Santos RB DR Ilha
Santos RB DR Ilha
Santos RB DR Ilha
FACULDADE DE ENGENHARIA
CAMPUS DE ILHA SOLTEIRA
Ilha Solteira
2018
RODRIGO BORGES SANTOS
Ilha Solteira
2018
Santos An Alternative Approach
Ilha
toSolteira06/03/201895
Design Periodic Rods Sim Tese (doutorado)
Engenharia Solid
Mecânica
Mechanics
Sim
FICHA CATALOGRÁFICA
Desenvolvido pelo Serviço Técnico de Biblioteca e Documentação
Em primeiro lugar, gostaria de expressar minha profunda gratidão ao meu orientador, prof.
Dr. Douglas D. Bueno, por sua amizade, ensinamentos, direcionamentos, encorajamento, su-
porte e paciência.
Meu sincero obrigado ao meu co-orientador Prof. Dr Michael J. Brennan pela generosa
contribuição nesta tese com muitas sugestões, direcionamentos e suporte técnico. Também,
agradeço ao meu co-orientador Prof. Dr. Vicente Lopes Júnior pela sua confiança, amizade e
por abrir as portas para o desenvolvimento desta tese.
Eu também sou muito grato à minha família por sempre me apoiar, incentivar e aconselhar
nos meus objetivos. Em especial, agradeço a minha esposa Maria Regina Ap. de Lima por estar
sempre do meu lado em todas as situações de minha vida. A concretização desta tese é um
sonho realizado. Obrigado Maria Regina por ter contribuído na realização deste sonho.
Finalmente e mais importante, eu agradeço a Deus por ser o guia de minha vida e por me
dar proteção, esperança, oportunidades, sabedoria, paciência, força, paz e saúde. É através Dele
que todas as coisas são possíveis em minha vida.
“Sonhos determinam o que você quer,
Ação determina o que você conquista.”
Aldo Novak
ABSTRACT
The reduction of structural vibration has been an important topic for many engineering applica-
tions. In traditional projects different passive control techniques involving viscoelastic materials
and dynamic absorbers and, more recently, active control methodologies including actuators and
sensors have been successfully employed. Different researches have demonstrated that vibra-
tion reduction can be obtained using the concept of periodicity. The periodic structures involve
identical elements or parts connected repeatedly. The design of periodic structures can be em-
ployed to get frequency band without elastic waves propagation, i.e., stop bands, introducing an
effect similar to the filter. In this context, the present work introduces an alternative approach
for designing periodic rods. This alternative involves the modeling of an infinite hybrid type
periodic rod in which a finite periodic structure is connected between two semi-infinite rods.
It is used a methodology that relates state vector and wave amplitudes. The main proposal
of this work is to develop a relation between the transmitted and incident longitudinal waves
amplitudes in terms of physical and geometrical properties of a generic candidate structure to
simplify the process of designing. Based on this approach is shown that a periodic rod can be
designed to satisfy requirements of a vibration suppression. A hypothetical problem is proposed
and numerical and experimental results show the stop bands obtained to solve the problem. It
shows that this approach is an important tool for designing this type of structures.
A redução de vibração estrutural tem sido um importante tópico para muitas aplicações de
engenharia. Nos projetos tradicionais, diferentes técnicas de controle passivo envolvendo mate-
riais visco-elásticos e absorvedores dinâmicos e, mais recentemente, metodologias de controle
ativo incluindo atuadores e sensores têm sido empregado com sucesso. Diferentes pesquisas
tem demostrado que redução de vibração pode ser obtida usando o conceito de periodicidade.
As estruturas periódicas envolvem elementos idênticos ou partes conectadas repetidamente. O
projeto de estruturas periódicas pode ser empregado para conseguir bandas de frequências em
que não há propagação de ondas elásticas, denominadas de "stop bands", introduzindo um efeito
similar ao de um filtro. Neste contexto, o presente trabalho apresenta uma abordagem alterna-
tiva para o projeto de barras periódicas. Esta alternativa envolve a modelagem de uma barra
periódica do tipo híbrida infinita na qual uma estrutura periódica finita é conectada entre duas
barras semi-infinitas. Para isto, é utilizada uma metodologia que relaciona vetor de estados e
amplitude de ondas. A principal proposta deste trabalho é desenvolver uma relação entre as
amplitudes de ondas longitudinais transmitidas e incidentes em termos das propriedades físi-
cas e geométricas de uma genérica estrutura periódica para simplificar o processo do projeto.
Usando esta formulação mostra-se que uma barra periódica pode ser projetada para satisfazer
os requisitos de uma supressão de vibração. Um problema hipotético é proposto e resultados
numéricos e experimentais mostram os "stop bands" obtidos para resolver o problema. Isto
mostra que esta abordagem é uma importante ferramenta para o projeto deste tipo de estruturas.
Figure 4 Illustration of the propagation constants for the longitudinal wave, where
PB and SB respectively represent the pass and stop band. . . . . . . . 25
Figure 23 Zoom in stop band of figure 22 for different number of cells consid-
ering Lca = 1, Sca = 5, Scb = 5, Sab = 1, ρcb = ρca = ρab = 1 and
Ecb = Eca = Eab = 1. . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Figure 24 Minimum values of the transmission for different values of Sca and
number of cells considering Lca = 1, Sab = 1, ρcb = ρca = ρab = 1 and
Ecb = Eca = Eab = 1. . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Figure 26 Bandwidth for different values of Sca and number of cells considering
Lca = 1, Sab = 1, ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1. . . . . 57
Figure 28 Stop bands for different values of Lca < 1 considering N = 3, Sca = 5,
Scb = 5, Sab = 1, ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1. . . . . 59
Figure 34 Zoom in stop band of figure 33 for different number of cells consid-
ering Lca = 1, Sca = 5, Scb = 5, Sab = 1, ρcb = ρca = ρab = 1 and
Ecb = Eca = Eab = 1. . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Figure 35 Maximum values of the reflection for different values of Sca and num-
ber of cells considering Lca = 1, Sab = 1, ρcb = ρca = ρab = 1 and
Ecb = Eca = Eab = 1. . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Figure 48 Stop bands for different values of Lca considering La = 432 mm (steel
case) or La = 421, 9 mm (aluminum case) or La = 135 mm (nylon case),
N = 1, Sca = 2, Sab = 1000 and Scb = 2000. . . . . . . . . . . . . . . 73
Latin Letters
A wave amplitude
b term of the transmission and reflection equation
C coefficient of power
c constant value
dx length of differential rod element
dB Decibel unit
E Young’s modulus
F external longitudinal force
√
j imaginary number ( −1)
k longitudinal wavenumber
L length of a segment
m positive or negative integer number
N number of cells
P average power transfered per cycle
p instantaneous power
q body force per unit volume
S cross-section area
t time variable
T temporal period
u longitudinal displacement
u̇ longitudinal velocity
v wave propagation velocity
x space variable and direction of wave propagation in one-dimensional
Z mechanical impedance
Greek Letters
ω angular frequency
Ω non-dimensional frequency
λ wavelength
ψ Bloch parameter (or propagation constant)
Λ eigenvalue of the transfer matrix
φ phase angle of the wave
σ axial stress
ε axial strain
ρ mass density
α term of the transmission and reflection equations
β term of the transmission and reflection equations
η term of the transmission and reflection equations
τ ratio of propagation times
κ ratio of impedance
ϕ angle relative to the phase velocity curve
θ angle relative to the group velocity curve
ξ damping loss factor
Superscripts
Subscripts
Acronyms
BW Bandwidths
PB Pass Band
SB Stop Band
WFEM Wave Finites Elements Method
WSEM Wave Spectral Element Method
Special Notations
[ ] matrix
{ } vector
| | absolute value of a number
√
square root of a number
min( ) minimum value of a number
max( ) maximum value of a number
det( ) determinant function
ln( ) natural logarithmic of a number
∂( )
∂t first-order partial differential relative to time variable
∂( )
∂x first-order partial differential relative to spatial variable
∂ 2( )
∂t 2
second-order partial differential relative to time variable
∂ 2( )
∂ x2
second-order partial differential relative to spatial variable
CONTENTS
1 INTRODUCTION 18
1.2 OBJECTIVES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.1 METHODOLOGY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.2 EQUATION FOR WAVE TRANSMISSION . . . . . . . . . . . . . . . . . . . . 47
4 NUMERICAL SIMULATIONS 51
6 CONCLUSIONS 79
REFERENCES 82
APPENDIX A - EQUATION FOR TRANSMISSION AND REFLECTION 85
1 INTRODUCTION
This chapter presents a brief description of the context in which this work is based on, as
well as publications regarding periodic structures whose aim is to attenuate structural vibration.
Furthermore, this chapter presents the objectives, contributions achieved, and how the text of
this thesis is organized.
The reduction of structural vibration has been a very important topic for many engineering
applications. In traditional projects, different passive control techniques involving viscoelas-
tic materials and dynamic absorber design and, also, active control methodologies involving
actuators and sensors have been successfully employed.
Different researches have demonstrated that vibration reduction can be obtained using the
concept of periodicity. These works have designed periodic structures based on geometrical and
physical properties. The structures involve identical elements or parts connected repeated times.
This idea has been applied for designing panels of satellites, aircraft fuselage and wings, truss
structures, pipelines, train tracks, and others (NARISETTI, 2010; MEAD, 1996). The design
of periodic structures can be employed to get frequency band without elastic wave propagation
i.e, stop bands, introducing an effect similar to the filter.
In engineering, some vibration problems consist of a vibration source with higher dominant
intensity, such as seismic events (XIONG et al., 2012). Other common sources are industrial
machines (electric generators, turbines, compressors, pumps, among others) that in some cases
are assembled upon supports or structures made of rods and produce vibrations and noise at
high levels, which may even be harmful to human health. In the context of this work, these
applications are called “unilateral source problems”, since, particularly, a formulation for rods
is developed. Besides, many of these problems present predominantly unidimensional vibration
propagation, as shown by (SINGH et al., 2004; ASIRI et al., 2005; XIUCHANG et al., 2011;
CHEN; WU, 2016), among others.
inclusion of a periodic structure in a floating raft isolation system (extensively applied in ships
and submarines) can result in significant attenuation of vibration and noise caused by resonant
substructures (SONG et al., 2015; WANG; MAK, 2018). In the automotive industry, vehicles
can use periodic viscoelastic materials to minimize vibration transmission from their engines
to the chassis. This type of structure keeps comfort also working in higher rotational levels in
comparison with conventional designs (ASIRI, 2007; JUNG et al., 2010). In addition, in the
civil engineering, periodic foundations can be used to achieve a much more effective seismic
isolation (GAOFENG; ZHIFEI, 2010; XIANG et al., 2012; CHENG; SHI, 2017).
In this context, the present work introduces an approach to design periodic rod type struc-
tures. The main proposal is to present a relation between transmitted and incident longitudinal
waves amplitudes in terms of physical and geometrical properties of a generic candidate struc-
ture shown in figure 1. The approach considers incident waves from the left side structure
and a problem with an one side source of vibration and a dominant one dimensional vibration
transmission and the objectives and contributions of this work are shown below.
Figure 1 - Scattering of longitudinal waves by a periodic structure. Incident (Ai ), reflected (Ar )
and (At ) transmitted longitudinal waves.
1.2 OBJECTIVES
The main objective of this thesis is to develop a methodology that involves the amplitude
ratio of incident and transmitted longitudinal waves that propagate in a periodic rod structure.
In addition, the secondary objectives are presented below:
• to model a periodic rod referred to as hybrid infinite in this work, consisting of a periodic
finite structure embedded in an infinite rod;
• to evaluate the transmission and reflection of waves that propagate in structures whose
periodicity is geometrical;
• to find out, through numeric simulations, the number of periodic cells needed to represent
the dominant dynamics of a structure with periodic cells along all its extension;
• to make explicit the influence of physical properties and structural geometrical parameters
on stop bands formation, as well as their characteristics, such as: location, bandwidth and
attenuation levels, in order to assist the process of designing this structure.
The following presents some of the contributions of this work in the context of periodic
structures design meeting requirements of permissible vibration levels, frequency bands with
stop bands, as well as geometrical constraints:
• scalar equations for transmission and reflection of longitudinal waves are developed as an
alternative to classic solutions regarding matrix systems of equations;
• some practical aspects of periodic structures design are discussed considering explicit
physical and geometrical parameters in order to assist the analyst’s understanding;
• a strategy to apply the formulation developed in this work is presented for infinite struc-
tures in practical engineering problems (finite structures).
21
Chapter 1 presents the context of this work, the objectives and the contributions achieved
in this research. The basic theoretical foundation involved in the study of the harmonic motion
of waves is described in chapter 2. Basic concepts are presented, such as periodic structures
filtering properties, and dispersion in periodic systems. Furthermore, the longitudinal wave’s
equation is obtained, the concept of wave impedance is defined and the analysis of transmission,
reflection and waves energy flow in a discontinuity is presented.
Chapter 4 presents the results for the numeric simulations using the transmission and re-
flection relations proposed. The simulations are presented for different structure configurations
with geometrical periodicity.
This chapter aims to present the basic theoretical foundation involved in this work. Ini-
tially, it approaches the concept of periodic structures, as well as their filtering properties, and
dispersion in periodic systems. Afterwards, important concepts regarding the harmonic mo-
tion of waves are described, such as some types of waves, wave number and length, phase
and group velocity, among others. At last, the longitudinal wave’s equation is obtained, the
wave impedance is discussed, and the analysis of transmission, reflection and energy flow in a
discontinuity is presented.
The design of period structures can be used to generate frequency ranges in which waves do
not propagate or are highly attenuated through the structure. The characteristics of these systems
make them potential candidates to act as passive vibro-acoustic filters. The filtering properties
exhibited by these ones as well as the frequency ranges positioning within a particular range
depend on the structural conception and arrangement of the periodic cells. The frequency ranges
in which these structures allow waves propagate are called pass bands (PB) and, in another
hand, frequency ranges in which there is waves attenuations are called stop bands (SB) or band
gaps (BRILLOUIN, 1953). This effect of attenuation is generated due internal reflection of
waves caused by the impedance variations (difference of material properties or geometrical
discontinuities).
The frequency pass and stop bands are formed through the constructive and destructive
waves interference inside a unit cell, respectively. According to Manktelow (2013) the inci-
dent wave is reflected and transmitted (partially) at an interface and each one of them repeat
the same process. At frequencies within the stop band, the waves reflected act to cancel the
incident wave resulting a transmitted wave with significant attenuation. On the other hand, at
23
frequencies within the pass band, the transmitted waves are not significantly attenuated (SZEFI,
2003a). A typical example of responses obtained from periodic structures analysis are showed
by vibration transmission curves. Figure 2 shows an illustration of this curve obtained by the
relation between the displacements uR /uL measured at the ends of the periodic structure when
a external force F is applied at the left end, where uR and uL are respectively the displacement
spectral amplitudes in the right and left sides of the periodic part of the rod.
PB SB PB SB PB SB PB
uR
uL
1
Transmission,
Frequency
Source: elaborated by the author.
Consider a periodic structure composed of N cells arranged along of the axis x as shown in
figure 3, where AR represents right-going waves while AL indicates the left-going waves.
AR AR AR AR
... ...
x
AL AL AL AL
According to the literature, the Bloch’s theorem for a periodic system shows that wave am-
plitudes between two consecutive cells n and (n + 1) are given by (MENCIK, 2010; GAZALET
et al., 2013)
Λ = eψ (2)
and A is the wave amplitude, ψ = ln(Λ) is the Bloch parameter (or also called propagation
constant), ln ( ) is the natural logarithmic and Λ is the eigenvalue of the transfer matrix (root
of the characteristic Eq. 98), discussed in section 3.4. For each frequency can be extracted a
pair of Λ such that Λ1 Λ2 = 1 (HVATOV; SOROKIN, 2015). Therefore, the eigenvalues can be
classified by two groups, i.e., |Λ1 | ≤ 1 which corresponds to waves traveling to the right and
|Λ2 | ≥ 1 which corresponds to waves traveling to the left side of the structure.
This theory states that for any structure with repetitive identical units, the change in wave
amplitude across a unit cell does not depend upon the location of the unit cell within the struc-
ture, i.e., a wave propagating in a periodic structure can be described by the motion of a single
cell. Thus, as result of this theorem, it is possible to understand the wave propagation though an
entire structure just by considering the wave motion in a single unit cell (PHANI et al., 2006) .
The Bloch parameters are complex numbers given by ψ = ψ Re + iψ Im . This one controls
the nature of elastic wave propagation in periodic structures. The real part ψ Re is the attenuation
factor and represents the amplitude decay of a wave propagation from one cell to the next one.
The imaginary part ψ Im is the phase constant and represent the phase difference in two adjacent
cells, i.e, is a measurement of the phase changing across one unit cell. If ψ is purely imaginary
(real part is zero) the absolute value of Λ is unit (|Λ| = 1) and waves propagating without
attenuation (zone of pass band). However, if a real part exists (|Λ| 6= 1) the attenuation of the
wave’s amplitude is observed (zone of stop band) (GAN et al., 2016; SOLAROLI et al., 2003;
SINGH et al., 2004).
The relationship between frequency and Bloch parameter, described by ω(ψ), is named the
dispersion relation for a periodic system. An important characteristic of Bloch wave disper-
sion relationships is the periodicity such that ω(ψ) = ω(ψ + 2mπ), where m is a positive or
negative integer number (BRILLOUIN, 1953). This means that the frequency ω is a periodic
function of ψ and also leads to the definition of a first Brillouin zone that corresponds to the
domain conventionally defined by ψ ∈ [−π, π]. Moreover, the dispersion relationship also fol-
lows the symmetry relationship ω(ψ) = ω(−ψ) that leads to the definition of an irreducible
25
Brillouin zone that corresponds to ψ ∈ [0, π]. A common representation for a dispersion curve
is illustrated by figure 4.
Figure 4 - Illustration of the propagation constants for the longitudinal wave, where PB and SB
respectively represent the pass and stop band.
associated with 1 associated with 2
PB SB PB SB PB SB PB
Eigenvalue
1
Attenuation factor
( Real )
0
Phase contant
( Imag )
Frequency
Source: elaborated by the author.
26
An elastic wave motion is a phenomenon in which a physical quantity (for instance, energy
or pressure) propagates through a medium, without transport of matter (FAHY; GARDONIO,
2007). The main types of waves propagating in a one-dimensional structure are illustrated in
figure 5.
In longitudinal or compression wave the particles motion of the medium are parallel to the
direction of wave propagation, the particles vibrate about their individual equilibrium positions.
In this type of wave motion there is some out-of-plane displacement due to the Poisson ratio
effect (i.e. lateral strains) as showed in figure 5 (a). In the transversal or shear wave the
particles motion are perpendicular to the direction of wave propagation, see figure 5 (b). The
other type of wave is the flexural or bending waves in witch there is translational displacement
and rotation, see figure 5 (c). In this wave motion, bending moment and the shear force govern
the deformation of medium. Note that, shear and bending waves are examples of out-of-plane
waves.
Figure 5 - Illustrative representations for the deformation patterns of three types of wave mo-
tions in one-dimensional structures.
Direction of wave propagation
x
(a)
Longitudinal
(b)
Shear
(c)
Bending
underformed deformed
Source: adapted from Fahy and Gardonio (2007).
27
The harmonic wave motion in time and space can be represented by (BRENNAN et al.,
2016)
or
where A is the amplitude of the wave, k is the wavenumber, ω is the angular frequency, x is the
space variable and t is the time variable. The wavenumber is spatial frequency and is inversely
proportional to the spatial period, i.e., to the wavelength λ (CREMER et al., 2005)
2π
k= (5)
λ
Also, the angular or temporal frequency is inversely proportional to the duration T (period)
of a cycle
2π
ω= (6)
T
The analogy between temporal frequency and the wavenumber is illustrated in figure 6.
The wavenumber is fundamental for understanding the harmonic wave motion. Physically, it
represents the phase change per unit increase of distance, in the same way as ω represents the
phase change per unit increase of time.
28
Time
T = temporal period
Temporal frequency ω= 2
T
Space
λ = spatial period
(wavelength)
Spatial frequency k= 2
λ
(
Consider a wave motion represented by equation (3) or (4). The argument of those functions
is designate as phase φ of this wave, i.e., φ = ωt ± kx. When the second term of that argument
is negative (−kx) as time increases, x must increase positively in order to keep a constant φ and,
thus, a wave propagating through the positive x direction. Otherwise, for +kx, x must increase
negatively and the wave propagating in the negative x direction. Figure 7 shows an illustration
of wave propagation in the positive and negative direction for a particular wavenumber.
29
t1
-x +x
t2
-x +x
t3
-x +x
Considering a wave propagating to the right side structure, i.e., phase this wave is φ =
ωt − kx. For the constant phase (ωt − kx = constant), we can be write (OSTACHOWICZ et al.,
2012)
ω constant ω
x= t− = t + ck (7)
k k k
The relationship between temporal frequency ω and spatial frequency k of the propagating
waves is defined as phase velocity (v p ) of the wave
ω
vp = (8)
k
Thus, a point of constant phase moves with this velocity and if an observer travelling in the
direction of wave propagation at this velocity sees no change of phase (FAHY; GARDONIO,
2007). The harmonic wave propagating to the right with phase velocity v p is presented in figure
8.
30
Space
-A
vp ( t2 - t1 )
Source: elaborated by the author.
Consider two waves propagating to the right side structure, with amplitudes A, but different
frequencies and wavenumber, i.e., waves propagation with phase velocity v p1 and v p2 , respec-
tively, as presented in figure 9 (a).
ω1 + ω2 k1 + k2
u(x,t) = Ū sin t− x (10)
2 2
where
ω1 − ω2 k1 − k2
Ū = 2A cos t− x = 2A cos [(∆ω)t − (∆k)x] (11)
2 2
In equation (10), one term is associated with modulation (cosine term) and another is as-
sociated with carrier wave (sine term). The superposition of a carrier wave and a modulation
wave is called of wave packet or group of waves, as presented in figure 10.
31
Figure 9 - Illustration of sum of waves propagating in the same direction. (a) two waves prop-
agating with same amplitudes, different frequencies and wavenumber (b) wave
propagating packet with velocity vg .
direction of waves propagation
first wave
with
vp1
(a)
second wave
with
vp2
vg
(b)
vp
Figure 10 - Wave packet as the superposition of a carrier wave and a modulating wave.
The phase velocity of the combined wave, figure 9 (b), is give by (BRENNAN et al., 2016)
ω1 + ω2
vp = (12)
k1 + k2
32
The waves packet does not propagate at the phase velocity. It propagates at the group
velocity that is defined by propagation velocity of modulation wave (Ū term). For a constant
phase (∆ωt − ∆kx = constant), we can be write (OSTACHOWICZ et al., 2012)
∆ω constant ∆ω
x= t− = t + c∆k (13)
∆k ∆k ∆k
Thus, the group velocity, which is the velocity at which the energy in the combined wave is
transported, is given by
ω1 − ω2 ∆ω
vg = = (14)
k1 − k2 ∆k
dω
vg = (15)
dk
Note that the group velocity is the rate of change of frequency with respect to the wavenumber.
A non-dispersive medium presents a group velocity that is equal to the phase velocity that is
frequency independent (such as tension waves in strings and compressional waves in rods). In
this case, the wavenumbers are proportional to frequency. In the dispersive medium the group
and phase velocities are different and they are frequency dependent (such as bending waves in
beams and plates). In this case, for each frequency the wave packet travels in a different velocity
and deform during the propagation.
33
Figure 11 - Illustration of the dispersion curves for a dispersive and non-dispersive medium.
vp= tan( )
vp= vg = tan( )
= constant
φ
k k
Source: elaborated by the author.
Consider axial stress in a long, slender and supported rod as shown in figure 12(a) and
assuming that lateral inertia effects associated with lateral contraction-expansion (or Poisson’s
effects) can be neglected. The coordinate x refers to cross-section of the rod and longitudinal
displacement of that section is represented by u(x,t). The body force q(x,t) per unit volume is
also considered and σ is the axial stress acting on a differential element of the rod. The free
body diagram of that rod element of length dx is presented in figure 12 (b).
Using Newton’s Second law, the equation of longitudinal motion for an element is given by
(GRAFF, 1991)
∂σ ∂ 2u
−σ S + (σ + dx)S + q(x,t)Sdx = ρSdx 2 (16)
∂x ∂t
where ρ is the material density and S is the cross-section area of the rod.
∂σ ∂ 2u
+ q(x,t) = ρ 2 (17)
∂x ∂t
34
Figure 12 - Illustration of a thin rod: (a) coordinate x and displacement u(x,t) of a section of
rod; (b) free body diagram of a rod element.
σ = Eε (18)
∂u
ε= (19)
∂x
Substituting equation (18) and (19) into equation (17) and assuming a homogeneous rod (E
and ρ not vary with x) is obtained
∂ 2u ∂ 2u
E + q(x,t) = ρ (20)
∂ x2 ∂t 2
∂ 2u ∂ 2u
E = ρ (21)
∂ x2 ∂t 2
or
∂ 2u 1 ∂ 2u
= (22)
∂ x2 v2p ∂t 2
35
where
s
E
vp = (23)
ρ
Equation (22) is the wave equation for a rod. Note that the phase velocity is independent of
frequency and because of this a rod is non-dispersive medium. In this case, the group velocity
is equal to the phase velocity according to described in previous sections of this work.
Replacing the equation (24) into equation (22) results in the following second-order ordi-
nary differential equation
∂ 2 u(x)
+ k2 u(x) = 0 (25)
∂ x2
ω
where k = is the wavenumber and its solution is given by
vp
Note that the term AR e− jkx represents a right-going propagating wave with amplitude AR
and the term AL e jkx represents a left-going propagating wave with amplitude AL (BRENNAN
et al., 2016).
36
An important concept in this field of wave propagation is the wave impedance or the me-
chanical impedance of a structure. Consider a right-going propagating wave u(x,t) = AR e j(ωt−kx)
in a semi-infinite rod as shown in figure 13.
x=0 x
u(0,t) u(x,t) , S, E
F(t)
+
Source: elaborated by the author.
For generate such motion, a force F(t) is applied on the left boundary (x = 0) and it is given
by
∂ u(0,t)
F(t) = −SE = jkSEAR e jωt (28)
∂x
and the velocity in this point of excitation can be expressed as
∂ u(0,t)
u̇ = = jωAR e jωt (29)
∂t
The wave impedance Z is defined as the ratio between the applied force and the correspond-
ing velocity at the point of excitation. In other words, wave impedance relates the velocity of the
particles in a medium and the corresponding force (or stress) required to produce that velocity.
Thus, it is given by (HAGEDORN; DASGUPTA, 2007)
The impedance in the semi-infinite rod is a real quantity which does not vary with frequency
and thus it can be seen as a linear viscous damper (BRENNAN et al., 2016). Similarly, it can
be define the specific impedance ZS as the ratio between the stress at the end of the rod and its
37
ZS = ρv p (31)
Figure 14 - Incident, reflected and transmitted waves at a junction between two semi-infinite
rods.
x=0
, S1 , E 1 u(0,t) 2 , S2 , E2
1
Ai At
- +
Ar
Let a positive-traveling longitudinal wave in the left side of discontinuity, when this wave
achieve the discontinuity (x = 0) it is partly reflected back into rod-1 and partly transmitted to
rod-2. Thus, the longitudinal displacement fields on either side of the discontinuity are set to
given by (CREMER et al., 2005)
where Ai , Ar and At are, incident, reflected and transmitted wave amplitudes, respectively. Note
that the frequency ω is the same for the waves in the left and right rods, however, as the wave
38
velocities are different in the two semi-infinite rods (once different material is assumed for each
one of them), the corresponding wave numbers are different, i.e, k1 6= k2 . At the junction (x = 0),
there are the following requirements of displacement and force continuities (HAGEDORN;
DASGUPTA, 2007; GRAFF, 1991)
and
For which the right side of the junction the force is distributed over a larger surface such
the stress is reduced. The stresses intensities are determined from the wave fields and Hooke’s
law
∂ u1 (0,t)
σ1 (0,t) = E1 = − jk1 E1 Ai e jωt + jk1 E1 Ar e jωt = σi (0,t) + σr (0,t) (36)
∂x
and
∂ u2 (0,t)
σ2 (0,t) = E2 = − jk2 E2 At e jωt = σt (0,t) (37)
∂x
Substituting equations (36) and (37) into equation (35) and using the equation (34), the
amplitudes of the transmitted and reflected waves are obtained by
2k1 E1 S1 2
At = Ai = Ai (38)
k1 E1 S1 + k2 E2 S2 1 + (Z2 /Z1 )
and
k1 E1 S1 − k2 E2 S2 1 − (Z2 /Z1 )
Ar = Ai = Ai (39)
k1 E1 S1 + k2 E2 S2 1 + (Z2 /Z1 )
where Z1 = ρ1 v1 S1 and Z2 = ρ2 v2 S2 are wave impedances for rod-1 and rod-2, respectively.
From equations (36) to (39) the following relationships for transmitted and reflected stress
intensities can be obtained (GRAFF, 1991; ROSE, 2014)
ρ2 v2 S2 − ρ1 v1 S1 (Z2 /Z1 ) − 1
σr = σi = σi (41)
ρ2 v2 S2 + ρ1 v1 S1 (Z2 /Z1 ) + 1
The average power (or energy flux) transfered per cycle for a harmonic force F(t) acting
through a collinear particle velocity u̇ is given by (HAGEDORN; DASGUPTA, 2007; FAHY;
GARDONIO, 2007)
Z T 2π
1
Z
ω ω
P= p(t)dt = F(t)u̇dt (42)
T 0 2π 0
were p(t) is the instantaneous power and T is the period of the harmonic motion.
Using equations (32) and (33), the component displacements in real form can be written as
follows (YILDIRIM, 1994)
From Hooke’s law (Eq. 18) and using equations (43) to (45), the component forces at the
junction (x = 0) are given by
Substituting the equations of forces (46) to (48) and the equations of velocities (49) to (51)
into the equation (42) we obtain as solution the average power carried by the incident, reflected,
and transmitted harmonic waves, respectively
1 1
Pi = − k1 E1 S1 ωA2i = − ρ1 v1 S1 ω 2 A2i (52)
2 2
1 1
Pr = + k1 E1 S1 ωA2r = + ρ1 v1 S1 ω 2 A2r (53)
2 2
1 1
Pt = − k2 E2 S2 ωAt2 = − ρ2 v2 S2 ω 2 At2 (54)
2 2
Based on these equations is possible to define the coefficients of power reflection CPR , and
power transmission CPT as
2
Pr Ar
CPR = = (55)
Pi Ai
2 2
Pt k2 E2 S2 At ρ2 v2 S2 At
CPT = = = (56)
Pi k1 E1 S1 Ai ρ1 v1 S1 Ai
41
From the equations (55) and (56) and using equations (38) and (39) is possible to show that
(HAGEDORN; DASGUPTA, 2007)
" 2 2 #
ρ2 v2 S2 At Ar
Pr + Pt = + Pi = Pi (57)
ρ1 v1 S1 Ai Ai
This last equation shows that incident average power is partially reflected and partially
transmitted at the junction of the rod, and the sum of the scattered average power is equal to the
incident average power, i.e., in this scattering process there is energy conservation. Also, for
convenience that equation can be rewritten by
In this chapter, an approach to design periodic rods is developed using the methodology
based on state vectors. This methodology has been employed and validated in the study of the
coupling dynamics between piezoceramic actuators and a beam (BRENNAN et al., 1997), as
well as in studies of active vibration control in unidimensional structures (BRENNAN, 1994).
As a result of the present proposal, equations for wave transmission and reflection are obtained
in periodic rods written explicitly in terms of physical and geometrical structural properties.
The approach considers a finite number of cells embedded in an originally homogeneous rod.
The transfer matrix eigenvalue problem for this new structure is solved and, finally, it is possible
to obtain the dispersion curves from the roots of an analytical equation, which is regarded as an
important simplification in the process of designing this kind of structures.
3.1 METHODOLOGY
Initially, the model consider a hybrid rod with a single cell connected to two semi-infinite
rods (parts b with cross-sectional area Sb ) as shown in figure 15. The unit cell consists of two
segments a and c with lengths La and Lc and cross-sectional area Sa and Sc , respectively. Points
1 to 3 indicate the location of each junction.
Figure 15 - Hybrid rod composed by a single cell connected to two semi-infinite rods.
unit cell
Ai Sc
- Sb rod rod Sb +
a Sa c
b 1 2 3 b
Ar At
La Lc
The relationship between a state vector and the longitudinal waves vector is given by
(BRENNAN et al., 1997; BRENNAN, 1994)
The state vector h(ω), the transformation matrix H(ω) and the vector of wave amplitudes
a(ω) are given by
( ) " # ( )
u 1 1 AL
h(ω) = H(ω) = a(ω) = (60)
F jkES - jkES AR
√
where u is axial displacement, F is axial force, j is −1, E is Young’s modulus, S is the cross-
sectional area, AL and AR are left and right going longitudinal wave amplitudes as previously
defined. For simplify the notation in this text, it is considered h(ω) = h, H(ω) = H and a(ω) =
a.
aR = TaL (61)
where subscripts L and R denote the left and right hand segments and T is the spatial transfor-
mation matrix given by
" #
e jkL 0
T= (62)
0 e− jkL
where L is the length of segment. The force balance and continuity of displacement at the
junctions can be applied to give the relationships between the state vectors at the junctions
where the subscripts b denotes the sections of the homogeneous parts of the rod (before and after
of the periodic part), a-c are first and second parts of the rod cell and 1-3 denote the location of
junctions.
Equations (63) to (65) can be transformed into wave-mode coordinates using equation (59)
such that
where
T̃cell = Hc Tc H−1 −1
c Ha Ta Ha (73)
Based on equation (72) it is possible to note the periodic part of this proposed hybrid rod is
represented by the transfer matrix of that single cell (Eq. 73). In this case, a hybrid rod with N
cells as illustrated in figure 16 can be defined by the following relation
where the global transfer matrix T̃G for a periodic structure composed by N cells can be obtained
by
or, similarly
Ai
- rod ... ... rod +
a c a c a c
b 1 3 b
Sb Sa Sb
Sc
Ar La Lc At
Equation (74) can be rearranged in terms of incoming and outgoing waves. The periodic
structure can therefore be represented by the block diagram shown in figure 17.
The expression relating the waves is a classical notation from the literature such that
( ) ( )
A(3b)L 0
ainc = = (78)
A(1b)R Ai
and
( ) ( )
A(1b)L Ar
aout = = (79)
A(3b)R At
Note that the term A(3b)L = 0, i.e., is assumed that there are no incident waves from the right
side of the structure, according to previously presented in this text (see section 1.1).
and
where H̃ = T̃G Hb and the subscripts 1?2 denote the columns of the respective matrices given in
the brackets.
After some mathematical manipulations (see details in appendix A), from equation (77) we
can obtain the following relationship for transmitted At and incident Ai amplitudes waves
At j2 2
= (α + β 2 + αβ b(Ω)) (82)
Ai η(Ω)
where
√
r
1 ρca
η(Ω) = α[(−Sab Eab ρab − √ ) sin(πΩ) cos(πΩLca )+
Sab Eab ρab Eca
√
r (83)
1 ρca
(−Scb Ecb ρcb − √ ) cos(πΩ) sin(πΩLca )] + j[−αb(Ω) − 2β ]
Scb Ecb ρcb Eca
√
r
ρca 1
b(Ω) = −2 cos(πΩ) cos(πΩLca ) + (Sca Eca ρca + √ )×
Eca Sca Eca ρca
r (84)
ρca
× sin(πΩ) sin(πΩLca )
Eca
ΛN N
1 − Λ2 Λ2 ΛN N
1 − Λ1 Λ2
α= , β= (85)
Λ1 − Λ2 Λ1 − Λ2
N is the number of cells and Λ1,2 are the eigenvalues obtained from T̃cell matrix (see details in
section 3.4).
Lc
Lca = (86)
La
Sa Sc Sc
Sab = Scb = Sca = (87)
Sb Sb Sa
48
Ea Ec Ec
Eab = Ecb = Eca = (88)
Eb Eb Ea
ρa ρc ρc
ρab = ρcb = ρca = (89)
ρb ρb ρa
are ratios of geometric and physical parameters and
ωLa
Ω= (90)
πva
is non-dimensional frequency. From the equations (87) to (89) it is easy to show that
For clarity, equation (82) is a new proposition of this work to study the hybrid rod defined
herein . Based on this relation and using the very convenient physical parameters normalization
(Eqs. 86 to 89) it is possible to design a hybrid rod to satisfy a project engineering requirements.
49
The proposed formulation also allows to study a relation for the reflected (Ar ) and incident
(Ai ) waves amplitudes as shown in appendix A. According to the results (section 4.1.2), this is
an equivalent way for analysing the vibration suppression problem involving this kind of hybrid
rods.
√
r
Ar 1 ρca
= {−α[(−Sab Eab ρab + √ ) sin(πΩ) cos(πΩLca )+
Ai Sab Eab ρab Eca
√
r
1 ρca
(−Scb Ecb ρcb + √ ) sin(πΩLca ) cos(πΩ)]− (94)
Scb Ecb ρcb Eca r
√ 1 ρca
jα[(Sca Eca ρca − √ ) sin(πΩ) sin(πΩLca )]}/η
Sca Eca ρca Eca
( ) ( )
uR uL
= T̃cell (95)
FR FL
The periodic condition (given by equation 1) for the displacements and forces in a section
implies to the following equation (BRILLOUIN, 1953)
( ) ( )
uR uL
=Λ (96)
FR FL
From equations (95) and (96) the following eigenvalue problem is defined
( )
uL
(T̃cell − ΛI) =0 (97)
FL
where I is the identity matrix and Λ is the eigenvalue which depends on the frequency.
The nontrivial solution of this equation requires to solve det(T̃cell − ΛI) = 0. In this case
50
Λ2 + b(Ω)Λ + 1 = 0 (98)
where b(Ω) is give by equation (84) and det() indicates the determinant function.
Therefore, the roots of the equation (98) are the eigenvalues of the matrix Tcell . Similar
results for this equation can be found in reference (NIELSEN; SOROKIN, 2015). Although
these authors consider a different formulation it is possible to use their results to verify the
eigenvalues obtained from this proposed approach.
51
4 NUMERICAL SIMULATIONS
In this chapter, the results for the numerical simulations using the proposed transmission
and reflection relations (equations 82 and 94), respectively, are presented. Structures with geo-
metrical periodicity are considered in order to evaluate the responses of transmission, reflection,
and energy flow of waves that propagate longitudinally. Furthermore, the locations and shapes
of the stop bands are discussed, and the analysis of the dispersion curves, obtained by the
eigenvalues and propagation constants calculated from the equations (98) and (2), respectively,
is carried out.
Consider the structure with geometrical periodicity illustrated in figure 18. In this case, the
values of the physical properties (modulus of elasticity and density) are considered the same for
the whole hybrid structure. The influence of the number of cells N, cross-sectional area ratio
Sca , and length ratio Lca related to the cells composing parts, is evaluated on the responses of
wave transmission and reflection, as well as on the stop bands formation and energy flow. The
results are presented through curves of transmission and reflection defined in terms of frequency
and the geometrical parameters of the periodic structure.
Ai
-
... +
Sb Sa Sc Sa Sc Sa Sc Sb
Ar La Lc
At
Figure 19 shows the curve obtained for the wave transmission through a periodic structure
whose parameters are: N = 3, Lca = 1, Sca = Scb = 5, Sab = 1, ρcb = ρca = ρab = 1 and Ecb =
Eca = Eab = 1. This curve results from the amplitude ratio of transmitted and incident waves,
as defined by equation (82). The stop bands is clearly observed, that is, the frequency bands
in which the transmitted waves are significantly attenuated. On the other hand, for the same
structure, figure 20 shows the curve for reflection, that is, the amplitude ratio of reflected and
incident waves, as defined by equation (94). Also, as expected, the existence of frequency
bands in which reflection increases towards the unity is observed. Note that, for both curves,
the stop bands arise periodically, presenting multiple repetitions of integer numbers of the non-
dimensional frequency Ω. Depending on the values of Lca , groups of stop bands arise within
a frequency band, as shown in figure 21. Note that there are five groups of stop bands (SB)
in the frequency range considered, each one containing three stop bands. For simplicity and
clarity of understanding, in the next sections only the first group of stop bands will be presented
in the results. Nevertheless, the same analysis can be applied to the remaining cases since the
response has a periodic behavior.
Figure 19 - Transmission for N = 3, Lca = 1, Sca = 5, Scb = 5, Sab = 1, ρcb = ρca = ρab = 1
and Ecb = Eca = Eab = 1.
0.9
0.8
0.7
0.6
|At /Ai |
0.5
0.4
0.3
0.2
0.1
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
F requency Ω
Source: elaborated by the author.
53
Figure 20 - Reflection for N = 3, Lca = 1, Sca = 5, Scb = 5, Sab = 1, ρcb = ρca = ρab = 1 and
Ecb = Eca = Eab = 1.
0.9
0.8
0.7
0.6
|Ar /Ai |
0.5
0.4
0.3
0.2
0.1
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
F requency Ω
Source: elaborated by the author.
Figure 21 - Transmission for N = 3, Lca = 3, Sca = 5, Scb = 5, Sab = 1, ρcb = ρca = ρab = 1
and Ecb = Eca = Eab = 1.
0.9
0.8
0.7
0.6
|At /Ai |
0.5
0.4
0.3
0.2
0.1
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
F requency Ω
Source: elaborated by the author.
54
Figure 22 shows the results obtained considering different numbers of cells for the periodic
structures. The remaining values of the physical and geometrical parameters are considered
fixed and are described in the titles of each figure in this text. To better visualize the results,
figure 23 shows a close-up of figure 22. It can be seen that the attenuation increases substan-
tially as the number of cells changes from N = 1 to N = 4. From four cells on, no significant
attenuation is observed. On the other hand, the increase in number of cells does not cause the
stop band width to change significantly.
Figure 22 - Transmission for different number of cells considering Lca = 1, Sca = 5, Scb = 5,
Sab = 1, ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1.
1
N = 1
0.9 N = 2
N = 3
0.8 N = 4
N = 5
0.7 N = 10
0.6 N
|At /Ai |
Increases
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
F requency Ω
Source: elaborated by the author.
55
Figure 23 - Zoom in stop band of figure 22 for different number of cells considering Lca = 1,
Sca = 5, Scb = 5, Sab = 1, ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1.
0.4
N = 1
0.3 N = 2
N = 3
N
|At /Ai |
N = 4
N = 5 Increases
0.2 N = 10
0.1
0
0.3 0.4 0.5 0.6 0.7
F requency Ω
Source: elaborated by the author.
The minimum values of the transmission curves of figure 22, within the stop bands, are
calculated and shown in figure 24 for different numbers of cells and Sca . It can be seen that for
Sca = 1 the structure becomes a homogeneous rod (non-periodic), and there is no stop band.
Therefore, the transmitted wave’s amplitude is equal to the one from the incident wave. For
values of Sca different from one there are stop bands, and thus an attenuation of the transmitted
wave is observed. Note that if the number of cells N and the area ratio Sca increase the attenu-
ation level of the transmitted waves increase too. It is important to point out that, for the cases
in which the project cannot assume certain values of Sca (due to space limitations, for instance)
to attain the desired attenuation, lower values of Sca alongside a higher number of cells can be
used as an alternative. For example, if a transmission value of 0.03 is desired, a structure with
N = 6 and Sca = 2 can be used instead of choosing N = 3 and Sca = 4 (see figure 24). Therefore,
this is a very practical process when designing such structures.
56
Figure 24 - Minimum values of the transmission for different values of Sca and number of
cells considering Lca = 1, Sab = 1, ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1.
100
Sca = 1
Sca = 2
Sca = 3
10-1 Sca = 4
0.03
Sca = 5
Sca = 10
min |At /Ai |
10-2
10-3
10-4
Sa Sc
1 2 3 4 5 6 7 8 9 10
N umber of cells
Source: elaborated by the author.
Unlike the number of cells, the geometrical parameter Sca causes an important effect on
the stop band widths. To evaluate such effect, these frequency bandwidths are calculated for
different values of Sca and an attenuation reference line of −3 dB1 is considered. The difference
between the two points of intersection with the transmission curve is calculated, as shown in
figure 25. The results obtained for the bandwidths (BW ) are shown in figure 26. In this figure,
a significant change in the bandwidth can be seen as Sca increases or decreases. Note that
the bandwidth has a symmetrical behavior for inverse values of Sca , i.e., the same bandwidth
is obtained for Sca or 1/Sca . Moreover, as the number of cells increases, a slight bandwidth
reduction is observed since an increase in its value is followed by an increase in the amplitude
attenuation of the stop bands, as shown in figures 22 and 24.
1 This
√ √
reference represents the bandwidth at 1/ 2 since 20 log10 (1/ 2) ≈ −3 dB.
57
Figure 25 - Transmission and bandwidth for different values of Sca considering N = 3, Lca = 1,
Sab = 1, ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1.
1.1
1
0.9
bandwidth
0.8
0.7
|At /Ai |
0.6 Sca
Increases
0.5 Sa Sc
0.4
Sca =1
0.3 Sca =2
Sca =3
0.2 Sca =4
0.1 Sca =5
Sca = 10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
F requency Ω
Source: elaborated by the author.
Figure 26 - Bandwidth for different values of Sca and number of cells considering Lca = 1,
Sab = 1, ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1.
N =3 N =5 N = 10
1
0.9
0.8
0.7 N
Increases
0.6
BW
0.5
0.4
0.3
0.2
Sa Sc Sa Sc
0.1
0
10-2 10-1 100 101 102
Sca
Source: elaborated by the author.
58
Figure 27 shows the effect of the geometrical parameter Lca on the stop bands formation
for integer values of this parameter. It can be seen that the stop bands arise within the frequency
band 0 6 Ω 6 1, and the number of unities is equal to the values of Lca ; that is, if Lca = 1 there
is only 1 stop band, if Lca = 2 there are 2 stop bands, and so on. On the other hand, figure
28 shows the stop bands formation for values of Lca lower than the unity. In this case, it can
be seen that they do not arise within a unit non-dimensional frequency band, but rather within
a frequency band whose maximum value is the inverse of Lca . Also, it can be seen that the
number of stop bands is the inverse of Lca . For example, if Lca = 1/2 there are 2 stop bands that
are within the frequency band 0 6 Ω 6 2; if Lca = 1/4 there are 4 stop bands that are within the
frequency band 0 6 Ω 6 4, and so on.
Figure 27 - Stop bands for different values of Lca 1 1 considering N = 3, Sca = 5, Scb = 5,
Sab = 1, ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1.
0.9
0.8
0.7
F requency Ω
0.6
0.5
0.4
0.3
La Lc
0.2
0.1
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Lca
Source: elaborated by the author.
59
Figure 28 - Stop bands for different values of Lca < 1 considering N = 3, Sca = 5, Scb = 5,
Sab = 1, ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1.
10
7
F requency Ω
3
La Lc
2
0
0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
Lca
Source: elaborated by the author.
The transmission curve for Lca = 3 is shown in figure 29. It can be seen that there is a group
consisting of three stop bands within the frequency band 0 < Ω < 1. In addition, the stop band
widths for different values of Sca are shown in figure 30. One notices that the width of the first
and the third stop bands are the same, but both are narrower than the second one.
The transmission curve for Lca = 1/3 is shown in figure 31. It can be seen that, in this
case, there is a group consisting of three stop bands within the frequency band 0 ≤ Ω ≤ 3.
Figure 32, in turn, shows the stop band widths for different values of Sca . One notices the same
characteristic on these widths, similar to the shape for Lca = 3.
60
0.9
bandwidth
0.8
0.7
0.6
|At /Ai |
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
F requency Ω
Source: elaborated by the author.
Figure 30 - Bandwidths for different values of Sca considering N = 3, Lca = 3, Sab = 1, ρcb =
ρca = ρab = 1 and Ecb = Eca = Eab = 1.
0.3
0.25
0.2
BW
0.15
0.1
0.05 Sa Sc Sa Sc
0
10-2 10-1 100 101 102
Sca
Source: elaborated by the author.
61
0.9
bandwidth
0.8
0.7
0.6
|At /Ai |
0.5
0.4
0.3
0.2
0.1
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
F requency Ω
Source: elaborated by the author.
Figure 32 - Bandwidths for different values of Sca considering N = 3, Lca = 1/3, Sab = 1,
ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1.
0.9
0.8
0.7
0.6
BW
0.5
0.4
0.3
0.2
Sa Sc Sa Sc
0.1
0
10-2 10-1 100 101 102
Sca
Source: elaborated by the author.
62
The numerical results obtained for the reflection of longitudinal waves are presented in this
section. Figure 33 shows the reflection curves for periodic structures using different numbers of
cells. The remaining values of the physical and geometrical parameters are fixed, as indicated
in the figures. To better visualize the results, figure 34 shows a close-up of figure 33. It can
be clearly seen that the amplitude of reflected waves increases significantly as the number of
cells in the structure changes from N = 1 to N = 3. From three cells on there exists already
a frequency range in which full reflection of the incident wave occurs. On the other hand, the
increase in number of cells does not cause any significant change in stop band widths.
Figure 33 - Reflection for different number of cells considering Lca = 1, Sca = 5, Scb = 5,
Sab = 1, ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1.
1.1
1
0.9
0.8 N
0.7 Increases
|Ar /Ai |
0.6
0.5
0.4
N = 1
0.3 N = 2
N = 3
0.2 N = 4
0.1 N = 5
N = 10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
F requency Ω
Source: elaborated by the author.
63
Figure 34 - Zoom in stop band of figure 33 for different number of cells considering Lca = 1,
Sca = 5, Scb = 5, Sab = 1, ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1.
0.99
0.98
N
N = 1
0.97 Increases
N = 2
N 3
|Ar /Ai |
=
0.96 N = 4
N = 5
0.95 N = 10
0.94
0.93
0.92
The maximum values of the reflection curves of figure 33, within the stop bands, are cal-
culated and shown in figure 35 for different numbers of cells and cross-sectional area ratios
Sca = 1. It can be seen that, for Sca = 1, the structure becomes a homogeneous rod (non-
periodic) and, therefore, since there is no stop band, the amplitude of the reflected wave is zero.
As for the values of Sca that are different from the unity, there exist stop bands and, therefore,
reflection of the incident wave is observed. Note that the ratio Sca and the number of cells N
cause an increase in the reflection as these values get higher. It is important to point out that, for
specific cases in which the project cannot assume values of Sca in order to guarantee the desired
level of reflection, it is possible to reach such level by applying lower values of Sca alongside a
higher number of cells, as previously observed for the transmission curves. This observation is
presented in section 4.1.1 for the transmission analysis.
Like transmission, it can also be seen that, for reflection, the geometrical parameter Sca
causes an important effect on the stop band widths. To evaluate such effect, similarly to the
transmission analysis, these frequency bandwidths are calculated for different values of Sca and
√
amplitude ratio of 1/ 2, as indicated by the black circles on this reference line (Fig. 36). The
results obtained for the bandwidths (BW ) are shown in figure 37. Note that the bandwidths are
the same for reflection and transmission. Moreover, figures 38 and 39 show both the reflection
and transmission curves for Lca = 1 and Lca = 1/3, respectively, in which one can clearly note
√
that their intersection occurs at 1/ 2. This fact shall be discussed in the next section.
64
Figure 35 - Maximum values of the reflection for different values of Sca and number of cells
considering Lca = 1, Sab = 1, ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1.
1.1
1
0.9
Sca =1
0.8 Sca =2
Sca =3
0.7
max |Ar /Ai |
Sca =4
0.6 Sca =5
Sca = 10
0.5
0.4
Sa Sc
0.3
0.2
0.1
0
1 2 3 4 5 6 7 8 9 10
Number of cells
Source: elaborated by the author.
Figure 36 - Reflection and bandwidth for different values of Sca considering N = 3, Lca = 1,
Sab = 1, ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1.
1
0.9 bandwidth
0.8
0.7
|Ar /Ai |
0.6 Sca
Increases
0.5
0.4 Sca =1
Sca =2
0.3
Sca =3
0.2 Sca =4
Sca =5
0.1 Sca = 10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
F requency Ω
Source: elaborated by the author.
65
Figure 37 - Bandwidths for different values of Sca considering N = 3, Lca = 1, Sab = 1, ρcb =
ρca = ρab = 1 and Ecb = Eca = Eab = 1.
0.9
0.8
0.7
0.6
BW
0.5
0.4
0.3
0.2
Sa Sc Sa Sc
0.1
0
10-2 10-1 100 101 102
Sca
Source: elaborated by the author.
1
0.9 bandwidth
0.8
0.7
Amplitude
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
F requency Ω
Source: elaborated by the author.
66
1
0.9
bandwidth
0.8
0.7
Amplitude
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
F requency Ω
Source: elaborated by the author.
The numerical results obtained for the mean power analysis of transmitted and reflected
waves in periodic rods are presented in this section. The main goal is to verify power distri-
bution and energy conservation during the process of wave dispersion, as presented in section
2.3.3. Figure 40 shows the coefficients of transmitted power CPT and reflected power CPR , re-
spectively, for the periodic structure whose configuration is defined in its title. In this figure, it
can be seen that the curves intersect at CPT = CPR = 0.5, that is, the point where the incident
power is equally divided into transmitted and reflected power. Note that, for CPT = 0.5, the cor-
√
responding transmission value is equal to 1/ 2, which therefore means that the intersection of
both transmission and reflection curves (see figures 38 and 39 ) occurs when the energy flow of
transmitted and reflected waves is the same. Figure 42, in turn, shows that the sum of transmis-
sion and reflection powers is equal to the incident power, that is, the energy flow is conserved,
as expected by equation (58).
67
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
F requency Ω
Source: elaborated by the author.
Figure 41 - Coefficients of transmission power versus transmission for different values of cells
considering Lca = 1, Sca = 5, Scb = 5, Sab = 1, ρcb = ρca = ρab = 1 and Ecb =
Eca = Eab = 1.
N =3 N =5
1
0.9
Coefficient of P ower (CP T )
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
T ransmission |At /Ai |
Source: elaborated by the author.
68
Figure 42 - Coefficient of total power considering N = 3, Lca = 1/3, Sca = 5, Scb = 5, Sab = 1,
ρcb = ρca = ρab = 1 and Ecb = Eca = Eab = 1.
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
F requency Ω
Source: elaborated by the author.
As discussed in section 2.1.2, the dispersion curves can be used to predict in which fre-
quency bands the stop bands arise. These curves can be analyzed using the eigenvalues from
the unit cell’s transfer matrix (which are the roots of Eq. 98), or using the propagation constants
obtained through the Bloch parameters ψ. Figures 43 and 44 show the dispersion curves us-
ing the eigenvalues obtained for the structures whose results of transmission and reflection are
shown in figures 38 and 39. Note that the stop bands exist for values of |Λ| = 6 1, and are located
in the corresponding frequency bands observed in the transmission and reflection curves. It is
important to point out that similar results for this kind of analysis were obtained and discussed
in greater detail in the reference (NIELSEN; SOROKIN, 2015).
69
|Λ2 | |Λ1 |
4
Amplitude
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
F requency Ω
Source: elaborated by the author.
Figure 44 - Eigenvalues considering N = 3, Lca = 1/3, Sca = 5, Scb = 5, Sab = 1, ρcb = ρca =
ρab = 1 and Ecb = Eca = Eab = 1.
|Λ2 | |Λ1 |
4
Amplitude
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
F requency Ω
Source: elaborated by the author.
Likewise, figures 45 and 46 show the dispersion curves, but now using the propagation
70
constants. It can be seen that the stop bands are located in those frequency bands where the
values of ψ have a real part (attenuation factor) and an imaginary part (phase factor) with values
that are constant and equal to π or 0. On the other hand, for values of ψ with only an imaginary
part, the wave propagation occurs without attenuation, thus defining the pass band.
Real ψ Imag ψ
3.5
2.5
Amplitude
1.5
0.5
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
F requency Ω
Source: elaborated by the author.
Real ψ Imag ψ
3.5
2.5
Amplitude
1.5
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
F requency Ω
Source: elaborated by the author.
71
In this chapter, practical aspects of periodic rod design are discussed. When designing
structures, engineers must meet certain requirements, such as frequency band, attenuation am-
plitudes at vibration levels, as well as geometric constraints, for example. Therefore, for clarity
of understanding, practical details and a hypothetical condition of a vibration problem are pre-
sented as follows. For practical purposes, quantities that involve states (displacement, velocity
and acceleration) are usually used instead of wave amplitudes. The following results are thus
converted from wave amplitudes to a displacement relation between the right and left sides,
uR and uL , respectively, of the periodic part (see figure 47). From equations (59) and (60), it
follows that
uL = Ar + Ai (99)
and
uR = At (100)
The relationship between equations (100) and (99) results in the transmission written in
terms of displacements,
At
uR Ai
= (101)
uL Ar
+1
Ai
Practical engineering problems involve finite structures, it is thus also necessary to adapt
the model developed in this thesis (infinite periodic rod) such that it represents a finite periodic
rod. For this purpose, it must be considered that the cross-sectional area of the homogeneous
rod Sb assumes a lower value in relation to the other cross-sectional areas, as illustrated in figure
47. Therefore, in the simulations presented in section 5.2, the following values are considered:
Sab = 1000 and Scb = 1000 × Sca .
72
Figure 47 - Illustration of the approximation considered from the infinite model (theoretical)
to the finite model (real), in which the cross-sectional area Sb tends to zero.
Lt
Sb ... Sb +
Sa Sc Sa Sc Sa Sc
uL La Lc uR
Consider the practical vibration problem in which the goal is to design a rod with only
geometrical periodicity, where the material is the same for the whole structure, to meet the
following requirements:
3. Maximum area ratio of Sca = 2 due to a hypothetical space constraint for the structure;
In this section, three candidate solutions for the design of the periodic rod corresponding
to the requirements from section 5.1 are presented. To better illustrate this process, initially,
the first two solutions do not meet all requirements. On the other hand, a discussion of how to
interpret the results is carried out and, in this case, how to choose the appropriate material for
the project as well.
Initially, suppose that the steel (E = 210 Gpa and ρ = 7800 kg/m3 ) is the material available
for manufacturing the periodic rod. From the results obtained in this work (section 4.1.1),
73
requirements 1, 2 and 3 (location, bandwidth and area ratio) are met for values of La = 430 mm,
Sca = 2 and N = 1 in the condition that Lca = 1, as seen in figure 48. Note that, for Lca = 3 and
Lca = 5, requirement 1 is met, but requirements 2 and 5 are not, since for these two conditions
of Lca the lengths of a single cell are 1720 mm and 2580 mm, respectively, and thus higher than
the maximum Lt required.
To evaluate requirement 4 (attenuation level), the effects of the number of cells are eval-
uated, as shown in figure 49. Note that the requirement is met when N=3. However, for
such number of cells the condition related to the total length of the periodic rod is not: Lt =
N × Lcell = N × (La + Lc ) = 3 × (432 + 432) = 3 × (864) = 2592 mm > 1000.
Conclusion for the steel:: requirements 1, 2, 3 and 4 are met, but requirement 5 is not.
Therefore, the material must be changed.
Figure 48 - Stop bands for different values of Lca considering La = 432 mm (steel case) or
La = 421, 9 mm (aluminum case) or La = 135 mm (nylon case), N = 1, Sca = 2,
Sab = 1000 and Scb = 2000.
5000
4500
4000
F requency [Hz]
3000
central frequency
2500
2000
1500
1000
1 2 3 4 5
Lca
Source: elaborated by the author.
By inspection of equation (90), for the same value of La , lower values for the propagation
speed va (lower E/ρ) result in lower values for the dimensional frequency ω and, as a conse-
quence, the stop bands are located in lower frequencies. On the other hand, lower values of
La cause the stop bands to arise in higher frequencies. Therefore, an alternative to meet the
requirements imposed is to use a material whose propagation speed is lower than the one of
steel, which renders possible the reduction of length La .
74
103
N = 1 (BW = 1750 H z)
N = 2 (BW = 1701 H z)
N = 3 (BW = 1593 H z)
102
T ransmission |uR /uL |
101
bandwidths
100
0.2
10-1
0 1000 2000 3000 4000 5000 6000
F requency [Hz]
Source: elaborated by the author.
Consider the aluminum (E = 70 Gpa and ρ = 2700 kg/m3 ) as the candidate material
for manufacturing the periodic rod. Repeating all the simulations for this material, for La =
421, 9 mm, and keeping the values of N, Sca and Lca , equivalent results are obtained, as shown
in figures 48 and 49. Thus, for this case, requirements 1, 2, 3 and 4 are also met, but requirement
5 is not: Lt = N × Lcell = N × (La + Lc ) = 3 × (421, 9 + 421, 9) = 3 × (843, 8) = 2531, 4 mm >
p
1000. Note that the aluminum’s wave propagation speed EAl /ρAl ≈ 5091 m/s is close to the
p
one of steel Esteel /ρsteel ≈ 5188 m/s, which does not allow a considerable reduction in length
La .
Conclusion for the aluminum: requirements 1, 2, 3 and 4 are met, but requirement 5 is
not. Therefore, the material must be changed.
In the case of Nylon (E = 3 Gpa and ρ = 1130 kg/m3 ), the propagation speed is around
p
Eny /ρny ≈ 1630 m/s, which is quite different from those of steel and aluminum. Therefore, a
considerable reduction in length La is expected in relation to the two previous cases. Repeating
the analysis for this material, for La = 135 mm, and keeping the values of N, Sca and Lca , equiv-
75
alent results to those shown in figures 48 and 49 are obtained. For this case, the requirement
of total length of the rod is met, since Lt = N × Lcell = N × (La + Lc ) = 3 × (135 + 135) =
3 × (270) = 810 mm < 1000 mm.
Therefore, considering a cylindrical periodic rod with a segment diameter c of 50 mm, the
full design is given as follows:
• Number of cells, N = 3;
Figure 50 shows a periodic rod whose dimensions are specified above. Therefore, to val-
idate the design approach proposed in this work, the experimental results for the periodic rod
defined are presented in the next section.
Figure 50 - Illustration of a cylindrical periodic rod whose dimensions are obtained using ny-
lon as material.
50,0 mm
35,4 mm
135,0 mm 135,0 mm
810,0 mm
The experimental setup is illustrated in figure 51. In this figure can be seen a right hand
side higher diameter part which is used to fix the accelerometer and avoid to damage it with the
impact hammer. The equipments used consists of: two uniaxial accelerometers (PCB 352A25
miniature), an impact hammer (PCB 086C04), acquisition system (LMS SCADAS Mobile, 8
channels), and a computer (DELL Intel(R) Core(TM) i7-2600, CPU 3.4 GHz, 16GB RAM).
The periodic rod is hung by elastic bands to simulate the free-free condition, as shown in figure
52 and, in addition, it is excited by an impact force at the end whose diameter is the smallest.
Then, the vibrations are measured by two accelerometers fixed at both ends of the rod. The
measurements are obtained and post-processed using the software LMS Test.Lab Impact Testing
version 14A with the following acquisition settings: 6400 Hz of bandwidth, 8192 spectral lines,
1,28 acquisition time and 0,78125 of resolution.
Figure 51 - Schematic of experimental setup showing the periodic rod, accelerometers, impact
hammer, acquisition system and computer.
aquisition system
computer
accelerometers
101
bandwidth
0
10
0.2
10-1
0 1000 2000 3000 4000 5000
F requency [Hz]
Source: elaborated by the author.
79
6 CONCLUSIONS
This work has presented an alternative approach to design periodic rod structures. Different
analysis techniques concerning this kind of structure have been developed in the literature. In
general, they discuss how to find the stop bands for a definite structure geometry. However,
the practical problem in engineering is, rather, to define a structure that results in vibration
attenuation in certain frequency bands of interest. In addition, analysts must understand other
requirements, such as dimensions and types of materials. Therefore, the present work introduces
a contribution based on this context.
According to the alternative method for designing periodic rods presented in this work, as
well as to the theoretical and experimental results obtained, one can conclude that:
• it is possible to model a periodic infinite hybrid rod by connecting a finite periodic struc-
ture between two semi-infinite rods. To do so, the methodology that relates state vectors
and wave amplitudes – already known in the literature – is employed;
• the simulations performed indicate that when designing a periodic rod as shown herein, 3
to 4 cells can be used such that the attenuation levels attained represent what is obtained
for a periodic structure in its entire extent. Other geometrical and material combinations
can provide different results;
• the geometrical parameters of the periodic structure (length ratio Lca , area ratio Sca and
number of cells N) have an important influence on stop bands formation, as well as on
their characteristics, such as: location in the frequency spectrum, bandwidth and attenua-
tion levels;
√
• the transmission and reflection curves intersect at the amplitude of 1/ 2, and it occurs
when the energy flows of transmitted and reflected waves are the same (CPT = CPR = 0, 5).
Also, for this amplitude value, the stop bands bandwidths for transmission and reflection
are equal. Furthermore, it has been shown that in the process of wave dispersion in non-
dissipative periodic rods, there exists energy flow conservation, that is, the energy from
the incident wave is partially divided into reflected and transmitted energies;
Also, practical details for defining a structure that meets hypothetical requirements of a
vibration suppression problem are presented in this work. For this case, to illustrate the design
technique’s application, three candidate solutions are presented for the design of the periodic
rod based on the desired requirements, and it has been shown that one of them meets such
requirements. Moreover, the material adopted has demonstrated to have an important influence
on the stop bands location, that is, materials whose wave propagation speed is lower cause the
stop bands to arise in lower frequencies. Additionally, on the other hand, smaller lengths La
cause the stop bands to arise in higher frequencies.
To verify the proposed formulation, a periodic rod has been designed applying the method-
ology developed in this work. Experimental tests have been carried out to verify the stop bands
predicted by the formulation. Using a nylon rod with 3 cells and 0,81 meters in length, the stop
band was identified between 2311 and 3782 Hz, which is consistent with the computer simu-
lations. Also, it has been considered an non-dimensional ratio to include structural damping
through the complex modulus of elasticity. A value of η = 0, 08 has been chosen by exhaus-
tive search (arbitrary choices) to represent the dynamics of the first three modes with the visual
similarity criterion between the curves.
Finally, it must be stressed that the formulation developed seems favorable to its application
in this kind of structure, especially because the equations are scalar and, thereby, no solutions
for matrix systems equations are required, as seen in formulations using finite elements methods
and spectrum elements. Furthermore, compared to the classic parametrization by impedance ra-
tio, the parametrization of equations by physical and non-dimensional geometrical properties
broadens the understanding of the designer. In addition, vibration suppression using this ap-
proach is more appropriate to big structures when metallic materials are employed (owing to
their high wave propagation speed). On the other hand, the use of materials whose wave propa-
gation speed is low may make feasible the design of small structures with vibration suppression
in low frequencies.
81
• to develop the design graphics (figures 24, 26 and 27) for cells of material periodicity,
mixed (geometry and material), among other variations;
• to adapt the methodology proposed in this thesis to the design of periodic beam structures,
as well as make use of combined solutions of longitudinal, flexural and shear waves prop-
agation;
• to discuss the use of this formulation on the design of truss structures, seeking the inclu-
sion of rods with periodic cells in strategical positions;
• to seek strategies to include damping in the formulation developed in order to adjust each
mode of interest.
82
REFERENCES
ASIRI, S.; BAZ, A.; PINES, D. Periodic struts for gearbox support system. Journal of
Vibration and Control, v. 11, n. 6, p. 709–721, 2005.
BRENNAN, M. J.; TANG, B.; ALMEIDA, F. C. L. Waves motion in elastic structures. In: JR.
V. ANS STEFFEN JR., V. L.; SAVI, M. A. (Ed.). Dynamics of smart systems and structures,
concepts and applications. [S.l.]: Springer, 2016.
BRILLOUIN, L. Wave propagation in periodic structures: electric filters and crystal lattices.
[S.l.]: Dover Publications, 1953.
CHEN, R.; WU, T. Vibration reduction in a periodic truss beam carrying locally resonant
oscillators. Journal of Vibration and Control, v. 22, n. 1, p. 2170–285, 2016.
CHENG, Z.; SHI, Z. Composite periodic foundation and its application for seismic isolation.
Earthquake Engineering & Structural Dynamics, v. 47, n. 4, p. 925–944, 2017.
DUHAMEL, D.; MACE, B. R.; BRENNAN, M. J. Finite element analysis of the vibrations of
waveguides and structures periodics. Journal of Sound and Vibration, v. 294, p. 205–220,
2006.
FAHY, F.; GARDONIO, P. Sound and structural vibration: radiation, transmission and
response. 2nd. ed. [S.l.]: Elsevier, 2007.
GAN, C.; WEI, Y.; YAN, S. Longitudinal wave propagation in a multi-step rod with variable
cross-section. Journal of Vibration and Control, v. 22, n. 3, p. 837–852, 2016.
GAOFENG, J.; ZHIFEI, S. A new seismic isolation method and its feasibility study.
Earthquake Engineering and Engineering Vibration, v. 9, n. 1, p. 75–82, 2010.
83
GAZALET, J.; DUPONT, S.; KASTELIK J., C.; ROLLAND, Q.; DJAFARI-ROUHANI, B.
A tutorial survey on waves propafation in periodic media: Electronic, photonic and phononic
crystals. perception of the bloch theorem in real and fourier domains. Wave Motion, v. 50, p.
619–654, 2013.
GRAFF, K. F. Wave motion in elastic solids. New York: Dover Publications, 1991.
HVATOV, A.; SOROKIN, S. V. Free vibrations of finite periodic structures in pass- and
stop-bands of the counterpart infinite waveguides. Journal of Sound and Vibration, v. 347, p.
200–217, 2015.
JUNG, W.; GU, Z.; BAZ, A. Mechanical filtering characteristics of passive periodic engine
mount. Finite Elements in Analysis and Design, v. 46, p. 685–697, 2010.
MACE, B. R.; DUHAMEL, D.; BRENNAN, M. J.; HINKE, L. Finite element prediction of
wave motion in structural waveguides. Journal of the Acoustical Society of America, v. 117,
p. 2835–2843, 2005.
MENCIK, J. M. On the low- and mid-frequency forced response of elastic structures using
wave finite elements with one- dimensional propagation. Computers and Structures, v. 88, p.
674–689, 2010.
. New advances in the forced response computation of periodic structures using the wave
finite element (wfe) method. Computational Mechanics, v. 54, n. 3, p. 789–801, 2014.
NIELSEN, R. B.; SOROKIN, S. V. Periodicity effects of axial waves in elastic compound rods.
Journal of Sound and Vibration, v. 353, p. 135–149, 2015.
NOBREGA, E. D.; GAUTIER, F.; PELAT, A.; SANTOS, J. M. C. D. Vibration band-gaps for
elastic metamaterial rods using wave finite element method. Mechanical Systems and Signal
Processing, v. 79, p. 192–202, 2016.
OSTACHOWICZ, W.; KUDELA, P.; KRAWCZUK, M.; ZAK, A. Guided waves in structures
for SHM: the time-domain spectral element method. [S.l.]: John Wiley & Sons, 2012.
PHANI, A. S.; J., W.; FLECK, N. A. Wave propagation in two-dimensional periodic lattices.
Acoustical Society of America, v. 119, n. 4, p. 1995–2005, 2006.
SOLAROLI, G.; GU, Z.; BAZ, A.; RUZZENE, M. Wave propagation in periodic stiffened
shells: Spectral finite element modeling and experiments. Jornal of Vibration and Control,
v. 9, p. 1057–1081, 2003.
SONG, Y.; WEN, J.; YU, D.; WEN, X. Suppression of vibration and noise radiation in a
flexible floating raft system using periodic structures. Journal of Vibration and Control,
v. 21, n. 2, p. 217–228, 2015.
SZEFI, J. T. Helicopter gearbox isolation using periodically layered fluidic isolators. Thesis
(PhD) — Pennsylvania State University, Department of Mechanical and Nuclear Engineering,
Pennsylvania, 2003a.
. Design and testing of a compact fluidic layered isolator for high-frequency helicopter
gearbox isolation. In: AIAA/ASME/ASCE/AHS/ASC STRUCTURES, STRUCTURAL
DYNAMICS & MATERIALS CONFERENCE. 45., 2004. Conference... [S.l.: s.n.], 2004.
XIONG, C.; SHI, Z.; XIANG, H. J. Attenuation of building vibration using periodic
foundations. Advances in Structural Enginnering, v. 15, n. 8, p. 1375–1388, 2012.
XIUCHANG, H.; AIHUA, J.; ZHIYI, Z.; HONGXING, H. Design and optimization of
periodic structure mechanical filter in suppression of foundation resonances. Journal of Sound
and Vibration, v. 330, p. 4689 – 4712, 2011.
This appendix describes the mathematical development to obtain the transmission (Eq. 82)
and reflection (Eq. 94) presented in sections 3.2 and 3.3, respectively. Mathematical manipula-
tions are developed using the relation between the outgoing aout and incoming ainc waves given
by (see section 3.1)
where
N
H̃ = T̃G Hb = T̃cell Hb (105)
T̃cell = Hc Tc H−1 −1
c Ha Ta Ha = T̃c T̃a (106)
To develop the equation (102) it is necessary to manipulate the equations (106) to (103) as
shown:
Consider
" #
1 1
Hc = (107)
jkc Ec Sc −jkc Ec Sc
and
1 1
" #
1 −jkc Ec Sc -1
H−1
c = = ... = 21 j2kc Ec Sc (108)
det(Hc ) −jkc Ec Sc 1 1
2 −j2kc Ec Sc
where
1 1
" #" #
1 1 e jkc Lc 0 2 j2kc Ec Sc (110)
T̃c = Hc Tc H−1
c = 1 1
jkc Ec Sc −jkc Ec Sc e− jkc Lc
0
2 −j2kc Ec Sc
1 1
" #
e jkc Lc e− jkc Lc 2 j2kc Ec Sc
T̃c = 1 1 (111)
jkc Ec Sc e jkc Lc −jkc Ec Sc e− jkc Lc
2 −j2kc Ec Sc
1
cos(ka La ) sin(ka La )
T̃a = Ha Ta H−1
a = ... =
Za (113)
−Za sin(ka La ) cos(ka La )
where
Za = ka Ea Sa ; Zc = kc Ec Sc (114)
" #
T̃11 T̃12
T̃cell = T̃c T̃a = ... = (115)
T̃21 T̃22
where
Za
T̃11 = cos(kc Lc ) cos(ka La ) − sin(kc Lc ) sin(ka La ) (116)
Zc
1 1
T̃12 = cos(kc Lc ) sin(ka La ) + sin(kc Lc ) cos(ka La ) (117)
Za Zc
Zc
T̃22 = − sin(kc Lc ) sin(ka La ) + cos(kc Lc ) cos(ka La ) (119)
Za
In this work is considered the following relation for power of a 22 matrix (ANDREESCU,
2014)
N
T̃cell = α T̃cell − β I (120)
88
where
ΛN N
1 − Λ2 Λ2 ΛN N
1 − Λ1 Λ2
α= , β= (121)
Λ1 − Λ2 Λ1 − Λ2
N is the number of cells, Λ1,2 are first and second eigenvalues obtained from T̃cell matrix and I
is 22 identity matrix. Using the relationship (120), is written the following equation
" #" #
N (α T̃11 − β ) α T̃12 1 1
H̃ = T̃cell Hb = (122)
α T̃21 (α T̃22 − β ) jZb − jZb
" #
(α T̃11 − β + jZb α T̃12 ) (α T̃11 − β − jZb α T̃12 )
H̃ = (123)
(α T̃21 + jZb (α T̃22 − β )) (α T̃21 − jZb (α T̃22 − β ))
where
Zb = kb Eb Sb (124)
" #
1 −(α T̃11 − β − jZb α T̃12 )
µ = [(Hb )1 | (−H̃)2 ] = (125)
jZb −(α T̃21 − jZb (α T̃22 − β ))
" #
−(α T̃11 − β + jZb α T̃12 ) 1
γ = [(−H̃)1 | (Hb )2 ] = (126)
−(α T̃21 + jZb (α T̃22 − β )) − jZb
and also
" #
1 − jZb −1
γ −1 = (127)
det(γ) (α T̃21 + jZb (α T̃22 − β )) −(α T̃11 − β + jZb α T̃12 )
89
where
Substituting the equations (125) and (127) into the equation (102), it is possible to write
where
" #
G11 G12
G = −γ −1 µ = ... = (130)
G21 G22
j2Zb
G11 = (131)
η
Using the equations (78) and (79) and the equation shown above is possible to write
( ) " #( )
Ar G11 G12 0
= (135)
At G21 G22 Ai
Note that is assumed that there are no incident waves from the rigth side of the structure. Thus,
90
At Atransmitted
= = G22 (136)
Ai Aincident
Ar Are f lected
= = G12 (137)
Ai Aincident
Substituting the equations (116) to (119) into equations (132) and (134) and considering
ωLa
that Ω = is non-dimensional frequency, the non-dimensional form is obtained
πva
j2[α 2 + β 2 + αβ b(Ω)]
G22 = (138)
η
and
1 1
G12 = {−α[(−κab + ) sin(πΩ) cos(πτΩ) + (−κcb + ) sin(πτΩ) cos(πΩ)]
κab κcb (139)
1
− jα[(κca − ) sin(πΩ) sin(πτΩ)]}/η
κca
where
1 1
η = α[(−κab − ) sin(πΩ) cos(πτΩ) + (−κcb − ) cos(πΩ) sin(πτΩ)]
κab κcb (140)
+ j[−αβ b(Ω) − 2β ]
1
b(Ω) = −2 cos(πΩ) cos(πτΩ) + (κca + ) cos(πΩ) sin(πτΩ) (141)
κca
and
Za Zc Zc
κab = κcb = κca = (142)
Zb Zb Za
are impedance parameters and
Lc va
τ= (143)
La vc
is ratio of propagation times parameter.
The equations (138) and (139) can be rewritten in terms of physical and geometric param-
91
j2[α 2 + β 2 + αβ b(Ω)]
G22 = (144)
η
√
r
1 ρca
G12 = {−α[(−Sab Eab ρab + √ ) sin(πΩ) cos(πΩLca )+
Sab Eab ρab Eca
√
r
1 ρca
(−Scb Ecb ρcb + √ ) sin(πΩLca ) cos(πΩ)]− (145)
Scb Ecb ρcb Eca r
√ 1 ρca
jα[(Sca Eca ρca − √ ) sin(πΩ) sin(πΩLca )]}/η
Sca Eca ρca Eca
where
√
r
1 ρca
η = α[(−Sab Eab ρab − √ ) sin(πΩ) cos(πΩLca )+
Sab Eab ρab E ca
√
r (146)
1 ρca
(−Scb Ecb ρcb − √ ) cos(πΩ) sin(πΩLca )] + j[−αb(Ω) − 2β ]
Scb Ecb ρcb Eca
and
√
r
ρca 1
b(Ω) = −2 cos(πΩ) cos(πΩLca ) + (Sca Eca ρca + √ )×
Eca Sca Eca ρca
r (147)
ρca
× sin(πΩ) sin(πΩLca )
Eca
where
Lc
Lca = (148)
La
Sa Sc Sc
Sab = Scb = Sca = (149)
Sb Sb Sa
Ea Ec Ec
Eab = Ecb = Eca = (150)
Eb Eb Ea
ρa ρc ρc
ρab = ρcb = ρca = (151)
ρb ρb ρa
are ratios of geometric and physical parameters.
92
This appendix describes the mathematical development to obtain the equation (98). Based
on the section 3.4 the relation between the left and right hand sides structure is given by
( ) ( )
uR uL
= T̃cell (152)
FR FL
The periodic condition ( Eq. 1) for the displacements and forces in a section implies to the
following equation (BRILLOUIN, 1953)
( ) ( )
uR uL
=Λ (153)
FR FL
From equations (152) and (153) the following eigenvalue problem is defined
( )
uL
(T̃cell − ΛI) =0 (154)
FL
where T̃cell is given by equation (115), I is the identity matrix and Λ are the eigenvalues which
are associated to the frequency. The nontrivial solution fo this equation requires to solve the
determinant function det(T̃cell − ΛI) = 0.
" #
T̃11 − Λ T̃12
det(T̃cell − ΛI) = det =0 (155)
T̃21 T̃22 − Λ
or
Substituing the equations (116) to (119) from appendix A into equation (156), is obtained
93
Λ2 + b(ω)Λ + 1 = 0 (157)
where
Za Zc
b(ω) = −2 cos(kc Lc ) cos(ka La ) + ( + ) sin(kc Lc ) sin(ka La ) (158)
Zc Za
The roots of the equation (157) are the eigenvalues of the matrix Tcell .
ωLa
Considering that Ω = is non-dimensional frequency, it is obtained the equation (158)
πva
in non-dimensional form
1
b(Ω) = −2 cos(πΩ) cos(πτΩ) + (κca + ) cos(πΩ) sin(πτΩ) (159)
κca
where
Zc
κca = (160)
Za
are impedance parameters and
Lc va
τ= (161)
La vc
is ratio of propagation times parameter. Note that the equation (159) is a terms of the equa-
tions proposed to relation of transmission and reflection shown in appendix A. In addition, this
equation is rewritten in terms of physical and geometric parameters of the structure, i.e.,
√
r
ρca 1
b(Ω) = −2 cos(πΩ) cos(πΩLca ) + (Sca Eca ρca + √ )×
Eca Sca Eca ρca
r (162)
ρca
× sin(πΩ) sin(πΩLca )
Eca
where
Sc Lc Ec ρc
Sca = Lca = Eca = ρca = (163)
Sa La Ea ρa