7 Summary ........................................................................................................... 28
Chapter 4 Reduced Order modelling of Aircraft Wing ............................................. 29
4.1 Major Components of Aircraft Wing.............................................................. 29
4.2 Equivalent 2D Beam-Frame Model of an Aircraft Wing ............................ 30
4.3 Segments of the 2D Beam-Frame Model .................................................... 31
4.4 Formulation of dynamic stiffness matrix of one segment of 2D beam-
frame model................................................................................................................. 33
4.5 Global dynamic stiffness matrix of 2D beam-frame model ....................... 40
4.6 Summary ........................................................................................................... 40
Chapter 5 Results and Discussions ........................................................................... 41
5.1 Dynamic Response of an Isotropic Rod ...................................................... 41
5.2 Dynamic Response of an Isotropic Beam ................................................... 46
5.3 Dynamic Response of an 2D Frame ............................................................ 49
5.4 Frequency Response analysis of an 2D beam-frame model of aircraft
wing ............................................................................................................................ 54
5.4.1 case1: straight wing ................................................................................. 54
5.4.2 case2: tapered wing ................................................................................. 61
Chapter 6 Conclusion and Future Scope of work .................................................... 69
Chapter 7 References .................................................................................................. 70

iii
List of Figures

Figure 2-1 Schematic view of Aircraft with its Parts. ................................................. 6

Figure 2-2 Cut-Out View of Boeing 777 Commercial Aircraft. ................................. 7
Figure 2-3 Wing Structure............................................................................................ 12
Figure 2-4 Fuselage structure ..................................................................................... 13
Figure 3-1 Outline of FFT Based Spectral Finite Element Method. ...................... 18
Figure 3-2 Nodal Loads and Degrees of Freedom for Longitudinal
Spectral Elements. .................................................................................... 19
Figure 3-3 Nodal Loads and Degrees of Freedom for Flexural Spectral
elements. .................................................................................................... 22
Figure 3-4 Nodal loads and degrees of freedom for spectral element of
2D frame .................................................................................................... 28
Figure 4-1 Schematic Top View of Aircraft. ............................................................. 29
Figure 4-2 Schematic Front View of Reduced order model of Aircraft Wing...... 30
Figure 4-3 Distributed Spring Between node-1 and node-2.................................. 31
Figure 4-4 2D Beam-Frame model of a Segment of the wing. ............................. 32
Figure 4-5 A Zoomed-in View of 2D Beam-Frame model of one segment
of the wing. ................................................................................................. 32
Figure 4-6 One Segment of the 2D Beam-Frame model of the Aircraft Wing. ... 33
Figure 5-1 Semi-Infinite Rod Modeled with two Elements...................................... 41
Figure 5-2 Axial Velocity history at Node-3, due to loading
(inset of the plot) in the Axial direction................................................... 42
Figure 5-3 Axial Velocity history at Node-3, due to loading
(inset of the plot) in the Axial direction................................................... 43
Figure 5-4 Axial Velocity history at Node-3, due to loading
(inset of the plot) in the Axial direction................................................... 44
Figure 5-5 Axial Velocity history at Node-3, due to loading
(inset of the plot) in the Axial direction................................................... 45
Figure 5-6 Semi-Infinite Beam Modeled with Two Elements. ................................ 46
Figure 5-7 Transverse velocity history at node 3 due to loading in the
inset of the plot. .......................................................................................... 47
Figure 5-8 Transverse velocity history at node 3 due to loading in the
inset of the plot. .......................................................................................... 48
Figure 5-9 2D Frame. ................................................................................................... 49
Figure 5-10 Gaussian Pulse Loading-1 and 2. ......................................................... 50
Figure 5-11 Transverse frequency response at node 1 due to
Gaussian pulse_1. ................................................................................... 50
Figure 5-12 Transverse frequency response at node 2 due to
Gaussian pulse_1. ................................................................................... 51

iv
Figure 5-13 Axial frequency response at node 1 due to Gaussian pulse_1........ 51
Figure 5-14 Axial frequency response at node 2 due to Gaussian pulse_1........ 52
Figure 5-15 Transverse frequency response at node 1 due to
Gaussian pulse_2. .................................................................................. 52
Figure 5-16 Transverse frequency response at node 2 due to
Gaussian pulse_2. .................................................................................. 53
Figure 5-17 Axial frequency response at node 1 due to Gaussian pulse_2....... 53
Figure 5-18 Axial frequency response at node 2 due to Gaussian pulse_2....... 54
Figure 5-19 Isometric View of CATIA model of Straight Wing. ............................. 55
Figure 5-20 Beam-Frame Model of Straight Wing. ................................................. 55
Figure 5-21 Axial velocity v/s Frequency at node 30 due to the
loading (inset of the plot) in the Transverse direction. ..................... 56
Figure 5-22 Transverse velocity v/s Frequency at node 30 due to the
loading (inset of the plot) in the Transverse direction. ..................... 56
Figure 5-23 Axial velocity v/s Frequency at node 30 due to the
loading (inset of the plot) in the Transverse direction. ..................... 57
Figure 5-24 Transverse velocity v/s Frequency at node 30 due
to the loading (inset of the plot) in the Transverse direction. .......... 57
Figure 5-25 Axial velocity v/s Frequency at node 20 due to the
loading (inset of the plot) in the Transverse direction. ..................... 58
Figure 5-26 Transverse velocity v/s Frequency at node 20 due
to the loading (inset of the plot) in the Transverse direction. .......... 58
Figure 5-27 Axial velocity v/s Frequency at node 20 due to the
loading (inset of the plot) in the Transverse direction. ..................... 59
Figure 5-28 Transverse velocity v/s Frequency at node 20 due
to the loading (inset of the plot) in the Transverse direction. .......... 59
Figure 5-29 Axial velocity v/s Frequency at node 30 due to the
Gaussian pulse-1. .................................................................................. 60
Figure 5-30 Axial velocity v/s Frequency at node 30 due to the
Gaussian pulse-2. .................................................................................. 60
Figure 5-31 Transverse velocity v/s Frequency at node 30 due to the
Gaussian pulse-1. .................................................................................. 61
Figure 5-32 Transverse velocity v/s Frequency at node 30 due to the
Gaussian pulse-2. .................................................................................. 61
Figure 5-33 Isometric view of CATIA model of tapered wing. ............................. 62
Figure 5-34 Beam-frame model of tapered wing. .................................................. 62
Figure 5-35 Axial velocity v/s Frequency at node 30 due to the
loading (inset of the plot) in the Transverse directio n. .................... 63
Figure 5-36 Transverse velocity v/s Frequency at node 30 due
to the loading (inset of the plot) in the Transverse direction......... 63

v
Figure 5-37 Axial velocity v/s Frequency at node 30 due to the
loading in (inset of the plot) in the Transverse direction.................. 64
Figure 5-38 Transverse velocity v/s Frequency at node 30 due
to the loading (inset of the plot) in the Transverse direction. .......... 64
Figure 5-39 Axial velocity v/s Frequency at node 20 due to the
loading (inset of the plot) in the Transverse direction. ..................... 65
Figure 5-40 Transverse velocity v/s Frequency at node 20 due
to the loading (inset of the plot) in the Transverse direction. .......... 65
Figure 5-41 Axial velocity v/s Frequency at node 20 due to the
loading (inset of the plot) in the Transverse direction. ..................... 66
Figure 5-42 Transverse velocity v/s Frequency at node 20 due
to the loading (inset of the plot) in the Transverse direction. .......... 66
Figure 5-43 Axial velocity v/s Frequency at node 30 due to the
Gaussian pulse-1. .................................................................................. 67
Figure 5-44 Axial velocity v/s Frequency at node 30 due to the
Gaussian pulse-2. .................................................................................. 67
Figure 5-45 Transverse velocity v/s Frequency at node 30 due to the
Gaussian pulse-1. .................................................................................. 68
Figure 5-46 Transverse velocity v/s Frequency at node 30 due to the
Gaussian pulse-2. .................................................................................. 68

vi
Abstract

This thesis presents a reduced order modeling of aircraft wing; a 2D beam-

frame model is formed by combining 1D beam elements. The formulation of
dynamic stiffness matrix of 2D beam-frame model is performed using Spectral
finite element method.

Dynamic response analysis for 1D rod elements, 1D beam elements and 2D

frame has been performed using spectral finite element method. Dynamic
response analysis for 2D Beam-Frame model of the wing for two different cases
is also been performed.

vii
Chapter 1

Introduction

1.1 Introduction

Studies involving the monitoring, detection and arrest of the growth of

flaws such as cracks constitute Structural Health Monitoring (SHM). It is a type of
inverse problem where in the presence of damage needs to be detected from t he
known input and the measured output. Small size damage affects only the higher
order modes leaving the lower modes unchanged. Effects of damages of smaller
dimensions can be captured only at higher frequencies. To assess the presence
of small size damage, one needs a mathematical model that can capture the high
frequency response of the damaged structure with small problem sizes.
Conventional Finite element modeling is not suitable for high frequency analysis.
To solve the wave equation exactly the time domain PDE is transformed to
frequency domain ODE. Wave propagation [1] based diagnostic tool is suitable
for this kind of analysis. Spectral finite element method (SFEM) is the most
efficient computational method for wave propagation analysis in structures [2].

Identification of damages in structure under dynamic loading requires

efficient numerical model based tools to solve the resulting inverse problem.
Often it is possible, as is the case in the present approach, to estimate the
location and severity of damage by solving a forward problem. One of such
solution to the problem is termed in our previous works as the Damage Force
Indicator (DFI) approach.

The extension of the previously reported computational and experimental

developments [1, 2] based on the DFI approach is given in [3]. It is essentially
concerned with the use of a sophisticated computational model, namely the
dynamic stiffness matrix with essential spectral properties and real-time
measurement using a sensor-actuator network. It consists of a model-free
approach, meaning the model of the damaged structure is not essential, such
that a dynamic stiffness of the baseline structure is sufficient to identify the
location of damage and its influence on the force transmission characteristics.

In the present work we attempt to address various modeling and

computational issues in constructing the base-line structural dynamic model data
for complex structures, starting with a 2D frame and then 2D Beam-Frame model
with higher geometric complexities. In practical situations, the DFI, which is a
measure of the load transmission capacity of the damaged structure, can be

1
obtained easily from the measurement of dynamic responses at the nodal
locations.

In many large-scale engineering problems efficient handling of the discretized

computational model becomes essential. Use of a reduced -order model (ROM)
derived from finite element (FE) model of the original system is of particular
interest in such problems. In computational structural mechanics, applications of
ROM is useful in global/local analysis, reanalysis and structural optimization,
eigenvalue problem, structural vibration and buckling, sensitivity studies and
control parameter design, model update and damage detection. Different non-
structural problems related to heat transfer, fluid-structure interaction and other
linear and non-linear steady-state boundary value problems have also been
analysed using ROMs. A detail review on the above aspects can be found in [19],
Woerkom [20] carried out an elaborate survey of order reduction methods for
application in flexible spacecraft dynamics. An applications of ROM in high
frequency vibration of linear time invariant flexible elastic structures is found in
[10].

The computational issue here is to reduce the on-line computational cost

as much as possible by taking advantage of the parallelism that exists between
off-line model construction and real-time data acquisition from a sensor network.

In the DFI approach, the diagnostic analysis is performed over a chosen

frequency band. This means that we do not separate out the steady-state
vibration from the propagating waves. This simplifies the tasks of signal
generation and processing to a great extent. Also we can take note of the fact
that the interacting wavelengths must be in the order of the size of the damages
or smaller. This size versus wavelength effect leaves the possibility that the DFI
can be computed over many frequency sub-bands in order to identify the
damage size.

Identification of the location and size of the damage using acoustic wave
have been demonstrated successfully [4-7]. In [4-7] the damage problem is
posed as a inverse problem, Here the damage model is generated as a
assembly of 1D beam elements and solved for dynamic response characteristics
of the damaged model.
For structures with known material properties, geometry and boundary
conditions, a practical and reliable prediction based on the spectral analysis can
be made by selective use of the excitation/measurement frequency band over
the wave dispersion curves. However, these identified parameters alone cannot
provide sufficient information for estimating the damage severity, unless a
second set of simulation/experiment has to be performed. This task may be
difficult due to time constraints in the operation of a SHM system. Therefore, it
would always be advantageous if one can use an integrated scheme, where a

2
numerical model can be directly used to correlate the acoustic sensor signals
with the fracture (or any other form of damage) parameters under the in-service
loading conditions.

Since it is useful to deal with the frequency spectra obtained from

broadband sensors, one can use the DFI or similar frequency domain measures,
which can be computed quickly from spectral response at various locations of the
structure. From the DFI vector, if it is identified that the damage exists between
two sensor locations, then the same set of data can be used to correlate, for
example, the Strain Energy Release Rate (SERR) and further details regarding
precise location from the arrival time of high frequency signals.

In the design of vehicles, such as automobiles, aircraft, spacecraft, or

submarines, it is important to be able to accurately predict dynamic behavior of
the structure. With the extremely high cost of building physical prototypes of
these vehicles, there is a growing emphasis on analysis of computer models. In
this dissertation, Spectral analysis for a reduced order model of structurally
complex aircraft wing is undertaken to explore issues associated with frequency
response analysis to practical structures. Of primary interest is the trade-off
between computational efficiency and accuracy. Comparisons of the dynamic
response measured at the tip of the wing for varying joint stiffness of one of the
joint is laid out.

Dynamic response analysis tools used by the aerospace industry rely

heavily on a linear approach. Commercial finite element codes make a modal
frequency response of a complicated structure straightforward, even for large
degree-of-freedom systems. However, this approach is not suitable for predicting
the response of aircraft structures when they respond to high dynamic loading in
a geometrically nonlinear manner. A nonlinear dynamic analysis in physical DoFs
is computationally prohibitive, especially for non-deterministic problems involving
long simulation times. It is common practice to reduce the system size by
modeling just a portion of the structure e.g. instead of performing analysis for the
whole aircraft wing, we can do it for a two 2-dimensional models one along the
span of the wing and another along the chord wise direction of the wing. In this
dissertation our emphasis will be to go for a reduced order model of aircraft wing,
a reduction in one dimension. However, with such an approach, important global
dynamics are lost because the assumption taken while modeling the reduced
order model will not be valid in the practical scenario and also the boundary
conditions cannot be accurately modeled in a spectral analysis.

3
1.2 Research Objectives

The primary purposes for this study are to:

1. Formulate a reduced order model of the aircraft wing which should be

amenable to solve for dynamic characteristics of aircraft wing with scope
of incorporating structural damage and related simulation.
2. Formulation of dynamic stiffness for the reduced order model of aircraft
wing using spectral finite element method.
3. Perform frequency response analysis for the reduced order model of
aircraft wing.

1.3 Outline of thesis

Chapter 2 provides background information for the study of aircraft structures

with the component wise discretization.

Chapter 3 includes a detailed understanding of formulation for spectral finite

element method for 1D elements and 2D frames.

Chapter 4 focuses on the reduced order modeling of aircraft wing and its
formulation using spectral finite element method

Chapter 5 discusses the results obtained using the spectral finite element
method for 1D elements, 2D frame and also results obtained for beam-frame
model of the aircraft wing using the spectral finite element method.

Chapter 6 Provides the summary and benefits of the study.

4
Chapter 2

Review of Aircraft Structural components in context of SHM

This chapter presents the background information on the aircraft structural

design and its components. It also gives an insight of load transfer from one
component to other.

2.1 Aircraft structural design and its components

Aircraft structural design is a subset of structural design in general,

including ships, land vehicles, bridges, towers, and buildings. All structures must
be designed with care because human life often depends on their performance.
Structures are subject to one-way and oscillating stresses, the latter giving rise to
fatigue. Metal structures are subject to corrosion, and some kinds of corrosion
are accelerated in the presence of stress. Aircraft structures are designed with
particular attention to weight, for obvious reasons.

The first aircraft had two wings made of light weight wood frames with cloth
skins, held apart by wires and struts. The upper wing and the struts provided
compression support while the lower wing and the wires supported tension loads.

In the 1920s, metal began to be used for aircraft structure. A metal wing is a
box structure with the skins comprising the top and bottom, with front and back
formed by I-beams called spars, interior fore-aft stiffeners called ribs, and in-out
stiffeners called stringers. In level flight, the lower skin is in tension while the
upper skin is in compression. For this reason, this design is referred to as
stressed skin construction. During turbulence, upper and lower skins can
experience both tension and compression. This box structure is able to support
the above-mentioned moments, making single wing aircraft possible. The
elimination of the struts and wires so dramatically reduced air drag that aircraft
were able to fly twice as fast as before with the same engine.

While automobile structures are spot welded and ships are arc welded,
bridges, buildings, and aircraft are riveted or bolted together. Rivets are the
preferred fastening method in bridges and buildings mainly because such joints
provide some structural damping via internal friction in the rivet-hole and plate-
plate interfaces. This damping reduces vibrations and oscillations.

Spot welding is practical for automobile bodies. It is fast, repeatable, and

strong. Laser welding is sometimes used on long edge-edge joints in auto
5
subassemblies because the weight of the overlapped regions of conventional
joints is saved and such joints are easy for a laser to access via line-of-sight. Arc
welding is rarely used in aluminum because the region around such joi nts
reaches a high temperature and this destroys desirable material properties
created by prior rolling and heat treatment. Spot welding is also rare in aluminum
because the ever-present tough aluminum oxide on the surface prevents good
electrical contact.

As a result, rivets and bolts are used exclusively for aircraft structural joints.
Rivet and bolt joints in aircraft are the critical element in aircraft integrity. Great
care is expended on creating these joints because they are subject to high
stresses. The holes are drilled with keen attention to making their axes normal to
the skin surface and their diameters correct. In highly stressed regions of the
wing, each hole is manually reamed out to pre -stress the region around the hole.
Each rivet or bolt is compressed or torque precisely in order to achieve the
stress-carrying capability intended by the structural engineers. Rivet diameter
and compression are calculated to ensure that the installed rivet not only
completely fills the hole but also creates compressive stress in the surrounding
material. If there is any possibility that drilling a hole will leave a burr on the back
side, this burr must be manually removed because it could puncture the
corrosion-resisting paint when the skins are pulled together by the fastener.

Figure 2-1 Schematic view of Aircraft with its Parts.

6
 Structural engineers take care to choose the size of the fastener to
support the stresses it is expected to bear. The same is true of skin thicknesses
as mentioned above. On an aircraft wing, the skin may be as much as ten times
thicker at the root than it is at the tip. The diameter of fasteners varies similarly,
with diameters as large as your thumb at the root and as small as 3 or 4 mm at
the tip. Such specialization raises the cost because it reduces economies of
scale in purchasing and inventory control, but it saves considerable weight.

The major aircraft structures are wings, fuselage, and empennage. The
primary flight control surfaces, located on the wings and empennage, are
ailerons, elevators, and rudder. These parts are connected by seams, called
joints.

All joints constructed using rivets, bolts, or special fasteners are lap joints.
Fasteners cannot be used on joints in which the materials to be joined do not
overlap - for example, butt, tee and edge joints.

Figure 2-2 Cut-Out View of Boeing 777 Commercial Aircraft.

2.2 Aircraft wings

Aircraft wings have to be strong enough to withstand the positive forces of

flight as well as the negative forces of landing. In this dissertation emphasis is

7
given to metal wings, as the formulation and results produced are confined to
metal wings.

Metal wings are of two types: Semi-cantilever and full-cantilever. Semi-

cantilever, or braced, wings are used on light aircraft. They are externally
supported by struts or flying wires which connect the wing spar to the fuselage. A
full cantilever wing is usually made of stronger metal. It requires no external
bracing or support. The skin carries part of the wing stress. Parts common to
both wing designs are spars, compression ribs, former ribs, stringers, stress
plates, gussets, wing tips and wing skins.

2.2.1 Components of the aircraft wing

2.2.1.1 Wing spars

Two or more spars are used in the construction of a wing. They carry the
main bending and torsional load from wing root to tip. Both the spar and a
compression rib connect the wing to the fuselage.

2.2.1.2 Compression ribs

Compression ribs carry the main load in the direction of flight, from leading
edge to trailing edge. On some aircraft the compression rib is a structural piece
of tubing separating two main spars. The main function of the compression rib is
to absorb the force applied to the spar when the aircraft is in flight.

2.2.1.3 Former ribs

A former rib, which is made from light metal, attaches to the stringers and
wing skins to give the wing its aerodynamic shape. Former ribs can be classified
as nose ribs, trailing edge ribs, and mid ribs running fore and aft between the
front and rear spar on the wing.

2.2.1.4 Stringers

Stringers are made of thin sheets of preformed extruded or hand-formed

aluminum alloy. They run front to back along the fuselage and from wing butt to
wing tip. Riveting the wing skin to both the stringer and the ribs gives the wing
additional strength.

2.2.1.5 Stress plates

Stress plates are used on wings to support the weight of the fuel tank.
Some stress plates are made of thick metal and some are of thin metal

8
corrugated for strength. Stress plates are usually held in place by long rows of
machine screws, with self-locking nuts, that thread into specially mounted
channels. The stress-plate channeling is riveted to the spars and compression
ribs.

2.2.1.6 Gussets

Gussets, or gusset plates, are used on aircraft to join and reinforce

intersecting structural members. Gussets are used to transfer stresses from one
member to another at the point where the members join.

2.2.1.7 Wing tips

The wing tip, the outboard end of the wing, has two purposes: To
aerodynamically smooth out the wing tip air flow and to give the wing a finished
look.

2.2.1.8 Wing skins

Wing skins cover the internal parts and provide for a smooth air flow over
the surface of the wing. On full cantilever wings, the skins carry stress. However,
all wing skins are to be treated as primary structures whether they are on braced
or full cantilever surfaces. Wing skin carry shear load coming onto the wing.

2.3 Aircraft fuselage

The largest of the aircraft structural components, there are two types of
metal aircraft fuselages: Full monocoque and semimonocoque. The full
monocoque fuselage has fewer internal parts and a more highly stressed skin
than the semimonocoque fuselage, which uses internal bracing to obtain its
strength.

The full monocoque fuselage is generally used on smaller aircraft, because

the stressed skin eliminates the need for stringers, former rings, and other types
of internal bracing, thus lightening the aircraft structure.

The semimonocoque fuselage drives its strength from the following internal
parts: Bulkheads, longerons, keel beams, drag struts, body supports, former
rings, and stringers.

9
2.3.1 Components of the aircraft fuselage
2.3.1.1 Bulkheads

A bulkhead is a structural partition, usually located in the fuselage, which

normally runs perpendicular to the keel beam or longerons. A few examples of
bulkhead locations are where the wing spars connect into the fuselage, where
the cabin pressurization domes are secured to the fuselage structure, and at
cockpit passenger or cargo entry doors.

2.3.1.2 Longerons and Keel Beams

Longerons and keel beams perform the same function in an aircraft

fuselage. They both carry the bulk of the load traveling fore and aft. The keel
beam and longerons, the strongest sections of the aircraft, tie its weight to other
aircraft parts, such as power plants, fuel cells, and the landing gears.

2.3.1.3 Drag Struts and Other Fittings

Drag struts and body support fittings are other primary structural
members. Drag struts are used on large jet aircraft to tie the wing to the fuselage
center section. Body support fittings are used to support the structures which
make up bulkhead or floor truss sections.

Former rings and fuselage stringers are not primary structural members.
Former rings are used to give shape to the fuselage. Fuselage stringers running
fore and aft are used to tie in the bulkheads and former rings.

2.4 Aircraft Empennage Section and primary flight control surfaces

The empennage is the tail section of an aircraft. It consists of a horizontal

stabilizer, elevator, vertical stabilizer and rudder. The conventional empennage
section contains the same kind of parts used in the construction of a wing. The
internal parts of the stabilizers and their flight controls are made with spars, ribs,
stringers and skins.

2.4.1 Aircraft empennage components and flight control surfaces

2.4.1.1 Horizontal Stabilizer and Elevator

The horizontal stabilizer is connected to a primary control surface, i.e., the

elevator. The elevator causes the nose of the aircraft to pitch up or down.
Together, the horizontal stabilizer and elevator provide stability about the

10
horizontal axis of the aircraft. On some aircraft the horizontal stabilizer is made
movable by a screw jack assembly which allows the pilot to trim the aircraft
during flight.

2.4.1.2 Vertical Stabilizer and Rudder

The vertical stabilizer is connected to the aft end of the fuselage and gives
the aircraft stability about the vertical axis. Connected to the vertical stabilizer is
the rudder, the purpose of which is to turn the aircraft about its vertical axis.

2.4.1.3 Ailerons

Elevators and rudders are primary flight controls in the tail section.
Ailerons are primary flight controls connected to the wings. Located on the
outboard portion of the wing, they allow the aircraft to turn about the longitudinal
axis.

When the right aileron is moved upward, the left one goes down, thus
causing the aircraft to roll to the right. Because this action creates a tremendous
force, the ailerons must be constructed in such a way as to withstand it.

Flight controls other than the three primary ones are needed on high-
performance aircraft. On the wings of a wide-body jet, for example, there are as
many as thirteen flight controls, including high and low-speed ailerons, flaps, and
spoilers.

2.4.1.4 Flaps and Spoilers

Wing flaps increase the lift for take-off and landing. Inboard and outboard
flaps on the trailing edge of the wing travel from full up which is neutral
aerodynamic flow position to full down causing air to pile up and create lift.
Leading edge flaps - Krueger flaps and variable-camber flaps - increase the wing
chord size and thus allow the aircraft to take off or land on a shorter runway.
Spoilers, located in the center section span-wise, serve two purposes. They
assist the high-speed ailerons in turning the aircraft during flight, and they are
used to kill the aerodynamic lift during landing by spreading open on touchdown.

2.4.1.5 Trim Tabs

Connected to the primary flight controls are devices called trim tabs. They
are used to make fine adjustments to the flight path of an aircraft. Trim tabs are
constructed like wings or ailerons, but are considerably smaller.

11
2.5 Load transfer

The wing and fuselage are the two major components of an aircraft. The
horizontal and vertical tails bear close resemblance to the wing. A wing and
fuselage structure consists of a collection of basic structural elements. Each
component, as a whole, acts like a beam and a torsion member.

In reality, aircraft loads are in the form of air pressure on the skin,
concentrated loads from the landing gear, power plants, passenger seats, etc.
these loads are to be collected locally and transferred to the major load-carrying
members. Without proper care, these loads may produce excessive local
deflections that are not permissible from aerodynamic considerations.

2.5.1 Wing structure

The main function of the wing is to pick up the air and power point loads
and transmit them to the fuselage. The wing cross -section takes the shape of an
airfoil, which is designed based on aerodynamic considerations. The wing as a
whole performs the combined function of a beam and a torsion member. It
consists of axial members in stringers, bending members in spars and shear
panels in the cover skin and webs of the spars.

The spar is a heavy beam running span wise to take transverse shear
loads and span wise bending. It is usually composed of a thin shear panel (the
web) with a heavy cap or flange at the top and bottom to take bending. Wing ribs
are planar structures capable of carrying in-plane loads. They are placed chord
wise along the wing span. Besides serving as load distributers, ribs also hold the
skin stringer to the designed contour shape. Ribs reduce the effective buckling
length of the stringers and thus increase their compressive load capability.

Figure 2-3 Wing Structure

12
2.5.2 Fuselage

Unlike the wing, which is subjected to large distributed air loads, the
fuselage is subjected to relatively small air loads. The primary loads on the
fuselage include large concentrated forces from wing reactions, landing gear
reactions and payloads. For airplanes with carrying passengers, the fuselage
must also withstand internal pressures.

The fuselage structure is a semimonocoque construction consisting of a

thin shell stiffened by longitudinal axial elements (stringers and longerons)
supported by many transverse frames (bulkheads) along its length. The fuselage
skin carries the shear stresses produced by internal pressures. The stringers
carry bending moments and axial forces; they also stabilize the thin fuselage
skin.Fuselage frames are used to maintain the shape of the fuselage and to
shorten the span of the stringers between supports in order to increase the
buckling strength of the stringer.

Figure 2-4 Fuselage structure

2.6 Summary

To perform dynamic analysis for aircraft wing, a reduced order model is

modeled which is amenable for using spectral finite element method. The
components of aircraft wing which are sufficient to capture the dynamic behavior
are spars, ribs and skin. Using these parts a reduced order model of the aircraft
wing modeled and the formulation of this model is done using spectral finite
element method.

13
Chapter 3

Wave Propagation and Spectral finite element method

This chapter gives an introduction to wave propagation and spectral analysis. It

also includes the formulation of 1D elements (rod and beam) using spectral finite
element method.

3.1 Wave Propagation

Wave propagation [1] is a transient dynamic phenomenon res ulting from

short duration impact loads of high frequency content. Impact loads are very
often encountered in aircraft structures in the form of gusts, bird hits, tool drops
during refurbishment and other transient environments during flight. Structures
are an important application Here, the wave propagation studies are used to
understand the dynamic behavior of structure subjected to impact loads. Here,
the presence of even minute defects and cracks, which occur at initiation of
damage and may lead to structural failure, can be detected with help of
diagnostic waves. In aircraft structures, the undesired vibration and noise
transmit from the source to other parts in the form of wave propagation, and the
reduction of this vibration is also another important application of wave
propagation studies.

The difference between structural dynamics and wave propagations in

structures is the high frequency excitation in the latter problems. If the temporal
variation of load is large (of the order of seconds), the frequency content of
loading will be low and such problems come under the category of structural
dynamics. Since the low frequency load excites only the first few modes, the
response of the structure can very well be approximated by the first few normal
modes. So the structure can be idealized with fewer degrees of freedom.
However, when the duration of the load is of the order of micro to milliseconds,
the wave starts propagating in the structure with a finite velocity. So the degrees
of freedom required for such models will be many orders higher than structural
dynamics problems and they are categorized as wave propagation problems.

In wave propagation, the group velocity of propagatio n, wave numbers,

level of attenuation of the response and the wavelengths are important in the
analysis. Therefore, phase information of the waves, which is of no major
concern in structural dynamic problems, become really importa nt in wave
propagation analysis.

14
 In wave propagation problems, analysis method based on frequency
domain is normally used. This is because of the complexities involved in solving
the multi-model phenomena problem in the time domain. Here, all the governing
equations, boundary conditions and variable are transformed to the frequency
domain using Fast Fourier Transforms, which are the numerical implementation
of discrete Fourier transforms.

3.2 Spectral Analysis

Spectral analysis is the synthesis of waveforms from the superposition of

many frequency components, used to obtain the local behavior of different
waveguides and their characteristics. The local behavior is synthesized to obtain
the global wave behavior. The phase changes incurred during wave propagation
can be expressed using spectral description of waves, by making use of the
governing differential for the problem.

In spectral analysis, wave numbers and group speeds are important to

understand the wave mechanics in a waveguide. They enable us to know
whether the wave mode is propagating or attenuating or a combination of these
two. For a propagating wave, the wave numbers will let us understand whether it
is dispersive or non-dispersive. If a wave profile changes its shape as it
propagates, the wave is said to be dispersive, otherwise it is called non-
dispersive.

3.2.1 Waveguide

Any structural element which guides the wave in a definite pattern is a

waveguide. Rods are axial or longitudinal waveguides, as they support only axial
motion. Beams are called flexural waveguides, as they support only twisting
motion. In 2D frames, axial and flexural motions both appear simultaneously and
are not coupled with each other.

In general, a waveguide supporting n different motions will have n highly

coupled governing partial differential equations and hence n different wave
modes, each of which will have unique wave numbers.

3.2.2 Wave number and Spectrum relation

The wave number (k) is frequency dependent parameter which

determines whether the wave propagates or not, or will propagate after certain
frequency. If a wave changes its shape during propagation, it is dispersive wave
and if the shape of the wave is retained it is non-dispersive.

15
 If the wave number (k) is a linear function of the frequency of the
frequency (), as k = C , (C is constant) then the waves will be non-dispersive
in nature, e.g. rods. But if the wave number (k) is a non-linear function of the
frequency (), as k = C n, then the waves will be dispersive in nature e.g. in
beams and plates. The relation between wave number and frequency is called
spectrum relation.

In general, the wave number can be written in complex form as k kR kI ,

at a given frequency. The real part of the wave number ( k R ) represents the
propagating wave, while the imaginary part ( k I ) represents the spatially decaying
damped wave mode.

3.2.3 Phase speed, Group and Dispersion relations

Phase speed ( CP ) represents the speed of the individual particles in a

wave packet and does not associate with transfer of mass, energy or momentum.
If the phase speed is constant and independent of frequency, the wave is non-
dispersive in nature. Phase speed is computed as

Cp (3.1)
kR

Group speed ( Cg ) represents the speed of propagation of a group of

waves having same frequency and is associated with the transfer of energy and
should be bounded. The time of arrival of all the waves depend on this
frequency. If the group speed is constant and independent of frequency, the
wave is again non-dispersive in nature.

d
Cg (3.2)
dk R

The relation between phase/group speeds and frequency is called

dispersion relation and the respective curves are generally called as dispersion
curves. For a non-dispersive system, the phase speed and group speed will be
same.

3.2.4 Relation between phase speed and group speed

From Eqn. (3.1), we get

16
kR (3.3)
Cp

Substituting Eqn.(3.3) in Eqn.(3.2), gives the following form as

1 1 1
d dC p 2
dC p
Cg d d d C p Cp (3.4)
Cp Cp C p2 d

From Eqn.(3.4), it is obvious that,

When,

dC p
0, Cg Cp
d Also,

dC p
, cut off Cg 0
d

3.2.5 Cut-off frequency

Cut-off frequency corresponds to the frequency, Here a purely imaginary
wave number changes to a purely real or a complex value with a non-zero real
part or vice-versa. Physically, it represents the frequency at which a completely
damped non-propagating wave-mode transforms to propagating mode or vice-
versa.

3.3 Spectral finite Element Method (SFEM)

Spectral finite element method (SFEM) [2], is considered as the most suitable
technique for studying wave propagation in structural waveguides subjected to
high frequency content loading. For this type of loading, the wave propagation
analysis by conventional finite element method (CFEM) is computationally
expensive, since the element size should be of the order of wavelength. However
in SFEM, the governing equation is transformed first to frequency domain using
discrete Fourier transform (DFT). In doing so, the governing partial differential
equation (PDE) is reduced to a set of ordinary differential equations (ODE) with
constant coefficients, where the time coordinate is removed from the formulation
and the frequency is introduced as a parameter. Hence the time doesnt appear
explicitly, but is replaced by phase relationships among the field variables which
are complex in nature. The resulting ODEs can be solved exactly and the
elements are formulated using the exact solution of the governing ODEs as

17
interpolating functions. The size of the global dynamic stiffness matrix is
appreciably smaller than that involved in the CFEM. The steps to be followed in
SFEM are as follows:

a) The exact dynamic stiffness is generated and assembled for each

sampled frequency and the force-displacement relationship is solved for
unit impulse to determine the system transfer function (also called the
frequency response function, FRF)
b) The FRF is then convolved with the Fourier coefficient of the load.
c) Next, Inverse fast Fourier transform (IFFT) is used to get the time history
of the response.

Figure 3-1 Outline of FFT Based Spectral Finite Element Method.

3.4 Spectral Element for Rods

In the aircraft structural assembly the rod like components are stringers in
the wing assembly and longerons in the fuselage assembly.

Because the dynamic behavior of a structure is of great importance in

engineering, it is necessary to be able to accurately predict the dynamic
characteristics of the structure. The finite element method (FEM) has been used
extensively in structural dynamics. The finite element model may provide
accurate dynamic characteristics of a structure if the wavelength is large
compared to the mesh size. However, the finite element solutions become
increasingly inaccurate as the frequency increases. Although the accuracy can
be improved by refining the mesh, this is sometimes prohibitively expensive in
18
two- and three-dimensional problems. The conventional finite element (mass and
stiffness) matrices are usually formulated from assumed frequency- independent
polynomial shape functions. Because the vibrating shape of a structure varies
with the frequency of vibration in reality, the FEM requires subdivision of the
structure into finite elements (or a mesh) for accurate solutions. Alternatively, if
the shape functions are frequency dependent, then the subdivision may not be
necessary. This leads to the dynamic stiffness matrix method.

The dynamic stiffness matrix is formulated from the frequency-dependent

(or dynamic) shape functions that are exact solutions of the governing differential
equations [15]. This frequency-dependent matrix has both mass and stiffness
properties of the structural element embedded in it. Because the dynamic
stiffness matrix is obtained by directly solving the governing differential
equations, all assumptions, being within the limits of the classical theory of
differential equations of motion only, are less severe than those made in the FEM
and other approximate methods. Thus, the dynamic stiffness matrix method
certainly provides more accurate predictions of structural dynamic
characteristics. Because the dynamic stiffness matrix method does not require
subdivision of a structural member into finite elements, it may eliminate
discretization errors and is capable of predicting an infinite number of Eigen
solutions by means of a minimum number of degrees of freedom (DOF). This
improves the solution accuracy and lowers the computation cost significantly.
Because the dynamic stiffness matrix is stiffness formulated, it can be assembled
at a specified frequency in a completely analogous way to that used for the FEM.
Thus, the assembly feature of the FEM can be equally applied to the dynamic
stiffness matrix method.

The dynamic stiffness relation is found via dynamic shape functions, these
are essentially interpolation functions between the element ends, but instead of
being simple polynomials, they are the exact displacement distributions.

(a) Finite Element (b) Throw-off Element

Figure 3-2 Nodal Loads and Degrees of Freedom for Longitudinal Spectral Elements.

19
3.4.1 Shape functions
Let be the general longitudinal displacement for a rod, be the longitudinal
displacement at nodal point 1, be the longitudinal displacement at nodal point
2 and be the member forces for a rod.

Let be the wave number for the longitudinal spectral element, be the
youngs modulus of the rod, be the cross sectional area of the rod and L be
the length of the rod.

The general longitudinal displacement for a rod is

u x Ae ik1 x
Be ik1 ( L x )
(3.5)

Here

A, B are constants determined from the boundary conditions on the

element.

The displacement end conditions for a finite element, shown in figure 3.2(a), are

u 0 u 1 A Be ik1L
, u L u 2 Ae ik1L
B (3.6)

After solving for A, B in terms of the nodal displacements, the longitudinal

displacement at an arbitrary point in a finite rod is

(3.7)

Here

are the frequency-dependent rod shape functions, are

given by

And

The significance of the shape functions is that the complete description of

the element is captured in the two nodal degrees of freedom . Since the
spectral element can be very long , using the shape function and the nodal values
response between nodes is computed.

20
The displacements and the member forces are

(3.8)

(3.9)

The displacement end conditions for a throw-off element, shown in figure 3.2(b),
are

The displacement at an arbitrary point for a throw-off element is

(3.10)

The shape function is frequency dependent and is complex even in simple

rod case with no damping.

3.4.2 Dynamic stiffness for rods

The member loads for finite element at each end of the rod are related to the
displacements by

(3.11)

(3.12)

Equation (3.11) and (3.12) are written as,

(3.13)

Equation (3.13) in short notation is written as,

(3.14)

Here is the frequency dependent dynamic element stiffness matrix for the rod,
it is symmetric as seen from the explicit matrix form

(3.15)

The stiffness relation for throw-off element is

(3.16)

Equation (3.15) gives the stiffness relation for finite element and equation (3.16)
gives the stiffness relation for throw-off element
21
3.5 Spectral elements for beams

In the aircraft structural assembly the components which can be

approximated as beam elements are spars in the wing assembly. The wing
assembly and fuselage assembly separately as a whole resemble their behavior
with the beam; they can be approximated as beam elements.

The approach followed here for beams is much the same as in the previous
section for rods but since beams are two-mode systems, then there must be two
degrees of freedom at each node making the stiffness matrix of order [44].

(a) Finite element (b) Throw-off element

Figure 3-3 Nodal Loads and Degrees of Freedom for Flexural Spectral elements.

3.5.1 Shape functions

Let be the general transverse displacement, be the general transverse
rotation degree of freedom for flexural spectral element.

Let be the two wave numbers for two-mode system of the flexural spectral
element, L is the length of the flexural spectral element.

The general displacement for a Timoshenko beam is

(3.17)

x Ae ik1 x
Be ik2 x
Ce ik1 ( L x )
De ik2 ( L x )
(3.18)

Here

A,B,C,D - Coefficients determined from the boundary conditions on the

element.

- Amplitude ratios given by

(3.19)

22
For throw-off element with C=D=0, the end conditions require that

(3.20)

The Coefficients A and B can be re-written in terms of the nodal displacement

and rotation .

The displacement can be written as

x g x 1 g x 1, x g 1 ( x) 1 g 2 ( x) 1 (3.21)
1 2

Here

are the frequency dependent beam shape functions and are given
by

g 1 ( x) [ R1e ik1 x
R2e ik2 x
]/
g 2 ( x) [ R1 R2e ik1 x
R1R2e ik2 x
]/
(3.22)
g 1 ( x) [e ik1 x
e ik2 x
]/
g 2 ( x) [ R2e ik1 x
R1e ik2 x
]/

R1 R2

The complete description of this element has now been captured in the two nodal
degrees of freedom .

For a finite element, the end conditions on the element are

(0) 1 , (0) , ( L) 2 , ( L) (3.23)

1 2

The displacement solution is

T ik1 x ik2 x ik1 ( L x ) ik2 ( L x )
( x), ( x) 1
Ae 2
Be 3
Ce 4
De (3.24)

Here

i
Are the respective amplitude ratios

Equation (3.24) is written as

T
( x), ( x) u ( x) [ A ][e( x)] A [ B ][e( L x)] B (3.25)

Here the two vectors of amplitude are

23
 ik1 x
A C e 0
A , B , [e( x)] ik2 x
(3.26)
B D 0 e

We have defined the amplitude ratio matrices as

[ A ] , [ B ] (3.27)
1 2 3 4

The matrix [ ] is modal matrix. It is fully populated matrix of order [22] and
typically is not symmetric. The matrix [e( x)] is diagonal and is the unit matrix
when the argument is zero.

Nodal displacement at x=0 and x=L is written in terms of vectors {A} and {B} as,

1 A
[ ][e(0)] [ ][e( L)] B
1 A B
(3.28)
2 [ A ][e( L)] [ B ][e(0)] C
D
2

Solving this as a system of matrices to get

A 1
B
[G ] 1
[G ] u (3.29)
C 2
D
2

The elements of [44] matrix [G ] are

G11 (r1 r2e2 ) / , G12 R2 ( r1 r2e2 ) /

G 13 (r2 r1e2 ) / , G14 R2 ( r1e2 r2 ) /
G 21 (r1 r2e1 ) / , G 22 R1 ( r1 r2e1 ) /
(3.30)
G 23 (r1e1 r2 ) / , 24 G R1 ( r1e1 r2 ) /
G 31 G13 , G 32 G14 , G 33 G11 , G 34 G12
G 41 G 23 , G 42 G 24 , G 43 G , G
21 G
44 22

Here

r1 ( R1 R2 )(1 e1e2 ), r2 ( R1 R2 )(e1 e2 ), r12 r22 (3.31)

24
 and
ik1L ik2 L
e1 e , e2 e .

Substituting for the coefficients in the assumed displacement field of equation

(3.17 and 3.18), the displacements and rotations can now be written in the
compact form as
T T
( x ) N [G ] u , ( x) N [ L1 ][G ] u (3.32)

T T
Here N [G ] and N [ L1 ][G ] give the shape functions for ( x) and ( x) ,

respectively. The {41} vector N is a function of x given by

T
N e ik1L
e ik2 L
e ik1 ( L x )
e ik2 ( L x )
(3.33)

While the matrix [ L1 ] is diagonal and contain the amplitude ratios

[ L1 ] diag[ R1 , R2 , R1 , R2 ] (3.34)

3.5.2 Dynamic stiffness for beams

The stiffness relation for the beam element is established by taking
appropriate space derivatives of the shape function.

For the throw-off element,

( x) g 1 ( x) 1 g 2 ( x) 1, ( x) g ( x) g ( x)
1 1 2 1 (3.35)

The nodal loads are expressed in terms of the displacement degrees of freedom
through the relationships for the structural resultants

d d2
M ( x) EI , V ( x) EI I 2
(3.36)
dx dx 2

Evaluation of these at x=0 leads to the dynamic stiffness as

V (0) V 1 k11 k12 1

EI (3.37)
M (0) M k k
1 21 22 1

Here the individual stiffness terms are

25
 k22 k12 i (k1 k2 ) i ( R1k1 R2 k 2 )
k11 , k12 k21 , k22 (3.38)
R1 R2 R1 R2 R1 R2

This symmetric throw-off stiffness matrix is complex and therefore the structural
stiffness matrix will also be complex after assembling.

The finite element looks the same as the conventional element except that
the mass is distributed instead of concentrated at the ends. It is assumed that
there are no applied loads between the ends.

Considering a beam waveguide of finite length L with loads applied only at the
ends. The shape functions are
T T
( x ) N [G ] u , ( x) N [ L1 ][G ] u (3.40)

Here

u is the collection of four degrees of freedom, This leads to the moments

and shear expressions.
T
M ( x) EI N ' [G ] u
T T
(3.41)
V ( x) EI N '' [G ] u I 2
N [ L1 ][G ] u

Here primes indicate differentiation. At x=0 and x=L equation (3.41) gives,

V (0) 1
M (0)
[ N ][G ] 1
(3.42)
V ( L) 2
M ( L)
2

Here the square matrix [ N ] contains the derivatives

T T
EI N ''(0) I 2
N (0) [ L1 ]
T
EI N '(0)
[ N ] T T
(3.43)
EI N ''( L) I 2
N (0) [ L1 ]
T
EI N '( L)

26
The relation between the loads and degrees of freedom is

F [k] u (3.44)

Here

[ k ] - Dynamic stiffness matrix,

F - Nodal load Vector with components V1 M 1 V2 M 2 , and

u - Nodal displacement vector with components 1 2

1 2

The individual stiffness terms of the matrix [ k ] are:

k 11 (k22 k12 )(r1 z22 r2 z21 ) / ,

k 13 (k12 k22 )(r1 z21 r2 z22 ) / ,
k 12 [ ik2 (r1 z11 r2 z12 ) ik1 (r1 z11 r2 z12 ) / ,
k 14 [ ik1 (r1 z12 r2 z11 ) ik2 (r1 z12 r2 z11 ) / , (3.45)
k 22 ( ik1 R2 ik2 R1 )(r1 z22 r2 z21 ) / ,
k 24 (ik1 R2 ik2 R1 )(r1 z21 r2 z22 ) / ,
k 23 k 14 , k 33 k 11 , k 34 k 12 , k 24 k 22 .

Here

r1 ( R1 R2 ) z11 ,
r2 ( R1 R2 ) z12 ,
(r12 r22 ) / EI ,
ik1L ik2 L
z11 1 e e , (3.46)
ik1 L ik2 L
z12 e e ,
ik1 L ik2 L
z22 1 e e ,
ik1 L ik2 L
z12 e e .

And the remaining terms are obtained using symmetry.

27
3.6 Spectral element for 2D frames

The 2D frames are the combination of 1-D beam elements connected

together, when beams are part of connected structures they additionally support
axial loads. As the transverse displacement and axial displacement are not
coupled with each other, the stiffness matrix for a 2D frame is taken as addition
of rod stiffness matrix and beam stiffness matri x. This gives the effect of
connected structures using 1D beam elements.

(a) Finite element (b) Throw-off element

Figure 3-4 Nodal loads and degrees of freedom for spectral element of 2D frame

Therefore the stiffness matrix for elements of 2D frames is of dimension

[66], here the active degrees of freedom are axial displacement, transverse
displacement and rotation per node.

3.7 Summary

This chapter has briefly discussed about the construction of the dynamic
stiffness matrix for 1D rod, 1D beams and 2D frames.

28
Chapter 4

Reduced Order Modeling of Aircraft Wing

This chapter covers the reduced order modeling of aircraft wi ng from a

three-dimensional model to a two-dimensional beam-frame model which is
amenable for frequency response analysis using spectral finite element method.

4.1 Major Components of Aircraft Wing

The major components of aircraft wing are spars, ribs and skin. To capture
the response at the joint of these structures, we take into account of the rivet
connections. Here the rivet modeling is not done; instead the effective stiffness of
the rivet is added at the respective joint nodes.

Figure 4-1 Schematic Top View of Aircraft.

The above figure illustrates top sectional view of the aircraft with attention
given to wing showing its major components spars, ribs and skin, right wing
being un-skinned showing spars running along the span-wise direction, ribs
placed along the chord-wise direction and its left wing being skinned.

29
4.2 Equivalent 2D Beam-Frame Model of an Aircraft Wing

Task is to capture the dynamic characteristics of aircraft wing. It has been

discussed that to model the three dimensional wing and perform analysis is too
expensive. So, here an equivalent two-dimensional beam-frame model of the
aircraft wing is modeled which will be amenable to solve for frequency response
analysis using spectral finite element method.

Instead of modeling whole aircraft wing a 2D model of the wing along its
span wise direction have modeled, which is capable to capture transverse
displacement accurately. The components which have to be taken into account
while a reduced order model of the wing is being modeled are spars, ribs and
skin.

In the aircraft wing, the spars bear the bending and torsional loads over the
wing. The ribs bear the in-plane transverse loads over the wing. The wing skins
bear the shear load over the wing.

Here is the schematic view of the aircraft wing in the reduced model as a
two-dimensional beam-frame model which is joined to fuselage at one end and
free at the other end.

Figure 4-2 Schematic Front View of Reduced order model of Aircraft Wing.

In the above figure the elements which are in the centre from wing root to
tip are replaced by spars, the two series of elements above and below the spar
elements are replaced by top and bottom skin respectively and the elements
normal to spar elements are replaced by ribs (each rib is modeled as a
combination of two elements).

30
 The aircraft wing is taken as a stiffened structure, using three of its major
components and rivets used to join them. To capture the dynamic characteristics
accurately the rivet connections are taken into account, which are used to join
ribs to the skin. These rivets are running from leading edge to trailing edge of the
wing joining the top and bottom skin with the ribs. The each line of rivets from
leading edge to trailing edge of the wing has taken as one rivet line connection,
the above connection is valid for the three dimensional wing but when it comes to
the reduced order model they have only two nodes. The rivet line connection is
replaced with a single distributed spring between these nodes.

The distributed spring is a combination of three springs one each in

longitudinal, transverse and rotational directions.

Figure 4-3 Distributed Spring Between node-1 and node-2.

Here
Kx Stiffness of the spring along longitudinal direction,
Ky Stiffness of the spring along transverse direction, and
K Stiffness of the spring along rotational direction.

4.3 Segments of the 2D Beam-Frame Model

To get the dynamic stiffness matrix for the whole 2D beam-frame model is
difficult, so the whole model is consider has an assemblage of segments. The
part of the model shown in figure 4-4 is a segment between two consecutive ribs.

31
 Figure 4-4 2D Beam-Frame model of a Segment of the wing.

Figure 4-5 shows the zoomed view of the segment of beam-frame model.
Considering the beam-frame model of one segment formed with seven elements,
fourteen distinct nodes and also four distributed springs at the respective joints.

Figure 4-5 A Zoomed-in View of 2D Beam-Frame model of one segment of the wing.

32
 Figure 4-6 One Segment of the 2D Beam-Frame model of the Aircraft Wing.

Figure 4-6 shows the 2D beam-frame model of one segment which goes
into the beam-frame model of the wing.

4.4 Formulation of dynamic stiffness matrix of one segment of 2D

beam-frame model

The main objective is to get global dynamic stiffness matrix of one segment
of 2D beam-frame model using spectral finite element method. Figure 4-5 shows
that there are seven elements in the model, which has modeled with 1D beam
elements, whose local stiffness matrix is similar to the elements of 2D frame with
three active degrees of freedom at each node.

The stiffness matrix for all the seven elements having three degrees of
freedom at each node can be assembled together to get the global stiffness
matrix. After assembling these elements there is no connectivity between some
nodes. This model without the distributed spring does not satisfy compatibility
condition; also the global stiffness matrix will be singular.

The stiffness matrix due to the distributed spring is defined as follows.

Let ar Area of the single rivet among the rivet line connection on the wing,
d s Shank diameter of the rivet,

33
 n Numbers of rivets along the each rivet line connection,
Ar Effective area of the distributed spring in reduced order model,
Er Youngs modulus of the rivet material
Gr Shear modulus of the rivet material and
hr Length of the rivet
Then
Ar n ar (4.1)
d s2
ar (4.2)
4
Effective stiffness of the distributed spring is defined as,
Kx 0 0
K 0 K 0 y (4.3)
0 0 K
Here
Kx Stiffness of the spring along longitudinal direction,
Ky Stiffness of the spring along transverse direction, and
K Stiffness of the spring along rotational direction.
Considering each spring as an element the stiffness along longitudinal,
transverse and rotational direction are as follows
Gr Ar Gr nar
Kx (4.4)
hr hr
Er Ar Er nar
Ky (4.5)
hr hr
M K y hr
K K y hr2 Er nar hr (4.6)
1
hr
The stiffness due to distributed spring is dependent on material properties
of rivet, no. of rivets in the rivet-line connection, length of the rivet and shank
diameter.

From the figure 4-6, Effective stiffness of the distributed spring between node 1
and 6 is given by,

34
 K x(1) 0 0
K *
1 0 K y(1) 0 (4.7)
0 0 K (1)

Similarly between node 3 and 4,

K x(2) 0 0
K *
2 0 K y(2) 0 (4.8)
0 0 K (2)

Similarly between node 5 and 10,

K x(3) 0 0
K 3* 0 K y(3) 0 (4.9)
0 0 K (3)

Similarly between node 7 and 8

K x(4) 0 0
K *
4 0 K y(4) 0 (4.10)
0 0 K (4)

Considering the distributed spring between node 1 and 6, from the

equilibrium of nodal forces and moments at node 1

FX N 1 K X (u10 u60 )
FY V1 KY (v1 v6 ) (4.11)
M M 1 K ( 1 6 )

The above equations in matrix notation can be written as

N 1 KX 0 0 u10 u60
V 1 0 KY 0 v1 v6 (4.12)
M 1 0 0 K 1 6

f1 K * (u1 u6 ) (4.13)

From the equilibrium of nodal forces and moments at node 6

35
f6 K * (u6 u1 ) (4.14)

Similarly,

Considering the distributed spring between node 3 and 4

From the equilibrium of nodal forces and moments at node 3

f3 K * (u3 u4 ) (4.15)

From the equilibrium of nodal forces and moments at node 4

f4 K * (u4 u3 ) (4.16)

Considering the distributed spring between node 5 and 10

From the equilibrium of nodal forces and moments at node 5

f5 K * (u5 u10 ) (4.17)

From the equilibrium of nodal forces and moments at node 10

f10 K * (u10 u5 ) (4.18)

Considering the distributed spring between node 7 and 8

From the equilibrium of nodal forces and moments at node 7

f7 K * (u7 u8 ) (4.19)

From the equilibrium of nodal forces and moments at node 8

f8 K * (u8 u7 ) (4.20)

Considering the dynamic stiffness matrix of element wise waveguide of 2D

beam-frame model as follows,

For element (1)

K 11(1) K 12(1) u1 f1
(4.21)
K (1)
21 K (1)
22
u2 f
11

36
For element (2)

K11(2) K12(2) u2 f12

(4.22)
K (2)
21 K (2)
22
u3 f 3

For element (3)

K 11(3) K 12(3) u6 f6
(4.23)
K (3)
21 K (3)
22
u7 f
7

For element (4)

K 11(4) K12(4) u2 f2
(4.24)
K (4)
21 K (4)
22
u9 f
9

For element (5)

K 11(1) K 12(1) u4 f4
(4.25)
K (1)
21 K (1)
22
u5 f
5

For element (6) and

K11(2) K12(2) u8 f8
(4.26)
K (2)
21 K (2)
22
u9 f13

For element (7)

K 11(3) K 12(3) u9 f14

(4.27)
K (3)
21 K (3)
22
u10 f
10

After assembling wave guide elements and including the effect of the
distributed springs through the equilibrium equations the global dynamic stiffness
matrix of one segment of the 2D beam-frame model is,

37
 f1 K 1* (u1 u6 )
K11 K12 0 0 0 0 0 0 0 0 u1
f2 f11 f12
K 21 K 22 K 23 0 0 0 0 0 K 29 0 u2
0 K32 K 33 0 0 0 0 0 0 0 u3 f K * (u u )
3 2 3 4

0 0 0 K 44 K 45 0 0 0 0 0 u4 f4 K 2* (u4 u3 )
0 0 0 K54 K55 0 0 0 0 0 u5 f5 K 3* (u5 u10 )
0 0 0 0 0 K 66 K 67 0 0 0 u6 f6 K 1* (u6 u1 )
0 0 0 0 0 K 76 K 77 0 0 0 u7 f K * (u u )
7 4 7 8
0 0 0 0 0 0 0 K88 K89 0 u8
f8 K 4* (u8 u7 )
0 K92 0 0 0 0 0 K 98 K 99 K 910 u9
f9 f13 f14
0 0 0 0 0 0 0 0 K109 K1010 u10
f10 K 3* (u10 u5 )
(4.28)

Re-arranging the right hand side terms to left hand side, the global dynamic
stiffness matrix becomes as,

K11 K12 0 0 0 K16 0 0 0 0 u1 F1

K 21 K 22 K 23 0 0 0 0 0 K 29 0 u2 F2
0 K 32 K33 K34 0 0 0 0 0 0 u3 F3
0 0 K 43 K 44 K 45 0 0 0 0 0 u4 F4
0 0 0 K 54 K 55 0 0 0 0 K 510 u5 F5
(4.29)
K 61 0 0 0 0 K 66 K 67 0 0 0 u6 F6
0 0 0 0 0 K 76 K 77 K 78 0 0 u7 F7
0 0 0 0 0 0 K87 K88 K89 0 u8 F8
0 K 92 0 0 0 0 0 K 98 K 99 K 910 u9 F9
0 0 0 0 K105 0 0 0 K109 K1010 u10 F10

[k] u F

Here coefficients of the above global dynamic stiffness matrix are given by,

38
K11 K 11(1) K 1*
K 12 K (1)
12

K16 K 1*
K 21 K (1)
21

K 22 K 22(1) K 11(2) K 11(4)

K 23 K (2)
12

K 29 K 12(4)
K 32 K (2)
21

K 33 K 22(2) K 2*
K 34 K *2

K 43 K 2*
K 44 K (5)
11 K 2*
K 45 K 12(5)
K 54 K (5)
21

K 55 K 22(5) K 3*
K 510 K * 3

K 61 K *
1

K 66 K (3)
11 K 1*

K 76 K 21(3)
K 77 K (3)
22 K 4*
K 78 K 4*
K87 K *
4

K88 K 11(6) K 4*
K89 K (6)
12

K92 K 21(4)
K98 K (6)
21

K99 K 22(5) K 11(7) K 22(4)

K910 K (7)
12

K105 K 3*
K109 K (7)
21

K1010 K 22(7) K 3*

39
F1 f1
F2 f
2 f11 f12
F3 f3
F4 f4

F5 f5
F6 f6

F7 f7
F8 f8

F9 f9 f13 f14
F10 f 10

4.5 Global dynamic stiffness matrix of 2D beam-frame model

The stiffness matrix of the beam-frame model is taken into two instants one
is due to the beam elements and the other due to the distributed springs. So after
adding these two matrices, the global stiffness matrix of the 2D beam-frame
model of the wing is obtained.

4.6 Summary

Through this global stiffness matrix the frequency response analysis of the
2D beam-frame model has been performed. The response history at all nodal
points had obtained in the frequency domain due to the high frequency loading at
tip node.

40
Chapter 5

Results and Discussions

This chapter comprises of two major parts first half includes response
analysis for 1D rod, 1D beam and 2D frame using spectral finite element method.
The second half includes two different cases of 2D beam-frame model of aircraft
wing using the formulation from the previous chapter of this dissertation.

5.1 Dynamic Response of an Isotropic Rod

Consider a semi-infinite rod, fixed at one end and impacted at the other
end. Model has two elements, first element is of finite length and second element
is considered as extending to infinity (a throw-off element). A throw-off element is
introduced in order to avoid reflections from the boundary.

The 1D rod of aluminum has been taken with material properties as follows,

Elastic modulus: 70 GPa

Shear modulus: 26.9 GPa
Cross-section: 10 10mm2
Length of the rod: 1m
The axial load is applied at node 3 and axial velocity history is measured at the
same.

Figure 5-1 Semi-Infinite Rod Modeled with two Elements.

The following plots shows the axial velocity history at node 3 due to
high-frequency pulse in the axial direction at the same, the applied load history is
shown in the inset image of each plot.

41
The spectrum relation of an isotropic rod for an undamped case is

(5.1)

Here is linear in frequency.

For the dispersion relation, constant phase and group speeds of

(5.2)

Figure 5-2 Axial Velocity history at Node-3, due to loading (inset of the plot) in the Axial direction.

Using the material properties,

From equation (5.2),

Constant speed of the wave travelling through rod = 5103 m/s

Time taken by the wave to reach the interface from the free end = 0.210-3 sec
Time taken by the wave to reach back to the free end after reflecting from the
interface = Time of flight = 2 (0.210-3) = 0.410-3 sec

42
From the above plot,

Time of launch of the pulse = 0.1475 10-3 sec

Time of reflection of the pulse = 0.5510-3 sec

Time of flight of the pulse = 0.402510-3 sec

The time of flight of the pulse in the response history is equal to that predicted
using material properties.

Figure 5-3 Axial Velocity history at Node-3, due to loading (inset of the plot) in the Axial direction.

Using the material properties,

From equation (5.2),

Constant speed of the wave travelling through rod = 5103 m/s

Time taken by the wave to reach the interface from the free end = 0.210-3 sec
Time taken by the wave to reach back to the free end after reflecting from the
interface = Time of flight = 2 (0.210-3) = 0.410-3 sec
From the above plot,

43
Time of launch of the pulse = 0.15 10-3 sec

Time of reflection of the pulse = 0.547510-3 sec

Time of flight of the pulse = 0.397510-3 sec

The time of flight of the pulse in the response history is equal to that predicted
using material properties.

Figure 5-4 Axial Velocity history at Node-3, due to loading (inset of the plot) in the Axial direction.

Using the material properties,

From equation (5.2),

Constant speed of the wave travelling through rod = 5103 m/s

Time taken by the wave to reach the interface from the free end = 0.210-3 sec

Time taken by the wave to reach back to the free end after reflecting from the
interface = Time of flight = 2 (0.210-3) = 0.410-3 sec

44
From the above plot,

Time of launch of the pulse = 0.5910-3 sec

Time of reflection of the pulse = 0.9910-3 sec

Time of flight of the pulse = 0.410-3 sec

The time of flight of the pulse in the response history is equal to that predicted
using material properties.

Figure 5-5 Axial Velocity history at Node-3, due to loading (inset of the plot) in the Axial direction.

Using the material properties,

From equation (5.2),

Constant speed of the wave travelling through rod = 5103 m/s

Time taken by the wave to reach the interface from the free end = 0.210-3 sec

Time taken by the wave to reach back to the free end after reflecting from the
interface = Time of flight = 2 (0.210-3) = 0.410-3 sec

45
From the above plot,

Time of launch of the pulse = 0.3410-3 sec

Time of reflection of the pulse = 0.7410-3 sec

Time of flight of the pulse = 0.410-3 sec

The time of flight of the pulse in the response history is equal to that predicted
using material properties.

5.2 Dynamic Response of an Isotropic Beam

Consider a semi-infinite Euler-Bernoulli beam, fixed at one end and

impacted at the other end. Model has two elements, first element is of finite
length and second element is a throw-off element. A throw-off element is
introduced in order to avoid reflections from the boundary.

The 1D beam of aluminum has been taken with material properties as follows,

Elastic modulus: 70 GPa

Shear modulus: 26.9 GPa
Cross-section: 10 10mm2
Length of the beam: 1m
The transverse load is applied at the node 3 and transverse velocity history is
measured at the same.

Figure 5-6 Semi-Infinite Beam Modeled with Two Elements.

46
 The following plots shows the transverse velocity history at node 3 due
to high-frequency pulse in the transverse direction at the same, the applied load
history is shown in the inset image of each plot.

The spectrum relation of an isotropic beam for an undamped case is

(5.3)

The mode 2 behavior is entirely imaginary for the undamped case, hence there is
no propagation behavior for this mode. Considering only the mode 1 wave
motion gives the phase speed and group speed as

(5.4)

Figure 5-7 Transverse velocity history at node 3 due to loading in the inset of the plot.

47
Using the material properties,

From equation (5.4),

Group speed of the wave travelling through beam = 3.01110 3 m/s

Time taken by the wave to reach the interface from the free end = 0.33210-3 sec

Time taken by the wave to reach back to the free end after reflecting from the
interface = Time of flight = 2 (0.33210-3) = 0.66410-3 sec

From the above plot,

Time of launch of the pulse = 0.59510-3 sec

Time of reflection of the pulse = 1.26510-3 sec

Time of flight of the pulse = 0.6710-3 sec

The time of flight of the pulse in the response history is equal to that predicted
using material properties.

Figure 5-8 Transverse velocity history at node 3 due to loading in the inset of the plot.

48
Using the material properties,

From equation (5.4),

Group speed of the wave travelling through beam = 3.01110 3 m/s

Time taken by the wave to reach the interface from the free end = 0.33210-3 sec

Time taken by the wave to reach back to the free end after reflecting from the
interface = Time of flight = 2 (0.33210-3) = 0.66410-3 sec

From the above plot,

Time of launch of the pulse = 0.34510-3 sec

Time of reflection of the pulse = 1.0210-3 sec

Time of flight of the pulse = 0.6710-3 sec

The time of flight of the pulse in the response history is equal to that predicted
using material properties.

5.3 Dynamic Response of an 2D Frame

Consider a 2D frame which is taken as a combination of 1D beam

elements, for finite structures such as frame shown in figure () the propagating
pulse has many reflections

Figure 5-9 2D Frame.

49
 A characteristic of connected structure is that they introduce sharp
spectral peaks in the responses.

Figure 5-10 Gaussian Pulse Loading-1 and Gaussian Pulse Loading-2.

The following plots show the transverse velocity v/s Frequency at the node 1 ,
node 2 respectively, due to high-frequency pulse in the transverse direction at
the node 1, respective load case is mentioned. The load histories are plotted in
figure 5.10.

Figure 5-11 Transverse frequency response at node 1 due to Gaussian pulse_1.

50
Figure 5-12 Transverse frequency response at node 2 due to Gaussian pulse_1.

Figure 5-13 Axial frequency response at node 1 due to Gaussian pulse_1.

51
 Figure 5-14 Axial frequency response at node 2 due to Gaussian pulse_1.

Figure 5-15 Transverse frequency response at node 1 due to Gaussian pulse_2.

52
Figure 5-16 Transverse frequency response at node 2 due to Gaussian pulse_2.

Figure 5-17 Axial frequency response at node 1 due to Gaussian pulse_2.

53
 Figure 5-18 Axial frequency response at node 2 due to Gaussian pulse_2.

5.4 Frequency Response analysis of an 2D beam-frame model of

aircraft wing

The 2D Beam-Frame model is replaced with 3D model of the

aircraft wing for performing frequency response analysis using
spectral finite element method. There are two cases which are solved
as benchmark problems each separately using the formulation done
in the previous chapter of this dissertation.
5.4.1 Case1: Straight wing
Here a CATIA model is prepared with two spars, six ribs and skin covering
from top and bottom, the wing span is 1m and constant chord length of 0.85m
and constant thickness of the airfoil as 0.082m.

54
 Figure 5-19 Isometric View of CATIA model of Straight Wing.

This 3D model can be replaced with the following beam-frame model to

perform frequency response analysis using spectral finite element method.

Figure 5-20 Beam-Frame Model of Straight Wing.

The following plots shows the axial velocity v/s Frequency, transverse
velocity v/s Frequency at node 30, 20 due to high-frequency pulse in the
transverse direction at node 30, the load time history, frequency spectrum is
shown in the inset image of each plot.

55
 Figure 5-21 Axial velocity v/s Frequency at node 30 due to the loading (inset of the plot) in the
Transverse direction.

Figure 5-22 Transverse velocity v/s Frequency at node 30 due to the loading (inset of the plot) in the
Transverse direction.

56
 Figure 5-23 Axial velocity v/s Frequency at node 30 due to the loading (inset of the plot) in the
Transverse direction.

Figure 5-24 Transverse velocity v/s Frequency at node 30 due to the loading (inset of the plot) in the
Transverse direction.

57
 Figure 5-25 Axial velocity v/s Frequency at node 20 due to the loading (inset of the plot) in the
Transverse direction.

Figure 5-26 Transverse velocity v/s Frequency at node 20 due to the loading (inset of the plot) in the
Transverse direction.

58
 Figure 5-27 Axial velocity v/s Frequency at node 20 due to the loading (inset of the plot) in the
Transverse direction.

Figure 5-28 Transverse velocity v/s Frequency at node 20 due to the loading (inset of the plot) in the
Transverse direction.

The following plots show a comparison of four cases Here first is healthy
case, other three cases degradation of the stiffness matrix of the distributed
spring is considered with steps of 0.96, 0.92 and 0.88 at spring no.10
respectively.

59
Figure 5-29 Axial velocity v/s Frequency at node 30 due to the Gaussian pulse-1.

Figure 5-30 Axial velocity v/s Frequency at node 30 due to the Gaussian pulse-2.

60
 Figure 5-31 Transverse velocity v/s Frequency at node 30 due to the Gaussian pulse-1.

Figure 5-32 Transverse velocity v/s Frequency at node 30 due to the Gaussian pulse-2.

5.4.2 Case2: Tapered wing

This case is similar to case 2 except that the wing is tapered. Here a
CATIA model is prepared with two spars, six ribs and skin covering from top and
bottom, the wing span is taken as 1m and chord length varying from 0.85m at
root and 0.45m at tip and constant thickness of the airfoil of the wing as 0.082m.

61
 Figure 5-33 Isometric view of CATIA model of tapered wing.

This 3D model can be replaced with the following beam-frame model to

perform frequency response analysis using spectral finite element method.

Figure 5-34 Beam-frame model of tapered wing.

The following plots shows the axial velocity v/s Frequency, transverse
velocity v/s Frequency at node 30, 20 due to high-frequency pulse in the
transverse direction at node 30, the load time history, frequency spectrum is
shown in the inset image of each plot.

62
 Figure 5-35 Axial velocity v/s Frequency at node 30 due to the loading (inset of the plot) in the
Transverse direction.

Figure 5-36 Transverse velocity v/s Frequency at node 30 due to the loading (inset of the plot) in the
Transverse direction.

63
 Figure 5-37 Axial velocity v/s Frequency at node 30 due to the loading in (inset of the plot) in the
Transverse direction.

Figure 5-38 Transverse velocity v/s Frequency at node 30 due to the loading (inset of the plot) in the
Transverse direction.

64
 Figure 5-39 Axial velocity v/s Frequency at node 20 due to the loading (inset of the plot) in the
Transverse direction.

Figure 5-40 Transverse velocity v/s Frequency at node 20 due to the loading (inset of the plot) in the
Transverse direction.

65
 Figure 5-41 Axial velocity v/s Frequency at node 20 due to the loading (inset of the plot) in the
Transverse direction.

Figure 5-42 Transverse velocity v/s Frequency at node 20 due to the loading (inset of the plot) in the
Transverse direction.

The following plots show a comparison of four cases Here first is healthy
case, other three cases degradation of the stiffness matrix of the distributed

66
spring is considered with steps of 0.94, 0.88 and 0.81 at spring no.10
respectively

Figure 5-43 Axial velocity v/s Frequency at node 30 due to the Gaussian pulse-1.

Figure 5-44 Axial velocity v/s Frequency at node 30 due to the Gaussian pulse-2.

67
Figure 5-45 Transverse velocity v/s Frequency at node 30 due to the Gaussian pulse-1.

Figure 5-46 Transverse velocity v/s Frequency at node 30 due to the Gaussian pulse-2.

68
Chapter 6

Conclusion and Future Scope of Work

The frequency response analysis is performed for 2D Beam-Frame model

of aircraft wing, for the low-high frequency loading the response is measured at
all the other nodal points. The degradation in one of the distributed spring is
introduced with three different steps and it is compared with the healthy case.

The spectral finite element modeling and computation reported here are
aimed at the off-line computation, in which the reduced order dynamic stiffness
matrix of the healthy structure is obtained and stored in the computer memory.

In the future scope of work, a near real time scenario with successive data
acquisition from a sensor network and computation of DFI vector. In reality, the
successive data acquisition from the sensor network and its synchronization with
the computation time to obtain DFI is important, especially for damage growth
monitoring.

The DFI approach we can pose it as a forward problem with a more or

less global measurement and estimate the damage size. If this can be achieved
in the future experimental demonstration, then it would be possible to reach the
level-III SHM capability using the DFI approach.

69
Chapter 7

References

[1] J. Doyle, Wave Propagation in Structures, New York, 1997.

[2] S. Gopalakrishnan, A. Chakraborty, D. Roy Mahapatra, Spectral finite

element method- Wave Propagation, Diagnostics and control in anisotropic and
Inhomogenous Structures. Springer-verlag, London. 2008.

[3] D. Roy Mahapatra, S.Gopalakrishnan, N. Hu, Specral Finite Element

Modelling of Complex Structures for Applications in Real-time Structural Health
Monitoring, AIAA.

[4] D. Kumar, D. R. Mahapatra, S. Gopalakrishnan, A Spectral finite element for

wave propagation and structural diagnostics analysis of composite beam with
transverse crack, Finite Elements in analysis and Design 40(13-14) (2004).

[5] A. Nag, D. Mahapatra, S. Gopalakrishnan, Identifiction of delamination in a

composite beam using a damaged spectral element, Structural Health Monitoring
1(1) (2002) 105-126.

[6] A. K. Garg, D. Roy Mahapatra, D. Suresh, S. Golpalakrishnan, S. N. Omkar,

Estimation of composite damage model parameters using spectral finite lement
method and neural network, Composite Science and Technology, 64 (2004)
2477-2493.

[7] D. Roy Mahapatra, S. Suresh, S. N. Omkar, S. Goplakrishnan, Estimation of

degraded composite laminate properties using acoustic wave propagation model
and a reduction-prediction network, composite laminate properties, 22 (2005)
849-876.

[8] D.S. Kumar, D. Roy Mahapatra, and S. Gopalakrishnan, Estimation of

dynamic fracture parameters in a trans verse cracked composite beam using a
simplified diagnostic wave propagation model, Structural Health Monitoring, 2005

[9] A. K. Rao, D. P. K. Muthukumaraswamy, D. Roy. Mahapatra, S.

Gopalakrishnan, and M.S. Bhat, Experimental studies toward to the development
of a structural health monitoring system using damage force indicator, Proc. Int.
Conf. Smart Materials and Structures, 2005.

70
 [10] C.V.S. Sastry, D. Roy Mahapatra, S. Gopalakrishnan *, T.S. Ramamurthy,
An iterative system equivalent reduction expansion process for extraction of high
frequency response from reduced order finite element model, Comput. Methods
Appl. Mech. Engg. 192 (2003) 18211840

[11] Usik Lee, Joohong Kim, and Andrew Y. T. Leung, The Spectral Element
Method in Structural Dynamics, The Shock and Vibration Digest, 32(2000) 451-
465.

[12] C T sun, Mechanics of Aircraft Structures, wiley, 2006.

[13] Component design handbook of aeronautics no.2 , London, sir Isaac pitman
and sons ltd.

[14] P. Kudela, M. Krawczuk, W. Ostachowicz, Wave propagation modeling in 1D

structures using spectral finite elements, journal of sound and Vibration, 300
(2007) 88-100.

[15] Arkadiusz Zak, Marek Krawczuk, Wieslaw Ostachowicz, Propagation of in-

plane waves in an isotropic panel with a crack, Finite Elements in Analysis and
Design, 42 (2006) 929-941.

[16] B C Lee, W J Staszewski, Modelling of Lamb waves for damage detection in

metallic structures: Part I. Wave propagation, Smart Mater. Struct. 12 (2003)
804-814.

[17] Elias G, Abu-Saba, Ji Yao Shen, William M. Mcginley, Lumped mass

Modelling For the Dynamic Analysis of Aircraft structures, The Journal of
Aerospace Engineering.

[18] Nicole Apetre, Massimo Ruzzene, Sathyanaraya Hanagud, S.

Gopalakrishnan, A Wave-Based Damage Index for the Analysis of the Filtered
Response of Damaged Beams, Journal of Mechanics of Materials and
Structures, 3 (2008) 1605-1623.

[19] A.K. Noor, Recent advances and application of reduction methods, Appl.
Mech. Rev., Trans. ASME 47 (5) (1994) 125146.

[20] P.Th.L.M. Van Woerkom, Mathematical models of flexible spacecraft

dynamics: A survey of order reduction approaches, Control Theory Adv. Technol.
4 (1990) 609632.

71