Ultrasonic Wave Propagation in Solid Media:

Applications to Stress Monitoring, Imaging

And Material Characterization

Annamaria Pau
Department Of Structural and Geotechnical Engineering
Sapienza University of Rome

PhD in structural and geotechnical engineering

July 19-23-2021
Department Of Structural and Geotechnical Engineering
Sapienza University of Rome

JOHN BIRIR
[Company name]
[Company name] [Company address]
[Company address]
Ultrasonic wave propagation in solids – Lesson 1

Program of The course

Day 1

(1) Dispersion principles explained by means of waves in strings ;( 2) waves in

unbounded isotropic and anisotropic media (bulk waves): the acoustic tensor.
(3.1)Reflection, Transmission and (3.2) refraction of waves from interfaces. (4)
And finally see how Guided waves propagate in a plates these are called guided
waves and they are mainly two types of them which are shear weights and
Rayleigh-lamb Waves

Day 2

We will see some more numerical aspects concerning the propagation of waves in
solids with the deformable cross-section and these are typical problems that take
advantage of the semi-analytical formulation and we will also see some console
examples. We will also see the relationship between Propagating waves and global
modes of vibration. Introduce you to the theory of acoustoelasticity which
describes the propagation of waves in a pre-stressed medium and show how the
wave propagation velocity is altered by the presence of an initial pre-stress and
restrain according to the different approximations of this theory. It is possible to
model the change of the wave propagation velocity as well as the occurrence of
some nonlinear effects.

Day 3

Applications:

Some applications are

-Material characterization such as the determination of mechanical properties such
as stiffness, damping, smeared damage and material symmetries and the
characterization of the stress states like the tension in tendons or in restressing
cables

- Present image reconstruction of defects or inclusion. This is the technique which

exploits the propagation of waves to reconstruct the image by means of either array
of transducers placed on one side of some object or by using tomographic image
reconstruction.
Some References

The basics

J.L. Rose, Ultrasonic Waves in Solid media, Cambridge University Press, 1999

K.F Graff, wave motion in elastic solids, Dover Publications, New York, 1975

B.A Auld, Acoustic fields and waves in solids, Vol I, II, Krieger publishing
company, Malabar, Florida, 1973

For the Acoustic tensor

M.E Gurtin, An Introduction to continuum mechanics, academic Press, Elsevier,

2003

Spectral elements

J.F. Doyle, Wave Propagation in structures An FFT-based spectral analysis

technology, Springer verlag, New York. 1989

Nonlinearities and effects of stress

F.D. Murhaghan, Finite deformation of an elastic solid, John Wiley and sons, New
York, 1951

Imaging

A.Kak, M. Slaney, Principles of computerized tomographic imaging, IEEE press,

USA, 1987.
Framework

Being able to handle wave propagation is essential for different reason. First of it is
essential to understand some non-destructive evaluation techniques which refer to
both to the evaluation of mechanical properties and to image reconstruction.

Moreover the study of wave propagation can also improve the understanding of
many problems in vibration analysis. Infactr the propagation of waves can handle
both transient and steady state vibration phenomena which arise when waves
encounter many boundaries and reflect giving rise to a steady state vibration.

On the Contrary, Vibration solutions or vibration mode are able to handle weight
propagation with difficulty infact a very high number of vibration modes is
necessary to describe a phenomenon and are included in the solution.

We begin our study of wave propagation with the description of the nature of
dispersive and non-dispersive waves. These are encountered whenever wave
propagation velocity becomes a function of frequency.

In general in non-destructive techniques it is preferred to use the propagation of

waves in there if this exist. Non-dispersive range of frequency however that is s not
always possible and a certain amount of dispersion is often unavailable.

1. Dispersion Principles waves in taught string

Dispersion principles

Waves in taut string- governing equation

Let us consider an element of infinitesimal length which is part of an infinite string

subjected to a tension F
Let us assume that the horizontal displacements are negligible with respect to the
vertical deflections, which are assumed to be small.

∂y
ds = dx =sin ϑ = ϑ ϑ = ∂ x

Following Newton’s second law, the equilibrium equation in the absence of

external forces

(P(x, t) =0) is:

2
∂ v
F¿ ) = P ( 2
∂t

This is a second order hyperbolic partial differential equation that we better write
as

ϋ
V’= C2
0

Where co = √ F/p denotes the velocity of wave propagation

Dispersion Principles

Waves in Taut string – solution

The solution to the homogenous equation of motion can be written in exponential

form as the superposition of two waves harmonic in time, travelling in space as
shown below.
Dispersion Principles

Waves in taut string – Solution v(x, t) = Aеi (kx+ꞷx) + Bе i(-kx+ɷx)

In Principle, These waves can be of any frequency and any wave number k (The
inverse of the Wavelength), which are tied by the characteristics equation:

Ꞷ = C0K

The Characteristic equation is obtained when one requires that the wave solution
satisfies the homogenous equation of motion:

ϋ
Ѷ= C2
0

ꞷ
2
(kx+ꞷx)
2
AK е i = 2 Ae i (kx+ꞷx)
C0
Dispersion Principles

Waves in a taut string Solution v(x, t) = Aеi (kx+ꞷx) + Bе i(-kx+ɷx)

The quantity φ = -kx + ꞷt is called Phase angle

The propagation velocity of the constant phase angle is called phase velocity:

-kx0+ ꞷt0 = -kx1+ ꞷt1 K (x1- x0) =ꞷ (t1- t0)

√
( x 1−x 0) ꞷ
C0= (t 1−t 0) = K C0 = F = Const
P

In the string, the phase velocity equals the wave velocity, moreover, it is a
constant: such kind of waves in terms of wave are called nondispersive.

The properties of waves in terms of wave and phase velocity are summarized in the
dispersion diagrams.

Dispersion Principles

Waves in a taut string – dispersion diagrams

Dispersion diagrams represent wavenumber as a function of frequency or phase

velocity as a function of frequency.
Dispersion Principles

Waves in a taught string – Solution v(x, t) = Aеi (kx+ꞷx) + Bе i(-kx+ɷx)

Nondispersive wave means that a given disturbance propagating in space and time
in the taught spring maintains its shape unaltered.

Dispersion Principles

Waves in taut string – solution v(x, t) = Aеi (kx+ꞷx) + Bе i(-kx+ɷx)

When one sets up boundary conditions, waves are reflected at the boundary so that
the waves propagating rightward and leftward interact with each other giving rise
to standing waves. At the same time, the frequency of the wave is not
undetermined any more but is given by the infinite roots of the characteristic
equation of the related Eigen problem, which derives from the enforcing of the
boundary conditions.

These standing waves are the natural modes of vibration, with which we are not
dealing with in this course.

Dispersion Principles

Waves in a taut string on an elastic base – governing equation and solution

Let us consider now a different problem which may involve distortion of the
propagating disturbance. This is the case of a taut spring on an elastic base.
The one dimensional wave equation has now an extra term which is caused by the
elastic force of the base

−K ϋ
Ѷ= v= 2
F C0

Now, let us examine possible conditions for the propagation of harmonic waves.
Assume that

V(x, t) = Aеi (kx+ꞷx)

By substituting this tentative solution in the equation of motion we obtain

ꞷ
2
K
(-k2 – F + 2 ) Aеi (kx+ꞷx) = 0
C0

And we have nontrivial solutions when the characteristic equation is satisfied

K ꞷ
2
2
(-k – F + 2 ) = 0
C0

Dispersion Principles

Waves in a taut string on an elastic base – dispersion V(x, t) = Aеi (kx+ꞷx)

The solutions of the characteristic equation in different ways but the one i prefer is
wavenumber as a function of frequency because frequency is always a real
number:

k =±
√ ꞷ2−K
2
C0 F

This result is represented in a plot which contains real and imaginary parts of the
wave number

This equation has two roots for each frequency

ꞷ K
2

2
> If k is real and we have waves propagating leftward and rightward
C0 F

ꞷ K
2

2
> Otherwise, we have nonpropagating waves (also called evanescent
C0 F

2 K
Cutoff frequency ꞷ =C 0 F
2
Dispersion Principles

Waves in a taut string k, the phase velocity can be determined:

Ꞷ C

√K
Cφ = K = 1+ 2
0

Fk

In this case, the phase velocity on the frequency.

This has consequences on the

maintaining of the shape of disturbance
as it travels along the string

Dispersion principles

Waves in a taut string on an elastic base – Dispersion,

For instance, this is what occurs when two harmonic waves travel along a
dispersive waveguide
Dispersion Principles

Group velocity

Another important concept that has to be introduced in the group velocity. A wave
packet containing different frequencies travels at a speed that is not the phase
velocity, but the group velocity.

This phenomenon can be explained by considering two waves with slightly

different frequency and wavelength propagating in a medium

V(x, t) = cos (k 1x - Ꞷ 1t) + cos (k 2x - Ꞷ 2t)

After some trigonometric manipulations, this can be written as:

1 1
V(x, t) = 2 cos ( 2 ∆ k x− 2 ∆ ꞷt) * cos (k x - ꞷt)

This is the product of a low frequency term and high frequency term with k and Ꞷ
average wavenumber and frequency:

k 1 +k 2 ꞷ1 +ꞷ2
K= Ꞷ=
2 2
Dispersion Principles

1 1
Group Velocity V(x, t) = 2 cos ( 2 ∆ k x− 2 ∆ ꞷt) * cos (k x - ꞷt)

∆ꞷ dꞷ
Cg = ∆ k or, in the limit, Cg = dk is the group velocity

Calculating the group velocity makes sense only for dispersive waves. In
nondispersive waves, the group velocity equals the phase velocity.

This animation shows what occurs to two harmonic waves travelling along a
dispersive medium when they are alone or together. The blue and red dots move
with the phase velocity, while the black dot moves with the group velocity. The
individual harmonics travel with different phase velocity, but their envelope
(yellow and green) travels with group velocity

Dispersion principles

Summary of dispersive/ nondispersive Waves

NONDISPERSIVE WAVES

ONE DIMENSIONAL SSYTEMS

Rigid cross-section

Transverse waves in taught spring

Axial waves in rod

TWO DIMENSIONAL SYSTEMS

Shear wave mode SH0 in a plate

Low frequency symmetric and

Nonsymmetrical waves in a plate (S0 and A0)

THREE- DIMENSIONAL SYSTEMS

Bulk waves such as be pressure (P) or shear (S) waves

DISPERSIVE WAVES

ONE – DIMENSIONAL SYSTEMS

Rigid cross- section

Flexural waves in a beam

ONE DIMENSIONAL WAVEGUIDES WITH DEFORMABLE CROSS-

SECTION:

all waves(guided waves)

TWO DIMENSIONAL SYSTEMS

High frequency shear waves in a plate (beside the SH0 mode)

High frequency axial waves (Symmetric S-waves)\All the flexural waves

(Asymmetrical A-waves)

2. Waves in Unbounded Solids

Homogenous and isotropic/ anisotropic material

Waves in Plates

Lamb wave A0 in an aluminum plate

If we take the mode A0, the transverse displacement are dormant and the wave
mode is a flexural- like mode. As the frequencies increases, the deformation along
the height of the of the plates becomes increasingly important and the displacement
field differ from that of the bending of an Euler beam with linear μ1 and constant μ3

Unbounded solid media

Homogenous and isotropic solids

Waves that propagate in unbounded media are called bulk waves

A structure can be considered an unbounded medium for all practical for purposes
if the wavelength of the propagating wave is small with respect to the size of the
structure.

The equations of motion for a homogenous isotropic linear elastic solid for
infinitesimal displacements and the absence of body forces, maybe written as:

1
div σ =ρ ü with = σ = 2 μE + μr EI E = 2 (∇ u +∇ uT )

some identities of tensor analysis

div ((∇ u +∇ uT ) = ∆ u+ ∇÷u div (∇ u ) = ∆ u

trE = divu div(divu) = ∇÷u

Navier differential equations of motion

μ∆ u + (ԅ + μ) ∇÷u=ρ ü

Unbounded solid media

Homogenous and isotropic solids

We look for solutions that are harmonic progressive waves, that is

u (x, t) = aei (k*x-ꞷx)

u is longitudinal if a and k are parallel, transverse if a and k are orthogonal.

We want to determine the conditions which have to be satisfied for a progressive

wave to solve the equations of motion:

μ ∆ u+ ( ԅ + μ ) ∇÷u= ρii
By the chain rule, we obtain the following identities:

∆ u=∇÷u=∇ i a . k e = - (a*k) k e i (k*x-ꞷx) = -k2 a e i (k*x-ꞷx)

i ( k∗ x−ꞷ x )

i (k∗x−ꞷ x) i(k∗x−ꞷ x)
¿ ∇ u=i÷a k e =−( a ⨂ k ) . k e = -(k⨂k)ae i(k∗x−ꞷ x)

ü = -ꞷ2 ae i(k∗x−ꞷ x)
u ⨂ v ij=ui v j
u.v ∑ ui v i
i

This leads to μ k 2 a+ ( ԅ + μ ) ( k ⨂ k ) a=ρ ꞷ2 a

Unbounded solid media

Homogeneous and isotropic solids

The equation of motion for travelling waves

μ k 2 a+ ( ԅ + μ ) ( k ⨂ k ) a=ρ ꞷ2 a

Can be rewritten in a simpler form

2
μ k Aa=ꞷ a
2
or Aa=c2a

With k=kn k = √ k 2x + k 2y +k 2z |n|=1

1
And where A = ρ ( μ I + (ԅ + μ ) n ⨂ n) is the acoustic tensor.

The equation of motion has nontrivial solutions only if c2 and a are respectively
eigenvalues and eigenvectors of the acoustic tensor.

Unbounded solid media

Homogenous and isotropic solids

A simple manipulation shows that:

ԅ+2 μ μ
A=( ρ
) n⨂n + ρ
(I- n⨂n)

Which is a spectral decomposition of the acoustic tensor. The spectral theorem

enables the conclusion that a harmonic progressive wave with velocity c is a
solution of the equation of motion if:

√
c= ԅ +2 μ
ρ
and the wave is longitudinal or if: c= μ
√ ρ
and the waves is

traverse

a. n=|a| a. n =0

Displacements occur in the same Displacements occur are orthogonal

direction as the direction of wave to the direction of the wave
propagation (one eigenvector) propagation (two eigenvector)

Unbounded solid media

Homogenous and isotropic solids

A variety of terminology exists for these two wave types

Longitudinal waves are also called dilatational, irrotational, pressure and primary
(P) waves.

√
c= ԅ +2 μ
ρ

Transverse waves are also called shear, equi-voluminal, distorsional and secondary
(S) waves
 √
c= μ
ρ

In a homogenous and isotropic solid, the wave velocity does not depend on the
direction of propagation.

Unbounded Solid media

Anisotropic solids

An example of waves propagating in infinite elastic anisotropic solids in the case

of waves propagating in pure crystals, where homogeneity and pure anisotropy are
assumed.

We will consider indicial notation. Newton’s law

∂ σ ik
=ρii ,
∂ xk

Constitutive equation strain- displacement relation

1
σ ik= C iklm ε lm ε lm= ¿
2

Navier equations:
2
∂ um
ρ iii= C iklm
∂ xk ∂ xl

Unbounded solid media

Anisotropic solids
Let us assume plane harmonic travelling waves and see if a solution of this form

Ui= Ae i (kjxj-ꞷx) with Ai = Aα i α idirection cosine of particle displacement

Is possible for Navier equations:

2
∂ um
ρ iii= C iklm
∂ xk ∂ xl

ρ ꞷ u1 =C iklm k k um
2
This gives an eigenvalue problem:

Which can be expressed in its most popular form, that of the Christoffel equation

( ρ Ꞷ 2 δ ℑ- C iklm k k k l ) um = 0 ui = δ ℑ u m

Unbounded solid media

Anisotropic solids

Consider that the wavenumber is a vector whose components

Kx = knx ky=kny kz=knz

Can be expressed as a function of its modulus k and direction cosines of the normal
to the wavefront ( nx, ny, nz), therefore the christoffel equation can be expressed as
a function of k and ꞷ only, considering that

C iklm k k k l = C iklm nk nl k 2= ┌ ℑ k 2

Therefore: ( ρ Ꞷ 2 δ ℑ−┌ ℑ k 2 ¿ um=0

The components ┌ ℑ coincide with those of the acoustic tensor. For a nontrivial
solution, the determinant of the coefficient matrix must be null.

Characteristic equation | ρ Ꞷ 2 δ ℑ−┌ ℑ k 2 ¿∨¿ 0

For a given direction of the wavefront n, the characteristic equation provide three
roots k 2

Unbounded solid media

Characteristic equation | ρ Ꞷ 2 δ ℑ−┌ ℑ k 2 ¿∨¿ 0

A wave mode shape components in terms of displacements can be determined by

calculating the eigenvectors, which are the solutions of the following system of
algebraic homogeneous equations:

{ {
α 1} 0 }
¿ α 2} = 0 }
α3 } 0 }

In anisotropic solids there may be directions of propagation such that wave modes
are purely parallel (P waves) or orthogonal (S waves) to the normal of the
wavefront, similar to what happens with waves in isotropic materials.

Unbounded solid media

Anisotropic solids

The results presented up to now can be expressed by using slowness surfaces. In

the 3D space, slowness surfaces are s polar plot where, for each angle (direction of
the wavefront) a ray length k/ω is reported.
Three slowness surfaces exist for a given material, one tied to one another of the
Eigen values.

The slowness surface of an isotropic material is a sphere.

The following figures are examples of slowness surfaces for an intermetallic alloy
NiAl

The picture taken from anisotropy and phonon focusing in NiAl: Atomic. Fu, Hou,
Fu, Mab.intermetallics, 42, 2013.

3.1 Reflection and Refraction

Normal and Oblique incidence

Reflection and Refraction corresponds to the normal and to the oblique incidence
of a wave front against the surface.

Wave reflection

Normal beam incidence

Wave reflection considerations are fundamental to study the propagation of waves

in solids.
Let us first consider incidence of a P wave with a plane wavefront. This occurs
when a plane wave of amplitude I (black) encounters the interface between two
media with different mechanical properties. Different mechanical properties result
in different wave velocities in the two media.

uI (x, t) = Ie i ¿¿

ꞷ ꞷ
uR(x, t) = Re i ¿¿ k1 = c k2 = c uT= Te i(k x +ꞷ x)
2

L1 L2

It occurs that some energy is reflected backward (red), and some energy is
transmitted (blue). The amplitude of the reflected and transmitted wave is a
function of the properties of media.

The amount of reflected and transmitted energy can be determined by requiring

that continuity equations are satisfied at the interface.

Wave reflection

Normal beam incidence

In a P- wave with a plane wavefront, the only displacement component different

from zeros the longitudinal one, parallel to the direction of propagation. The only
stress component different from zero is the normal component

σ xx ( x , t )=( ԅ+2 μ ) ∂ u∨∂ x=ik ( ԅ+2 μ ) u(x ,t )

If we set up continuity
Ie i ¿¿ - Re i ¿¿ = Te i(k x +ꞷ x)
2

And equilibrium conditions

ik1 (ԅ ¿ ¿ 1+ 2 μ1 )¿ Ie i ¿¿ + ik1 (ԅ ¿ ¿ 1+ 2 μ1 )R e i ¿¿ ¿ = ik2 (ԅ ¿ ¿ 2+ 2 μ2 )T e i¿ ¿ ¿

Reminding that c= ԅ +2 μ
√ ρ

We obtain two algebraic equations I+R = T

ρ1 c 1 L ( I −R ) =ρ2 c2 L T

in these equations I is given is known because it is the amplitude of the incident

wave and r and t which are the amplitude of the reflected and transmuted waves
are to be determined by the solution of the system of equations which is a very
simple system of equation to solve we can derive the amplitudes of a reflected
and transmitted waves as a function of properties of the two materials.

Wave Reflection

Normal beam incidence

Whose solution gives the reflected and transmitted wave amplitudes.

R ρ1 c 1 L −2 ρ2 c2 L T 2 ρ1 c1 L
= =
I ρ1 c 1 L + ρ2 c2 L I ρ1 c1 L + ρ1 C2 L

The amount of reflection, and transmission depends on the ratio between acoustic
impedances ρc , which, in turn, depend on the material properties

r= ρ2 c 2/ ¿ ρ c
1 1L ¿

R 1−r T 2
I
= 1+r I
= 1+ r

If the second material has the same properties of the first one (r=1), no reflection
occurs and all of the wave is transmitted.
If the first material is stiffer and /or heavier than the first (r<1), the wave is partly
reflected and partly transmitted. Because it has got higher weight propagation
velocity or heavier .

r=1 means that the two materials have the same properties

The wave is partly transmitted for amounts which depend on the ratio between
impendences.

If the second material is stiffer and or heavier than the second (r>1), meaning that
the row is greater than 1 in the limit the whole wave will be reflected and there is
no transmission because we see that the amplitude of the transmitted wave tends to
zero for r going to infinity.
Wave reflection

Normal Beam incidence.

It may seem there is a mistake that is T > 1

Is actually correct: at the interface between two materials with mechanical

properties that differ mum from each other concentrations of stress may occur.
This may cause debondings between layers.
The calculations fit perfectly if we check conservation of energy.

Conservation of energy is calculated if we calculate the work done by internal

forces per unit time and unit area normal to the directions of propagation

σ u +σ u = σ u
R R T T 1 1

We actually obtain 1:

( 1−r
1+r )
2
+r (
2
1+ r
2
) =1
Wave reflection

Double change of material

The procedure described can be implemented for a P-wave which encounters a

double change of material properties.

This expression is a representation of this problem which is made in a very simple

form. It can be more complete way just to see that it has are more than two
constants to determine.

u1 (x, t) = Ie i(k x+ꞷ x) + R1e i(−k

I I x+ꞷ x)
x=o x=d u3(x, t) = T3e i(k (x−d)+ꞷ x)
3

ꞷ ꞷ ꞷ
k1 = c k2 = c k3 = c
L1 L2 L3

1 2 3

u2 (x, t) = T2e i(k x +ꞷ x) +R2e i(−k (x−d)+ꞷ x)

2 2
This time, equilibrium and continuity conditions have to be imposed at x=0 and
x=d. The number of unknowns to be determined is greater:

R1, R2, T2, T3

When you have two changes of material properties you have to determine more
constants. This simplified version can be expressed as a function of two constants
and determines the two constants. Similarly the same as what was done before by
imposing continuity and equilibrium equations at two different objects x=0 in the
place where the first change of material properties and second is x=d represents
just the span the extension of the second material.

Wave reflection

Double change of material

The time- history in more complex cases will actually be a summation of many
delayed reflected and transmitted waves coming from the first and second
discontinuity all these extending to infinity or the summation mean in the absence
of damping.
∝
2
R(x, t) = ∑ R Aj e iꞷ (t+ ∁ o ([ x d−x +d ( i−1 ) ])
j=1

∝
1
T(x, t) = ∑ T Bj e iꞷ (t+ ∁ o ([ x+2 d ( j−1 ) ])
j=1
 In practice every reflection reduces the amplitude of the wave. It will be sufficient
to extend the summation of the first to the for the reflected wave or to just one
wave for the transmitted waves (for R) or 1 wave (for T) and they can be
calculated like series of products which accounts for the different reflection and
transmission that occurs are the two different discontinuity
2 j −3 2 ( j−1)
J=1 RA1 = RA TB1 = TATB J>1 RA1 = TA R B TB TBi = TA R B TB

r−1
RB =
1+r
2
TA =
1+ r
1−r
RA =
1+r
2r
TB =
1+ r

Wave reflection

Normal beam incidence

Similar reflection and transmission is observed when other waves and kinds of
discontinuities are encountered very similar properties are exhibited by;
  Axial waves in bars
 Shear waves in plates
 Discontinuities like sharp changes of cross section or height. In these cases, r
is the ratio between the cross section area in the damaged or reduced part
and the undamaged cross- section area

r = A2 / A1

The response can be exploited to identify the properties of a discontinuity based on

the response to a known incident wave.

Some numerical and experimental tests for simple case of a PVC bar

The reflected and transmitted response can be exploited also to identify the
properties of the discontinuity which cannot only be some damage but may have
the need inquire for instance geometrical properties of something which is
embedded in some material and which one can not access, This is typical of power
plants for instance if we did some experiments about that in our laboratories on a
PVC bar as depicted.

Wave reflection

Double change of material/ sharp change of cross section

If a bar is excited with an impulse force at the end, the response can be described
using D’ element solution. Amplitudes of transmitted and reflected wave are the
same as those obtained for harmonic waves.

Wave reflection

Double sharp change of cross-section

To understand how the response the reflected response changes when the extension
of the cross section changes can be done. the extension of the discontinuity
influences the shape of the reflected wave. Depending on the minimum wavelength
contained in the excitation constructive or destructive interference can be observed.
The lambda is the wavelength of the incident wave and when lambda and r a which
are extension of the change in cross section (the maximum length, wavelength
involved) in the phenomenon is the same order of magnitude or smaller than the
extension of the defect or of the part with reduced cross section, it is possible to see
very clearly to distinguish clearly the two reflections but when the wavelength
starts to increase, the two responses get closer and in the end they have a
destructive interference so that you may no longer identify and understand
correctly what is the distance between the two changes of cross-section. This is a
phenomenon which occurs in all wave phenomena meaning that when one want to
observe something one has to use a wavelength which is the same size as of the
object or of the thing you want to observe. A phenomenon occurring in optical
waves.
Wave Reflection

Double sharp change of cross section/ Experimental results

Change in reflected and transmitted response following the interaction of the

incident wave with changes of cross section of different length and depth.

When the wavelength is smaller than the extension of the changing cross-section
one can see clearly the two coming waves from the two changes in the cross
section while the wavelength increases on start to see a destructive intereference.

Wave reflection

Double sharp change of cross- section/ experimental set up.

This is a sketch of the experimental set us, performed on a PVC bar with different
depths and extension of the double sharp change of the cross- section.
Table 1: Geometrical and mechanical properties of the the PVC bar

E[ MPα ρ¿ L[cm] D[sm] co¿

4150 1400 200 2 1721.71

Table 2: Geometrical and mechanical properties of the PVC bar

D1 D3 D4
Dd [cm] 1.7 1.4 1.4
r 0.72 0.50 0.50
α [cm] 10 10 15

Wave reflection

Double sharp change of cross- section/ results

The parameters to be identified are determined by minimizing two objective

functions that measure the difference between the analytical model and the
experimental response.
The parameters of discontinuity (diameter, position and extension) are identified
with low errors, in general smaller than those obtained using a procedure based on
the frequency variation.

n n
G 1 ( r )=∑ (T ¿ ¿ i ( r )−T i)2G 2( α , x d)=∑ (R i (¿ α , x d )−Ri )2 ¿ ¿
i=1 i=1

Table 3: identified damage characteristics

Dd Ddi %err xd x di %err α αi

%err
D1 1.7 1.6 5.9 35 36.8 5.2 10
10.7 6.7
D2 1.4 1.2 14.3 35 36.7 9.9 10
13.7 9.9
D3 1.4 1.2 14.3 35 37.4 6.9 15
14.5 2.8

Table 6: identified damage characteristics based on the frequency variation

Dd Ddi %err xd x di %err α αi

%err
D1 1.7 1.9 11.8 115 114 0.9 10
18 80.0
D2 1.4 1.7 17.7 115 106 7.8 10
29 190.0
D4 1.4 1.6 14.3 115 113 1.7 15
 29 93.3

These methods of identification based on the comparison on the minimization of

an objective function which is built by minimizing the difference between the
analytical model and the experimental response. There is comparison to methods in
table 3 to methods based on the frequency variation. In all cases all the parameters
of the structure are identified with smaller errors than the case with frequency
variations

3.2 Wave Refraction

Wave reflection

Oblique incidence – refraction

Refraction occurs when an inclined wave front encounters an interface between

two materials on some inclined angle. When we have the wave front which
encounters discontinuity with an inclined angle. First of all, refraction occurs in
much the same way as it occurs for optical light waves. The refracted angle is
calculated by Snell’s law:

Segment AB: inclined wave front

 Basically refraction means that the angle of the wave front is modified by the
interaction between the wave and the interface

Wave reflection

Oblique incidence- refraction

We can imagine that from each point of segment AB spherical waves are generated
(according to Huyghens). The difference in wave propagation velocity between the
two media, along with the difference in arrival times of the points of wave front,
modify the direction of the normal to the wave front.

Segment AB: inclined wave front

AB AD
sin θ1= sin θ2=
AC AC

C 2 sin θ1 = C 1 sin θ2
AD AB
AC = =
sin θ2 sin θ1

t AD
C2 t C1

= AB
sin θ2 sin θ1

tAD = tAB
The spherical AB is covered by the propagation of the way wave in the same time
as the segment Bc. These two times are equal but the segment is different with the
wave propagation velocity therefore by equating these two times and also by
expressing the length of AD and B as the function of the two angles of the length
AC which is basically a geometrical procedure one can derive snells law

Wave Reflection

Oblique incidence – refraction- critical angles

Snell’s law

C 2 sin θ1 = C 1 sin θ2

Snells law is very important because it enables us to conclude that there are some
critical angles of instance.

We have two critical angles of incidence:

−1 C1
θ σ 1=sin sin θ2=1 , θ2=90
C2 L
0
1). with

In this case, all of the longitudinal energy is converted to an interface wave, only
shear waves remain in the second material.
 −1 C1
2). θ σ 2=sin with sin θ2=1 , θ2=900
C2 S

In this case no significant energy is propagated through the second material: all of
the energy is either reflected or transmitted into interface waves.

θ σ 1 >θσ 2

This is because C2s<C2L

Generally the shear weight velocity is smaller than the longitudinal weight velocity
which causes the critical angle numbe1 being greater the critical angle to.

Wave reflection

Oblique incidence – mode conversion

Mode conversion also occurs: the incident energy is portioned into one part
associated with a shear wave and another part associated with P waves. This is
because, like a force, the incident wave has one normal and one tangential
component with respect to interface.
4. Waves in Plates

Guided waves Vs bulk waves

Presented here are the governing equations of elastodynamics in traction-free

homogenous and isotropic plates.

The procedures we will introduce are similar to those used in a countless number
of guided wave problems, which includes bars, pipes, multiple layered plates etc.

The difference between the bulk waves and guided waves is that bulk waves
describe the propagation in the bulk of the material away from boundaries while
guided waves arise in a medium which has boundaries that enforce a specific
propagation regime in between these boundaries and the solution is represented by
a function which is the product of a traveling part and of the stationary part. The
stationary part describe the motion on the cross section and is basically what is
called wave mode shape.

Both bulk waves and Guided waves are governed by the same field equations,
however for bulk waves there are no boundary conditions that have to be satisfied
by the assumed solution while in guided waves problems the governing equations
include also some boundary conditions

Waves in plates

Guided waves

This the equation of motion of a homogenous and isotropic way plate. The field
equations are
 μ ∆ u+ ( ԅ + μ ) ∇÷u= ρii

Which have to be satisfied together with the boundary conditions of free stress on
the upper and lower surfaces. This is the same equation for plates. The difference
in plates is that boundary conditions have to be satisfied and these have to be
satisfied on the part when x sub3 equals h on the upper surface and h sub 3 equals
to minutes h

σ n 3=( 2 μ E+ ԅtr EI ) n 3=0 on x 3=± h

That is {σ 31, σ 32, σ 33} = 0 on x 3=± h

There are two main procedures for solving this problem one is the method of
Helmholtz potentials which is an elegant method but its applications limited to
isotropic plates and the other method is the partial wave techniques which is a
more general technique and applicable to more complex cases for instance plates
made of composite material . This is basically a scross section of a plate.
Waves in Plates

The partial wave technique

In the partial wave technique we search for solutions of the simple exponential-
type propagating along the plane x1, x2 and with a plane wave front contained in
the plane x2, x3
t (k 1 x 1+k 3 x3−ꞷ x) i (k 1(x 1+a x 3)−ꞷ x)
u=U ( x3 ) e ¿ U ( x3 ) e

Withα =¿ k3/ k1 = k3/k

If we substitute this equation in the field equations

μ ∆ u+ ( ԅ + μ ) ∇÷u= ρii

We obtain an eigenproblem whose eigenvalues are the possible ratios α , while the
related eigenvectors are the displacement components U1, U2, U3 that sustain such
types of waves

[ ][ ] [ ]
k 1 [−ԅ−( 2+α ) μ ] +ꞷ ρ 0−a(ԅ + μ) U 1
2 2 2
0
0−k 1 ( 1+ a ) +ꞷ ρ0
2 2 2
U2 = 0
U 0
−a ( ԅ+ μ ) 0 k [−μ−a ( ԅ+2 μ ) ] +ꞷ ρ
2 2 2
3
1
The wave solutions decouples the field equations. The first and third equations
refer to Lamb waves (u1, u2, u3 = 0), with in- plane displacements. The second
equation represents shear waves (u2, u1, = u3 = 0) with out-of- plane displacements.
When we have displacements when we have displacements along u1 and u3.

Waves in Plates

The partial wave Technique

The eigenvalue Problem:

[ ][ ] [ ]
k 1 [−ԅ−( 2+α ) μ ] +ꞷ ρ 0−a(ԅ + μ) U 1
2 2 2
0
2
( 2
)
0−k 1 1+ a +ꞷ ρ0
2
U 2
= 0
0
−a ( ԅ+ μ ) 0 k [−μ−a ( ԅ+2 μ ) ] +ꞷ ρ U 3
2 2 2
1

It has nontrivial solutions when its determinant is null, that is, when

−k 21 ( 1+α 2 ) μ+ ꞷ2 ρ=0

Which has two solutions representing shear waves:

a 1, 2=±
√
c2 ρ
μ
−1 ¿ ±
c2
2
cT √
−1

Or when (k 21 ( 1+α 2 ) μ +ꞷ2 ρ)(k 12 ( 1+α 2 ) (ԅ +2 μ )−ꞷ2 ρ)=0

Whose solutions are a 3 ,4 =±

√ c2
2
cT
−1 or a 5 ,6=±
√ c2 ρ
ԅ +2 μ √ c2
−1 ¿ ± 2 −1
cL
That the Lamb waves. This eigenvalue problem has 6 roots and must be solved
repeatedly for different values of c=Ꞷ /k 1 between frequency and wave number.

Waves in Plates

The partial wave technique

At this point, we have to require that the boundary conditions are satisfied on the
free surfaces

σ n 3=( 2 μ E+ ԅtr EI ) n 3=0 on x 3=± h

To do so we have first to express the possible wave solution and components of

stress on the free surface as a function of the previously determined eigenvalue and
eigenvectors for shear waves and for lump weights you see that everything is
expressed as a function of alpha in a summation where the indexes is q and it is
extended in this case for shear waves 1,2 the two alpha sub q
2
i [ k ( x +a x3 )−Ꞷ x ]
{u1 u2 u 3 }=∑ { 0 ,1 , 0 } U q e 1 1 q

q =3

2
{σ 13 σ 23 σ 33 }=∑ ik 1 { 0 , μ a q , 0 } U 2 q e
i [k 1 ( x 1+aq x 3)−Ꞷ x]

q=3

And for lamb waves from three to six which are the indexes referring to the
different routes.
6 6

{u1 u2 u 3 }∑ {1 , 0 , R (aq )} U 1 q e {σ 13 σ 23 σ 33 }=∑ ik 1 { D1 q , D 2q , D3 q } U 1 q e

i [ k1 ( x 1+ aq x 3)−Ꞷ x] i[ k 1( x 1+ aq x 3)−Ꞷ x]

q=3 q=3
With { D1 q , D2 q , D3 q }= { μ ( R ( a q ) +a q ) , 0 , ¿

ԅ+ ( 2+a2q ) μ−c 2 ρ
R ( a q )=
a q (ԅ+ μ)

Waves in Plates

The partial wave technique

Applying boundary conditions that is requiring that

2
{σ 13 σ 23 σ 33 }=∑ ik 1 {0 , μa , 0 } U 2 q e
i [k 1 ( x 1 ±aq h )−Ꞷx ]
={0 , 0 , 0 , }
q=1

For shear waves

And that
6
{σ 13 σ 23 σ 33 }=∑ ik 1 { D1 q , D2 q , D3 q } U 1 q e [
i k 1( x 1 ± aq h)−Ꞷ x ]

q=3

For lamb waves,

Gives two separate eigenvalue problems

[ ][ ] [ ]
ik 2 a1 h ik 1 a2 h
μa1 e μa2 e U 21 = 0
μa 1 e−ik 2 a h
1
μa2 e−ik a h
1
U 22
2
0

The first one is for shear waves and the second is for Lamb Waves
[ ][ ] [ ]
D13 E3 D14 E 4 D15 E 5 D16 E6 U 13 0
D33 E 3 D33 E 4 D35 E5 D35 E 6 U 14
=0
D13 E3 D14 E 4 D15 E 5 D16 E6 U 15 0
D13 E3 D34 E 4 D35 E 5 D36 E6 U 16 0

ik1 aq h −ik1 aq h
Eq =e Eq =Ẽ e

{ D1 q , D2 q , D3 q }= { μ ( R ( a q ) +a q ) , 0 , ¿

These Eigen value problems are associated with different characteristic equations
which in the case of shear waves have analytical solution while lamb waves have
to be solved numerically.

Waves in plates

The partial wave Technique

The solution to the two last eigenvalue problem finally enables the determination
of the wavenumbers as a solution of trigonometric equations:

nπ
Ω ( k 1) =sin ( 2 hk 1 a1 )=0 Which for shear waves, have an analytical solution k 1= 2 ha
1

4a a
And Ω ( k 1) =tan¿ ¿ 3 5 = 0 for lamb waves have to be solved numerically
¿¿

Ω ( k 1) =tan¿ ¿ ¿ ¿ = 0

The frequency is given by the product Ꞷ = c k1 where k1 is the solution of the

second eigenvalue problem and c is the value with which we have entered the
first eigen problem
The character equations Ω ( k 1) can be used to determine the group velocity
considering that

d Ꞷ ∂ Ω/∂ k
1
Vg = dk =
1 ∂ Ω/∂ Ꞷ

Waves in Plates

The partial wave technique

Is there some plane strain/stress state in the solutions we have previously looked
at?

For shear waves there is not.

Note that for shear waves, we don’t have neither the state of plane stress nor a state
of plane strain.
2
i [ k 1 ( x1 +aq x 3 )−Ꞷ x ] Direction of
{u1 u2 u 3 }=∑ { 0 ,1 , 0 } U 2 q e propagation x1
q =1

If you look at this picture, you see that x sub 1 is

The direction of propagation. And E12, E21 , E23, E33

Are the strings related to shear waves? This is not the state plane stay straight.

But for the Lamb waves yes, there is a state of Plane stress strain but not plane
stress
6
Direction of
{u1 u2 u 3 }∑ {1 , 0 , R (aq )} U 1 q e
i [ k1 ( x 1+ aq x 3)−Ꞷ x]
propagation x1
q=3

In this picture there is direction of propagation

And the strains are contained in the same

Plane of the cross section. Basically they are

Orthogonal to the plane of the plate

Waves in Plates

Shear waves: example for a steel plate.

These are very complicated diagrams which can be obtained by the dispersion
curves which can be obtained for shear waves.

Looking at the phase velocity there is the sh0 wave which non-dispersive, it has
got constant phase velocity. Constant group velocity and has also a mode shape
basically known as a wave mode shape in which all the points translate
orthogonally to the direction of the propagation of the same quantity.
Waves in Plates

Shear waves- SH0 mode

To better understand how these waves are made, this animation depicts the
propagation of an SH0 wave in the plate. You can see all the points translate the
same quantity orthogonal to the direction of propagation.

All the higher modes shapes which are called sh1, sh2, xh3 and so on have cut off
frequencies which coincides with the above frequency below which for instance
sh1 mode cannot see. It exists only locally but it does not propagate because the
wave number is just complex.

The old mode shapes are non-symmetric with respect to the plane of the plates
while even mode shapes are symmetric with to the plane of the plate and as you go
up to consider higher modes, these displacement fields are more intertwined.
This wave does not work with the laser.

Waves in Plates

Lamb waves: an Aluminum plate.

The picture below shows an example for an aluminum plate of the dispersion
curves for the lamb weights.

In Lamb Waves, the direction of wave propagation (x1) lays in the same plane as
the displacements u1 and u3)

In these case for lamb weights, there are those which are called a0 and s modes,

The a modes are kind of bending like modes while s modes are kind of longitudinal
waves modes. Also in these case all the modes have cut off frequencies meaning
that there are frequencies below which they do not exist.
They are in generally all dispersive although there are some frequency ranges in
which dispersion is limited for instance if take the part of A0 wave one can
approximately say that the phase velocity is nearly constant and the same happens
for group velocity. In case of shear waves, the wave mode remains an altered
when you change the frequency but it is not the same for S waves and A waves.

Waves in Plates

Lamb waves S0 in an aluminum Plate

The S0 mode is the simplest mode one can imagine meaning that when one is at
the part with the red dot means low frequency .Unlike shear waves, the mode
shapes of Lamb waves with frequency for instance at S0 mode , at low
frequencies , the longitudinal displacements are dominant.. As the frequencies
increases, the Poisson effect becomes increasingly important.

From the above diagram the displacements are mainly along direction x sub1
while displacement along the direction a3 exists and are limited, this is called
Poisson effect. When you go up with the frequency the wave mode shape changes
and the Poisson effect become increasingly important.

Waves in Plates

Lamb wave A0 is an aluminum plate.

In an analogous way the A0 mode means a symmetric mode are basically bending
modes and the displacements are mainly constant along the cross section meaning
that the cross section moves orthogonal to the direction of propagation.

If we take the A0 mode, the transverse displacements are dormant and the wave
mode is a flexural –like mode. As the frequency increases, the deformation along
the height of the plate becomes increasingly important and the displacement field
differ from that of the bending on an Euler beam with linear u1 and constant u3.

Nevertheless, there is an interesting u displacement along u1 which is again

bending and is totally analogous to the bending of an Euler Bernoulli beam with
the difference that in the Euler Bernoulli beam, this u1 displacement field is a line .
When you go higher frequencies like in the case of the blue dot, the u1
displacement becomes more and more complex which means that it resembles less
to align which is in the case with Euler Bernoulli.

THANK YOU!