I
• О
Ultrasonic Flaw
Detection for
Technicians
3rd Edition
J.C.Drury
Copyright© J. C. Drury / Silverwing Limited
All rights reserved. No part of this publication may
be reproduced, stored in a retrieval system,
or transmitted, in any form or by any means,
electronic, mechanical, photocopying, recording
or otherwise, without the prior permission
of Silverwing Limited
Designed and printed by
Imex Group Limited
Darcy Business Park, Llandarcy
Neath, SA10 6EJ
Cover Design: Lara Griffiths
CONTENTS
Chapter 1	History of Ultrasonic Flaw Detection	5
Chapter 2	Basic Principles of Sound	9
Chapter 3	Properties of Sound	18
Chapter 4	Transducers for Generating Sound Waves	35
Chapter 5	Proble Construction	48
Chapter 6	The Pulse-Echo Flaw Detector	55
Chapter 7	The Ultrasonic Beam	64
Chapter 8	Calibration and Reference Standards	74
Chapter 9	Compression Wave Techniques	79
Chapter 10	Shear Wave Techniques	91
Chapter 11	Surface Wave Techniques	108
Chapter 12	Immersion Techniques	111
Chapter 13	The Examination of Steel Castings	118
Chapter 14	The Examination of Forgings	133
Chapter 15	The Examination of Welds	145
Chapter 16	Defect Sizing and Evaluation Techniques	192
Chapter 17	Assessing the Performance of Equipment	242
Chapter 18	Report Writing	249
FOREWORD
In the twenty-five years since the first edition of ‘Ultrasonic Flaw Detection
for Technicians’ was published, there have been a number of advances
in transducer technology and the flaw detection instruments. The gradual
acceptance by the industry that the sizing of weld defects by intensity drop
was not as accurate as had been claimed led to the development of the
TOFD technique. Modern digital flaw detectors and computer technology
allow far more information to be store by the operator. I felt that it was time to
give the book a thorough review and to try to address some of the advances
in this second edition.
Over the years, so many knowledgeable people have helped to keep me
informed of changes or provided me with pictures, drawings, techniques and
data that it would be impossible to remember all their names. To miss just
one name could give offence and so I can only thank all of them here.
To those of you who are just starting on a career in ultrasonic flaw detection
and who read this book, I wish you every success for the future and maybe
someday you will quote an old friend of mine who after thirty plus years in the
business claimed that he’d never done a day’s work ‘It had all been fun’. That
only goes to prove that there are masochists in every walk of life.
John Drury
Swansea, March 2004
CHAPTER 1
HISTORY
People have probably used the natural resonance of fabricated solid objects
to make sure they are “Sound”, meaning free from serious imperfections,
as long as they have been making those objects. We talk about “The ring of
truth”, “Sound as a bell” and use similar phrases to denote honesty or quality
of manufacture.
Every solid object, whether it is a piece of pottery or china, a cast bell or
forged sword, has a natural resonant frequency (pitch) when given a sharp
tap. The presence of a large void, crack or similar discontinuity will cause the
resonance of the affected object to differ from that of the standard object. If
the difference in pitch, or duration of ringing is big enough, the human ear
will detect it. The limiting factor with simple acoustic testing is reached when
the critical size of discontinuity that will ultimately lead to failure is too small
to cause a change that can be detected by the most sensitive ear.
The property of sound that governs detectability is wavelength, and if a
discontinuity has a major dimension that is less than half a wavelength
sound will tend to wash around the discontinuity rather than be reflected
by it. In metals, the wavelength of sound at audible frequencies is relatively
large and so only large discontinuities can be detected by ear. Much higher
frequencies are needed to detect the small imperfections that are critical in
modern highly stressed components and it was not until the late nineteenth
and early twentieth centuries that the technologies existed to generate and
detect such high frequencies.
Lord Rayleigh in “The theory of sound” published in the 1870’s described the
fundamental principles defining the nature and behaviour of sound. This was
followed by the discovery of the piezoelectric effect by the Curie brothers
in 1880 with further work by Lippmann in 1881. They found that certain
naturally occurring crystals, cut in a certain way, developed an electrical
5
potential across the faces of the material when subjected to mechanical
pressure and that a mechanical distortion occurred if an electrical potential
was applied across those faces. The piezoelectric effect was eventually
exploited to generate and detect sound waves at the frequencies required
for modern flaw detection.
After the Titanic disaster in 1912, it was suggested that perhaps underwater
sound waves could be used to detect icebergs at sea at a range that
would allow the ship to take avoiding action. The idea became even more
important for the detection of submarines during World War 1 and lead to
the development of a pulse echo system by the end of the war. In the years
immediately following the armistice the pulse echo system found peaceful
uses in hydrographic surveys to chart the ocean depths, and in the fishing
industry to detect shoals of fish. It is interesting that the pulse echo principle
was not adopted for flaw detection in metals until the early part of World War 2.
It was in 1929 that Sokolov, in Russia, first described some work that he had
carried out on cast steel using sound waves at high frequency generated
by quartz crystals. During these experiments he had detected defects in
castings that were too thick to be examined by radiography. His early work
used a quartz crystal to generate a continuous sound wave through the metal
and a pool of mercury on the opposite side to display the arrival of the sound
at that surface. The sound conveyed to the mercury set up a vibration pattern
on the surface of the mercury in much the same way as a pattern develops
on the surface of a cup of tea placed on a vibrating surface. In Sokolov’s
case, a change in pattern indicated an internal change in the casting.
In 1935 he describe a more practical design for flaw detection in metals
in which he used a second quartz crystal to detect the transmitted sound
instead of the mercury pool. He also described his method for coupling the
sound between metal and crystals. This new method of non-destructive
testing was named “Supersonic” flaw detection until the word supersonic
became more readily associated with high speed flight and in the late 1950’s
6
the name was changed to “Ultrasonic flaw detection”. Both names were
intended to indicate that the vibrations were at a frequency that was too high
to be detected by the human ear.
Many other workers especially in Germany and Russia adopted Sokolov’s
continuous wave technique. The sound arriving at the receiver crystal
generated a voltage proportional to the intensity (loudness) of sound
reaching the receiver. The presence of a void or other discontinuity in the
sound path decreased the amount of sound transmitted and the receiver
voltage would therefore be lower. Theory predicted that only a portion of the
energy reaching the far side of the metal would be transmitted to the receiver
crystal and that the remaining portion would be reflected back towards the
transmitter. In when testing cast structures, the reflected energy was usually
too weak to complete the return journey and the technique worked well.
However, it was found that when the technique was applied to fine grained
structures such as forgings or rolled plate, the reflected energy was strong
enough to reach the transmitter, reflect again and join in with the continuing
transmitted waves.
These reflected waves might join in with the new waves in phase,
constructively, thus increasing the intensity of sound, or join in out of phase,
destructively, which decreased the sound intensity. Since the technique used
the received voltage to indicate the condition of the test object, it can be seen
that the interference of the reflected energy destroyed the effectiveness of
the test.
During the early 1940’s, several workers on the field, notably Sproule in
the UK, Trost and Gotz in Germany and Firestone in the USA, began to
use short pulses of ultrasound instead of continuous waves. This approach
had two distinct advantages over previous methods; firstly, it was possible
to wait until all the multiple echoes from one pulse had died away before
sending the next pulse, thereby avoiding interference. Secondly, by placing
the receiver crystal on the same side of the test piece as the transmitter,
7
it was possible to display the thickness of the material and depth of any
discontinuity. Sproule used separate crystals for transmitting and receiving,
whereas Firestone used only a single crystal, the crystal acting as a receiver
during the interval between pulses. Both approaches had advantages over
the other under certain conditions and so flaw detectors by 1950 allowed the
use of either single or twin crystal operation.
Until 1947, ultrasonic flaw detection was restricted to the detection of defects
that were parallel to the scanning surface. Attempts to introduce the sound
into metals at angles above about 10° were thwarted because the beam
underwent mode conversion as well as refraction. The second mode was a
shear wave that can only exist in solids. This wave refracted at a different
angle from the compression wave and travelled at about half the speed. The
presence of two beams with different angles and speeds made interpretation
of signals very difficult. Sproule overcame the problem in that year by
increasing the angle of incidence until the compression wave was eliminated
and he introduced a range of “Shear wave” probes with beam angles in steel
of 45°, 60°, and 70°. This opened up the field for many new applications of
ultrasonic flaw detection in aerospace, welding and other industries.
Since 1950, there have been advances on many fronts in transducer
materials, electronics, and data handling and storage, but the same basic
principles remain in use for many applications. Several techniques for
estimating the size of discontinuities were developed during the 1960’s and
1970’s, but none of these proved to be accurate enough for the new science
of fracture mechanics to reliably predict the likelihood of failure. Sproule
had described the diffraction signal originating from the tip of a reflector in
the 1950’s but it was Silk, in 1977, who first described a practical technique
for using diffraction signals from the top and bottom of a discontinuity to
measure its through thickness dimension. His technique, known as “Time of
Flight Diffraction” (TOFD) offers greater accuracy of defect sizing and has
become widely used in critical weld inspection.
8
CHAPTER 2
BASIC PRINCIPLES OF SOUND
Sound waves are vibrations of the particles of solid liquid or gas through
which the sound is passing. Each particle oscillates about a mean position
and in doing so causes a similar vibration to be taken up by its neighbour.
The resulting disturbance radiates out from the source as a sound wave.
Sound waves are therefore a form of mechanical energy that can only
exist in a solid liquid or gas and not in a vacuum. Essentially, there are
two requirements for sustaining a vibration: there must be something to
vibrate and some force that will always try to return that ‘something’ to its
original position. In other words, there must be MASS and ELASTICITY.
This is illustrated in figure 2.1a below. A weight is suspended from a beam
by a spring. The weight (W) provides the MASS and the spring provides the
ELASTICITY. At rest, the force of gravity (G) acting on the weight is balanced
by the tension (T) in the spring.
If the weight is pulled downwards from its rest position (A) to position B, the
tension in the spring will increase. When the weight is released, the weight
will accelerate back towards position A reaching its maximum velocity at
position A when the forces T and G are again equal. The momentum of
9
the weight travelling at speed will cause the weight to overshoot position
A. Immediately the tension in the spring is less than the force of gravity
and the weight will begin to decelerate until it comes to rest at position C.
Because force G is now larger than T, the weight will start to descend again;
overshooting position A again until the increasing tension in the spring
eventually stops the downward movement. At this time, the whole cycle
of events starts again and continues until friction and air resistance losses
gradually bring the oscillations to a stop.
Figure 2.1b is a graph of the displacement of the weight, during this up
and down motion, against time. In the diagram, two points on the graph
are shown where the weight is doing the same thing, travelling upwards
and passing through position ‘A’ on consecutive passes. The distance
(time) between these two points represent one complete cycle of the
oscillation. The number of cycles of oscillation completed in a given period
of time (usually one second) is called the ‘Frequency’ of the oscillation. The
maximum displacement of the weight from its normal rest position is called
the ‘Amplitude’ of the oscillation.
One of the best examples of an oscillating source of sound that can be
used later in describing the action of an ultrasonic test probe is the guitar.
The strings of a guitar are elastic and pre-tensioned to produce a particular
frequency of vibration. Each string is distorted by the guitarist to stretch
the string and then released. As soon as it is released, the string begins to
oscillate about its mean position at the resonant frequency of that string.
Shortening the string using a finger to hold the string against one of the frets
can change the frequency. The human ear recognises the frequency as the
‘Pitch’ of the note produced. The ‘Loudness’ of the note depends on how
far the guitarist distorted the string, in other words, the ‘Amplitude’ of that
distortion.
The mass of woodwork to which the string is attached amplifies the sound
and adds its own harmonic frequencies to produce a range of notes to give
10
the characteristic richness of tone to the instrument. The band of frequencies
produced is called the ‘Bandwidth’ of the sound in ultrasonics.
THE ACOUSTIC SPECTRUM
Sound waves are described above as the oscillation of particles of solids,
liquids or gases. The human ear can only detect a small range of possible
vibration frequencies, roughly between 16 cycles per second and 20,000
cycles per second. In theory, however, there is a limitless spectrum of
frequencies and that are possible even if humans can’t hear the whole
range. The spectrum is illustrated in figure 2.2 below: -
0.5MHz	20MHz
Typical test range
1000	10,000	100,000	1,000,000 10,000,000 100,000,000
Acoustic Spectrum
Fig. 2.2
The unit used to denote frequency is the Hertz, abbreviated as Hz, where
1Hz is one cycle per second. One thousand Hz is written as 1KHz (Kilo
Hertz) and one million Hz as 1MHz (Mega Hertz). The part of the spectrum
from zero to 16Hz is below the range of human hearing and is called the
‘Subsonic Range’. From 16Hz to 20KHz is known as the ‘Audible Range’
and above 20KHz as the ‘Ultrasonic Range’. Ultrasonic flaw detection uses
vibrations at frequencies above 20KHz.
Most flaw detection takes place between 500KHz and 20MHz although
there are some applications, for example in concrete, that use much lower
frequencies and there are special applications at frequencies above 20MHz.
In most practical applications in steels and light alloys, frequencies between
2MHz and 10MHz predominate. Generally the higher the test frequency,
11
the smaller the minimum detectable flaw, but it will be shown in following
chapters that higher frequencies are more readily attenuated by the test
structure. Choosing an appropriate test frequency becomes a compromise
between the size of flaw that can be detected and the ability to get sufficient
sound energy to the prospective flaw depth.
MODES OF PROPAGATION
Sound energy travels, or ‘propagates’, outwards from the source of the
vibration as the oscillation of a particle of solid, liquid or gas disturbs the
neighbouring particles so that the neighbour takes up the oscillation. It will
take time for the disturbance, called the ‘sound wave’, to reach a given
distance from the source. This is a measure of the velocity of sound in a
given medium. It will be shown that this velocity varies depending on the
characteristics of each material and the way in which the disturbance is
transmitted from one particle to the next. The different ways in which the
disturbance may be transmitted are known as the ‘Modes of Propagation’.
The different modes of propagation come about because solids, unlike
liquids and gases, have a modulus of rigidity as well as a modulus of
elasticity. Figure 2.3 shows a column of air trapped inside two cylinders, each
closed at one end, and with the open end of one fitting perfectly into the open
end of the other. If the two cylinders are pushed together, the pressure of the
trapped air increases and when the applied force is removed, the cylinders
will spring back to their original positions. Similarly, if the two cylinders are
pulled apart, the pressure will decrease, and on release, the partial vacuum
will restore the cylinders to that position.
12
A similar resistance to compressing or stretching the column would be
experienced if the air were to be replaced by water, but the resistance would
be stronger. If the cylinders were to be replaced by a single cylinder of steel,
the resistance to stretching (tension) or compression would be very strong
indeed! These hypothetical columns of a gas, a liquid and a solid could be
represented by a spring attached to the inside of the cylinders as shown in
figure 2.4. The strength of the spring would in turn represent the value of
Young’s Modulus of Elasticity (‘E’) for the material. Solids, liquids and gases
all have this resistance to compression and tension.
Spring representing
Young’s Modulus of
Elasticity (‘E’)
Fig. 2.4
The Modulus of Rigidity (‘G’) is the material’s resistance to a shear load
and this is illustrated in figure 2.5. This shows two cylinders fitting perfectly
across the open ends. If a force is applied to slide the top cylinder to the left
and another to slide the bottom cylinder to the right it is clear that there would
be little resistance to this shear load if the space was filled with air or water,
but considerable resistance in the case of a rigid body like steel. For solids,
this rigidity could be represented by another spring across the column at
right angles to the modulus of elasticity (figure 2.6).
Spring representing the
Modulus of Rigidity (‘G’)
Fig. 2.6
13
COMPRESSION WAVE MODE
Because liquids and gases have no modulus of rigidity, sound waves can only
propagate by using their resistance to tension and compression. This type of
sound wave is called the ‘Compression Wave’. Compression waves can exist
in solids, liquids and gases because they all have elasticity. Compression
waves are also known as ‘Longitudinal’ waves, and sometimes as ‘Plane’
waves The individual particles of the solid liquid or gas oscillate about their
mean position, and the direction of propagation of the compression wave is
in the same plane as the particle motion as shown in figure 2.7.
Particle Motion
◄-----------►
GKJJEED
Direction of Propagation
Fig. 2.	7
SHEAR WAVE MODE
Shear waves only exist in solids and rely on the modulus of rigidity of the solid
under test, they can exist on their own or co-exist with compression waves
and surface waves. Shear waves are also sometimes called ‘Transverse’
waves. Again, the individual particles of the solid oscillate about their mean
position, but the direction of propagation of the shear wave is at right angles
to the particle motion. This is illustrated in figure 2.8.
Direction of Propagation
Fig. 2.	8
14
SURFACE WAVE MODE
At the surface of a solid, a complex mode of oscillation can exist in which
the particle motion is mainly perpendicular to the direction of propagation
as with the shear wave, and partly in the same plane as the direction of
propagation as with the compression wave. This mode of propagation is
called the ‘Surface wave’ or ‘Rayleigh wave’. Surface waves only affect the
surface layer of the solid to a depth of about one wavelength, and have the
advantage that they follow the surface contour of the object and only reflect
at an abrupt change such as a corner or crack. For the surface wave, the
particle motion is elliptical with the major axis of the ellipse at right angles to
the direction of propagation. This is shown in figure 2.9.
Elliptical
Particle
Motion
Direction of Propagation
Fig. 2.	9
LAMB WAVE MODES
Lamb waves, like Surface waves, propagate parallel to the test surface
and have an elliptical particle motion. They occur when the thickness of the
test material is only a few wavelengths at the test frequency and where the
test piece is of uniform thickness. Lamb waves fill the wall thickness and
propagate along the major axis of the component. They can travel several
meters in steel, so they can be used for rapid scanning of plate tube and wire.
Recent developments for rapid corrosion monitoring in buried pipes use Lamb
waves under the name ‘Guided Waves’. The wall of the component flexes
so that the sound ripples along the material distorting both surfaces. Figure
2.10 illustrates a type of Lamb wave where the crests of the wave on the near
and far surfaces coincide. These are called Symmetrical Lamb Waves. Figure
15
2.11 shows another type of Lamb wave where the crest on one side coincides
with a trough on the other. These are called Asymmetrical Lamb Waves.
Fig. 2.10
These waves are generated at incident angles that depend on the test
frequency and material thickness. These parameters also determine the
number of modes of Lamb wave can exist in the test material. In order to
generate a Lamb wave the velocity at which the incident compression wave
in the Perspex sweeps along the interface must coincide with the velocity
of the Lamb wave in the material, this is achieved by adjusting the angle of
incidence i°. This velocity can be calculated from: -
v=-^~
sinz"
Where: -
Vp = the velocity of the incident wavefront along the test surface
Vc = the incident compression wave velocity in Perspex
i° = the angle of incidence in the Perspex
16
Figure 2.12 illustrates the above formula.
Fig. 2.12
CREEPING (LATERAL) WAVES
There is a special type of compression wave called a ‘Creeping’ or ‘Lateral’
wave. It sneaks along the surface rather like a surface wave, its use is
described under TOFD techniques.
SUMMARY
There are several modes of propagation that can be sustained in solids, but
only Compression waves can exist in solids, liquids and gases.
17
CHAPTER 3
PROPERTIES OF SOUND WAVES
1.	VELOCITY
Sound travels at different speeds through different materials. This is
noticeable when, for example, a railroad worker is observed from a distance
striking a rail with a hammer. Since the speed of light is much faster than that
of sound, the observer first sees the hammer strike the rail. If the observer is
also close to the rail, the next event is the sound of the blow coming out of
the rail and finally the airborne sound is heard.
This tells us that the speed of sound in the rail is faster than the speed of
sound in air. It is true that sound travels faster in liquids than in gasses and
faster in metals than in liquids. However, it is also true that sound travels at
different speeds in different metals. There is a distinct speed of sound for
each material and in ultrasonics this is called the VELOCITY of sound for
that material. This being so, it would be useful to have an understanding of
the reasons for the difference.
Imagine two pairs of identical steel balls, one pair joined by a strong compression
spring and the other pair by a weak spring. If one of each pair is moved towards
its partner at a constant speed, the spring joining the pair will start to compress.
Eventually there will be enough compression in the spring to overcome the
inertia of the second ball and it will start to move. As shown in figure 3.1, the
second ball will move sooner for the pair connected by the stronger spring.
P = Constant force for both pairs
Mi = Movement of strong spring second ball
h = Time taken to movement Mq
M2 = Movement of weak spring second ball
t2 = Time taken to movement M2
Fig. 3.1
18
In the analogy, the balls represent the particles of solid, liquid or gas through
which the sound wave is propagating and the springs represent Young’s
Modulus of elasticity ‘E’. The suggestion made by the analogy is that the
disturbance will pass more quickly from one particle to the next in a material
having greater elasticity. In other words, the velocity of a compression wave
will be higher for greater values of elasticity. This is generally the case but
there is another main factor affecting velocity, and that is the density of the
material.
Consider another situation in which pair of aluminium balls and a pair of lead
balls replace the steel pairs in the above analogy but with each pair joined by
springs of equal strength. The inertia of the lead ball is greater than that of
the aluminium ball and this time it will take longer to get the lead ball moving.
This suggests that the compression wave velocity will be lower for high-
density materials than for low-density materials. Density and elasticity are
the dominant factors affecting velocity, but there is another one, which plays
a relatively minor, but none the less significant, role, and it is called Poisson’s
Ratio. During a tensile test, to measure the strength of a metal sample, the
diameter of the sample reduces as the sample is stretched. The change in
diameter divided by the change in length is Poisson’s Ratio. Considering
all these factors, the velocity of a compression wave in a material can be
calculated from the following formula: -
jz = p’.	* ~G
VP (1+0X1-20)
Where
- Compression wave velocity
E - Young’s Modulus of Elasticity
P - Material Density
° - Poisson’s Ratio
19
Shear waves are able to exist in solids but they do not travel at the same
velocity as the compression wave in a given material. This is because it is the
Modulus of Rigidity, rather than Young’s Modulus, that dictates the velocity,
and the modulus of rigidity is lower than the modulus of elasticity. This means
that the shear wave velocity is always slower than the compression wave
velocity in a material. As a rule of thumb, the shear wave velocity is roughly
half the compression wave velocity. The velocity can be calculated from: -
V =---------------r Or, alternatively И = —
1 Vp 2(1+0)	VP
Where
К = Shear wave velocity
G = Modulus of Rigidity
P = Material Density
(5 = Poisson’s Ratio
Surface (Rayleigh) waves also have their own particular velocity, which is
generally taken to be approximately 90% of the shear wave velocity.
Although the velocity for each of these modes of propagation can be
calculated, it requires a precise knowledge of all the parameters, and these
are not usually available to the ultrasonic practitioner. Parameters such as
density and strength vary with alloying, heat treatment, casting, rolling and
forging processes - all of which make it difficult to know that the correct
values are being used. Instead, it is more normal to carry out a routine called
‘Calibration’ during the setting up procedure for an ultrasonic inspection. In
the calibration procedure the flaw detector time-base is adjusted to give a
convenient scale against a calibration sample of known thickness and made
of the same material as the work to be tested. Table 1 at the end of this
chapter lists the compression and shear wave velocities for a number of
materials.
20
2.	WAVELENGTH
While the particles are completing each cycle of their oscillation, the sound
wave is moving outwards in the direction of propagation at the characteristic
velocity for that material. It follows that during the time taken to complete one
cycle of vibration, the sound wave will move a certain distance depending
on the velocity in that material. For a given sound frequency, this distance is
relatively small for liquids and gasses compared to that in metals, because
velocities are higher in metals. The distance travelled by the sound wave
during one cycle of vibration is called the WAVELENGTH. In general, if the
maximum dimension of a reflecting surface is equal to or greater than half a
wavelength, the reflection will be detectable. It follows that calculation of the
wavelength will help in the choice of test frequency for a specific application.
Wavelength is given the Greek symbol A. (lambda) and for any material and
sound frequency, wavelength can be calculated from the equation: -
Where
A.	= wavelength
V = Velocity
f = frequency
Example 1
Calculate the wavelength of a 5MHz compression wave in Steel, given
that the velocity of sound in mild steel is 5,960 metres per second (M/sec).
,. = r
.	5.960 ,,
A, =---------Meters
5,000.000
л X = 0-00192M
21
It would be better to express such a small distance in millimetres (mm) by
multiplying the answer by 1,000: -
A, = 0.00192 x 1,000
X = 1.192mm
At ultrasonic frequencies, the wavelength of sound in metals is relatively
short and so it is usual to express the wavelength in millimetres. This is
done at the start of the calculation by changing the velocity from meters to
millimetres a second by multiplying the value in M/sec by 1,000.
Example 2
Calculate A for a 5MHz compression wave in Aluminium, given that the
velocity is 6,400 M/sec.
. 6,400x1,000
A =-----------mm
5,000,000
A, = 1.22>mm
Example 3
Calculate the wavelength of a 2MHz shear wave in aluminium given that the
shear wave velocity is 3130 M/sec.
3130x1000
----------mm
2000000
A, = 1.565/»/»
22
з.	ACOUSTIC IMPEDANCE
Acoustic impedance of a material is the product of the material’s density and
velocity. At the interface between two materials, the acoustic impedances
either side of the interface will determine what proportion of the incident
sound wave will reflect and what proportion will transmit into the second
material. The symbol allocated to acoustic impedance is ‘Z’ and for a given
material, Z = pxV
4.	REFLECTION
Incident sound
.Reflected sound
A
Material 1
Acoustic Impedance Zi
Interface
Material 2
Acoustic Impedance Z2
Transmitted sound
Fig. 3.2
Figure 3.2 shows the interface between two materials whose acoustic
impedances are Z, and Z2 respectively. In the example, part of the energy
is transmitted into Material 2 and part is reflected back into Material 1. The
percentage of the incident energy that is reflected can be calculated from the
equation: -
RE =
Iji+A
x!00%
Where: -
RE is the reflected energy
2i &Z2 are the acoustic impedances
23
Example 4
Calculate the percentage of the incident energy that would be reflected at a
‘steel to water’ interface given that Zsteel = 46.7 and Zwater = 1.48.
f 46.7-1.48 ?
1^46.7 + 1.48 J
X1OO%
xl()()%
RE = 88%
Note that the remaining 12% is transmitted into the water.
If the example had been given as a ‘water to steel’ interface, the second line
of the calculation would have shown a negative value inside the brackets.
However, the square term outside the bracket would restore the answer to a
positive value and the answer would have been the same 88% reflected, this
time in the water, and 12% would have been transmitted into the steel.
When the interface is between two solids, as in the case of a brazed joint
between two pieces of steel, the reflected energy is much smaller, most of
the energy passing across the braze and into the second steel layer. There
are also examples of two very different materials that have the same acoustic
impedance such as Ro-cee rubber and water. Sound travelling through water
and then encountering this particular rubber compound will carry on through
the rubber as if the interface did not exist. Table 1 at the end of this chapter
shows the acoustic impedance for a number of materials.
5.	COUPLANT
Acoustic impedances for metals tend to have high values whereas those
for gasses are low. From the above example it is clear that at a solid to gas
interface, the proportion of energy reflected is going to be very high. That is
useful because it means that a discontinuity such as a crack or a void in a metal
24
object will reflect almost all the sound back to the test surface. However, it is
also a nuisance because it means that air between the ultrasonic probe and the
test surface will prevent the sound from entering the component. A couplant is
a liquid or paste used between the probe and the test surface to try to match
the acoustic impedance of the probe to that of the test material. It is not a very
efficient process because the best couplants, for example glycerine, only allows
about 15% of the sound to enter the component, and only the same proportion
of any energy coming back to the test surface can enter the probe to give an
echo. At best, then, only a little over two percent of the energy generated at the
probe ever gets back to the display.
There are specially formulated couplants for use in flaw detection as well
as water, oils, greases, glycerine and pastes such as wallpaper paste. The
most important considerations when choosing a couplant are firstly that it is
not hazardous to the individual and secondly that it will not adversely affect
the component.
6.	REFRACTION
Figure 3.2 shows the incident sound as if it were a single ray of energy, but
of course it is really a beam that has some width, rather like a torch beam. If
the incident beam is directed at an interface between water and steel at an
angle other than normal, the angle taken up by the transmitted beam in the
steel will be greater than the incident angle in water. The advancing wave
front in a sound beam can be defined as the plane in which all the oscillating
particles are ‘in phase’, or at the same position in their oscillating cycle. The
bottom edge of the beam shown in figure 3.3 arrives at the interface first
and immediately takes up the faster velocity of the steel. As the rest of the
wave front reaches the interface, so the transmitted beam gradually takes
up steel velocity. By the time that the top edge of the beam enters the steel,
the sound from the bottom edge has already travelled four times further than
it would have in water. Joining up the ‘in phase’ points on the wave front at
the instant the top edge enters the steel shows the wave front advancing at
a new angle. The beam of sound is said to have undergone ‘Refraction’ as it
crossed the interface and the new angle is called ‘the angle of Refraction.
25
Fig. 3.3
The refraction occurs because of the difference in velocity on either side of the
interface and the proportions of energy reflected in the water and transmitted
into the steel remain the same as it would be for normal incidence. Figure 3.4
shows the incident, reflected and refracted angles. These angles are always
measured from the Normal to the interface. In the diagram, Pis the angle of
incidence, r°is the angle of reflection and R° is the angle of refraction.
Fig. 3.4
26
The angles and velocities are related and the relationship is expressed in
Snell’s Law such that: -
Sini° Sinr° _ SinR°	Where: -
К	J 2	i° = Angle of Incidence r° = Angle of Reflection R° = Angle of Refraction V, = Velocity in Medium V2 = Velocity in Medium
7.	MODE CONVERSION
If Mediuml is a liquid and Medium 2 a solid, some of the energy in the
solid will change to the Shear Wave mode. This change is known as
Mode Conversion. For small angles of incidence the proportion of energy
changing to shear wave mode is small and can be ignored. However as the
angle of incidence increases the proportion increases and the shear wave
becomes significant so that there can be two types of wave in medium 2
at the same time, both of which can reflect from surfaces within the object.
Since they both travel at different speeds, and Snell’s Law tells us that they
will refract in different directions, the results can be very confusing. This was
a restricting factor in ultrasonics until Sproule developed the first Shear wave
angle probes in 1947. Until then it was unsafe to rely on angles of refraction
greater than about 10° since echoes from the compression wave could not
be discriminated from the shear wave reflections. Because of this ambiguity,
ultrasonics tended to be restricted to the detection of discontinuities with
surfaces parallel to the scanning surface such as laminations and cavities.
Attempts to detect, for example, weld defects such as lack of sidewall fusion
and root cracks by angling the beam were not reliable.
Sproule realised that the compression wave refracted angle would always
be about double the shear wave refracted angle because the shear wave
velocity is about half the compression velocity. Therefore if the angle of
incidence were to be increased progressively there would be a critical
angle of incidence at which the compression wave would refract through
27
90°. Any increase in angle of incidence beyond this critical angle would
leave only a shear wave in medium 2 and the compression wave would
undergo total internal reflection in Medium"!. With only a shear wave in
medium 2 travelling at a known velocity and at a known angle, the field
was open for many new applications of ultrasonics. The critical angle at
which the compression wave is refracted through 90° is called the first
critical angle. For a water to steel interface the first critical angle is about
15° and for a Perspex to steel interface the angle is about 28°. At these
critical angles, the remaining shear wave is at an angle of refraction just over
30°. Increasing the angle of incidence above the first critical angle causes the
shear wave refracted angle to increase so that transducers can be produced
at a suitable angle to detect particular defect propagation directions.
Eventually a second critical angle of incidence will be reached at which
the shear wave will be refracted through 90°. The shear wave at this second
critical angle will again mode convert, this time to become a Surface
(Rayliegh) wave. This new wave travels at 90% of the shear wave velocity,
only penetrates to a depth of about one wavelength, will follow the surface
contour of the object and will only reflect at an abrupt change in surface
direction such as a corner or a crack. If the angle of incidence is increased
beyond the second critical angle, no sound will be transmitted into medium
2. Ultrasonic transducers having refracted angles between 0° and 10° are
likely to be compression wave probes and those with refracted angles
between 35° and 80° will be shear wave probes. Surface wave probes have
a refracted angle of 90°. Between 10° and 35°, and 80° to 90° it would be
possible to have two simultaneous modes existing in Medium 2 and so it is
unusual to find transducers in these two ranges - exceptions to this rule will
be discussed in a later chapter.
28
Fig. 3.5
Refracted angle Steel
Fig. 3.6
Figures 3.5 and 3.6 show the relationship between the incident angle and
refracted angle for water to steel and Perspex to steel interfaces. The graphs
show that the second critical angle for water to steel is about 28° and for
Perspex to steel about 58°. These values would be different if medium 2
were to be aluminium or some other solid than steel.
Example 5
An incident compression wave in water meets a steel interface at an incident
angle of 19°, calculate the shear wave refracted angle in the steel given that
the compression wave velocity in water as 1480m/s and the shear wave
velocity in steel as 3240m/s.
29
From Snell’s Law
Sini _ SinR
Therefore: -
SinR =
V2xSini

o o 3240x0 3256
SinR =-----------
1480
SinR = 0.7128
R = 45.46°
From a practical point of view it is more usual to know the refracted angle
needed in the test material in order to detect a particular discontinuity, and
so the calculation would be to find the necessary angle of incidence, in water
for immersion testing, or in Perspex for contact scanning. Example 6 shows
this version of the application of Snell’s Law.
Example 6
Calculate the angle of incidence required in Perspex in order to produce 45°
Shear wave in steel given that the compression wave velocity in Perspex is
2680m/s and the shear wave velocity in steel is 3240m/s.
From Snell’s Law
Sini _ SinR
Therefore: -
Sini =
2680x07071
3240
Sini = 0.5849
Incident angle = 35.8°
30
8 REFLECTIVE MODE CONVERSION
Mode conversion also takes place when an ultrasonic beam reflects at
internal surfaces in solids whether these are boundary surfaces, machined
features, or discontinuities. The relationship between incident angle of a
given beam and the relative amplitude of the reflected and mode conversion
beams for steel is shown in the following graphs. They allow an assessment
to be made of the potential confusion in any given situation and can be used
to determine an alternative test angle to be chosen to avoid the problem.
a) Incident Compression Wave
Air
Fig. 3.7
A compression wave incident on a steel to air interface will reflect as a
compression wave together with a mode converted shear wave. At first
glance, the graph in figure 3.7 looks a bit crowded and confusing, so it is
worth looking at three areas of the graph to help understand its use.
If we go to the 10° angle of incidence (a) on the base line and project upwards
to the mode conversion angle (B) we can see that the shear wave travels at
an angle just less than 5° (we could calculate this from Snell’s Law). If we
continue the projection until we meet the shear wave amplitude curve (S)
31
and read across to the right hand scale we see that the relative amplitude
of the shear wave is about 25%. Continuing upwards to the reflected
compression wave amplitude curve (C), we find that the relative amplitude of
the compression wave is about 95%. So for an incident compression wave
at 10° the shear wave mode conversion is still relatively small compared with
the reflected compression wave.
Working in the same way at an incident angle (a) of 30° we find that 0 is
around 15° but the relative amplitudes of the shear wave and compression
wave are 90% and 70% respectively. Both will give strong signals if they
reach the receiver.
Lastly the extreme case, where a is around 60° and (3 around 30° we find that
the relative shear wave amplitude is 90% but the reflected compression wave
amplitude has fallen to only about 10%. For greater angles of incidence than
60°, the shear wave rapidly decreases in amplitude and the compression
wave recovers. Clearly we need to take care in our interpretation of signals
if we see that a compression wave in steel is likely to meet a known
reflecting surface in that part of the graph where the shear wave amplitude
is significant.
b) Incident Shear Wave
Air
Fig. 3.8
к
32
д shear wave will reflect as a shear wave together with a mode converted
compression wave. Using the graph in figure 3.8 as we did before, we can
see that the most severe case is when the incident shear wave meets a
steel to air interface at about 30°. The reflected shear wave amplitude is very
low and the mode-converted compression wave is very strong and almost
perpendicular to the test surface.
If the incident shear wave grazes a surface, in other words the incident
angle is around 90°, there will be a mode conversion to Rayleigh wave. This
can happen when a shear wave grazes the bore of a machined hole in the
specimen. In that case the Rayleigh wave will follow the bore surface and will
reflect if it encounters a sharp changes to the bore such as a keyway. If you
are not aware of the possibility, you may assume that there is a discontinuity
in a false position. An example is shown in figure 3.9 below.
Assumed reflection point
Rayleigh wave reflects from keyway
Fig. 3.9
33
Materials	Velocity (C)	Velocity (S)	Density	Acoustic Impedance
Units	m/s x 103	m/s x 103	kg/m3	
Air	0.33	-	1.2	0.0004
Aluminium	6.40	3.13	2700	17.3
Barium Titanate A	5.26	-	5700	30.0
Barium Titanate В	5.53	-	5700	31.5
Beryllium	12.89	8.88	1800	23.2
Brass	4.37	2.10	8450	37.0
Cast Iron	3.5 to 5.6	2.2 to 3.2	7200	25 to 40
Copper	4.76	2.33	8930	42.5
Glass (Crown)	5.66	3.42	3000	17.4
Gold	3.24	1.20	19300	63.0
Iron	5.96	3.22	7850	46.8
Lead	2.40	0.79	11300	27.2
Lead Niobate	2.76	-	5800	16.0
Lead Zirconate Titanate A	3.00	-	7600	22.8
Lead Zirconate Titanate В	3.00	-	7500	22.5
Lithium Sulphate	5.45	-	2060	11.2
Magnesium	5.74	3.08	1720	9.90
Mercury	1.45	-	13550	19.6
Molybdenum	6.25	3.35	10200	63.7
Nickel	5.48	2.99	8850	48.5
Oil	1.44	-	900	1.3
Perspex	2.68	1.32	1200	3.2
Platinum	3.96	1.67	21400	85.0
Polystyrene	2.35	1.12	1060	2.5
Steel (Mild)	5.96	3.24	7850	46.7
Steel (Stainless)	5.74	3.13	7800	44.8
Silver	3.70	1.70	10500	36.9
Tin	3.38	1.61	7300	24.7
Titanium	5.99	3.12	4500	27.0
Tungsten	5.17	2.88	19300	100.0
Tungsten-Araldite	2.06	-	10500	21.65
Tungsten-Carbide	6.65	3.98	10000 to 15000	66.5 to 98.5
Uranium	3.37	2.02	18700	63
Water	1.48	-	1000	1.48
Zinc	4.17	2.48	7100	29.6
Zirconium	4.65	2.30	6400	29.8
Table 1
34
CHAPTER 4
TRANSDUCERS for generating and detecting sound waves
SWINGING THE LEAD
There is an amusing story about a nautical gentleman at the time when
wooden ships were being superseded by iron ones. This sailor thought up
a new way to determine the depth of water under the hull to replace the old
lead weight on a rope method. He decided that it should be possible, with a
large hammer and a stopwatch, to bang on the iron bottom of the ship with
the hammer and time the return echo from the seabed with the stopwatch.
The measured time could be used to calculate depth using the speed of
sound in water. Fired with enthusiasm, he gathered together a number of
marine dignitaries in the bilges of his ship, passed a large sledgehammer
to a muscular boatswain, took out his stopwatch and ordered the ‘swain to
wallop the floor. This he did with such vigour that the hull boomed for ten
minutes and the assembled observers were deafened for a month! Nobody
heard an echo.
There are parallels with ultrasonic flaw detection in the story; we need our
sound pulses to be ‘loud’ enough to penetrate to the depth of the anticipated
flaw and we need the duration of the pulse to be short so that it does not
mask any returning echoes. We also need the sound frequency of the pulse
to produce a wavelength short enough to detect the smallest reflector that
must be detected to ensure safety. In this chapter we will discuss various
ways of generating and detecting suitable pulses and some of the limitations
we face in terms of penetration and flaw sensitivity.
ULTRASONIC TRANSDUCERS
A transducer is a device that will change one form of energy into another.
An electric motor changes electrical energy into mechanical energy and an
alternator does the reverse. Ultrasonic transducers change electrical energy
into mechanical energy (sound waves) or vice versa. There are several
35
methods used to generate and detect ultrasonic pulses in modern flaw
detection and the most common of these makes use of the Piezo Electric
effect found in certain materials. Other methods, such as the Electro Magnetic
Acoustic Transducer (EMAT) and Laser technology will also be described.
PIEZO ELECTRIC TRANSDUCERS
In 1880 the Curie brothers discovered that slices cut in a particular way
from certain crystal materials would generate an electrical potential across
the faces of the slice when distorted by a mechanical force. They called
this phenomenon ‘Piezoelectricity’ from the Greek words for ‘Pressure’ and
‘Electricity’. A year or so later Lippman reported that the reverse was true;
that a voltage applied across the slice would produce a mechanical distortion.
Quartz was the prime example of a piezoelectric material, but Rochelle salts
and Tourmaline crystals also displayed the same effect. Modern piezoelectric
gas igniters use a cam pressing on a quartz crystal to produce a high voltage
that creates a spark.
For the first thirty years of ultrasonic flaw detection from Sokolov in
1929, until the end of the nineteen fifties, quartz was the most common
transducer material. Appropriate slices were cut from a single crystal. Later
new polycrystalline materials were developed that had lower electrical
impedance (resistance to high frequencies) and gave better ultrasonic
performance, as much as 60 to 70 percent more efficient than quartz. In the
raw state, these materials do not display an overall piezoelectric effect. This
is because, although the small individual crystals making up the material are
piezoelectric, their arrangement within the bulk is haphazard so and tends to
cancel out any overall distortion or voltage. In order to produce an effective
piezoelectric disc, the material has to be ‘Polarised’. During polarization
the individual crystals align themselves in the same direction so that their
combined effect is coherent. The polarisation process involves heating the
discs in an oil bath to a critical temperature called the ‘Curie Temperature’,
applying a strong electrostatic field across the disc and then allowing the
temperature to cool slowly. Figure 4.1 illustrates the polarising process.
36
Electrostatic field
Heated oil
++++++++++++F+
Polycrystaltine disc
Fig. 4.	1
The Curie temperature differs for each of the common materials used in
ultrasonics, so that the oil bath will need to be heated to a suitable temperature
for the material in use. For Barium Titenate the Curie temperature is around
120°C whereas for various grades of Lead Zirconate Titenate (PZT) the
temperature is from 190° to 350°C and for Lead Metaniobate (PMN) it is
about 400°C. If the material is subsequently heated to a temperature near
to the Curie temperature, the disc will ‘depolarise’ and lose its piezoelectric
properties. It follows that care needs to be taken to avoid depolarisation
when testing hot materials and this will sometimes influence the choice of
transducer material.
MODE OF VIBRATION
Whether the transducer disc is made from a naturally occurring piezoelectric
crystal, or one of the polarised polycrystalline materials, we usually refer to
the disc as ‘the crystal’ when talking about probe construction. The crystal
‘disc’ or ‘plate’ may be round or rectangular and for some applications may
be curved plates or concave discs to focus the sound. The way in which
the plate vibrates when stimulated by an electrical pulse depends upon the
‘cut’, in the case of quartz, or the direction of polarisation in the case of
polycrystalline materials. Figure 4.2 is a drawing of a typical quartz crystal
showing the three axes defined by crystallographers, and two plates cut from
a crystal, one an X-cut plate and the other a Y-cut plate.
37
X-cut crystal
Fig. 4.	2
An X-cut plate is taken from the quartz crystal so that the X-axis is
perpendicular to the plate and the Y-cut plate has the Y-axis perpendicular
to the plate. If a voltage is applied across the faces of these plates, an X-
cut crystal will distort in the thickness mode whereas a Y-cut crystal will
distort in shear mode. Figure 4.3 illustrates the changes in shape when an
alternating voltage is applied to an X-cut crystal and Figure 4.4 shows the
shape changes for a Y-cut crystal. The same two modes of vibration can be
obtained using the polycrystalline materials by polarising across the faces of
the plate (equivalent to X-cut), or parallel to the faces of the plate (equivalent
to Y-cut)
Fig. 4.3
Fig. 4.4
38
The X-cut crystal is the one most commonly used in ultrasonic flaw detection,
it can generate and detect compression waves, and can therefore transmit
sound through the liquid couplant we use. Since shear waves cannot exist
in liquids or gases, the only way in which a Y-cut crystal could be used to
generate shear waves in a metal object would be to use a solid couplant; in
other words we would need to glue the crystal in position. This is done in a
few very special applications.
METHOD OF PULSING AND FREQUENCY
When we generate a short pulse of sound with our ‘crystal’, we don’t ‘drive’
the crystal with an alternating voltage of suitable frequency; instead, we
‘pluck’ the crystal with a short sharp electrical shock and allow the crystal to
‘ring’ at its natural resonant frequency. This is rather like ‘plucking’ a guitar
string that also vibrates at its natural frequency. The string is stretched by
the finger and only produces sound when it is released; the greater the initial
stretch, the louder the sound that is produced. In the case of the piezoelectric
plate, the crystal stretches as the voltage is applied and only produces sound
when the voltage is rapidly cut off. To increase the amplitude (loudness)
of the ultrasound we increase the peak voltage (pulse energy) applied to
the crystal. With the guitar string we can change the resonant frequency
by making the effective length longer or shorter by placing a finger on a
different fret. The frequency of our ultrasonic transducer is determined by
the thickness of the crystal. As the crystal is made thinner, so the resonant
frequency increases. Quartz crystals are split in the appropriate plane to
produce X-cut plates, shaped as rectangles or discs and then lapped to the
correct thickness for the required frequency. The polycrystalline materials
are made as slurry that is moulded and compacted under pressure and then
sliced and lapped to the required thickness.
The required thickness for a given frequency can be calculated from the
frequency-thickness constant for the crystal material to be used. Since
this depends on the velocity of a compression wave in that material it can
be seen that the thickness for a given frequency will not be the same for
39
PZT and quartz, for example. The frequency-thickness constant is defined
mathematically as: -
fxt = —
2	Where: -
f = the desired frequency
t = the crystal thickness
v = the compression wave
Example 7
Calculate the required thickness of a PZT crystal to produce a resonant
frequency of 5MHz given that the compression wave velocity for PZT is
ЗОООМ/s.
2xf
3000
2x5000000
xIOOOmm
t = 0.3mm
CONTROL OF PULSE LENGTH
In ultrasonic flaw detection we measure the time taken for each echo to
arrive at the receiver after entering the scanning surface of the object. If we
know the velocity of sound in the material we can determine the distance
travelled by the sound wave. Suppose that a crack has grown from a bolthole
in the object as in figure 4.5; some of the sound will reflect from the top of the
bolthole, and a little while later, some will reflect from the crack. The arrival of
the two echoes at the receiver will be separated by a short interval of time (T2
- TJ. If the ringing time of the crystal (pulse length) is longer than this interval
of time, then we may not be able to distinguish the crack from the top of the
40
bolthole - we may miss the crack. We say that we have not ‘resolved’ the two
echoes or that the resolution is poor. In order to improve resolution we need
to ensure that the pulse length is as short as possible.
Fig. 4.5
TiT2
In an orchestra, if a drum needs to produce a very short but loud sound,
the drummer gives the drum a hefty bang to make the sound loud, and
immediately puts a hand on the drum skin to stifle the note. In ultrasonics
we shorten the pulse duration by applying a weight to the back of the crystal
known as the ‘damping’ or ‘backing’ slug. The damping slug is often made
of a mixture of tungsten powder in an epoxy resin. The amount of damping
applied to the crystal will govern the resolution of the probe. There are
several practical ways of measuring resolution that we will describe later,
but we can also express the resolution in terms of the number of cycles in
the pulse. A short pulse probe will have only one or two cycles whereas a
longer pulse probe may have from three to five cycles. An undamped crystal
may have twelve or more cycles in the pulse. For a given number of cycles
in a pulse, the duration or space occupied by the pulse will depend on the
wavelength, which in turn depends on the probe frequency and the velocity
of sound in the material being inspected. We can say that: -
Pulse length = Number of cycles in the pulse multiplied by the
wavelength.
41
It is obvious that one way to improve resolution would be to increase the
test frequency, however in a later chapter we will see that the penetration
of sound into the object decreases as the frequency increases. Choosing a
suitable test frequency is often a compromise between resolution penetration
and flaw sensitivity and sometimes we will be faced with the situation where
ultrasonics will not be able to detect a particular discontinuity at the critical
depth. While resolution is an important consideration in many applications, it
is not always the case and we sometimes prefer to use a longer pulse. One
example of an application where we might choose to use a long pulse probe
could be the examination of a long shaft such as a railway axle. The screen
on our flaw detector may only be 75mm wide and the display may represent
the length of the shaft, say 2.5m; a short pulse of 2 cycles will occupy such a
small part of the screen that it is too feint to see and it would be better to use
a longer, more visible pulse
PIEZO-COMPOSITE TRANSDUCERS
In a more recent development of the piezoelectric transducer, the active
plate in the test probe is made by slicing piezoelectric crystals into small
squares and assembling them into a matrix separated with an epoxy or a
rubber compound as shown in figure 4.6. The main advantages of this type
of construction are firstly, lower acoustic impedance allowing better matching
to the couplant and more sound into the specimen. This is an advantage
when testing castings and stainless steel. Secondly, resolution - they tend
to provide very short pulses, and thirdly, the absence of additional damping
means that the probes have a very low profile.
Plan view
Side view
Fig. 4.6
□□□□□□□
□□□□□□□
□□□□□□□
□□□□□□□
Epoxy
Crystal
42
pOLYVINYLIDENE FLUORIDE (PVDF) TRANSDUCERS
pVDF was also found to display Piezo electric characteristics and has been
used in ultrasonic flaw detection. These thin plastic films have advantages
and limitations compared with conventional crystals. On the plus side, they
can be easily shaped to focus sound, they produce very short pulses and they
give good transmission into water because the acoustic impedance is similar
to water. Against these advantages, the films are fragile and cannot be used
in contact scanning, and the power output is relatively low compared with
ceramic crystals. The main application is in high resolution immersion testing.
ELECTROMAGNETIC ACOUSTIC (EMAT) TRANSDUCERS
EMAT transducers provide a non-contact alternative to piezoelectric
transducers. Sound waves are generated in the surface of a conductive test
object by an electrical pulse applied to a flat coil that is positioned between
a strong magnet and the test piece. The interaction between the magnetic
field generated in the coil by the electrical pulse and the fixed magnetic field
of the magnet causes a rapid ‘shock’ deformation at the surface of the test
piece and an ultrasonic wave travels through the metal object. The EMAT
probe needs to be close to the test surface, but does not need to touch
it. Returning echoes arriving at the scanning surface cause the surface to
vibrate in the magnetic field. This generates eddy currents in the test surface
and the coil detects the eddy currents. Figure 4.7 illustrates the set-up for an
EMAT probe.
EMAT Magnet
Test object
Fig. 4.7
43
EMAT probes can be used with an air gap when testing hot surfaces and
on surfaces coated with non-conducting material such as rubber, paint and
fibreglass because the sound wave does not have to travel through the gap
material. The probes can be configured to generate horizontally polarised
shear waves directly into the test object. This is an advantage when testing
austenitic welds, castings and other materials with dendritic grain structure
because the shear wave does not mode convert when it meets a reflecting
surface that is parallel to the direction of polarisation. Because shear waves
travel at roughly half the velocity of compression waves and have shorter
wavelengths, it is possible to obtain better near surface resolution and this
can be an advantage when testing thin materials.
However, there are some disadvantages with EMAT probes, they are relatively
large and inefficient compared with conventional probes and they cannot be used
on non-conducting test objects unless a conducting coating is applied.
LASER TRANSDUCERS
Another non-contact method of generating ultrasound uses laser technology.
A short burst of a laser beam on the surface of the test object causes a thermal
shock with rapid local expansion of the surface. The sudden distortion of the
surface causes an ultrasonic pulse to travel through the test object. The
returning echo distorts the test surface and this distortion can be detected
by a separate laser interferometer without a couplant, or can be detected
with a conventional piezoelectric crystal and couplant. The gap between
the transducer and test surface can be greater than is possible with EMAT
probes and can be as much as 250mm (10 inches). Typical applications
include the inspection of composite materials in the aircraft industry.
Q’ FACTOR AND BANDWIDTH
Up to this point we may have gained the impression that our transducer
produces a pure note at the calculated frequency, but this is not true. In
fact the sound wave produced contains a band of frequencies related to the
44
thickness of the crystal, its diameter or length and width plus the effects of the
damping medium. In addition the electrical characteristics of the transducer
and associated circuits affects the overall spectrum of frequencies. We refer
to this spectrum as the ‘Bandwidth’ of the probe. In a well-designed probe,
the centre of this band should be the desired probe frequency and the lower
and upper limits are usually defined as the frequencies at which the amplitude
is reduced by a given factor. Some people use 30% (-3dB) and others 50%
(-6dB) as the factor we will use 50% in the following examples. Figure 4.8
illustrates the bandwidth of a 5MHz probe in which the -6dB bandwidth is
equal to the centre frequency, in other words, from 2.5MHz to 7.5MHz.
Fig. 4.8
Figure 4.9 shows the bandwidth for another 5MHz probe, but this time the
bandwidth is only from 3.75 to 6.25MHz.
Fig. 4.9
45
The probe shown in figure 4.8 can be described as having a broad bandwidth
whereas the probe in figure 4.9 has a narrower bandwidth. In practice, short
pulse probes have a broad bandwidth and long pulse probes are narrow
bandwidth. For a given crystal size, material and frequency damping not
only reduces pulse length, but also reduces pulse amplitude, so the narrower
bandwidth probes will have longer pulses but more amplitude in the pulse
therefore giving deeper penetration.
Another way of expressing bandwidth that is also common in other branches
of electronics is the ‘Quality Factor’ or ‘Q’ of the probe and is defined by the
formula: -
Where: -
f0 = the centre fequency
f, = the upper - 6dB frequency
f2 = the lower - 6dB frequency
Example 8
Calculate the Q factor for the probe illustrated in figure 4.8.
7 5-2 5
Q = 1
46
example 9
Calculate the Q factor for the probe illustrated in figure 4.9.
5
Q~ 6 25-375
Q = ^
2.5
(2 = 2
Undamped crystals can have a Q factor as high as 20,000 but for ultrasonic
flaw detection the Q factor is normally in the range 1 to 10.
47
CHAPTER 5
PROBE CONSTRUCTION
COMPRESSION WAVE PROBES
Standard compression wave probes can be for contact scanning or for
immersion testing. The contact scanning probes are either single crystal or
twin crystal (dual) in construction. The construction of a typical single crystal
contact probe is shown in the diagram in figure 5.1.
Co-axial
connector
Fig. 5.1
The thickness of the crystal determines the operating frequency as we
described in the previous chapter and the faces of the crystal are coated in
silver to make electrical contact.
The damping slug is cast onto the rear of the crystal and bonds to it as the
epoxy sets. The amount of damping used determines the pulse length. A fine
wire is soldered to the back of the crystal, using a solder that melts at low
temperature, before adding the damping slug.
The wear face is glued to the front face of the crystal to protect it during
contact scanning. The thickness of the wear face is important. It is made
to be one quarter of the wavelength at the test frequency for the velocity of
sound in the wear face material. This thickness gives maximum transmission
48
of sound out of the probe into the test sample. Some wear faces are made
from shim steel, others from a hardwearing ceramic material. The steel wear
faces can be used to earth the front face of the crystal to the probe housing
and are less fragile if you drop the probe, but are inclined to stretch and
disbond from the crystal with use. If a non-conductive wear face is used, an
alternative earthing method must be used.
The wear face, crystal and damping slug assembly are then fitted into the
housing, the other end of the centre wire is soldered to the centre terminal of
the connector and the cap and connector fitted to the housing. Figure 5.2 is
a photograph of a typical single crystal compression wave probe.
Twin crystal, or ‘dual’ probes are used to eliminate the ‘dead zone’ occupied
by the transmission pulse with a single crystal probe. In this type of probe
one crystal acts as a transmitter, the other as a receiver and the amplifier
is isolated from the transmitting crystal. The two crystals are mounted on
acrylic or polystyrene wedges these components are illustrated in figure 5.3.
An acoustic barrier, usually made of cork, is fitted between the wedges and
crystals to prevent cross talk between the transmitter and receiver. Figure
5.4 shows a typical twin crystal probe.
Wedges
Fig. 5.3
49
Fig. 5.4
Immersion probes are similar in construction to that shown in figure 5.1
except that it is not necessary to fit a wear plate and so the silvered face of
the crystal is usually visible. Probes can be focussed and this is achieved by
fitting a plastic or epoxy lens to the front of the crystal, or by making a curved
sectioned crystal. Figure 5.5 shows a 20MHz immersion probe with a small
diameter spherically focussed crystal.
Fig. 5.5
The lens or curvature can also be cylindrical as illustrated in figure 5.6. The
cylindrical version is often referred to as a ‘Paintbrush probe’ because it
allows a wide scan.
50
Focussing can also be achieved using a technique called ‘Phased Array’,
although not with conventional ultrasonic sets. The phased array probe
contains a number of small crystals and the pulsing circuit is designed to be
able to apply a pulse to all crystals simultaneously to produce a conventional
zero degree compression wave, or to pulse each crystal separately with a
small time interval between each. In the diagram shown in figure 5.7, the outer
elements are triggered first and a progressive delay is used to pulse inner
elements, the centre crystals being the last to be triggered. The result is that
the ultrasonic wavefront reinforces in the curved way shown in the diagram
to focus at a region determined by the delay intervals. By changing these
intervals, the focal length can be changed. The principles of constructive and
destructive reinforcement will be dealt with later in chapter 7.
Fig. 5.7
Single crystal ‘Delay line’ probes are sometimes used in contact scanning
to reduce the ‘Dead Zone’ below the beam entry surface occupied by the
transmission pulse and probe noise. The delay line is usually Perspex or a
similar material and provides a stand off just like the water path in immersion
testing. The length of the delay line must be sufficient to allow one or more
backwall echoes in the specimen depending on the application. Figure 5.8 is
an example of a delay line probe.
Fig. 5.8
51
SHEAR WAVE PROBES
Since shear waves cannot travel through liquids or gases, angled beam
probes use compression waves in the incidence wedge in contact probes.
The incident angle will be an angle between the first and second critical
angles so that we only have the mode converted shear wave in the test
material. Figure 5.9 is a sketch of the typical arrangement.
We not only get a mode converted, refracted shear wave in the test piece,
but we also have a reflected compression wave in the wedge. If this internal
reflection manages to get back to the crystal face as it bounces around the
wedge, we would have a standing echo that would be confusing. Several
methods of avoiding this problem have been used over the years. The
earliest probes used a long Perspex path shaped ‘Cusp’ so that the reflection
would be absorbed before it could return to the crystal. The Cusp made a
rather unwieldy probe and the next design used ‘V’ shaped grooves in the
front and top surfaces of the incident wedge to scatter the internal reflections.
Some had a plastic material moulded onto these grooves to further damp the
reflection. In the latest, most compact, designs the wedge is surrounded by
a material that has a good acoustic match to Perspex, but a much higher
absorption of sound. The internal reflections are transmitted easily into this
layer and then absorbed. Figure 5.10 shows examples of the three designs
and illustrates their relative size.
52
Fig. 5.10
Figure 5.11 is a photograph of a sectioned shear wave probe, showing
the crystal, incidence wedge and the blocking medium for the internal
reflections.
Fig. 5.11
Phased Array transducers, such as the one already discussed (figure 5.7),
are also used to generate angled shear waves in the test piece. These
transducers have the advantage that the phase delay between the crystal
elements can be varied to give different angles of refraction. The delays can
be swept through a range of values to give a shear wave beam that sweeps
through a desired range of shear wave angles rather as a Radar scanner
sweeps the skies.
In the last chapter, we said that EMAT probes could generate compression
or shear waves, but that shear waves were often used because they can
be directed perpendicular to the test surface (that is a 0° probe). That has
advantages in resolution, because the wavelength for a shear wave is about
half the wavelength for a compression wave and because the velocity of
the shear wave is about half that of the compression wave, we are able to
measure thinner sections than we can with conventional 0° probes of the
53
same frequency. The EMAT probe shown in figure 5.12 is a radially polarised
shear wave probe operating broadband between 1-10MHz, with a centre
frequency of about 5MHz.
Courtesy of Ultrasonics Group,
Dept of Physics, University of Warwick
Fig. 5.12
54
CHAPTER 6
PULSE-ECHO FLAW DETECTOR
The ultrasonic flaw detector is required to provide the voltage pulse to
activate the probe crystal, to amplify received signals from the probe and to
display those signals so that the relative time of arrival and amplitudes of the
signal train can be viewed and interpreted. In order to display the very short
intervals of time involved in testing metals, the early pulse echo systems
used a cathode ray tube (CRT) as the display module. More recently,
equipment manufacturers have turned to digital technology and used LCD
panels for the display. The result has been the manufacture of much smaller
and lighter ultrasonic equipment. Ultrasonic sets in the early 1960’s used
thermionic valves (vacuum tubes) and weighed 25 to 30 Kg (50 - 60 lbs).
From the late 1960’s, transistor technology and smaller CRT’s meant that the
flaw detectors became smaller and lighter weighing between 5 and 10 Kg
(10-20 lbs). In the new millennium, the weight has come down to around 3
Kg. Figures 6.1 to 6.3 show the progression.
Circa 1980
Courtesy of Sonatest PLC
Fig. 6.2
Circa 1960
Fig. 6.1
Circa 2000
Courtesy of GE Inspection Technologies
Fig. 6.3
55
Figure 6.4 is a block diagram of a typical analogue flaw detector showing the
main components and the controls associated with each component.
Fig. 6.4
The clock or ‘timer’ is the heart of the flaw detector. It feeds an electrical
pulse to the Pulse Generator and simultaneously to the Timebase Generator.
This timer pulse causes the pulse generator to send a short, high voltage
pulse to the crystal and at the same time triggers the timebase generator to
begin to sweep the electron beam in the CRT tube from left to right between
the ‘X’ plates at a constant speed.
As soon as the high voltage pulse at the transmitter crystal is cut of, the
crystal starts to vibrate and an ultrasonic pulse propagates into the test
piece. While this sound pulse travels through the material, the CRT sweep
continues to track the time as it moves towards the right hand side of the
display. Reflections from internal surfaces arrive at the receiver crystal,
generate a voltage in the crystal and this voltage is amplified and passed
56
to the ‘Y’ plates where it causes a vertical deflection of the electron beam
proportional to the amplitude of the received signal.
When the electron beam reaches the extreme right hand side of the CRT it
flies back to the left hand side and waits for the next trigger pulse from the
clock. This whole sequence of events takes place so quickly that we wouldn’t
be able to see the trace. The clock repeats the sequence many times a
second and the result is a flicker free trace that increases in brightness
the more times we repeat the process each second. The number of trigger
pulses per second is known as the ‘Pulse Repetition Frequency’ (PRF),
or ‘Pulse Repetition Rate’ (PRR). It is important that we allow enough time
between pulses for all the multiple echoes within the specimen to die away
or we will see the tail end of these echoes showing as ‘Ghosts’ at confusing
positions on the timebase. For this reason the PRF is controlled by the Depth
Range Coarse control in the timebase generator circuit. However, some flaw
detectors have an additional manual control that the operator can use. Ghost
echoes are most likely to be encountered when testing fine-grained light
alloy forgings that have very low attenuation of sound.
The voltages developed in the receiver crystal are very small and need
to be amplified. The ‘Amplifier’ circuit needs to be tuned to accept the
frequency of the ultrasonic pulse and this can be by way of switched bands
for example, 1-3Mhz, 3-7MHz, 7-10MHz & 10-15MHz, or it could be a
wideband amplifier with the range 1 -15MHz. If the former, the set will have a
‘Frequency’ selector switch that should be switched to the appropriate band
for the probe in use just as you would use the tuning dial to select the desired
radio programme.
The ‘Gain’ or ‘Sensitivity’ control allows the amplification to be increased
or decreased depending on the strength of the received signals much like
the volume control on a radio. The Gain control is usually calibrated in
decibels (dB) and is sometimes called the ‘Attenuator’. Strictly speaking,
an attenuator should be calibrated such that increasing the dB reduces the
57
signal amplitude, but this is seldom the case over recent years. The ‘Bel’
is a unit that is commonly used in electronics to compare the ratio between
two power or voltage values and is a logarithmic unit so that large ratios
can be expressed concisely. The intensity of sound in a received pulse is a
measure of the power or energy in that pulse, and that mechanical energy
is converted into electrical energy by the piezoelectric crystal. If the power
increases from P1 to P2, then the gain can be expressed as: -
p,
Gain = Log10 —Bels
However, the Bel is too large a unit for the values we shall encounter
in ultrasonics and so we use a unit of one tenth of a Bel or decibel. The
equation then becomes: -
p
Gain - 10Z_og10 —dB
Pi
The CRT measures voltage and electrical power is proportional to the square
of the voltage: -
Gain = 10Log10
And,removing the brackets: -
Gain = 20Logw -^~dB
The height of a signal on the CRT is proportional to the voltage applied to the
‘Y’ plates and so we can change the equation so that it is in terms of signal
height: -
Gain = 20Log10 ^-dB
i de
V,
58
Example 10
Calculate the gain ratio in dB between a signal that is 60% full screen height
and one that is only 30% full screen height.
Gain = 20Logw ^^8
Gain = 20Log^02dB
Gain = 20x0 301 OdB
Gain = 6.02dB
When we measure depth or thickness from the timebase, we use the left hand
flank of the signal on the screen. Sometimes surface roughness, material
grain size, or electronic ‘noise’ create noise signals (grass) that obscure the
point where the flank meets the timebase and it is difficult to make the correct
reading. In these circumstances, we can use the ‘Suppression’ or ‘Reject’
control to remove the grass a little like the way we use a tone control on a
radio to cut out ‘hiss’. Because this control can also cut out small relevant
signals and make the gain non linear, a warning light comes on when the
control is in use.
The last feature that we need to consider in the amplifier circuit is the one
that controls the degree of rectification and smoothing of the pulse. The
received signals are, of course, a few cycles of alternating voltage. We can
display these as they are - ‘Unrectified’ - but it is not so easy to measure
amplitude directly from the screen. It is more usual to display these signals
as ‘Rectified’ and smoothed signals in which the negative half cycles are
inverted and the signal envelope smoothed out. On some equipment, we
may also have the choice to only display the ‘Positive’, or ‘Negative’ half
cycles and this may give a sharper flank to the signal. Figure 6.5 illustrates
the four conditions, but unsmoothed to illustrate the principle.
59
Un rectified
Full
Positive
Negative
Fig. 6.5
The ‘Timebase’ circuit controls the sweep speed and delay functions. The
sweep speed will determines the thickness range that can be displayed on
the CRT A high sweep speed (fast timebase) may only allow a return path
from a 10mm thickness in the test piece and at the other extreme, a low
sweep speed (slow timebase) may allow a return path from 5 metres or
more. Two controls achieve the desired thickness range, the ‘Coarse Depth’
or ‘Range’ control switches the range in steps (10mm, 50mm, 100mm,
500mm, 1m & 5m for example) and the ‘Fine Depth’ or ‘Range’ control is a
continuously variable control that allows fine adjustment during calibration
to allow for the specific material velocity. The fine depth range control is
sometimes labelled ‘Material’ or ‘Velocity’.
There are times when we don’t want the timebase generator to begin the
sweep when the crystal is pulsed. For instance, when we are carrying out
an immersion test we want the timebase to start when the sound enters
the specimen so that the left hand end of the timebase represents the top
surface of the test piece. Another example might be when we are testing a
long shaft and we want to look in more detail at, say, the last 200mm of the shaft.
In either case, we can delay the start of the sweep with the ‘Delay’ control.
The last component to consider is the display module, the CRT The image
created by the electron beam (the trace) must be displayed so that the
baseline is aligned with the graticule, extends beyond the left and right hand
60
ends of the graticule, is bright enough to see in the test environment and
is in focus. There are four controls for these functions, the ‘X-shift’ and ‘Y-
shift’ controls position the trace, the ‘Brightness’ control can be adjusted
for indoor or outdoor viewing, and the ‘Focus’ control sharpens the trace.
On many flaw detectors, only the focus and brightness controls are provided
for operator adjustment.
Digital flaw detectors provide the same PRF, Amplifier and Timebase
functions but these are usually controlled using a combination of menu
selection and so called ‘Smart Knobs’ through the controlling CPU. Figure
6.6 is a representative block diagram for a digital instrument.
One of the real advantages of the digital instruments is the facility to store
calibrations for a number of inspection procedures and probes, to store
whole traces complete with the calibration data for each trace and to create
databases to store thickness readings. Because the instruments are based
on computer technology, it is possible to connect the flaw detector to a PC
through a serial cable and download stored data, for reporting purposes.
The LCD display also has advantages over the CRT. It consumes less
power than the CRT; it can be backlit for viewing in low light conditions and
at the same time is easy to see without backlighting in daylight. In difficult
conditions, the trace can be ‘Frozen’ so that the operator can move to a
more comfortable position before reading the timebase.
61
Fig. 6.6
Many flaw detectors, both analogue and digital, have gating circuits that
allow signals to be monitored by the instrument and the output used to trigger
audible or visual alarms, or to be connected to chart recorders or computers.
The monitor gates may be displayed in one of two ways. The timebase may
be raised over the gate distance as shown in figure 6.7, or a separate ‘bar’
may be used as shown in figure 6.8.
1	1	1	1	1	
	Gate
uiihiu. in il 1щ,ннкш.ш1|щ jjiiIhji	
01	23450700 10
Fig. 6.7
Fig. 6.8
There are four main functions controlling the gate, these are: -
Gate Start	Gate Level or Threshold
Gate Width	Gate sense (Rising or Falling Signal)
62
The gate ‘start’ control positions the left hand edge of the gate, the first
depth that you want to start monitoring. The gate ‘width’ control then allows
you to set the right hand edge of the gate, the last depth that you want to
monitor. Any signal within that depth range is said to be ‘in the gate’. You
may only want signals exceeding a predetermined amplitude to ‘trigger’ the
gate alarm and you do this using the gate ‘level’ or ‘threshold’ control. For
those gates that look like figure 6.7, you set a signal in the gate at the desired
amplitude, and adjust the ‘threshold’ until the alarm just triggers. For those
gates that look like figure 6.8, you simply adjust the gate ‘level’ control until
the gate is at the desired screen height. For some inspections, such as when
you are using the ‘through transmission’ technique, you may wish to monitor
for a decrease in signal amplitude. The gate sense can be changed using
the ‘sense’ control. When ‘falling signal’ has been selected, the alarm does
not trigger as long as there is a signal in the gate that exceeds the threshold
level. Instead, the alarm operates as soon as the signal drops below the gate
threshold.
Some flaw detectors have more than one gate. Two gates can be used in
several ways; one can monitor backwall echo amplitude (falling signal) and
one can be used to monitor part of the timebase for discontinuities (rising
signal). The two gates can be used to monitor consecutive backwall echoes
and the difference (gate 1 minus gate 2) can be output as the thickness of
the object. The ‘menu’ of a digital flaw detector may allow you the choice to
monitor either signal amplitude or ‘time of flight’ (depth). This is also possible
with some analogue flaw detectors by the appropriate pin selection on the
output connecting lead. Generally, the voltage range for the output signal
is about OV to 5V; this means that the vertical or horizontal (amplitude or
timebase) scales of the display will be proportional to the output range. If
monitoring and recording amplitude, for example, a full screen echo height
will output 5V and a half screen height signal will output 2.5V.
63
CHAPTER 7
THE ULTRASONIC BEAM
The beam of sound waves emerging from an ultrasonic probe is rather like
the beam of light from a torch. The beam will spread out into an elongated
cone shape, and the further away you go from the source, the weaker will be
the beam. So in order to know just how this beam affects our inspection, we
need to study the shape of the beam in detail, and to study the changes in
intensity of the beam along its axis and across the beam.
As a general principle, we have said that the beam gets weaker as we
get further from the transducer. This weakening, or decrease in intensity
represents a loss of energy, we say that the beam is attenuated as it
progresses through a material. The sound beam suffers this attenuation for
the following reasons: -
ABSORPTION	- of the energy due to moving the vibrating molecules
SCATTER	- of sound waves reflecting from the grain boundaries
INTERFERENCE EFFECTS - close to the transducer
BEAM SPREAD	- the energy spreads over a larger area with distance
The amount of energy lost through ‘Absorption’ depends upon the elastic
properties of the material being tested so that steel and aluminium have less
absorption than lead, or Perspex. ‘Scatter’ also depends upon the material
being tested, the larger the grain size, the greater the scatter (see figure
7.1). Forged and rolled materials generally give less scatter than castings or
forgings. Heat treatment may reduce grain size and therefore reduce scatter,
making testing easier. Faced with a material that presents either, or both,
high absorption and scatter, you have to resort to a lower test frequency
to overcome the problem. We can either say attenuation (absorption and
scatter) decreases as test frequency decreases, or penetration increases,
as frequency decreases. This is a well-known fact - whoever heard of a ship
fitted with a high-pitched foghorn?
64
Fig. 7.1
INTERFERENCE EFFECTS
Point Source: - If we consider a point source of sound energy, then the
disturbance (sound wave) will radiate outwards from the point in an ever
increasing circle, just like the ripples on a pond spreading out when you drop
a stone into it. So sound radiates in all directions from a point source, (see
figure 7.2).
Sound wave expanding
outwards from point source
Positive peak wave front
Negative peak wave front
Finite Source: - Our transducer, however, is not a point source, but a plate
of piezoelectric material of finite dimensions. In order to appreciate the way
in which sound spreads out from a finite source, and to help us understand
interference effects we will use Huyghens principle, Huyghens said that you
can consider a finite source to be made up of an infinite number of point
65
sources. When you energise the transducer, sound will radiate out from each
of these point sources, just as it did for the stone dropping into the pond.
Figure 7.3 shows sound radiating from just one of these point sources and
figure 7.4 shows sound radiating from several point sources.
Fig. 7.3
Fig. 7.4
It can be seen from figure 7.4 that a short time (t,) after the finite source
has been energised, the disturbances from each of the point sources will
have moved outwards by the same amount. Along a line equal to the radius
of the small circles, running parallel to the face of the transducer, these
disturbances re-enforce each other to produce a wave front moving out from
the transducer. Notice also, a little energy ‘diffracts’ around the edge of the
transducer and is ‘lost’. A little while later (t2), sound from each point source
will have travelled a little further and reinforce at a new distance in front of the
transducer, thus the sound wave progresses from the source (figure 7.5).
Time ta
Fig. 7.5
66
This wavefront may represent the initial expansion of the transducer as it
starts to vibrate (a positive going half cycle). It will tend to push particles of
the material away from the source. Shortly afterwards, the transducer will
contract as part of its vibration, and a wavefront, drawing particles into the
source (a negative going half cycle) will follow on behind the initial wave
front, followed by another push, then another pull and so on.
In Chapter 3 (figure 3.3), we discussed refraction of the beam as an angled
incident wave meets an interface. The bottom edge of the beam reaches the
interface first and takes up the new velocity. We can use Huyghens principle
to explain what happens. As each point along the incident wavefront reaches
the interface, each in turn takes the new velocity and in the new material, the
line of initial wavefronts will determine the direction of the refracted beam.
Similarly, in Chapter 5 (figure 5.7), we discussed phased array probes. The
shape of the beam and beam angle will be determined by the wavefront
where there is individual wavefronts are in phase.
Now consider a point reflector ‘P’ just in front of the probe centre. Let us
consider how this reflector is affected by just three of the point sources, one
in the centre and two at the edges of the transducer (figure 7.6).
I* *	*1
• p
Fig. 7.6
We energise the source, and a split second later sound from the middle point
source reaches P, and gives it a push (figure 7.7). Notice that energy from
the edges of the probe has not reached P yet. This will take longer because
P is further from the edges than from the centre.
Fig. 7.7
67
By the time sound from the edges of the transducer reaches P (figure 7.8)
and tries to push P away from the transducer, the energy from the centre
may be on the opposite half cycle of vibration, and be pulling P back towards
the transducer. The resultant force acting on point P will be the vector sum
of the forces acting from all parts of the crystal. In our example, the result is
that P doesn’t move at all (i.e. the sound intensity=O). The distance between
the solid arc (positive peak) and the dotted arc (negative peak) is half a
wavelength. If a different frequency had been used, it may have been that
the second positive half cycle from the centre of the crystal reached point
P at the same time as those from the edges of the crystal. In that case, the
forces would have reinforced and point P would have been given an extra
hard ‘push’.
Fig. 7.8
When two solid arcs cross, the forces from those two parts of the crystal
are both ‘pushing’ at the intersecting point and when two dotted arcs cross
the forces from that part of the crystal are both ‘pulling’ at the intersecting
point. In both cases we call the effect ‘constructive interference’. When a
solid arc cuts a dotted arc, the forces are in opposition and we call the effect
‘destructive interference’. Of course point P will not always be exactly
a multiple of half wavelengths away from the center and the edges, and
constructive interference happens when the relevant point sits anywhere
within the same half cycle. Destructive interference happens when the
relevant point is in dissimilar half cycles.
‘Interference’ occurs whenever energy arrives at different phase
(wavelength) intervals at a particular point. Whether the interference is
constructive, or destructive, is determined by the path difference between P
68
and the centre, and P and the edges. As P gets further away from the front
of the transducer, this path difference becomes negligible compared to the
wavelength (figure 7.9) and interference problems become insignificant.
Next half cycle
Initial wavefronts
Fig. 7.9
Variations in intensity due to interference effects occur for some distance in
front of the transducer, as we have just seen. This region is known as the
‘Near Field’ and the extent of the near field, known as the near field distance
can be calculated from: -
Where,
NF = Near Field Distance.
Crystal Diameter.
Wavelength
Example 11
Calculate the Near Field distance for a 10mm diameter, 5MHz crystal
transmitting into steel (Velocity 5960m/sec. .-. X = 1.192mm).
NF = 21mm (Approx.)
69
This means that for this probe, in steel, we can expect fluctuations in intensity
of sound for the first 21mm of steel depth due to interference effects. As a
result, it is unwise to rely solely on amplitude as the criterion for acceptance
or rejection of the part for discontinuities that are in the near field region.
The last item on our list of factors affecting attenuation of the sound as it
travels through a material is the ‘Beam Spread’. Because the beam spreads
out into a conical shape, intensity follows the inverse square law just as it
would for a beam of light or X-rays. If you double the distance from the probe,
the intensity drops to one quarter of its original value because of beam
spread. Of course, it will actually fall to less than a quarter, because we have
to add any absorption, scatter losses to the beam spread losses. We can
now plot a graph of intensity against distance from the probe, to summarise
the previous discussions. Figure 7.10 show amplitude on the vertical axis
and distance on the horizontal axis. Distance is shown in multiples of the
near field distance.
70
Fig. 7.11
The beam profile shown in figure 7.11 is very much a ‘theoretical’ beam
spread. Alongside there are three ‘slices’ through the beam showing that the
highest sound intensity is in the centre of the beam. The sound gradually
fades away towards the edge of the beam until there is no sound left. It is
often more convenient to define the beam to a theoretical edge where the
intensity of sound has fallen to one half (-6dB), or sometimes one tenth (-
20dB) of the intensity at the beam centre. We can consider three theoretical
edges; one defining the absolute edge of the beam, another defining the
6dB edge and the third defining the 20dB edge. These three edges can be
expressed mathematically: -
o. 0 1.22X
Sin— =------
2 D
Defines the absolute edge
71
0 0.56k
Sin— =—-—
Defines the 6dB edge
o- 6
Sin — =
2
1.08k
D
Defines the 20dB edge
It is often convenient to use the theoretical beam shape shown in figure 58
in order to explain some concepts in ultrasonic flaw detection. However it is
not good practice to use a calculated beam shape for sizing discontinuities
by one of the intensity drop methods. This is because practical beam shapes
seldom match the theoretical model closely enough. We will see later how to
plot a practical beam profile for each of our probes.
Example 12
Calculate the 20dB beam spread angle for a 5MHz compression wave in
steel from a 10mm diameter crystal.
0 1.08k
Sin—-------
2 D
0 1.08x1.192
Sin — =---------
2	10
0 1 28736
Sin— =--------
2	10
Sin— = 0.128736
2
— = 7.4°
2
0 =2x7.4°
= 14.8°
72
We have used three terms connected with the beam of sound in the test
material, namely ‘Dead Zone’, ‘Near Field’, and ‘Far Field’. The dead zone
is that part of the timebase occupied by the initial pulse when using a single
crystal contact probe. The near field is the distance in the material that suffers
from interference effects and the far field is the rest of the beam beyond the
near field. The trace shown in figure 7.12 is calibrated for 100mm of steel
return path using a single crystal 5MHz compression wave probe. The three
zones are shown on the trace.
Fig. 7.12
73
CHAPTER 8
CALIBRATION & REFERENCE STANDARDS
During practical sessions using the twenty basic exercises, it will become
apparent that neither the vertical nor the horizontal scales of the display have
any absolute meaning of themselves. The horizontal scale can be adjusted
to represent a great variety of different time intervals, and these, for a given
material and velocity, can be translated into depth values. The vertical scale
gives an indication of the amplitude of signal being detected, provided you
know how much ‘Gain’ you are using, but it does not necessarily tell you
much about the size of defect causing that reflection. The safest way to get
more information about the specimen from the display is to compare signals
from the specimen with those from specially machined blocks. These blocks
we normally classify under one of two headings, depending on the function
of the block.
The term ‘Calibration Block’ is defined in British Standard BS 2704 as:
- “A piece of material of specified composition, heat treatment, geometric
form, and surface finish, by means of which ultrasonic equipment can be
assessed and calibrated for the examination of material of the same general
composition.” Therefore, a calibration block may be a simple step wedge
in a particular material to allow the timebase to be calibrated for accurate
thickness measurement, or it may be a more complex block like the A.2
block described in BS 2704 which allows calibration of timebase, plus
calibration of probe index, angle, resolution etc.
The second heading, ‘Reference Block’ is defined in BS2704 as: - “An aid
to interpretation in the form of a test piece of the same material, significant
dimensions and shape as a particular object under examination, but not
necessarily containing natural or artificial defects”. So, for example, a
section of an aircraft wing forging may be prepared as a reference block so
that a technician may become familiar with the standard signal patterns from
the various changes in section and more easily recognise a defect quickly
74
when examining the component on an aircraft. More usually, the block would
contain artificial defects from which the sensitivity (gain) used in the test
could be set.
CALIBRATION BLOCKS
The BS 2704 A.2 Calibration Block, also known as the International Institute
of Welding (LLW.) block, or ‘V1 block’, is illustrated in Fig. 8.1 .The block can
be used for the following assessments: -
- Calibration of the timebase in terms of thickness.
- Assessment of Dead Zone.
- Checking linearity of the timebase.
- Checking linearity of the Amplifier.
-	Assessing overall sensitivity of probe and amplifier.
-	Determination of the angle of refraction.
-	Determination of Beam Characteristics.
Checking Resolution.
Determination of probe index
Finding the correct Zero Point.
The A.2 block was derived from the original ‘Dutch Block’ designed by RTD
Rotterdam and accepted by 11W as the ‘I IW V1 Block’. In its original form, the
deep slot at the center of the 100mm radius was a scribed line and a 25mm
radius slot was positioned as shown in figure 8.2. This design is still used
in some parts of the world, and has the advantage that shear waves can be
calibrated for ranges other than multiples of 100mm. In all other respects it
is the same as the A.2 block.
75
The BS 2704 A.4 Calibration block, also known as the ‘V2 block’, is a more
compact form of the ‘V1 block’ suitable for site use, although somewhat less
versatile in its functions. Figure 8.3 illustrates the A.4 block.
Fig. 8.3
The Institute of Welding (I.O.W.) Beam Profile calibration block is designed
primarily for beam profile measurement and has four 1.5mm diameter side
drilled holes giving eight depths from two scanning surfaces. These can be
examined by direct scan for probes of various angles, and at several more
ranges for each probe, using indirect scans by reflecting from the far surface.
There are two series of five holes on an inclined axis to measure shear
wave probe resolution and to simulate an inclined discontinuity. The block is
illustrated in figure 8.4.
76
Fig. 8.4
REFERENCE BLOCKS
Area / Distance reference blocks are mainly used for setting sensitivity levels
and accept/ reject levels for defect sizes by reference to echo height. Blocks
are produced in a range of scanning depths and each set of blocks contains
the same diameter flat-bottomed hole in each block. There are three sets
of blocks, a set with 3/64” diameter flat-bottomed holes, a set with 5/64”
diameter holes and one with 8/64” diameter holes. The scanning depths can
range from 1/г” to 22”, but at shop floor level, you would only have the few
blocks appropriate to your range of work. Figure 8.5 shows a typical block, in
this case a 3 x 5 block (3” scan depth, 5/64” flat bottomed hole).
A = Scan depth
В = Hole diameter
Fig. 8.5
77
Distance/Amplitude Correction (DAC) reference blocks are made from the
same thickness and grade of material as the work piece. They contain an
artificial flaw (a side-drilled hole). The change of echo height with changes
in scanning distance (multiple skips) is noted and plotted on the display as a
“DAC” curve so that a signal amplitude can be specified to cover all depths
within the working range for reporting, acceptance, or rejection purposes.
Figure 8.6 shows a typical ASME DAC block and figure 8.7 shows a DAC
curve.
Fig. 8.6
Fig. 8.7
78
CHAPTER 9
COMPRESSION WAVE TECHNIQUES
CALIBRATION OF TIMEBASE
The important thing to remember when calibrating the timebase for
compression waves is that the left hand end of the timebase (Zero) must
exactly correspond to the entry surface of the beam and the right hand end
represents a known thickness in the material being tested. The exceptions
to this rule are those occasions when you are using delay to expand some
distant portion of the material, or when you are using a multiple echo
technique and only noting the decay pattern. For single crystal probes, the
initial pulse contains two elements; the applied square wave voltage pulse
and the ringing of the crystal. The top surface is represented by the end of
the applied voltage pulse where the crystal ringing starts. Unfortunately, the
amplitude of this part of the initial pulse is so large that it is not possible to
identify the point at which the ringing starts, nor is it possible to tell from the
timebase line.
There is a similar problem when we calibrate using a twin crystal probe
because the initial pulse is at the start of the Perspex delay line and the sound
enters the work piece sometime later. In any case, because the amplifier is
deliberately isolated from the transmitter crystal, there is no signal to mark
the entry surface. Our calibration procedure, whether for a single or twin
crystal probe must find some other way to identify the true zero. The most
common way to achieve this is to use two echoes that are a known distance
apart, to set one at timebase zero and the other at the right hand end (10) of
the timebase. We do this in the following way: -
-	Use the delay control to position the first backwall echo from thedesired
range on our calibration block to zero.
-	Then usethedepth rangecontrolstoposition thesecond backwall echoto
10 on the timebase.
-	This may also move the echo from the zero position and so we need
79
to check and adjust this with delay again.
-	These two controls are used alternately until the two echoes are exactly
on 0 and 10. We now know that the timebase is exactly equal to the
calibration thickness
-	Once we are happy that we have that exact range on the timebase,
we lock the depth controls.
-	We then use delay to move the first backwall echo to the right until
it is exactly on 10. If the timebase is exactly a known range
and the first backwall echo is on 10; then zero must coincide with the
entry surface.
Figures 9.1 to 9.3 illustrate the procedure for calibrating the timebase for
100mm of steel on the A2 calibration block.
IstBWE Delay i	i	i	1	1	2nd BWE Depth 	► 	114	1
i	l	l	l	l	I	I	1	1	1
im 1				It	tilnu.Luiliiit iihIiiii	uttliHi tinliut iitilitiL itnliHi tiiilnii
01	23456789	10
Fig. 9.2
80
In this example we know that the first and second backwall echoes represent
100mm and 200mm of steel path time because the A2 block is 100mm
wide at this probe position. Therefore figure 9.2 represents exactly 100mm
timebase. This timebase remains constant as long as we do not alter the
depth control, so figure 9.3 represents zero to 100mm exactly.
CHOICE OF COMPRESSION WAVE PROBES
TWIN CRYSTAL PROBES
For conventional techniques twin crystal probes are generally used on
thicknesses below 50mm. They are also in general use for high temperature
thickness measurement, where a thermal insulating material is used instead
of Perspex, to protect the crystals.
SINGLE CRYSTAL PROBES
Single crystal probes are generally used on thicknesses in excess of 50mm.
They are also used below 50mm if resolution is an important factor, since
single crystal probes usually have shorter pulse lengths than twin crystal
probes. However, for conventional techniques they can only be used when
the transmission noise does not encroach upon the useful part of the
timebase for that job. As a guide, you can expect the shortest transmission
noise from high frequency, heavily damped probes.
81
PROBES FOR MULTIPLE ECHO TECHNIQUE
These are usually single crystal probes, although in some cases twin crystals
can be used. When dealing with thin walled material it is possible to get
resonance and anti-resonance conditions that will affect the decay pattern
and may give false indications. This can be avoided if you choose a probe
frequency such that the plate thickness is more than 1.5 x the wavelength of
the compression wave in the specimen material, and a pulse length that is
not more than 3 times the wavelength.
THICKNESS MEASUREMENT
One of the most important uses of ultrasonics is that of thickness
measurement. It is particularly useful because it is relatively quick, simple
and accurate, and access to only one surface of the specimen is required.
There are many types of equipment and techniques made exclusively for
thickness measurement. It is not intended to deal with all of them here. We
will only discuss the use of the pulse echo system with an А-scan display.
A-SCAN RECTIFIED DISPLAY
This is the most common display presentation for ultrasonic flaw detection
equipment. In chapter 8 we described the display for an unrectified trace and
various types of rectification.
a)	CALIBRATION
The basic calibration of the timebase should be carried out to ensure proper
positioning of the zero and backwall echo as described above. The calibration
block should be made of the same material as the work piece but corrections
for velocity can be made if the correct test block is not available provided that
the velocities in the material of both the work piece and the actual test block
used is known. The correction procedure is described below.
For best results the range chosen for calibration should be the shortest
range which allows the first back wall echo to be displayed. For example, if
the nominal wall thickness of the work piece is 9mm and your flaw detector
82
is capable of showing 10mm across the full graticule, then the 10mm range
should be used. Since the graticule of most flaw detectors can be divided
into 100 small units it follows that a timebase calibrated such that those 100
units represent 10mm gives you a reading accuracy of 0.1mm per division.
If on the other hand you calibrate such that 100 units represent 25mm, the
reading accuracy is 0.25mm per division.
b)	AMPLITUDE (GAIN SETTING)
The amplitude of the calibration echo and the amplitudes of thickness echoes
made on the work piece should be adjusted to the same predetermined
amplitude. This is normally between 1/3 and 1/2 full screen heights.
c)	READING THE THICKNESS (SINGLE ECHO)
The specimen thickness is determined from the left hand edge of the backwall
echo. This is normally a steep sloping line. If a small half cycle appears at
the left hand edge of the signal that was not present during calibration, this
may be removed by inserting a small amount of suppression or by choosing
‘positive’ or ‘negative’ rectification. (See figure 9.4).
Extra half cycle
After suppression or
rectification change
Fig. 9.4
83
d)	READING THE THICKNESS (MULTIPLE ECHOES)
If the specimen thickness and calibrated range are such that multiple echoes
are produced, the most accurate result can be obtained by reading the
thickness of the last multiple echo displayed and dividing the answer by the
number of backwall echoes. In the example shown in figure 9.5, the fifth
backwall echo shows at 22 mm. so the true thickness is 22 divided by 5 = 4.4
mm. In this case, a single crystal probe has been used and the initial pulse
is obscuring the start of the first backwall echo.
Sometimes the initial pulse obscures the entire first backwall echo and
maybe all or part of other back echoes. Figure 9.6 shows the same thickness
but with the first two back echoes obscured. Care must be taken to assess
the number of echoes that have been obscured.
Fig. 9.6
84
e)	VELOCITY CORRECTION
Supposing you are asked to measure the thickness of an aluminium alloy
forging of nominal thickness 20 mm (velocity 6400m/s), and you only have
a steel I.I.W. block for calibration (velocity 5960m/s). The procedure would
be as follows: -
(i)	Calibrate for 25 mm. of steel on the I.I.W. block.
(ii)	Measure thickness of aluminium specimen as if it were steel.
(Suppose in our example that the thickness indicated is 18.5mm).
(iii)	Correct for velocity by the following calculation: -
Indicated thickness x Compression wave velocity in work piece
T _	Compression wave velocity in calibration block
T_ 18 5x6400
5960
T - 19.866mm.
f)	USE OF TIMEBASE DELAY
Apart from its use to correct for Perspex path distance in twin crystal
compression wave probes, “Delay” can be used as an aid to more accurate
thickness measurement. For example, you may want to accurately measure
the thickness of a component whose nominal thickness is 80 mm. If you
calibrate the timebase so that 100 scale divisions represents 100 mm of
that material, each small division represents 1 mm. If instead, you calibrate
the timebase so that 100 scale divisions represent 25 mm of the test
material each division on the scale represents 0.25 mm instead of 1 mm.
The delay control can then be adjusted so that the third backwall echo
from the calibration block is set at O, and the fourth backwall echo at 100
scale divisions. The timebase would represent a thickness range of 75mm
85
to 100mm. The first back echo from the work piece (80mm) will appear at
approximately 1 /5th of the timebase range.
An even more accurate result would be obtained if you calibrated the
timebase for 150mm to 175mm to display the second back echo from the
work piece at about 160mm and then divide your answer by 2. The accuracy
of this reading would be 0.125 mm. It is important when using this technique
that you first check the nominal thickness by calibrating the timebase for
100mm and ensuring that the first back echo is at about 80mm.
A-SCAN UNRECTIFIED DISPLAY.
There are occasions in thickness measurement, particularly if the scanning
surface is rough, when a lot of unwanted signals, “noise “ or “grass” appear
on the CRT and make it difficult to determine the point at which the back
echo starts. If the ultrasonic set allows an unrectified trace to be selected,
then measurements can be made using the tip of a particular down going half
cycle instead of using the point at which the signal first leaves the timebase.
a)	Firstly, let us identify our measuring point. Figure 9.7 shows a back
echo from the 25mm range on the V1 block with the timebase calibrate
for 50mm using the conventional rectified display. The presentation
has then been changed to ‘unrectified’ and the vertical or ‘Y’ shift used
to raise the timebase to a level between 1/3 and 175th full screen
height. Gain has been adjusted so that the peak of the longest down
going half cycle just meets the graticule. In this case, it is the second
half cycle that is the longest, and we will use this half cycle as our
measuring point. (Note that sometimes a back echo from the work
piece may show the 1st or 3rd half cycle as the longest - despite this,
if you calibrate on the second half cycle you then always measure
from the second half cycle even if it is not still the longest,)
86
b)	Having identified the half cycle that you are going to use, you calibrate
the timebase so that this part of the signal coincides with the correct point
of the graticule. In the case shown in figure 9.7, if we wished to calibrate
for 50 mm we would use “delay” to move the second half cycle from 5.15
to 5.0 divisions and check that the second half cycle of the second back
echo coincides with 10.0 divisions (see figure 9.8).
c)	Calibration for other timebase ranges would be done in the conventional
way but using the second half cycle instead of the left hand edge of the
pulse, as your measuring point.
87
LAMINATION TESTING
Lamination testing of plates and pipes that are to be welded or machined is a
very common NDT task. It is also a simple application of compression waves
in ultrasonic flaw detection, but one that might give some problems when
examining thinner samples.
STANDARD PROCEDURE
a)	Calibrate the timebase to allow at least two backwall echoes to be
displayed.
b)	Place probe on the work piece and adjust the gain controls so that the
second backwall echo is at full screen height.
c)	Scan the work piece looking for lamination indications that will generally
show up at half specimen thickness together with a reduction in back
echo amplitude. In some cases, a reduction in the amplitude of the
second back echo may be noticed without a lamination signal being
not due to poor couplant or surface conditions.
MULTIPLE ECHO TECHNIQUE
Lamination testing of plate or pipe less than 10 mm. in wall thickness may
be difficult using the standard procedure because multiple echoes are so
close together that it becomes impossible to pick out lamination signals
between backwall echoes. In such cases, we can use a technique called the
“multiple echo technique” using a single crystal compression wave probe.
The procedure is as follows: -
a)	Place the probe on a lamination free portion of the work or calibration
piece.
b)	Adjust the timebase and gain controls to obtain a considerable number
of multiple echoes in a decay pattern over the first half of the time base.
Atypical example is shown in figure 9.9.
88
Uniaminated plate
Fig. 9.9
c)	Scan the work piece, the presence of laminations will be indicated by
a collapse of the decay pattern such as the one shown in figure 9.10. The
collapse occurs because each of the many multiple echoes is closer to its
neighbour in the presence of a lamination.
Fig. 9.10
EXAMINATION OF BRAZED AND BONDED JOINTS
Compression waves can also be used for the detection of areas of non-
adhesion in brazed or bonded (glued) joints,
a)	BRAZED JOINTS
If the wall thicknesses permit clear separation between back wall echoes,
brazed joints can be examined using the standard procedure for lamination
89
checking. However, since the braze metal separating the two brazed walls
will have a slightly different acoustic impedance to that of the parent metal,
a small interface echo may be present for a good braze. The technique,
therefore, is to look for an increase in this interface echo amplitude. See
figure 9.11. If the two brazed walls are too thin to permit clear back echoes,
a multiple echo technique can be used as described above.
Good braze
Brazed joint
Fig. 9.11
b)	BONDED JOINTS
These may include metal-to-metal glued joints and metal to non-metal glued
joints (e.g. rubber blocks bonded to steel plates). The technique used is a
multiple echo technique. Each time the pulse reaches a bonded interface; a
portion of the energy will be transmitted into the bonded layer and absorbed.
Each time a pulse reaches an unbonded layer, all the energy will be reflected. If
we look at the multiple echo pattern for a good bond, the decay will be relatively
short because of the energy loss at each multiple echo into the bond. However,
for an unbonded layer each multiple echo will be slightly bigger because
there is no interface loss, and the decay pattern will be significantly ;onger.
Figure 9.12 shows a good bond (position 1) and poor bond (position 2).
Rubber
Decay pattern for good bond
Fig. 9.12
90
CHAPTER 10
SHEAR WAVE TECHNIQUES
Shear waves at various angles of refraction between 35° and 80° are used to
locate defects whose orientation is not suitable for detection by compression
wave techniques. Some defects, of course, have volume and their shape
enables them to be detected by both compression and angled shear waves.
In this chapter, however, we will be dealing with planar defects whose
orientation is such that only angled shear waves can be used.
Because the beam is travelling through the test piece at a refracted angle
other than perpendicular, we need to distinguish between the beam path
length to a discontinuity and its depth below the test surface. When we
encounter a signal, we can measure the beam path length (range) from
the timebase, but we may want to calculate how far in front of the probe
(horizontal distance) and how far below the surface the reflector is located.
It is also important when using shear waves to know where along your
probe the beam enters the specimen (beam index). Knowing the beam
index position relative to some datum on the specimen, and the exact
beam angle allows you to calculate the horizontal and vertical distances.
There are standard terms for various distances when using shear waves
and these are illustrated in figure 10.1.
FSD = Full skip distance
HSD = Half skip distance
9	=	Beam angle
AB = Half skip BPL
ABCD= Beam path length
Fig. 10.1
91
Full skip and half skip distances are measured along the top surface and
beam path length (BPL), along the beam centre. To calculate these, knowing
specimen thickness (t) and probe angle (9) use the following formulae: -
a)	HALF SKIP DISTANCE = t x Tant)
b)	FULL SKIP DISTANCE = 2 xtxTanO
c)	HALF SKIP BPL = ——
'	Cose
d)	FULL SKIP BPL =
Cose
If the probe angle is exactly equal to the nominal angle i.e., if your 60° probe
really is 60°, not 59° or 62°, you can calculate these distances more easily
from the following formula: -
Distance required = t x F
Where F is the appropriate factor from table 2, below.
Probe angle (0)	35°	45°	60°	70°	80°
Half skip distance (HSD) factor	0.7	1.0	1.73	2.75	5.67
Full skip Distance (FSD) factor	1.4	2.0	3.46	5.49	11.34
HS Beam path length (BPL) factor	1.22	1.41	2.0	2.92	5.76
FS BPL factor	2.44	2.83	4.0	5.85	11.52
Table 2
92
Example 13
Calculate the Full skip distance for a 40° shear wave beam in a 20mm thick
steel plate.
FSD = 2xtxTanQ
FSD = 2x20xTan40
FSD = 2x20x0 8391
FSD = 33.564mm
Example 14
Calculate the Half skip beam path length for a 45° shear wave beam in a
20mm thick steel plate.
HSBPL 20x141
HSBPL = 28.2mm
ESTABLISHING THE TRUE PROBE BEAM INDEX
We need to find the exact beam index for any shear wave probe before
measuring the beam angle. This is because the beam index may not be the
one marked on a new probe - it may be a millimetre or so before or after
the marked index. Manufacturers use a standard drawing to make probes,
but the velocity of sound in Perspex varies from batch to batch, and with
temperature. Also the beam index and probe angle change as the probe
wedge or ‘shoe’ becomes worn with use. So, the establishing of beam index
and angle will be routine checks throughout the life of the probe. Finding the
beam index is a simple procedure carried out on the A2 or A4 calibration block.
The probe is positioned close to the edge of the calibration block and
beaming towards the 100mm radius (A2 block) or the 25mm or 50mm radius
(A4 block) as shown in figure 10.2 (A2 block) and figure 10.3 (A4 block).
93
The probe is moved backwards and forwards about the centre mark of the
radius with the probe kept parallel to the edge of the block. As the probe
moves, the signal will rise to a maximum and then fall again as shown in
figure 10.4. When the signal reaches the maximum amplitude, the beam
centre is meeting the tangent to the radius at right angles. This happens
when the beam centre is entering the block at the centre of the radius. The
true beam index is now in line with the centre mark of the 100mm radius. If
this does not coincide with the beam index marked on the probe, you would
then either mark the true index on the probe body, or, if the probe body has
a millimetre scale, make a note of the true position in front or behind the
marked index.
Fig. 10.4
94
MEASURING THE TRUE BEAM ANGLE
Once the true beam index is known, the true beam angle can be measured
on the A2 or A4 calibration block. The nominal probe angle is marked on the
probe and is the refracted angle for steel unless identified for another material.
A 45° shear wave probe made for aluminium would be marked ‘45AL’ and for
copper ‘45CLT. The actual angle for a new probe may be plus or minus two
degrees from the nominal angle because of the batch velocity variations in
Perspex, and will change with wear. Most of us have an inherent tendency to
wear the probe in a particular way, just as we do for shoes. We may wear the
heel of the probe down and so increase the actual angle, or wear the toe and
decrease the actual angle. For this reason the beam angle measurement is
also a routine probe check. If the probe is worn down towards one edge, the
beam will be thrown off towards that side and the condition is called ‘squint’
- we will look at how to assess squint in a later chapter.
Beam angle is measured on the A2 block by aiming the beam at the 45mm
diameter hole and on the A4 block at the small hole. The probe is positioned
on the block at a point near the nominal angle and the gain adjusted to give
a signal amplitude of about 50% full screen height. As the probe is moved
forward and back, the signal rises and falls just as it did when finding the
beam index. When the signal reaches its maximum amplitude, the beam
centre is aimed at the centre of the hole and the beam is hitting the tangent
to the hole at right angles. The true beam angle can be read against the
true beam index from the graticule on the calibration block. The example
shown in figure 10.5 has the beam index opposite an angle of about 43° and
the nominal angle is 45°. With this probe, we would have to use 43° in our
distance calculations and for defect sizing.
95
45
CALIBRATION OF TIMEBASE
The method of calibration of the timebase for shear waves depends on the
purpose of the inspection. If the inspection were to be volumetric, looking
for any discontinuities within the scanned volume of the test piece, then we
would calibrate for a suitable timebase range at shear wave velocity. On the
other hand, if the purpose is to look for a specific discontinuity such as a
fatigue crack, in a predicted location, we may well use a ‘Skip’ method or a
‘Reference Block’ method. The calibration for a known range will be dealt
with first, using the A2 block and then the A4 block.
USING THE A2 BLOCK
-	Place the probe on the A2 block as shown in figure 10.6.
-	Obtain a maximum echo from the 100mm radius.
-	Adjust the gain control to peak the signal at about 80% full screen
height.
-	Use the delay control to position the 100mm signal at zero on the
timebase.
-	Use the depth controls to place the second reflection from the
100mm radius at ten on the timebase.
-	Check that the left hand edges of the two signals are exactly at zero
and ten.
96
-	Lock the depth controls
-	Use delay to move the first signal from zero to ten.
-	The time base is now calibrated for 100mm at shear wave velocity,
and zero represents the top surface entry point below the beam index.
Fig. 10.6
Sometimes you may see part of the initial (transmission) pulse around the
zero, this will depend on the pulse length and gain setting as shown in figure
10.7.
Fig. 10.7
97
The slot that marks the 100mm radius on the A2 block is about 4mm deep so
that when the probe is aligned with the edge of the block, the slot makes a
corner for the returning echo to reflect part of the energy back to the 100mm
radius. This is why it is possible to obtain repeat echoes from the radius. If
the slot were not there, the reflected energy from the first returned signal
would reflect to the rear of the probe. Figure 10.8 shows an exaggerated
view of the beam path to illustrate the ‘corner’ effect.
Fig. 10.8
USING THE IIW BLOCK
-	Place the probe on the IIW block as shown in figure 10.9.
-	Obtain a maximum echo from the 100mm radius.
-	Adjust the gain control to peak the signal at about 80% full screen
height.
-	Use the delay control to position the 100mm signal at zero on the
timebase.
-	Use the depth controls to place the reflection from the 25mm radius
slot at ten on the timebase.
-	Check that the left hand edges of the two signals are exactly at zero
and ten.
98
-	Lock the depth controls
-	Use delay to move the first signal from zero to eight on the timebase.
-	The time base is now calibrated for 125mm at shear wave velocity,
and zero represents the top surface entry point below the beam index.
Fig. 10.9
The sound path in figure 10.9 shows the first return echo produces the
100mm signal and the reflected portion at the probe interface heads towards
the 25mm radius where it reflects back to the probe. However, the reflection
reaching the probe is not directed at the crystal and so there is no signal at
this time. Instead, the reflection goes back towards the 100mm radius where
it reflects again and this time reaches the crystal to make the second echo.
The total return path distance travelled in the calibration block by the time we
see the second echo is: -
100mm (first pass) + 25mm + 100mm (second pass) - 225mm
We move the first echo to eight on the timebase because the timebase is
locked at 125mm range and the first echo is from 100mm range. If the probe
is now turned around to face the 25mm radius we should obtain a signal at
2.0 on the timebase (one fifth of full scale).
USING THE A4 BLOCK
To calibrate for 100mm using the A4 block: -
-	Place the probe on the A4 block as shown in figure 10.10.
99
-	Obtain a maximum echo from the 25mm radius.
-	Adjust the gain control to peak the signal at about 80% full screen
height.
-	Use the delay control to position the 25mm signal at 2.5 on the
timebase.
-	Use the depth controls to place the second reflection (from the 50mm
radius slot) at ten on the timebase.
-	Check that the left hand edges of the two signals are exactly at 2.5
and 10.
-	Lock the depth controls
-	The time base is now calibrated for 100mm at shear wave velocity,
and zero represents the top surface entry point below the beam index.
01	23456789	10
Fig. 10.10
The sound path in figure 10.10 shows the first echo from the 25mm radius
and then the echo from the 50mm radius after reflecting at the scanning
surface down to the 25mm radius and back. The total return path is: -
25mm + 50mm + 25mm = 100mm.
Facing the 25mm radius on the A4 block, signals will arrive at 25, 100, 175,
250mm and so on, incrementing by 75mm each time. If the probe is turned
around to face the 50mm radius, signals will arrive at 50, 125, 200, 275mm
and so on, again incrementing by 75mm.
100
CALIBRATION USING THE SKIP’ METHOD
If the purpose of the inspection is to detect surface breaking flaws at the
bottom surface or top surface, we know that the echoes will arrive at exactly
the half skip or full skip beam path lengths. We could calibrate the timebase
for an exact range using one of the methods described above and calculate
the beam path lengths for half and full skip using the formulae. We would
then know exactly where to look on the timebase for the two conditions. We
do use this method to carry out the critical root scan in weld inspection.
However, in many cases there is a quicker and simpler method. Using a
piece of plate of the same wall thickness as the item to be inspected we can
point the probe at the end surface (position 1) and scan back as shown in
figure 10.11 until we see the echo from the bottom corner (position 2).
Scan
Fig. 10.11
The signal will rise to a maximum as the centre of the beam moves into
the corner. We can adjust the timebase and gain to make sure that we can
see that maximum point. As the maximum is reached, we would adjust the
timebase range to position the signal at some convenient part of the trace,
usually about ’4’. We would then continue moving the probe backwards until
the top corner reflection is seen (position 3). As this signal maximises, we
note its position along the timebase. Figure 10.12 shows a trace with the half
skip and full skip positions marked, and in this example, gates positioned
over the two critical locations so that the operator can listen for the alarm
rather than watch the display all the time.
101
Another point to note from figure 10.12 is that the position for full skip is at
‘9’ on the timebase and not at ‘8’ (twice ‘4’). This means that timebase zero
is not the top surface, and furthermore, we don’t know the exact timebase
range. However, for this inspection it doesn’t matter because we are only
interested to find out whether or not there is a bottom or top surface breaking
‘corner’.
Fig. 10.12
If the plate to be inspected has accessible edges, you don’t need a calibration
plate because you can use the corners on the test piece to set up the two
positions on the timebase. However, you have to be sure that there are no
laminations in the beam path because these might reflect the beam back to
the top without reaching the bottom. It is easy to check whether the signal
is coming from the anticipated corner because a shear wave meeting an
interface at an oblique angle is easy to ‘damp’. If you put an oily finger on
the expected reflecting corner, the signal will be seen to reduce in amplitude
significantly. In figure 10.11, if you ‘damp’ the bottom corner when the beam
is at the half skip position (2), the signal will fall and when the beam is at the
full skip position (3) you can damp the signal at the top corner and at the
reflecting point on the bottom surface.
102
PIPE WALLS
If you are going to scan a pipe wall in the longitudinal direction, then you can
use any of the above calibration procedures. However, if you are scanning
circumferentially the calculation of beam path length, and skip distances
is more complicated. If you have a segment of pipe of the same outside
diameter and wall thickness as a reference block, you can use the ‘skip’
method for finding the critical half and full skip positions on the timebase.
If you also need to look for discontinuities in the volume of the object, you
calibrate the timebase on the A2 or A4 block for an exact range, and then put
the probe on the ‘reference’ pipe segment and note the half and full skip ranges.
The wall thickness for any given outside diameter is important because
the normal range of angled shear wave probes (45°, 60°,and 70°), when
used on thick wall pipe may cut across to the outside surface again without
touching the bore. An example is shown in figure 10.13 where a 45° shear
wave only reaches about half way through the wall. In other words, for this
outside diameter, the thickest pipe wall we could test with a 45° probe is only
half that shown in the diagram. It follows that, when you are presented with
an unusually thick pipe wall for a particular outside diameter, you need to
choose your probe angle carefully in order to inspect the bore properly. For a
given angle, the maximum wall thickness that allows the centre of the beam
to reach the bore of the pipe can be calculated from: -
_ c/x(1 - Sine )
~	2
Where: -
t = Maximum possible wall thickness
d = Pipe outside diameter
0 = Probe angle
103
Fig. 10.13
Thickest wall that can be
tested with this 45° probe
This formula can be turned round so that you can calculate the best angle
given for a wall thickness, the formula becoming: -
Sine =1-—
d
Example 15
Calculate the beam angle that just grazes the bore of a 100mm outside
diameter pipe having a 20 mm wall thickness.
Sine =1-—
d
40
Sine =1-—
100
sine =1-0.4
Sine =o.6
= 37°
104
\Ne would choose the nearest standard angle for a probe, which is 35° for this
size of pipe. Since shear wave probes at angles below 35° are not available
because of the confusion that arises from the spurious compression wave,
you will see that there is a maximum wall thickness for any given outside
diameter that can be tested by the half skip technique. The maximum wall
thickness is where the ratio of OD to ID exceeds 4.5 to 1.
For convenience, the formula t = - - ~ has been modified t = dxf ,
2
where f is a factor for standard angle probes that has been calculated from:
--------. The factors are shown in table 3 below.
2
Probe angle (0)	35°	45°	60°	70°	80°
Probe factor (f)	0.213	0.146	0.067	0.030	0.0076
Table 3
Table 4 shows maximum wall thickness that can be tested for three standard
angles and a range of pipe diameters.
Pipe OD	Maximum pipe wall thickness for probe angles		
	35°	45°	60°
4” (100mm)	21mm	14mm	7mm
6” (150mm)	32mm	22mm	10mm
8” (200mm)	43mm	29mm	13mm
10” (250mm)	53mm	36mm	17mm
12” (300mm)	64mm	44mm	20mm
14” (350mm)	74mm	51 mm	23mm
16” (400mm)	85mm	58mm	27mm
18” (450mm)	96mm	66mm	30mm
20” (500mm)	106mm	73mm	33mm
Table 4
105
Once the correct angle for the pipe size and wall thickness has been chosen,
you can establish the skip and half skip positions using a section of pipe with a
drilled hole to produce the reguired ‘corner’ reflectors - as shown in figure 10.14.
Fig. 10.14
CALIBRATION OF THE GAIN
This is often called “setting the sensitivity”, and it means that we adjust the
gain so that a significant discontinuity will give a signal that is large enough
to see, but small surface scratches will not. Very often, we use a reference
block, similar in shape and material to the specimen, and containing either
a drilled hole or an artificial (machined) crack. The probe is aimed at this
reference hole or crack, to obtain an echo, this is then maximised by probe
movement, and then, the gain is adjusted to give the required signal height
known as the ‘reference level’ and the gain is then said to be ‘calibrated’. This
reference level may be 50% or 75% of full screen height, and is often used
as the basis for getting acceptance standards for the inspection. Hence, you
may find that you are working to a specification that says that any signal
equal to, or greater than the amplitude of the reference level is cause for
rejection of the component, whereas any signal lower than the reference
level may be ignored.
TESTING FOR OUTSIDE DIAMETER SURFACE FLAWS
Discontinuities that break the top surface such as the crack show in figure
10.15 will cause a reflection to occur at exactly the beam path distance for
the full skip if a suitable angle that will reach the bore is used. However, as
you can see from figure 10.16, if you are testing a thick wall tube or a solid
106
bar, the beam may reach the top surface without first reflecting from any
other surface. The beam path length at which a top surface defect will appear
in that case can be calculated from the formula: -
BPL = Deos®
Where: - D is the outside diameter and о is the probe angle.
In the sort of application illustrated in figure 10.16, if there is no crack, the
sound will carry on around the bar or pipe as shown in figure 10.17. Provided
there is enough sensitivity, you may only need to scan from position ‘A’ to
position ‘B’. The beam will sweep the entire circumference during the short
scan and as long as you have enough timebase and gain, echoes from any
discontinuities breaking the surface will appear at predictable positions.
Fig. 10.17
107
CHAPTER 11
SURFACE WAVE TECHNIQUES
Surface waves have been used very successfully for a great number of
applications, particularly in the Aircraft Industry. However, it is not so common
in the Steel Industry because surface finishes are often less smooth, and
magnetic particle inspection will find most defects detectable by surface
waves. Nevertheless, there are occasions when the use of surface wave
techniques can give the simplest and most positive results and so, in this
chapter we will discuss some general principles that can be applied when
considering a surface wave technique.
ADVANTAGES OF SURFACE WAVES
Surface waves will follow gentle contours without reflection, but will reflect
sharply from a sudden change in contour. Figure 11.1 shows a typical
example of a component having a complex shape that would make the
use of shear or compression waves difficult, if not impossible. Cracks may
develop anywhere along the leading or trailing edge of the blade out to about
two thirds of the blade length, or in the root radius. A surface wave probe
placed at the end of the blade, and directed towards the root will send a
beam along the surface, round the radius and reflect from the edge of the
root as shown. Cracks in the suspect areas will give reflections at an earlier
time than the root.
Fig. 11.1
108
The fact that surface waves only penetrate to a depth of about one
wavelength can be used to advantage when testing relatively thin wall
sections. Figure 11.2 shows a pipe with a change of section. We are told
that cracks may occur on the inside or outside diameter of the pipe in the
necked region. An angled shear wave probe might be used, but it would be
difficult to predict the skip points as the beam bounces around the section
change. However, if we choose a surface wave at a frequency for which the
wavelength is approximately equal to the wall thickness, then the surface
wave will fill the wall thickness, and follow the section change, reflecting for
a defect breaking either surface.
Fig. 11.2
LIMITATIONS OF SURFACE WAVES
The main limitation of the surface wave technique is that the beam is almost
immediately attenuated if the surface finish is rough, covered in scale, or
a liquid (such as the couplant), or has any pressure applied by another
object (such as your hand). For this reason it is normal to use grease as the
couplant for surface wave probes (it doesn’t run!), and to apply the grease
to the probe, place the probe on the job and scan forward (away from your
own grease trail). Ridges left in the couplant during scanning, and objects
resting on the test surface, often give spurious signals that might be taken to
109
originate from defects. For this reason it is normal to test such indication by
rubbing a cloth over the area indicated by the signal. If after this ‘cleaning’
operation, the signal disappears, then it was a spurious indication.
CALIBRATION & DEFECT LOCATION
It is not usual to calibrate the timebase for surface waves in the way we
would do for shear or compression waves. This is because we can normally
run a finger along the surface in front of the probe, when we find a defect
indication, until the signal is no longer ‘damped’. This happens as we pass
over the defect with our finger, however, there are occasions when access
is limited and we are directing surface waves to a region that is out of sight
and cannot be reached with the hand. In these cases, the timebase can be
calibrated using the same procedure as for shear waves on an A2 or similar
block.
The sensitivity can be set from drilled holes or spark-eroded slits in suitable
reference blocks. In the aircraft industry, these reference blocks are usually
sections of an actual component with a spark-eroded slit in the critical
location.
110
CHAPTER 12
IMMERSION TECHNIQUES
Immersion testing techniques are mainly used in the laboratory and for
large factory installations carrying out automatic ultrasonic inspection. It has
the advantages of giving uniform couplant conditions and simple changes
of beam angle without changing probes. The basic principles involved are
simple and in this chapter, we only intend to deal with these basic principles
because the detailed techniques are very specific to each application.
COMPRESSION WAVE TECHNIQUES
In figure 12.1, we show a simple set up for a compression wave inspection
of plate. The plate to be tested is immersed in water and the probe assembly
moved to a convenient part of the plate leaving a suitable gap between probe
and object. The compression wave probe is housed in a fully gimballed
housing often called the manipulator. Adjustment about the two main axes of
the manipulator is normally by micrometer screw.
Fig. 12.1
The first procedure when setting up the technique shown in figure 12.1
is to ensure that the beam of sound is perpendicular to the top surface of
the plate. This is done by adjusting each screw on the manipulator until a
111
maximum echo is obtained from the plate top surface. The screen should
appear as shown in figure 12.2.
You will notice that, although in figure 12.1, the water gap between probe
and plate is about the same as the thickness of the plate, in figure 12.2,
the timebase distance between the transmission pulse Signal 1) and the
first water specimen interface (2) echo is much bigger than the timebase
distance from the interface echo to first back echo (3). This is because the
velocity of sound in water is only about one quarter of the velocity in steel
or aluminium. To make sure that the next repeat of the water interface echo
(5) does not interfere with the first back echo in the specimen the minimum
water gap should be one quarter of the specimen thickness plus 6mm,
for steel samples. In figure 12.2 signals 4 and 6 are repeat echoes of the
backwall.
Fig. 12.2
The first water to specimen interface echo is called the ‘top surface echo’
because it represents the time at which the sound enters the specimen. We
would normally use the delay control to put this top surface echo at zero on
the timebase. Calibration of timebase for a suitable test range is normally
done with a contact probe on the A2 block (for steel) in the usual way. If
in our example the specimen were 35mm thick, the timebase calibrated to
100mm of steel at compression wave velocity, and the top surface echo
delayed to zero, the screen would be as shown in figure 12.3 - the signals
are numbered the same as for figure 12.2.
112
Fig. 12.3
The screen presentation shown in figure 12.3 now takes on an appearance
that is familiar to us in contact scanning. The top surface echo being
equivalent to our usual transmission pulse, except that it is cleaner than a
normal transmission pulse having no crystal reverberations in the trailing
edge, hence the dead zone is less than with contact scanning. In its simplest
form, the probe could be mechanically scanned in a zigzag pattern, known
as a ‘raster scan’, at a constant water gap, and the operator could watch the
screen in the usual way for defect indications. A more reliable method might
be to use a ‘Monitor’ to watch the timebase and then warn the operator when
a defect signal enters the ‘gate’. The gating circuits are able to react much
more quickly than a human and this allows for much faster scanning, and
more reliable detection than would be achieved with contact scanning.
When the gating circuit is switched ‘on’ the gate appears. It can be moved
across the screen to the left or right by using the gate ‘start’ control. We
would use the gate start control to put the left hand edge of the gate close to
the right hand edge of the top surface echo (2). The gate ‘width’ control is
then used to expand or contract the gate width (it is the right hand edge of
the gate which moves) so that the right hand edge of the gate is close to the
left hand edge of the first backwall echo (3). An echo of the right amplitude
which pops up in this gate, (i.e. between top and bottom echoes), will trigger
113
an audible or visual alarm. Figure 12.4 is the same as figure 12.3 with the
addition of a gate between top surface and backwall echoes.
Fig. 12.4
The gate ‘threshold’ control determines the height a signal has to reach
before it triggers the alarm system. It can be set so that the monitor ignores
small defect signals.
The alarm system in the monitor can also be used to operate a marking
device to mark the object with paint or letter stamp in the defective regions,
or it can be used in conjunction with a pen recorder to produce a plan view
map of the specimen known as a ‘С-Scan’, showing defective areas. Figure
12.5 is an example of the appearance of a simple C-scan of a plate sample.
The recorder is arranged to write a scan line when no discontinuity above
the threshold is in the gate, but stops writing when a signal exceeds the
threshold. In figure 12.5, the C-scan shows three defective patches. Note
that with the simple image, there is no depth information about the defects;
we can only measure the area. If we want to record the depth of each defect
we would have to move the manipulator over each defect, measure the
depth from the display and record the value manually.
The latest systems benefit from computer technology and can store both the
plan view and depth information. The map produced shows the area of the
flaw but colour coded to indicate depth.
114
Plate length
Fig. 12.5
SHEAR WAVE TESTING
One advantage with the immersion technique is that a shear wave of any
desired angle can be produced simply by tilting the compression wave probe
through the appropriate angle of incidence. Figure 12.6 shows the plate
used in figure 12.1 set up this time for a 45° shear wave scan. The probe
manipulator has been angled to the angle of incidence required for a water
to steel interface to produce the desired angle of refraction.
B-SCAN PRESENTATION
Because positional encoders are fitted to the system that give information
about the exact position of the manipulator in relation to the plate co-
115
ordinates and the probe angle, all the information is available to calculate
the position of any reflector within the plate. This is particularly easy to do
in real time now that we can connect the system directly to a computer.
Therefore, not only can we generate a C-scan, but we can also generate a
view of a ‘slice’ through the plate to see where the discontinuity is within the
depth of the plate. Such a ‘slice’ image is known as a ‘В-scan’. If we asked
the computer to generate a В-scan image along the line A - A’ in figure 12.5
above, we would be able to see the depth of the two flaws that fall along that
line. The resulting В-scan for this compression wave inspection might look
like the image shown in figure 12.7.
A<-------------------------------------kA
Too surface echo ----------------------------------
 Laminations
Backwall echo ------------ ----------- ----------
Fig. 12	.7
Note that the top surface echo generates a continuous line showing constant
coupling of the sound into the plate, but the backwall echo line is broken
under the shadow of the laminations in the plate volume. In this example, a
fainter repeat image of the discontinuity that is positioned at less than half
plate thickness can be seen. This image corresponds to the repeat echo that
appears in a conventional A-scan
THROUGH-TRANSMISSION TECHNIQUE
Some materials, particularly plastic, have a very high absorption of sound.
Sometimes, even at very low frequencies, it is not possible to get back wall
reflections because the sound cannot get to the back wall and all the way
back. The through-transmission technique detects the amount of sound
reaching the back wall, and indicates the presence of a defect by a reduction
in amplitude of this through transmission signal. The technique is simple to
116
carry out and is illustrated in figures 12.8 and 12.9. The other type of gate
is shown in these two examples and the gate has been positioned at the
threshold amplitude. The gate ‘sense’ control has been set to the ‘negative’
position so that it will trigger if the through transmission signal falls below the
gate.
Fig. 12	.8
Fig. 12	.9
117
CHAPTER 13
EXAMINATION OF STEEL CASTINGS
In the next three chapters, we will be dealing, in turn, with the examination of
castings, forgings and welds. The object will be to discuss the basic routines
and procedures for these examinations, because routine and self-discipline
are at least half the battle in ultrasonic flaw detection. However, don’t take
these notes as being the specification to which you always work. This must
change with each job, and will depend on the customer and purpose for
which each casting, forging or weld is to be used. The routines will always be
similar and details will change. Make sure you know which code, standard or
specification you must work to before you start any job.
PROBES
Both shear and compression wave techniques are widely used for the
examination of castings. Because the grain structure has an appreciable effect
on the attenuation of sound waves, the test frequencies used in the examination
of castings tend to be lower than we have used up to now. Frequencies
of 1 MHz to 2.5MHz are common and occasionally it is necessary to use
as low as 0.5MHz (500KHz) in order to penetrate to the far boundary. Composite
crystal probes often give better signal to noise ratio and better sensitivity.
EQUIPMENT
A pulse echo flaw detector having an А-scan presentation is required. The
equipment should cover the frequency-range 0.5MHz to 6MHz, and when
used with the probes selected for the job should have good resolution and
penetration characteristics. Penetration characteristics are assessed by
placing a compression wave probe on the Perspex insert of the A2 block,
setting gain controls to maximum, and counting the number of back wall
echoes from the Perspex. A result having two to four back wall echoes
indicates a low penetrating power for casting work, and six to ten back wall
echoes indicates a high penetrating power,
118
The most commonly used probes are compression wave (single and twin
crystal) and shear wave probes of 45°, 60° and 70°. Frequencies will depend
on the material and thickness of casting to be tested. Probe sizes will depend
to some extent on frequency. At 2.5MHz, crystals of 12 to 25mm diameter are
used, but larger ones may need to be used at lower frequencies to avoid very
wide beam spread. For instance, a 10mm diameter 1.25MHz compression
wave probe would have a beam spread of 71°. Clearly, this is too wide a
beam spread for most applications. On the other hand, a 23mm diameter
crystal of the same frequency would give a 29° beam spread.
HEAT TREATMENT
In order to obtain the most favourable grain structure to reduce attenuation,
it is desirable to heat treat the casting before carrying out the ultrasonic test.
By heat treatment, we mean that after removal from the mould, the casting is
heated to, and held at, a temperature above the transformation range before
re-cooling. It makes little difference to us in ultrasonics whether this process
produces castings said to be ‘Annealed’, ‘Normalised’ or ‘Hardened and
tempered’. For quality control purposes, it may be necessary to inspect a
casting before heat treatment, but it must be appreciated that in such cases
the inspection will be less effective.
SURFACE CONDITION
Again, to achieve optimum results the casting should be available for
ultrasonic inspection with a suitable surface finish. To achieve the desired
surface finish, co-operation between fettling shop, or machine shop, and the
ultrasonic inspector is necessary.
a)	Cast Surface - a good cast surface will transmit ultrasonic waves, but
shot blasting will improve the coupling efficiency. It may be necessary
to hand-grind all or part of the cast surface, but care must be taken to
ensure that the profile of the casting is retained. Cast surfaces should
not be fettled by hammering or peening if ultrasonic examination is to
be carried out as this creates surface lapping.
119
b)	Rough Machined Surface - it is quite common for the casting to be
supplied, in the rough machined condition, for ultrasonic inspection.
However, the final cut should have been made with a broad nosed
tool to ensure a flat and smooth surface finish. A ‘Gramophone record’
finish is undesirable because it may give rise to spurious echoes and
cause excessive probe wear.
PROCEDURE
With any ultrasonic inspection procedure, it is vital that you establish a
routine that is systematic and comprehensive. Going through an inspection
in a series of well defined steps makes sure that you don’t forget anything,
and that at any one time you don’t have too many factors cluttering up your
thinking and observing faculties. The steps you should take in any casting
inspection should follow this sort of sequence: -
a)	‘Information’ - Make sure you have all the information you need about
the method by which the casting was produced, its geometry
(engineering drawings), the location of critical areas, and the types of
defects which are most likely to occur in those areas.
b)	‘Equipment’ - Make sure you have the right equipment for the job, and
that your equipment is working properly. Calibrate it properly for your
initial scan.
c)	‘Inspection standards’-Make sure you know all the standards to which
you are expected to conform, and the acceptance limits for defects.
d)	‘Visual Inspection’ - Look at the component to see that it conforms
to the drawings and information you have been given. Check that the
surface is suitable for ultrasonic inspection, and look for obvious
surface breaking defects. If you can see a defect that makes the casting
unacceptable, there is little point in proceeding with the ultrasonic test!
e)	‘Penetration’ - Before starting your ultrasonic inspection make sure
that you can get sufficient sound into the specimen. For compression
wave probes, you can do this by producing a number of backwall echoes,
and for shear waves, you can obtain corner reflections for instance. Your
selection of probe and your success in penetration will depend on a
120
number of factors, those influenced by the specimen being: -
(i)	Casting thickness
(ii)	Shape of the casting and area of probe contact surface
(iii)	Surface finish
(iv)	Grain size and structure
f)	Initial scan’ - You then proceed to scan the surface of the specimen
systematically to ensure full coverage. A compression wave probe (or
probes) is normally used for this initial scan. Areas showing defect
indications can be marked for critical assessment at a later stage.
g)	‘Critical scans’ - If areas of the casting have been high-lighted as critical
and susceptible to certain types of defects, then those areas should be
scanned carefully with a suitable range of probes chosen to give optimum
response to the critical defects.
h)	‘Assessment of flaws’ - After the initial and critical scan we go back
to the areas marked down as containing defects or perhaps a better word
would be ‘indications’, and carry out a careful evaluation of those
indications using as many probes as necessary to obtain all the essential
information about the discontinuity. This evaluation includes: -
(i)	Precise location of the defect,
(ii)	Assessing the nature of the defect,
(iii)	Plotting the size of the defect,
i)	‘Reporting’ - Having gathered all the information you have about the
casting, you prepare your report. We will be dealing with reports in a later
section, but in general, a report should show: -
(i)	What you did, and how you did it,
(ii)	What you found,
(iii)	How that compares with the acceptance standard.
We will now discuss a number of defects that occur in castings, and deal briefly with
the technique for detection of each and the sort of signal you might expect to see. It
will be appreciated that in these notes, we cannot give an accurate picture of every
defect you are likely to see, but they should form a useful guide to interpretation
for your future work. The terms used are in accordance with B.S.2737.
121
DEFECTS DUE TO INADEQUATE FEEDING (SHRINKAGE DEFECTS)
Shrinkage defects are cavities formed during solidification, and are formed
through liquid to solid contraction. These defects are not normally associated
with the presence of gas, but high gas content can magnify their extent.
Shrinkage defects may occur in steel castings where there is a localised
variation in section thickness. However, they may also occur in parallel
sections where penetration of the liquid feed metal is difficult. Shrinkage
defects in steel castings can be considered as falling into three types,
namely: Macro-shrinkage, Filamentary shrinkage, and Micro-shrinkage.
Typical locations at which shrinkage cavities are most likely to occur are
shown in figure 13.1. Where there is a localised change of section thickness,
a hot spot will occur which cannot be adequately fed. This will lead to
shrinkage cavitation and should therefore, be avoided if possible. Acute
angle junctions (‘V’, ‘X’ and ‘Y’) are least satisfactory and T or ‘L’ junctions
are less of a problem.
122
MACRO-SHRINKAGE
A large cavity formed during solidification. The most common type of this
defect is piping which occurs due to an inadequate supply of feed metal. In
good design, piping is restricted to the feeder head. The technique used to
detect this defect depends on the casting section thickness. For sections
greater than 75m thick a single crystal compression wave probe can be
used, whilst for thicknesses below 75mm it is advisable to use a twin crystal
probe. The presence of a defect is shown by a complete loss of back wall
echo together with the appearance of a new defect echo. An angle probe
should be used to confirm and augment the information gained from the
compression wave probe. The scans are illustrated in figure 13.2.
Fig. 13.2
FILAMENTARY SHRINKAGE
This is a coarse form of shrinkage, but of smaller physical dimensions than
a macro-shrinkage cavity. The cavities may often be extensive, branching
and inter-connected. Theoretically, filamentary shrinkage should occur
along the centre line of the section, but this is not always the case and on
123
some occasions, it does extend to the casting surface. This extension to the
casting surface may be assisted by the presence of pinholes or wormholes.
Filamentary shrinkage can best be detected with a combined double probe if
the section is less than 75mm thick. Defect signals tend to be more ragged in
outline than for macro-shrinkage. The initial scan should be carried out with
a large diameter (23mm) probe and the final assessment with a smaller (10
- 15mm) diameter probe. See figure 13.3.
Fig 13.3
MICRO-SHRINKAGE
This is a very fine form of filamentary shrinkage due to shrinkage or
gas evolution during solidification. The cavities occur either at the grain
boundaries (inter-crystalline shrinkage), or between the dendrite arms (inter-
dendritic shrinkage). Using a compression wave technique, the indications
on the display from micro shrinkage will tend to be grass-like (see figure
13.4), that is a group of relatively small poorly resolved signals extending
over some portion of the timebase. The existence of a back wall echo in the
presence of defect signals will be to some extent dependent on the frequency
chosen. For instance, there may be no back wall echo when using a 4-5MHz
probe due to scattering of the beam. This might suggest a large angular type
of defect. However, a change to 1-2MHz may well encourage transmission
through the defective region to add a back wall echo to the defect echoes,
and disproving the ‘large cavity’ impression.
124
Fig. 13.4
DEFECTS ASSOCIATED WITH HINDERED CONTRACTION DURING
COOLING
a)	‘Hot tears’are cracks which are discontinuous and generally of a ragged
form. They are caused by stresses that develop near the solidification
temperature when the metal is at its weakest. The stresses arise when
the contraction of the cooling metal is restrained by a mould or core, or by
an already solid thinner section. In figure 13.5, we show some of the
causes and locations of this type of cracking.
b)	‘Cracks’ or ‘Stress Cracks’ are well-defined, approximately straight
cracks that are formed when the metal is completely solid.
The location of a hot tear can rarely be determined accurately using a
compression wave probe because of the orientation of the defects. The
most satisfactory technique is to use angled probes. In steel castings, the
best way to find the cracks or hot tears is magnetic flaw detection, using
ultrasonics to plot the depth of defects.
125
4444
4444
4444
A
4444
4444	4444
Hot tears due to mould resistance along direction A and В
<---------- C -------------->
Hot tear due to casting resistance along length 'C*
Hot tear due to change of section ‘D’
Fig. 13.5
DEFECTS ASSOCIATED WITH ENTRAPPED GAS
a)	‘Airlocks’ - When molten metal is poured into a mould, air may be
entrained in the metal stream that may appear in the subsequent casting
as a cavity or several cavities, just below and parallel to the casting
surface. They are normally best detected by a twin crystal compression
wave probe. See figure 13.6.
b)	‘Gas Holes’ - These defects are discrete cavities, usually greater than
1,5mm in diameter that are caused by the evolution of dissolved gases
from the metal during solidification. A ‘Blow hole’ is the name given to
a gas hole caused by gas evolved from the mould or core rather than
126
the metal. A ‘Worm hole’ is the name given to a tubular gas hole that is
usually perpendicular to the casting surface. Since gas holes may be close
to the surface, twin crystal compression wave probes are the most suitable.
See figure 13.7.
0	5	10
Probe position 1
1 BWE
2BWE
0	5	10
Probe position 2
Fig. 13	.6
Defects
Probe position 1
1 BWE
2 BWE I
0	5	10
Probe position 2
Fig. 13	.7
THE EXAMINATION OF CAST STEEL ROLLS
Defects which occur in these large components (See figure 13.8) are
referred to as stress ‘cracks’, ‘clinks’ or tears, tears being probably the most
accurate description: -
127
Cast form
Machined shape
300cm
Bottom end

Fig. 13	.8
PROCEDURE
The following procedure describes the technique to be applied to an un-
machined fully heat treated casting. Because of the size of the component
and the grain condition it has been found from experience, that the most
satisfactory test frequency for the initial compression wave scans along and
across the roll is 0.5MHz. Any discontinuity indications that are detected at
this stage can be explored in more detail using compression waves at 1-
2MHz and shear waves at 1 -5MHz.
a)	Compression wave test along the cast steel roll. Note that in figures
13.9 and 13.10 we are using two 0.5 MHz single crystal probes, one as a
transmitter and the other as a receiver - a sort of disconnected ‘dual’
probe. It is a useful technique because if you lose the back wall echo
but get no defect echo because the defect has an inclined surface, you
can leave the transmitter in one position and scan the receiver to try to
find the reflected sound.
b)	Carry out a compression wave test across the roll using the same two
probes scanning first with the probes separated circumferentially (figure
13.11), and then longitudinally (figure 13.12). The longitudinal scan
will detect circumferential defects and the circumferential scan will find
longitudinal defects.
c)	Explore defective areas with further probes such as twin crystal
compression wave probes of higher frequency and shear wave probes
between 45° and 70°.
128
Fig. 13.11
JTI 1 lRl lTl 2 lRl
r
Fig. 13.12
129
MEASUREMENT OF ULTRASONIC ATTENUATION
Measurement of ultrasonic attenuation in cast steel components will provide
useful information in the assessment of grain size and thereby on the
effect of heat treatment. In the ‘as cast’ condition, grain size is large and
the ultrasonic beam is scattered giving an increase in background noise
(grass) and reducing the sensitivity of the ultrasonic technique. In the fully
heat treated condition the ‘as cast’ grain structure is re-crystallised resulting
in grain refinement. The ultrasonic attenuation of the material following this
treatment is low.
The predominant factor in attenuation measurement is the relationship
between the ultrasonic wavelength and grain size. These are related as
follows: -
High attenuation occurs when the grain diameter is greater than the
wavelength (D>X)
Low attenuation occurs when the grain diameter is less than the wavelength
(D<X)
The following factors will influence attenuation measurements: -
a)	The frequency used should be as high as practicable i.e., between 4 - 6
MHz for compression waves and between 2-4 MHz for shear waves.
b)	More critical results are obtained if you use shear waves.
c)	For compression waves, the path distance should be between 50-200mm.
d) For shear waves, the path distance should be between 10-100mm.
e)	The roughness of the scan surface.
f)	The roughness of the back surface.
g)	The analysis of the cast steel
h)	The nominal heat treatment.
i)	The position of ingates, risers, feeder heads and test bars.
j)	The couplant
k)	The examination must be carried out on sound material in thicknesses
over 50mm, the presence of micro-shrinkage is unlikely to affect the
results
130
PROCEDURE 1
This simple method can be used to quickly check whether the attenuation
(and, therefore, grain size) varies around a casting. It cannot be used to
compare one casting with another or two parts of the same casting if the
thickness varies. The method is best described by the diagrams shown in
figure 13.13. A backwall echo or corner reflector is obtained from the casting
and the amplitude adjusted to a particular value, ‘A’ in the illustration. The
calibrated gain setting is noted as an indication of attenuation in the casting.
Obviously, the actual value is only usable if you have measurements from
other similar castings for comparison. Experience from other castings
would help you to decide whether, in this case, the casting could be
successfully inspected at the frequency of the probe you used to make the
measurement.
Fig. 13.13
PROCEDURE 2
The method illustrated in figure 13.14 is more useful and more accurate
because it gives an attenuation figure in dB per cm, and can be used to
compare areas of similar or different thickness on the same casting, or to
compare different castings. Using a compression wave probe, you obtain two
131
back wall echoes. You then adjust the amplitude of the first back wall echo
to a predetermined screen height (A,mm) and note the attenuator reading
(Y,). The second back wall echo is then brought up to the same height (A2)
and you note the new attenuator reading (Y2). You then subtract Y1 from Y2
to obtain the number of decibels absorbed over the distance between the
two backwall echoes. The attenuation of sound in dB/cm is the dB difference
divided by the sound path travelled. This is the energy lost in sound travelling
from the top surface to the bottom and back again, i.e. twice the material
thickness, so divide (Y, - Y2) by 2 x thickness and you have the attenuation
for that region of that casting in dB per cm.
M
(Y -Y }
Attenuation = —-----— db / cm
2t
Gain Yi
i /
i ।
i ।
i t
\i
Gain Y2
to reach
Ai
Fig. 13.14
132
CHAPTER 14
EXAMINATION OF FORGINGS
The testing of forging is in many ways more straightforward than the testing
of castings. For one thing, the grain is far more refined, giving much lower
attenuation and less noise, and allowing a higher test frequency to be used.
Secondly, defects such as cavities and inclusions in the original cast billet
are flattened and elongated during the forging, rolling or extrusion process to
become better reflectors, and largely parallel to the outer surface. The one
exception to this might be cracks that may not be parallel to the scanning
surface.
Much of the testing of forgings can be accomplished with compression
waves using single or twin crystal probes at frequencies between 4 - 6
MHz and occasionally up to 10 MHz. Angled shear wave probes are used
to explore defects detected by the compression waves, and to search for
defects that might not be suitably orientated for compression waves. In the
testing of forgings, particularly those that have been in service for some
time, it is very often possible to predict where defects will be, if they exist,
and for this reason many specifications only call for a limited scan looking
for one particular defect in one location.
DEFECTS IN FORGINGS
a)	‘Pipe’ - This defect is the remains of primary or secondary piping that has
not been removed from the original cast ingot (See figure 14.1). It is
usually situated along the centre line of the component, and its length
will depend upon the amount of elongation of the original ingot necessary
to produce the required size of product. As a secondary pipe has never
been exposed to the atmosphere, it is possible for some portions to ‘weld’
together during forging to produce an intermittent defect.
133
Primary pipe
Secondary pipe
Cast ingot
Fig. 14.1
b)	‘Inclusions’ - Inclusions of non-metallic matter present in the metal
because of impurities and the melting or refining process (oxides,
silicates, sulphides and phosphates) may be present in the cast ingot.
These may change shape during subsequent forging processes, if the
inclusions became plastic at the processing temperature. They may
also be broken down into many smaller parts. Larger defects are
simple to detect using compression waves, but as defects get smaller
in size, they become more difficult to find.
c)	‘Bursts’ - Bursting may result when forging processes are carried
out at too low a temperature, or when subjecting a metal mass to
drastic reduction. If this occurs at the end of a forging, it can be seen,
and ultrasonics used to plot its extent into the forging. However,
internal bursts can be formed, usually underneath a change in section,
which can only be found by ultrasonics.
d)	‘Thermal Cracks’ - Sudden changes in the rate of heating or cooling can
result in uneven stresses in the forging leading to crack formations. Unless
these cracks are relatively large, their random orientation may make them
difficult to detect ultrasonically. For steel forgings, magnetic flaw detection
is often the most suitable way of finding these defects.
134
e)	‘Hairline Cracks’ - Hairline cracks occur in certain grades of steel due
to the differences in solubility of hydrogen in the liquid and solid metal.
During solidification, hydrogen is thrown out of solution and diffuses
readily in the atomic state until it reaches some discontinuity such as
a microscopic inclusion. Here it combines into molecular form creating
enormous pressures, so forming the nucleus for fine cracks. These
cracks, however, have no preferred orientation but because of their
numberandthe random orientation, generally present suitable reflecting
surfaces for ultrasonics.
The ultrasonic inspection of any forging, as with castings requires a set
routine to ensure all problems are covered: -
a)	Make sure you know all about the component, its material, shape,
manufacturing process, heat treatment etc.,
b)	Know the major defects likely to occur in the component, and their most
probable locations - and the acceptance standard.
c)	Choose your equipment and probes, based on the information you have
gathered.
d)	Carry out a visual examination.
e)	Carry out the basic ultrasonic scan necessary to find all the defects.
f)	Carry out any supplementary scans that may be necessary to fully
describe the defects.
g)	Make your report fully and clearly.
INSPECTION OF COMPONENTS OF UNIFORM CROSS SECTION
In most rolled or forged materials where reduction has taken place uniformly
from a larger size to an elongated smaller size, defects will be parallel to the
outside surface. The examination of rolled plates for lamination has already
been dealt with in chapter 9 under ‘Lamination Testing’. Drawn bar reduces
secondary piping to a long, roughly cylindrical discontinuity along the axis of
the bar. This can be detected by a compression wave technique (figure 14.2)
and just in case small deviations from the cylindrical shape make it difficult to
detect from one direction, it is usual to make two scans 90° apart.
135
Rolled or forged bar may produce a flatter defect (figure 14.3) from the pipe
in the original ingot, and to ensure that these are detected, it is necessary
to make a number of scans along the length of the bar, with about 180°
between first and last scans.
Fig. 14.3
Scans from seven directions
in this example and scans 2
and 3 would give the best
results
Discontinuities in square sections may also be orientated in such a
manner that little reflected energy reaches the transducer. The application
136
of stresses to alternate faces during the forging process tends to induce
defects lying diagonally as shown in figure 14.4. In such cases, an angled
probe inspection will be more sensitive than a compression wave scan. For
adequate coverage, two faces 90° apart need to be scanned.
Compression waves 1 & 2
give poor responses due to
discontinuity orientation
Shear wave scan 3 gives a
better response
One point to watch when testing cylindrical components is the tendency for
only line contact to be made. The contact area gets smaller as the diameter
of the object decreases. This leads to an increase in beam spread and a
reduction of test sensitivity. To compensate for this the operator normally
turns up the gain and this has the effect of extending the probe transmission
noise giving an increased dead zone. The best solution to the problems of
line contact, if you have a lot of bar of the same diameter to test, is to use a
twin crystal probe and shape the shoe to fit the radius.
DEFECTS FROM THERMAL TREATMENT
Defects caused by the stresses set up by faults in heat treatment of a
component may occur in any plane and position within the component. To be
certain of detecting these defects, scan from as many surfaces as possible,
and use as many beam angles as possible. Often defects associated with a
particular component and faulty heat treatment technique, occur in the same
137
region of each component. As this becomes a clear ‘trend’ during production,
ultrasonic techniques are often simplified to basic scans for laminar detects
and inclusions, plus one scan aimed exclusively at finding the heat treatment
defect peculiar to that component.
DEFECTS PROPAGATING IN SERVICE
Some defects develop, or first show up in service after the component has
been subjected to its working loads for some time. Fatigue cracks and stress
corrosion cracks are typical of this problem. The defects initiate from small
imperfections in the original casting, forging, welding or heat treatment
processes which were not (and probably could not be) detected during the
manufacturing stages. They occur in areas of highest stress concentration
and their initiation points and direction of propagation is usually predictable
(if they are going to occur at all that is!). Often they are highlighted in fatigue
test programmes or become apparent through an analysis of in-service
failures. Inspection techniques are developed to examine critical areas for
a particular defect in a particular location. In the case of critical components
such as the main wing spar of an aircraft, designers and stress engineers
may say “If it is going to crack, it will crack from this point, and we would need
to detect the crack before its length exceeds ‘x’ mm”. We may then develop
an ultrasonic technique, using a reference block containing an artificial
defect somewhat smaller than ‘x’ mm (to be on the safe side) - and inspect
hundreds of aircraft every year for twenty years looking for this particular
defect, and never find one!
One example of in-service inspection is the routine examination of railway
axles (figure 14.5). It is an interesting application to look at briefly because
it illustrates a number of problems that tend to occur in the working life of
any Ultrasonic Inspector. The first obvious problem is that the component
is longer than many of us meet in the normal routine. The timebase has
to be compressed to represent maybe 7 or 8 feet of steel, and some flaw
detectors, adequate for other inspections, may not give sufficient timebase
range, or there may be problems getting both ends of the time base in focus
138
at the same time. There may not be sufficient pulse energy available to
penetrate that much steel. Probes need to be carefully chosen to have as
narrow a beam spread as possible, because the beam has plenty of time
to spread. For this reason probes of 20 to 25mm diameter and 1 - 4 MHz
frequency are normally used. Even then, it will almost certainly spread out to
touch the sidewalls and spurious echoes due to mode conversion will occur.
Signals are so compressed that they become difficult to see - especially with
short-pulsed (high resolution) probes.
Next, we have changes of shape that will give rise to signals on the display,
and it is near these section changes that defects are most likely to occur. The
reflections from section changes will give a standard signal pattern for each
type of axle. Each of these signals will need to be memorised so that you
are only looking for differences from the normal pattern. These differences
will have to be explored further in order to report fully on any discontinuity
you find.
Railway axle
Fig. 14.5
Another source of standard signals that may occur on railway axles, and in
other applications, is associated with bushes or bearing housings that are a
shrink fit to the axle. In such cases, a proportion of the sound from the beam
edge of the compression wave, meeting the interface between the axle and
bush at an angle, is transmitted and mode converted into the bush, (figure
14.6). The shear wave in the bush suffers multiple reflections giving a pattern
of echoes such as the one shown in figure 14.7. These standard signals
must also be identified and memorised.
139
Mode conversion in bush
Fig. 14.6
Fig. 14.7
It is quite common, when testing axles and shafts for some areas to be
masked from inspection because of section changes (figure 14.8). These
masked regions may be reduced by using compression wave probe angled
to give a refracted angle of not more than 10°. If a special probe is not
available, the result can be achieved by inserting a small Perspex wedge
between the probe and the scanning surface (figures 14.9 & 14.10).
An example of the ultrasonic testing of a particular shaft is shown in figure
14.11. Initially the shaft is scanned from either end using a compression
140
wave probe. You would tend to concentrate your search for defects from
each end of the shaft, to the half closest to the probe, but paying some
attention to signals arising in the second half. Supplementary scans of 45°
or 60° would be required to confirm the presence of cracking at changes in
section close to the journal.
Fig. 14.8
141
Scans 1 & 2 compression
wave.
Scans 3 & 4 shear wave to
confirm defect
Fig. 14.11
During the compression wave scan signals will appear and need to be
identified as: -
a)	Known changes in section.
b)	Spurious echoes due to mode conversion.
c)	Defect signals.
In the first instance, these can only be identified by your knowledge of your
equipment and the shaft. You need to know the beam spread of your probe,
and the calibration of your flaw detector. From this and accurate timebase
readings for each signal you should be able to identify each signal. In the
case of the shaft illustrated in figure 14.12, the display and interpretation
would be as follows: -
142
In Figure 14.12, signal 1 is the defect, 3, 4 and 7 are from known section
changes, 5 and 6 are from internal reflections and 2 and 8 are mode
conversions.
examination of lugs
The examination of lugs such as the one shown in figure 14.13, for fatigue
defects, is normally carried out using the angled shear wave probes. As with
the examination of thick wall tubes, you need to consider the penetration of
the beam to the bore of the lug. In the case shown a 35° probe is needed to
find the anticipated discontinuity. The probe wedge will have to be radiused
to fit the outer surface to ensure good coupling, maintenance of correct angle
and correct sensitivity. Stress analysis has predicted that failure will only
occur in the shaded portion of the sketch, radiating from the bore.
Fig. 14.13
'n the lug shown in figure 14.14 represents a rudder hinge fitting from an
aircraft, history has shown that if failure occurs it will be in a region close to
defect shown. This particular inspection is a very good example of the
rare occasion when it might be necessary to use an angle that means both
s^ear and compression waves will be present.
143
When we draw a line back to the scan surface at 90° to the defect, draw
in a normal from that point on the scan surface and measure the angle
of refraction needed to meet the defect at right angles, we find that angle
to be 24°. The problem is that we know that 24° is right in the middle of
that ambiguous zone where shear and compression waves co-exist in the
specimen. The inspection obviously needs a special angle probe, but do we
ask for a 24° shear wave probe or a 24° compression wave probe?
If we look at the situation where a 24° compression wave probe is chosen, we
will see that the unwanted shear wave will come out at about 11 ° travelling at
about half the compression wave velocity. It will strike the bore as shown in
figure 14.14, and because of beam spread a portion of the reflected energy
will arrive back at the probe, at about the same time as the anticipated defect
(compression wave distance to defect is twice as far but the velocity is twice
as much), and this will lead to confusion. If we choose a 24° shear wave,
then the unwanted compression will be at about 50° and will travel off into
the body of the fitting and not return at a confusing time. Hence, in this case
our choice would be the 24° shear wave. This is typical of the problems
encountered in the testing of this sort of forging, and great care must be
taken to plan the technique before applying a probe to a component.
144
CHAPTER 15
EXAMINATION OF WELDS
There have been some spectacular failures of plant and components in the
past, due to faulty welding processes or procedures. In many cases, these
have originated from weld defects that can most reliably be detected by
ultrasonic flaw detection. A painstaking procedure for the manual examination
of welds has evolved over the years. The work is often tedious, sometimes
uncomfortable, but always demanding of both skill and understanding. The
elements required to make a good ultrasonic welding inspector can be
summarised as: -
a)	A thorough understanding of flaw detection theory and practice
b)	A good working knowledge of welding procedures and the origins of
weld defects
c)	Experience, and a lot of patience
d)	Most of all he needs integrity
If you are working on a critical project, with a good design team, and a good
welding crew you may test many hundreds of feet of welding without seeing
any significant indications. Under these circumstances, the temptation to
relax your vigilance must be great. Nevertheless, the inspection only remains
valid as long as you devote all your attention to the job.
In the welding process, two pieces of metal are joined together. Molten
‘filler’ metal from the welding rod blends with the molten parent metal at the
prepared fusion faces, and fuses the two pieces together as the weld cools
and solidifies. Some of the defects occur because the fusion faces do not melt
properly or blend with the filler metal (lack of penetration and lack of fusion
defects). Some defects occur because the scale or slag which forms at the
top of each ‘pass’ of the welding rod, is not chipped away completely before
the next pass is made (Slag inclusions). Some defects occur because the
welding electrode dips into the molten weld and bits of copper or tungsten
drop into the weld (dense metal inclusions). Some defects occur as in much
the same way as casting defects (Porosity, piping, wormholes, shrinkage,
145
undercut etc.). Some defects occur because of the thermal stresses, set up
by having part of the component at molten temperature, and the rest (of the
parent material) at much lower temperatures (cracks, tears etc.).
Many of the defects that can occur in welds do not significantly alter the
strength of the weld; others do in varying degrees. However, planar defects
(cracks, lack of penetration or fusion) particularly those breaking the surfaces
of the welded joint, give rise to the most severe reductions of weld strength.
Our inspection procedure should be such that defects that will produce, or
lead to, an unacceptable reduction in weld strength, are detected.
INSPECTION PROCEDURE
As with any inspection procedure, you need to be systematic, and this
requires self-discipline. The temptation to scrub a probe across the test
surface, chasing every small signal that pops up, occurs with all of us, but
it must be resisted. At each stage of the inspection, you need to know what
you are looking for and which zone of the weld you are testing. The routine,
which you must adopt, is: -
a)	Find out all there is to know about the weld:-
(i)	Material
(ii)	Welding process and associated defects.
(iii)	Weld preparation design.
(iv)	Parent metal thickness adjacent to weld.
(v)	Any special difficulti experienced by the welder because of the
weld location on site
(vi)	Acceptance standards.
b)	Establish the exact location and size of the weld. Ideally, you should mark
the parent metal either side of the weld before welding commences so
that the exact centre line can be established after welding. In some
cases, where the weld re-enforcement has been ground flush with the
parent material, it may be necessary to etch the weld region to establish
the weld width. The centre line of single V. butt welds can be roughly
checked with acompression wave probe. Markthe centre line accurately
146
on the scanning surface.
c)	Carry out a visual inspection of the weld checking that the surface is
free from weld spatter, and smooth enough for scanning. Some defects
may show at the surface and be noticed during this visual examination
(undercut, cracks, crater pipes, burn through etc.,). If you can see these
defects and know that they are in excess of the acceptance standard,
thengetthosedefects remedied before you begin tocarry out ultrasonics.
In fact this is a good principle to observe at all stages of the inspection,
as soon as you see one, or a group of defects, which make the weld
totally unacceptable, stop! - There is no need to do any more.
d)	Carry out an ultrasonic inspection of the parent metal either side of the
weld over a band that extends as far as full skip for your shallowest
angled probe (usually a 70° probe) plus half the cap width. In this scan
you can use a compression wave, and assess material thickness as well
as locating laminar defects which might interfere with the passage of
shear waves during the weld examination.
e)	Carry out a critical root examination from both sides of the weld using a
suitable angled probe. This is because it is the root area in which defects
are most likely to occur and where their presence is most detrimental. It is
also the region in which a regular echo, from the weld penetration bead,
can be expected, and so it needs to be a carefully controlled scan. Note
regions in which defect indications occur.
f)	Carry out an examination of the weld body from both sides of the weld
using angled probes. The scan pattern should ensure that the total
volume of the weld is examined. Note regions in which defect indications
occur.
g)	If transverse cracking could occur with a particular weld design or
process, then a scan using angled probes, parallel to the weld axis must
be carried out. Defect indications should be noted.
h)	At this stage, if no defects have been found, the weld can be accepted.
If however, some defects have been noted, you now go back to those
areas and explore the defect as thoroughly as possible to determine: -
(i)	Its exact position in the weld.
147
(ii)	Its size along the weld axis (length of the defect).
(iii)	Its size through the weld thickness.
(iv)	The nature of defect (planar, volumetric, crack-like etc.,)
(i) Draw up a full report about your examination of the weld. The report
should be comprehensive enough forsomebody else tofind the weld, test
it using the same technique as you, and to the same test sensitivity, to
find the same defects, and using the same sizing technique as you did,
come up with the same conclusions.
BUTT WELDS IN PLATE AND PIPE
Figure 15	.1 illustrates the weld preparation for a typical single ‘V’ weld, and
the terms used to describe various parts of the prepared weld area.
Centre line
Preparation angle
Root gap
Fig. 15.1
Figure 15	.2 shows a cross section of the same weld after welding, showing
the original preparation, and the number of passes made to complete the
weld, in this case 8 passes: -
Weld cap (reinforcement)
Weld body
Fig. 15.2
Weld root bead
148
Figure 15	.3 shows several other weld preparations used in the fabrication of
pipe and plate butt welds.
‘LT preparation
Backing ring or strip
Double 'V preparation
‘EB’ insert
Square edge preparation
Fig. 15	.3
VISUAL EXAMINATION
When you approach the weld to begin your inspection, you should
already know what the welding preparation was what welding procedure
was followed, and what inspection standard you must follow. Your visual
inspection begins with a quick check to make sure that the weld is ready for
examination. The weld spatter (that is small splashes of molten-metal which
stick to the surfaces around the weld and then solidify) should have been
removed. The scanning surface either side of the weld cap should be free
from scale and corrosion pits, in other words smooth enough to move a probe
across, for at least a band extending to full skip plus the probe size. In some
cases, the parent metal will have to be smoothed off with a surface grinder to
achieve this finish. In the case of some critical welds the weld cap may also
be dressed to give a smooth contour, or ground flush with-the parent metal
to allow the probe to scan right across the centre line of the weld. If this is
the case, you must ensure that there are no humps or bumps in the profile
across the weld that will prevent probe movement or lift the probe to leave an
149
air gap underneath the contact face. Not all welds can have the cap ground
off. In some instances, the extra metal thickness in the cap region is needed
to strengthen the weld.
You also look for obvious weld defects, since this may make the weld
unacceptable without being ultrasonically tested. Defects such as undercut,
and cracks can often be seen quite easily. Figure 15.4 shows what we mean
by undercut. Note that this can also occur at the root, but will only be seen if
you have access to both surfaces.
Undercut at cap and root
Fig. 15	.4
Another fault, which may not always adversely affect weld acceptability, but
which might interfere with subsequent ultrasonic inspection is ‘misalignment’,
illustrated in figure 15.5. This fault occurs through poor setting up before
welding or when pipes that are not truly round, are butted together. The
welder will often try to disguise this by blending the cap in with the parent
metal on either side. The clue to misalignment is often, therefore, a widening
of the cap. A similar effect occurs when plate or pipe of different wall
thicknesses are welded. This is called ‘mismatch’: -
Misalignment
“VL______
Mismatch
Fig. 15	.5
150
compression wave inspection
The compression wave inspection of the parent metal, and, if the weld
cap is smooth enough, the weld itself forms a vital part of the procedure.
Firstly, by checking parent metal thickness it gives you actual thickness
values for subsequent shear wave calibrations, rather than the nominal
thickness obtained from the drawing. It also detects mismatch immediately.
The systematic scanning of the parent metal in the band on which the
subsequent shear wave scans are to be carried out, will detect laminations,
which though they may not affect the strength of the welded plate or pipe,
might interfere with a shear wave beam. Figure 15.6 illustrates this problem.
A large lamination causes the beam to reflect up to the cap giving a signal
that might be mistaken for a normal root bead, and at the same time, misses
the lack of penetration defect.
Fig. 15.6
The compression wave scan, if the weld cap is dressed, allows you to locate
the weld bead, and thus check the position of the centre line. An echo from
the weld bead, because of beam spread, will be accompanied by a back wall
151
reflection from the parent metal. The bead echo, at slightly greater range
than the back wall echo will maximise when the probe centre is over the
bead centre. The range difference between bead and back wall tells you
how prominent the weld bead is. Whilst scanning across the weld you will
also obtain echoes from weld defects such as slag, porosity, etc., which have
volume. These can be plotted out and confirmed later with shear waves.
SHEAR WAVE CRITICAL ROOT INSPECTION
The next step is to make a careful inspection of the weld root area. We
make this a separate operation because: -
a)	Defects in this area usually have the most serious affect on weld
strength.
b)	It is one region in which defects are very likely to occur.
c)	It is a region in which reflections occur (from the weld bead) in a good
weld, and root defect signals will appear very close to the standard bead
signal, i.e., it is the region where you are most likely to be confused.
Because of the critical importance of this part of the weld, and the possible
confusion between defect and bead signals, this root scan needs a high
degree of self-discipline to maintain a rigid procedure. We will see that this
part of the inspection can be broken up into several stages. At each stage of
the scan you will be looking for a specific defect; other signals may appear
and you may be tempted to ‘chase’ them to see where they originate. This
temptation is to be resisted, because in chasing stray signals you may miss
the defect you are looking for.
Before we look at the procedure for this scan, it is worth looking at the weld
root conditions you are likely to meet in Single ‘V’ butt welds. These are
illustrated in figure 15.7.
152
a.	Normal bead
____________TL_______
b.	Lack of Penetration
c. Incomplete root fusion
d. Excess penetration - root shrinkage
----------------
e. Root undercut
Fig. 15.7
PROCEDURE
The main aim of this scan, in the first instance, is to detect lack of penetration,
or incomplete root fusion (figures 15.7 b and c). That is when one or both
root faces have not been fused. To detect this defect we mark out a scanning
line at half skip distance back from the original root face, on either side of the
weld, (i.e. half skip plus half root gap from centre line}. We then place a guide
so that when the heel of the chosen angle probe is butted against the guide,
the probe index is on the scanning line as shown in figures 15.8a and 15.8b.
Flexible magnetic strips are very useful for this purpose.
Fig. 15.8a
153
Scan lines
Plan view
Fig. 15.8b
Next, we calibrate the timebase on the A2 or A4 block for a suitable range.
For parent metal thicknesses up to about 30 mm, a timebase range of 100
mm is suitable for this root scan. We calculate the beam path length for a
bottom corner reflector from the BPL factor or from the formula t-?Cos0=BPL.
Lack of penetration will give a signal at this range when the probe index is
placed on the scanning line that you have drawn on the parent metal surface.
With the probe index on this scanning line and the heel of the probe against the guide
to keep it there, we scan around the weld looking for a signal at our critical range.
With the probe in this position, we will of course see a reflection from the
weld bead if the weld is a good one, but this signal will be a small distance
(depending on how big the weld bead is) away from the anticipated spot for
a lack of penetration defect (figure 15.9). If there is some root shrinkage or
undercut, we will also see a signal from that, but at a slightly shorter range
154
Root shrinkage or undercut
Fig. 15.10
We can see that during this first scan there are three possible root conditions
that might show up. They are, NORMAL BEAD, LACK OF (or incomplete)
PENETRATION, or ROOT UNDERCUT, We need to be able to differentiate
between these three from the points that we have just discussed. Let’s look
at a specific case to help fix the principles in our minds. Consider a single ‘V’
butt weld in plate 20mm thick. The weld preparation is a 60° included angle
with a 2mm root gap and a 2 mm root face. (Figure 15.11.)
c/L
The scanning line for a 60° probe should be at half skip (34.6 mm) plus half
the root gap (1 mm) i.e. 35.6 mm from the weld centre line. The beam path
length to the root edge preparation is then 40 mm, i.e. lack of penetration
Would give a signal at 40 mm along the timebase. In the sketch the bead
signal would be 4 mm further away, i.e. 44 mm and root undercut about 2
fhm closer at 38 mm.
155
If the probe were positioned about 2 mm further back from the weld centre
line, the centre of the beam would be aimed at the corner made by the
undercut and parent plate, giving a maximum response from the undercut
and at the critical distance of 40 mm. In other words, we might mistake
undercut for lack of penetration, and so you can see the importance of
knowing the weld centre line, and root gap, and of marking these accurately.
If you have marked out accurately and have found a signal at 38mm which
you wish to confirm as undercut, you can do this after completing your initial
scan, by coming back to the suspect area, removing the guide strip, and
slowly moving the probe backwards. If the suspect signal rises in amplitude,
maximises at a range of 40 mm and then falls slowly, you can be reasonably
certain that you have root shrinkage or undercut.
Of course, all that we have said in the last three paragraphs depends on
knowing the actual centre line of the weld, and the root gap. We very often
say that this or that ‘must be’, only to find that in the field, it is not! In fact,
VERY OFTEN you won’t have enough accurate information about the weld
to make the job as simple as we have just suggested. However, we can still
usually come to the right conclusions. Firstly, you should always be able to
get parent metal thickness accurately from your compression wave scan.
Therefore, you can calculate the beam path length at which lack of penetration
should show up. You shouldn’t then confuse a normal root bead with lack of
penetration; the confusion will come between lack of penetration and root
undercut. This is where a knowledge of welding defects comes into its own,
because with lack of penetration there will be no weld bead signal, whereas
with root undercut there usually is. In addition, with site welding, unlike the
deliberate defects produced for training or examinations, it is unlikely that a
welder will produce inch after inch of good, uniform penetration bead that
suddenly stops for a few millimetres and then starts again. The weld bead is
likely to trail away, stop, and then trail in again. If he is a bad welder, there
will probably be excess penetration in places, inadequate or no penetration
in other places. In other words, if while you are doing your critical root scan
the weld bead signal varies a lot in amplitude and position, be careful! There
156
is a probability of defects! Figure 15.12 shows the sort of irregular weld bead
profile to be found in this situation.
Fig. 15.12
It is not always the welder’s fault that these defects are produced in the weld;
sometimes the access and environment problems are such that it is almost
a physical impossibility to do a good root run in a particular part of the weld.
It is often worth having a chat to the welder, therefore, to find out where the
awkward areas were, since those are the most likely to contain defects. If
you can’t see the welder, then try to look at the weld through his eyes to see
which areas might have been difficult.
Your choice of test sensitivity can help or hinder in this root scan. Too much
gain can give you a confusing jumble of signals in the root area, too little,
and you risk missing things. As a guide only, you are about right if you peak
the 100 mm echo from the A2 block to full screen height and then add 10dB
of gain for testing plate, and up to 20 dB for testing pipe welds. However,
since lack of penetration is a good corner reflector, and a normal weld bead
is often not quite as good a reflector, it is sometimes useful to try a quick
scan at 10 dB lower than these settings, because a major lack of penetration
will show up well at this setting and the bead won’t. However, the critical root
scan should then be repeated carefully at the higher setting.
When you have carefully examined the root, probing from one side of the
Weld centre line, you move to the other side of the centre line and go through
all again to confirm your findings from the new side. This scan from the
157
second side will also help you to interpret two other types of defect in the root
area that we have not discussed. The first one of these is shown in figure
15.13. It is a small slag inclusion, or porosity just above the root.
This defect might appear just short of the half skip beam path length when
doing scan 1, leading you to guess that it might be a root undercut. If this
were so, scan two should put it just further than the critical distance, but the
inclusion will show in about the same place, i.e. just short again. Furthermore,
from scan 1 we would expect undercut to give a rising signal as the probe
moves back 2 or 3mm, but the inclusion will give a rising signal as you move
forward. It will also give a rising signal for forward movement from side 2.
The second defect mentioned above is shown in figure 15.14. This shows a
crack starting from the edge of the root bead.
158
prom side 1 a large signal would appear just where you would expect to see
undercut. However, from side 1 the bead signal would be obliterated. From
side 2, however, it would be possible to get a bead signal as well as a defect
signal.
CHOICE OF PROBE ANGLE
This is perhaps not as critical for the root scan as it is for the remaining
scans. We normally choose from 45°, 60° or 70° and sometimes 80° probes,
to have the shortest beam path length to the root, and our choice is limited
by the condition of the weld cap. On thinner plates it may not be possible to
position a 45° or 60° probe so that the half skip beam points at the root gap,
without the toe of the probe riding up onto the weld cap. If the weld cap has
been dressed flush with the parent material, then we would probably use a
45° probe provided the material was not so thin that the critical beam path
length came into the probe ‘clutter’. For welds with cap in place we can
make the following recommendations about probe angle’s for various wall
thicknesses for the root scan: -
Parent metal thickness	Probe angle
6 - 15mm
15 - 35mm
Over 35mm
60° or 70°
60° or 45°
45°
SHEAR WAVE WELD BODY EXAMINATION
Once we have completed our examination of the root area, we can begin to
look at the fusion faces and weld body. Again, we need to mark out the parent
metal surface to fix the scan limits for the probe angle we have chosen. Our
main aim in this is to ensure that the whole of the weld volume is carefully
tested. Figure 15.15 shows the outer limits of the scan, which positions the
probe so as to produce full skip distance to the nearest edge of the weld cap.
That means that the probe index is at a distance from the weld centre line
64ual to full skip plus half the cap width.
159
The parent material is marked with a line, parallel to the weld centre line, at
this distance, both sides of the centre line. Two new parallel lines are also
drawn at the half skip limits if you have changed probe angle since the root
scan. Our scanning pattern is going to be between these half and full skip
limits. Figure 15.16 shows a plan view of the marked area.
160
PROBE ANGLE
The initial choice of probe angle for the weld body scan depends upon the
weld preparation angle (Figure 15.17). The angle should be chosen to meet
any lack of sidewall fusion at right angles, for maximum response. The exact
angle to meet this fusion face at right angles can be calculated from: -
g
Angle = 90	Where 0 is the weld preparation angle
Example 16
Calculate the most suitable probe angle for examining the fusion faces of a
weld with a 60° weld preparation angle.
60
Angle = 90~2
Angle = 90-30
Required probe angle - 60°
Example 17
Calculate the most suitable probe angle for examining the fusion faces of a
weld with a 45° weld preparation angle.
45
Angle 90 ^
Angle = 90-22.5
Required probe angle = 67.5g
In the first case, clearly you would use your 60° probe, but in the case of the
45° weld preparation angle, it is not likely that you will have a 67.5° probe, so
Уои would choose the nearest - a 70° probe.
161
The procedure, having selected the appropriate probe angle, is to scan in a
zigzag pattern between the marked scan limits (Figure 15 17)
Each forward scan should be at right angles to the weld centre line, and the
pitch of the zigzag should be half probe width to ensure full coverage.
USING SEVERAL ANGLE PROBES
This scan concentrates on the weld body. You have already assessed the
root area, and know the range at which the root bead appears as the probe
reaches the half skip marker. So the part of the timebase which is of interest
is between the root bead signal, and the calculated beam path length for full
skip. You will, of course, get some reflections from the weld cap, but these
will be at a range at, or in excess of, the calculated full skip beam path, and
occur as you approach the full skip limits of your scan.
The range at which you are testing, particularly when using a 70° probe
to suit the weld preparation angle, can be quite lengthy, and you may feel
that the sensitivity to other defects within the weld zone may be rather
162
low. In such cases, it is reasonable to use 45° or 60° probes to carry out
supplementary scans. Remember that this will not give favourable results for
lack of sidewall fusion defects. If the weld cap has been dressed, you will be
able to overcome the problem by scanning across the weld centre line from
half skip to the far edge of the original cap, instead of changing probe. Care
should be taken to ensure that any residual undulation, left when the cap is
dressed, is not severe enough to lift the probe index clear of the surface as
shown in figure 15.18. Ideally, the weld profile should be flat and flush with
the parent metal, however, come undulation is to be expected.
Air gap
unacceptable
Fig. 15.18
USE OF TWIN CRYSTAL ANGLE PROBES
When you are scanning directly over a dressed weld, some of the defects
might be very close to the top surface. If the transmission noise of a single
crystal probe lasts longer than the return time in the Perspex shoe, the noise
will obscure part of the timebase and mask the defect echoes close to the top
surface. In such cases twin crystal angle probes are available and may be
used, just as twin crystal compression probes are used for overcoming the
dead zone in thickness or lamination testing. Measurement of beam index,
and angle, and timebase calibration can be carried out in the same way as
for single crystal angle probes.
USE OF COMPRESSION WAVE PROBES
Also, if the weld is dressed useful confirmation of defects having volume can
be made by scanning a compression wave over the weld body.
163
PLOTTING WELD DEFECTS
Signals that appear on the screen need to be plotted out to determine their
position within the weld volume. This can be done using the plotting device
shown in figure 15.19. The beam centre on the slide for the probe is used to
determine defect position; the beam edges are used in defect sizing, and this
part of the operation will be described fully in Chapter 16.
Fig. 15.19
The plotter is used in the following way. Firstly, draw a scale diagram of the
weld preparation on the transparent cursor as shown in figure 15.20. This will
enable us to plot defects between top surface and half skip. In the example,
we show a single ‘V weld in 20mm plate.
164
Next, draw a mirror image of the weld below the first one. This will enable us
to plot defects between half and full skip. (See figure 15.21).
Now let us assume that we have picked up a defect in the weld and
maximised the signal from the defect from the position shown in figure 15.22.
Vou note the beam path length, from the timebase, and measure the surface
distance ‘s-d’ between the probe index and the weld centre line.
165
Let us assume that the beam path length was 20 mm and the surface
distance was 17 mm. We set 17 mm on the horizontal scale of the slide
against the T of the transparent cursor, as shown in figure 15.23. Place a
fine ‘x’ mark showing the position of the defect where the beam centre line
crosses the 20mm BPL arc. In this case, the position of the defect is correctly
plotted on the weld centre line at about half material thickness.
Fig. 15.23
166
If we look at another example, such as the one shown in figure 15.24, in
which the defect is being detected as the pulse travels between half skip and
full skip, we can see how to use the mirror image we have drawn on the slide.
C/L
Once again, we measure the surface distance and the beam path length; in
this example, we will say they are 60 mm and 64 mm respectively. With the
60 mm mark of the horizontal scale against the T of the cursor, we see in
figure 15.25 that 64 mm down the beam centre puts the defect on the side
wall of the weld nearest to the probe, and just about the middle of the plate
thickness.
Fig. 15.25
167
In the illustrations so far we have shown the beam on the slide running from
the top left hand corner, at the probe angle 60°, towards the right hand side,
and we have used this when scanning the weld from the left of the centre
line. Figure 15.19 showed that the probe card is drawn to show the beam
centre pointing to the right from one edge and from the left at the other edge.
Some people use a probe card that is only half the length of the one shown
in figure 15.19 but they draw on both sides of the card. One side scans from
the right and the reverse side scans from the left.
SCAN FOR TRANSVERSE CRACKS
Having examined both the weld root and the weld body, our next scan is to
detect transverse cracks, breaking either top or bottom surfaces. Magnetic
particle inspection is obviously a quick and effective method for detecting top
surface cracks, and so very often, you are only looking for cracks breaking
the bottom surface. If the weld is dressed, you begin your scan at the weld
centre line and scan along the line in each direction, you then make several
scans parallel to, and either side of the weld centre line, from each direction,
making sure you cover the entire weld region.
If the weld cap has not been dressed, as in figure 15.26, you will have
to scan parallel to the weld centre line, alongside the weld cap with the
probe inclined towards the centre line as shown. Since a crack tends to
have a jagged edge, it is likely that some energy will be reflected back to
the transmitter, but a safer technique would be to use a pair of probes, one
transmitting and one receiving; this is also shown in figure 15.26.
168
DEFECT identification
д|| the scans you have completed up to now have been confined to:-
a)	Finding the defects
b)	Establishing their positions in the weld volume.
You now know that the weld is:-
a) Free from defects and so is acceptable
or, b) Is so badly defective it is clearly unacceptable
or more likely
c)	Some defects exist, but you need to know more about their nature and
size so that you can compare them with an acceptance standard and
then make your report.
In most cases, the next job is to use one of the sizing techniques to determine
the length and through thickness dimension of each defect you have plotted.
At the same time, you will try to assess the nature of each defect. The various
methods of defect sizing are described in chapter 16. We will, however, in
this Section, look at some of the methods of assessing the nature of the
defect. By this, we would like to mean the interpretation from our ultrasonic
results of whether a defect is a slag inclusion, or porosity, or undercut lack of
fusion, crack, and so on. However, the flaw detector only gives us two pieces
of information at any one time; that is, firstly, the time interval between the
signal from some reflecting surface, and a known reference, and, secondly,
the amplitude of that signal. More information can be gleaned from the way in
which the signal changes in amplitude and time as we move the probe. We can
also see whether the signal is a single clean ‘spike’ or a ragged group of echoes.
It is rather like someone shining a narrow beam torch on a large wall on a
dark night. Is it a garden wall? Is it an out-building, or a house? All that can be
seen in the circle of light is a few bricks. To find out more, we could ‘explore’
the wall with the torch, scanning up and down, and from side to side, storing
•П our mind’s eye, all that had gone before, but never at one time able to see
a whole door, or window, or wall in the narrow beam. We might recognise
|	169
part of a door or a window, maybe enough to know that it is not just a garden
wall, but we may never be sure whether it is a house or a shed.
In ultrasonics, we are going to scan our probe beam over the defect to see how
time and amplitude change in relation to our probe movement; the change of
amplitude will be of major importance. The shape of the signal ‘envelope’, as
it is called, will give us a clue to the shape of the defect, and from this and our
knowledge of defects which are likely to occur in this welding or manufacturing
process, we can make an intelligent ‘guess’ as to the nature of the defect. There
is no real substitute for a thorough knowledge and experience in the welding or
manufacturing process if your guess is to be responsible, but always remember
that even at best, it is only an intelligent guess. The industry has recognised
this limitation over the years and it is now common practice only to categorise
discontinuities in general terms such as ‘volumetric’, ‘planar’, ‘crack-like’ and so on.
INITIAL ASSESSMENT FROM SIGNALS
We have seen that in certain parts of the weld, the position on the timebase
for a known position of the probe can give us our initial clue to the nature
of the defect - for example, the different root conditions in a single ‘V’ weld.
These can be confirmed by placing the probe in a similar position on the
other side of the weld centre line, as we have seen. We can also get some
‘negative’ information from a fixed signal. If we plot a defect’s position and
find it to be in the centre of the weld, then that defect cannot be lack of side
wall fusion, for instance, neither can it be any of the root defects.
The profile of a fixed signal can also give us some clues about the nature of
the defect. Consider the two signals shown in figures 15.27 (a) and (b): -
170
in both cases, we have a large indication at about 5 on the timebase. The
energy causing the signal in figure 15.27 (a), all originates from a depth
equivalent to 4.8 timebase divisions, in other words, we have a ‘clean’
break in the timebase. Compare this with the energy causing the signals in
figure 15.27 (b), where the signals originate at depths between 4.0 and 7.5
timebase divisions - we have a very ‘ragged’ signal. The defect causing (a) is
likely to be smooth and regular in its presentation to the beam, whereas the
signal at (b) might be caused by the very irregular outline of a slag inclusion
or a jagged crack, or it might be a cluster of smaller defects at different
depths such as slag inclusions, gas pores, lamellar tearing, etc. The fixed
signal has not told you what the defect is, but it has told you what it might be,
and what it probably is not. To go back to the torch and the dark night, if your
beam settles on a small patch of brick work, you know that the object is a
brick structure, you don’t know if it is a building or a bridge, but you do know
that it is not a car, or a bus, or a cow.
ASSESSMENT FROM PROBE ORBITING
Consider a planar defect running parallel to the weld axis (for instance, lack
of penetration in a single ‘V’ weld). Supposing you have found such a defect,
and your probe has been positioned to maximise the signal so that the set up
is rather like the one shown in figure 15.28, (position A),
Fig. 15.28
171
When you plot the defect, you determine that its position is ‘x’ mm in front of
the probe index. Imagine a circle, radius ‘x’ mm centred at the origin of the
reflection; swing the probe around this circle from A to В to C - D - E - F, and
back to A so that the beam centre always passes through the same point. At
A and D you will get a maximum reflection from the defect. This signal will
quickly disappear as you orbit away from A or D because the sound is no
longer striking the defect normally, and is therefore reflecting away from the probe.
Suppose the signal amplitude at A and D was 4 divisions, then at 3° rotation
before or after A and D it was 1.5 divisions, at 5° before or after A and D it
was 0.1 divisions, and elsewhere it was zero. We could draw a sort of polar
diagram for the signal amplitude from various directions of scan - it would
look rather like figure 15.29, showing clearly that planar defects are very
directional. In figure 15.29, the radius of the concentric circles indicates
signal amplitude and the scanning directions A to E are those shown in the
previous diagram.
Fig. 15.29
If the reflecting surface was a gas pore rather than a planar defect like the
one shown in figure 15.30, then because the gas pore is spherical in shape
it will present the same reflecting surface to the beam all the way round the
orbital scan. Then the polar diagram would look like the one shown in figure 15.31.
172
Fig. 15.30
D
Fig. 15.31
Supposing the polar diagram for a defect that you have detected looks like
the one in figure 15.32, what deductions can we make from the shape of the
polar diagram?
D
Fig. 15.32
173
Firstly, it is clear that the defective region has volume, because you can
detect a signal from any direction. Secondly, the reflecting surfaces within
the defect region are irregular in their presentation to the beam. This
would be typical of a large slag inclusion. In addition to the irregular polar
diagram, you may already have noticed variations in the timebase range and
possibly an irregular fixed signal such as the one we saw in figure 15.27 (b).
However, you must remember that a group of gas pores or lamellar tearing,
or a particularly jagged crack could give similar indications. The probabilities
favouring this or that defect type narrow as you gather more information.
ASSESSMENT FROM PROBE ROTATION
This technique gives similar information to the orbiting technique. Figure
15.32 represents the same situation as the one we saw in figure 15.28.
Instead of orbiting the probe about the defect, the probe is rotated about its
axis, by about 70° or so in each direction.
If we consider the three defect shapes again: -
a) Planar (e.g. lack of fusion)
b) Spherical (e.g. porosity}
c) Irregular (e.g. slag)
These might give the envelopes for signal amplitudes similar to those shown
in figure 15.33 a, b, & c.
174
Clockwise I Anticlockwise
a) Planar
л .
Clockwise | Anticlockwise
b) Spherical
Clockwise I Anticlockwise
c) Irregular
Fig. 15.33
assessment by traversing and lateral scans
Another good clue to the nature of some defects is the ‘lie’ of the defect and
this can be established by probe movement forward and back (traverse)
or parallel to the weld centre line (lateral), at the same time plotting probe
movement and timebase range changes.
Supposing you have established, by orbital or rotation scans that a defect is
essentially planar. You return the probe to a position that gives a maximum
echo, and plot the reflection point. You move the probe towards the weld
centre line and the signal moves to the left on the time base. Occasionally
you stop and measure timebase range and the surface distance between
probe index and weld centre line. From these measurements, you plot
several more reflecting points along the defect. You then move the probe
back (away from centre line) and plot further reflecting points. The resulting
Plot may look like figure 15.34, a line of points along the fusion face.
Maximum reflecting point ---------------
5	Fig. 15.34
1
175
Now you return to the maximum echo position, and placing a guide behind
the probe, scan sideways parallel to the weld centre line, first one way, then
the other. Again, you stop occasionally, make measurements, and plot a plan
view of the reflecting points (Figure 15.35). These plotted points show that
the defect lies parallel to the weld centre line.

C/L
Fig. 15.35
If you now fit together all the information from your various scans, you know:-
a) There is a reflecting source in the body of the material
b) It is planar in character
c) It lies along the fusion face
d)	It runs parallel to the weld centre line.
This information is compared with your knowledge of the welding process,
and of weld defects, and you draw the reasonable conclusion that our defect
is lack of sidewall fusion.
The case we have just considered was straightforward; so don’t run away
with the idea that this identification business is simple. Very often lack ol
fusion is associated with slag entrapment, and orbital scans tell you the
defect is ‘irregular’. Sometimes the edge preparation is damaged during
welding and the non-fusion face doesn’t plot out along the line shown on
your scale drawing as the fusion face. It can’t be said too often, the best yoi
can hope for in many cases is to shorten the list of possible defects for any
one indication.
SIZING AND REPORTING
Once all the defects have been identified as far as possible, and sized
your final job is to write a report of your findings, comparing them with some
176
acceptance standard if required. However, this is such an important topic, we will
devote a separate chapter to acceptance codes and reporting, that is chapter 17.
SUMMARY, SINGLE V’ WELDS
yye seem to have covered a lot of ground talking mainly about the routine for
inspecting one weld configuration, but much of the subject matter applies to
other weld configurations, or for that matter, to castings and forgings as well.
So before we leave single ‘V’ welds, and go on to look at other configurations,
let us just note the routine we have followed: -
a)	Visual examination
b)	Compression wave scans
c)	Critical root scans
d)	Weld body scans
e)	Transverse defect scan
f)	Defect interpretation and sizing
f) Reporting
DOUBLE V’ WELDS
The routine for double ‘V’ welds is basically the same as the one just described.
There are some differences in detail, in the critical root examination, and the
weld body scan, because of the differences in weld configuration.
CRITICAL ROOT SCAN
The typical weld preparation for a double ‘V’ weld is shown in figure 15.3. Figure
15.36 shows the theoretical ‘lack of penetration’ defect in this type of weld.
Fig. 15.36
177
It can be seen, that in theory at least, this defect, planar, vertical, and in the
middle of the weld volume, ought not to reflect sound back to the probe. In
practice however, there is often enough slag or distortion at the top or bottorr
of the defect, to give a reflection. It is usual therefore, to use a 70° probe
positioned at half skip distance from the weld centre line, to carry out the
critical root scan. The anticipated timebase range for an echo from lack of
penetration cannot be predicted as precisely as for single ‘V’ welds, but о
course you do not have the added problem of root bead or undercut signals
to contend with.
TANDEM PROBES FOR CRITICAL ROOT EXAMINATION
The classic method for detecting vertical reflecting surfaces within the volume
of the material is the tandem technique shown in figure 15.37. Although thi'
illustrates lack of root fusion in a double ‘V’ weld, it can be used for any welc
preparation having a vertical face.
Fig. 15.37
In figure 15.37, ‘6’ is the probe angle, ‘S’ is the separation between probe
indices, ‘d’ = depth of aiming point, and ‘t’ is the specimen thickness. Fc
double ‘V’ welds, we aim at the centre of the weld at half parent met&
thickness, and the probe separation, ‘S’, is equal to half skip distance fc
that probe angle. In other applications, we may wish, for instance, to explon
a fusion face that is vertical throughout the weld thickness. Our prob
separation for any depth can be calculated from the formula: -
S = 2(t-d)tanQ
178
fELD BODY EXAMINATION
pj-fie weld body examination is much the same as for single ‘V’ welds, but
l^iis time your scan starts at one quarter skip distance from the weld centre
ЦпС| goes back to full skip plus half weld cap width, (See figure 15.38). This
you have four fusion faces to examine, and you need to remember
^at the bottom weld cap will give reflections between half skip beam path
length to 3 or 4 mm beyond half skip beam path length. This cap will prevent
confirmation of the condition of the lower fusion face on the opposite half of
the weld.
C/L
WELDS WITH BACKING STRIPS (RINGS, OR EB’ INSERTS)
This type of weld is shown in figure 15.3. The inspection procedure only differs
from that for single ‘V’ welds in the detail of the critical root examination. In
the root examination of this type of weld, the prime object is to confirm that
fusion has taken place between the parent metal root preparation and the
backing strip or insert.
‘EB’ INSERT
When properly fused, this weld configuration is like a perfect single ‘V’ weld
With a constant root bead profile. Setting up, then, is exactly as we did for the
S|ngle ‘V’ root scan, and we expect to see a root bead signal at a particular
Place on the timebase, which remains constant in amplitude as we scan
along our probe guide (provided, of course, couplant and surface roughness
are also uniform). A drop in amplitude in the signal from the insert is a clue
that fusion may not be complete. The presence of an echo at exactly hah
skip beam path length would be positive evidence of non-fusion. Since the
insert gives a very strong signal as a rule, and that signal is often only 2 - ;
mm beyond the half skip position, a short length of non fusion only shows as
a half skip signal sliding up the front of the insert signal (i.e. poorly resolved
as shown in figure 15.39.
Fig. 15.39
Lack of fusion at the top of the insert (figure 140) can best be detected b\
compression wave probe. For this reason it is desirable for the weld cap t
the dressed to allow the compression wave scan. If this cannot, or has nc
been done, this defect can often be found as a signal originating from jus
above the root, when using a shear wave angle probe because of distortioi
or entrapped slag.
Fig. 15.40
180
Racking strips or rings
yyhen properly fused the weld cross section looks like the one shown in
figure 15.41.
Fig. 15 41
The shear wave root scan allows energy to pass through the root into the
backing strip. Reflections from within the strip will show as a pattern of
signals beyond half skip beam path length (see figure 15.42). A decrease in
amplitude or total loss of this pattern indicates non-fusion of the backing strip.
Again, it is desirable to have the weld cap dressed so that a compression
wave probe can be used to check the root fusion. With a compression wave
probe over the weld centre, an echo will be received from the back wall, and
from the backing strip. Loss of the backing strip echo indicates lack of fusion
(See figure 15.43).
Fig. 15.43
181
T’ WELDS
The examination of both T and Nozzle welds is somewhat different to the
weld configurations already studied. For complete inspection, scans from
several surfaces are required, and access to more than one surface may
not be available. In other words, you may often have to carry out a limited
inspection only. We will consider the ideal case where all surfaces are readily
accessible, remember that in practice you may not be able to carry out all
these scans.
T welds may be fully penetrated, or only partially penetrated by design.
The inspection procedure is much the same in either case, but for partial
penetration welds you need to monitor the non-fused portion to ensure that
it is not longer than the design permits. Full and partial penetration joints are
illustrated in figure 15.44.
Fig. 15.44
182
^je have, by now covered enough of the basic principles of welding inspection
Io concentrate on the scans to be made. For T welds, these are illustrated
|n figure 15.45. In the diagram three scans are indicated by numbers on the
probes: -
1 Scan 1 Compression wave - looking for Laminations, Lack of Fusion,
Lamellar Tea ring.
Scan 2 Shear wave - Weld Body Defects, Toe Cracks.
Scan 3 Shear wave - Fusion Faces, Weld Body.
As with previously discussed weld configurations, probe angles and
frequencies will be chosen to suit the geometry of the weld and accessibility.
For Scan 3 it is useful to choose a probe angle that will produce a beam
centre line parallel to the weld cap (See figure 15.46) to reduce the tendency
for confusing cap echoes. However, this may produce surface waves in the
cap that could also be confusing. Remember that surface waves can be
damped with an oily finger and this may help with confirmation of which wave
is causing a signal.
Fig. 15.45
183
Fig. 15.46
NOZZLE WELDS
Nozzle welds are those in which one pipe is joined to another as a branch,
at either right angles or some other angle. As with T joints, the weld may be
fully penetrated, or only partially penetrated. The branch may be let into the
main pipe to let liquids or gases in and out, for instance, or the branch may
simply be mounted on to a pipe that is not perforated, as in the case of a
bracing strut in a tubular structure. The two types are shown in figure 15.47
in which the shaded portion shows the pipe wall. The main difficulty in the
examination of nozzle welds is the fact that the weld profile is changing as
you scan around the weld. Access to all the desirable scanning surfaces is
also a problem, and it is rare to find completely free access. So once again,
you may only be able to carry out a limited inspection.


Branch pipe
Bracing strut
Fig. 15.47
184
Some typical weld preparations are shown in figures 15.48 to 15.52. It is
obviously not within the scope of this book to reproduce all the types of weld
preparation in use. We should be able, however, to look at some of the basic
principles. In the diagrams, we illustrate the wall of the main pipe or vessel
(called ‘shell’), and the wall of the branch, stub, or nozzle (called ‘branch’).
Fig. 15.48
Fig. 15.49
Full penetration ‘set through’ weld
Fig. 15.50
Partial penetration 'set through’ weld
Fig. 15.51
185
Fig. 15.52
FULLY PENETRATED SET ON’ NOZZLE
The scans to be carried out are shown in figure 15.53.
Fig. 15.54
Scans 1 & 2 are compression wave scans of branch and shell to determine:
a)	Thickness
b)	Lamination
c)	Fusion of shell wall, weld body.
186
|Scan 3 is a critical root scan against a probe guide. For a weld preparation
Jangle of 40° as shown, 65° is the optimum probe angle. Moving the probe
^back towards position 4 scans the fusion face and weld body.
^PARTIAL PENETRATION ‘SET IN’ NOZZLE
?The scans are similar to those shown in figure 15.54. However, we do need
to check the actual penetration achieved, and to make sure that the vertical
fusion face is fused (See figure 15.55).
Intended condition
Faulty fusion face
Fig. 15.55
We can find out whether the weld has penetrated far enough to fuse the
vertical preparation edge by very careful plotting of the root signals. It is
usual to plot both the maximum reflecting point, and, as confirmation, the
point at which the signal just disappears (i.e., beam centre and beam edge).
From an accurate drawing of the weld preparation, the intended point of
maximum penetration can be determined, and the range of this point, using
the beam centre and the beam edge, can be measured. The ‘Intensity Drop’
method of estimating the end of the intended non-fusion (using the beam
kdge) is described in chapter 16.
‘SET THROUGH’ NOZZLE
These are rather like full or partial penetration T’ welds, and can be scanned
h the same way as shown in figure 15.45 (scans for T welds). The main
I	187
complication arises from the fact that the fusion faces, when one pipe fits
into another in this way, lie along a line that looks rather like a ‘saddle’ (see
figure 15.56).
Fig. 15.56
The equivalent to scan 1 in figure 15.45, would be made from the bore of
the branch. (Compare figure 15.45 with figures 15.50 & 15.51). To determine
the weld limits, it is usual to scan the compression wave probe up and down
the bore (i.e., parallel to branch axis) noting the change of signal from wall
thickness to weld region, and carefully marking the probe position. A series
of points plotted in this way can be joined with a chalk or wax pencil line to
give you the weld limits.
FINDING THE WELD CROSS SECTION
The main problem with nozzle weld inspection is the changing geometry as
you scan around the weld. Since knowledge of the weld cross-section is a
vital part of the inspection procedure, it is important to be able to draw the
section at any point around the weld circumference at any point that you
might be testing. The problem is illustrated in figure 15.57 giving you a view of
the joint looking down the bore of the branch and in figures 15.58 and 15.59.
188
pf we look at a cross-section of the weld through the longitudinal axis of the
|rriain vessel (along the line 90° to 270°) the weld preparation would look like
^the one shown in figure 15.58.
However, if we were to look at a cross-section through 0° to 180° the weld
preparation would look like the one shown in figure 15.59.
truly longitudinal section and the truly circumferential are the easy
°hes to draw! You can of course, construct an accurate scale drawing
189
by conventional engineering drawing techniques, but this can be a time
consuming task. One alternative that gives acceptable results in practice is
to use a mimic gauge, (normally used for marking floor tiles so that they can
be fitted around door frames). The gauge consists of a stock through which
are fitted a number of equal length wires that are free to slide through the
stock. If we position the mimic gauge over the weld and adjacent shell and
branch, it takes up the external profile as shown in figure 15.60.
Fig. 15.60
We can trace around the mimic profile onto a sheet of paper as in figure 15.61.
Shelt wall
Fig. 15.61
190
VVe know the thicknesses of branch wall and shell, so we can mark points
on the drawing, parallel to the outer surfaces and at the appropriate wall
thickness (see figure 15.62).
Inner contour drawn
from branch and shell
thickness
Mimic
profile
Weld profile drawn
in from known shell
thickness
Fig. 15.62
From these marks, it is possible to construct the complete cross section of
the weld at that position. In this context is useful to remember that the root
gap and the distance between the branch outer surface and shell should be
constant, so you can deduce an approximate position for the fusion faces.
191
CHAPTER 16
DEFECT SIZING AND EVALUATION TECHNIQUES
The evaluation of defect size and nature are the two most controversial topics
in ultrasonic flaw detection. Sizing in particular can be a confusing subject
for the beginner, with various techniques being advocated or condemned
by the many experts in the field. In many ways, those of us who have beer
involved in the training of ultrasonic operators have been as much to blame
for the confusion as have the experts, and those who set the standards. Wt
have in the past taught students a particular sizing technique as if it were an
absolute measuring system and without giving a cautionary note on the limit:
of accuracy. In this section, it is hoped that we can redress the balance anc.
show the various techniques and their limitations. No one technique has ye
been established which gives a high degree of accuracy or repeatability in a1
circumstances. The ‘Time of flight diffraction’ (TOFD) system that has beei
developed over recent years has proved to be very accurate and fast in man\
instances. Nevertheless, TOFD can sometimes fail, especially for flaws tha'
originate at or near the scanning surface. The discerning operator will use <
combination of techniques to obtain the best result, but he will also be aware
that the result obtained may still be in error. To use the various method:
intelligently you need to appreciate the underlying philosophies involved.
In the various approaches to defect evaluation there are two basic philosophies:
a)	Those that attempt to deduce the actual dimensions of each defect
These techniques often allow the inspector some discretion in varying
the technique detail in order to achieve a more accurate result.
b)	Those that attempt to standardize technique detail so that a greate
uniformity of results is achieved, and so that ‘go’ or ‘no-go’ criteria
can be set as acceptance standards. These techniques evaluate th<
defect signal by comparison with ‘known’ reflectors, and do no
purport to give actual defect size.
192
Iln the former category, we have the Intensity Drop technique, the Maximum
Amplitude technique, and the TOFD technique, whilst in the latter category
^we have the DGS system and the American ‘DAC’ reference approach set
out in ASME Codes.
INTENSITY DROP TECHNIQUE
In order to build up an understanding of how this method can be used to size
defects, it might be informative as well as amusing to consider the following
analogy: -
You are acting on behalf of the town planning committee, who have been
informed that householder has built an extension to his house that is two
metres longer than was agreed in the planning application. The wicked
occupant owns a ferocious dog that prevents surveyors from entering the
premises to measure by conventional means. You find yourself outside the
property, on a dark night, armed only with a torch with a beam divergence of
5°, a pocket scientific calculator, and a walking stick with a series of marks
every 10 cm. From town plans, you know the fence to be 11,5 metres from
the building. With this equipment, you proceed to measure the wall of the
property.
You calculate that the beam diameter at a distance of 11.5 metres from the
torch is 90cm. You move up to the fence and aim your torch horizontally at the
middle of the front wall, then move the torch until the beam forms a perfect
Circle (the beam strikes the wall at right angles). You now shuffle sideways
maintaining the torch at right angles to the wall until the beam reaches one
end of the wall. You move more carefully until the edge of the beam is just
Clipping the end of the wall. Stop! Scratch a mark on the ground immediately
below the torch with the heel of your boot. Next, you shuffle back along the
fence shining your torch on the wall again until the other edge of the beam is
lust shining on the opposite end of the wall. Stop and make another mark. A
plan view of what you have done is shown in figure 16.1.
|	193
Fig. 16.1
The first mark is to the left of the edge of the wall by an amount equal to half
the beam diameter, and the second mark is to the right of the wall by the
same amount. In other words, the distance between the two marks is the
length of the wall, plus the beam diameter.
You use your walking stick to measure the distance between the two marks:
this turns out to be 13.4 metres. You quickly subtract the beam diameter
(13.4-0.90m.) and find that the length of the wall is 12.5 metres. You consult
the planning application documents and see that the proposed length of the
wall was 11 metres with a tolerance of 25 centimetres. Then the dog starts to
bark so you make a hasty retreat to make your report. Your report says that
the extension is not correct within the allowed tolerance.
What are the possible errors? Your entire mission has been fraught with
problems, not the least of which was the possibility of a confrontation with an
angry householder if the dog had barked earlier. However, from the accuracy
point of view, the following sources of error should be appreciated: -
a)	You were lucky that there was no fog to scatter the torchlight.
b)	Did you check that the extension was not to the rear of the house?
c)	Are you sure that your torch beam width is 5°? An error of +1° a
194
this range would mean that the wall was 18 centimetres shorter than you
calculated, whilst an error of -1° would make the wall 18 centimetres
longer than calculated.
d)	Did you estimate the range properly? There may have been some
road widening and error of +50 cm. would mean that the wall was 13.5cm
shorter than you calculated, whilst an error of -50cm. would make the
wall 13.5cm longer than you estimated.
e)	Was the torch beam at right angles to the wall? An error of 5° will make
the wall 3cm. shorter than you calculated.
f)	How wide was the mark made with your heel? Did you measure from
equivalent edges of the heel print? The error could be of the order of
up to 8cm. unless you were wearing boots with Stiletto heels.
g)	Are you sure that the last edge that you found was the house wall and
not the dog kennel alongside the house. That might mean a possible
error of 1 metre!
If you were unlucky with all these possible sources of error, the wall might
be 1,45m, shorter than the maximum permitted and the town council will be
made to look silly - they won’t like that! So you need to be careful, and to
double check each step in the exercise.
(Back to defect sizing! We are going to look at a system that uses the same
(basic principle as our story, that is, we will use range, beam spread, and
(probe movement information to estimate defect dimensions. The key to the
whole operation is an accurate knowledge of your beam characteristics. We
will assume that you have carefully measured the probe beam index and
angle, following the standard procedure. The next thing to do is to plot the
beam spread. You will note that I have said plot, not calculate. Why go to this
trouble when we said earlier that beam spread could be calculated from: -
Sin— =
2
kxk
d
Where: 0 - beam spread angle
X = wavelength
к = a constant
d = crystal diameter
195
However, the crystals we use in ultrasonic flaw detection are not perfect
vibrators, each will have its own peculiarities that make it vary from the
calculated beam shape, and the variation will be too large for the accuracy
we require in sizing. Because of this, we choose to plot, from known
reflectors, the edges of a zone within the beam where the sound intensity
has fallen by some arbitrary amount, usually 20 dB, or 6dB from its maximum
intensity at that range.
PLOTTING THE BEAM SPREAD
We will now look at the plotting of a 20 dB beam spread diagram for a 45°
shear wave probe. For the plotting of beam spread diagrams we normally use
the IOW beam profile block if we are working with steel; for other materials,
a similar block in the appropriate material would need to be made. The IOW
block contains four drilled holes for beam spread measurement. These holes
are 1,5mm diameter, and are drilled 22mm deep. There is also a group of five
similar holes and this group is used to measure the resolution of shear wave
probes. The beam profile block is illustrated in figure 16.2.
<	305
Surface A
, '	"	з , 1
ni ' ,,	О	2	,, I
© o	•	Side view	t I
I %	Surface D	4	| Й
й	й j
; ..	I
Surfaces
Surface C
8 I 2	> PlanVieW 4	81
2	Surface A	3 4	X 8
Й	! й I
i	: Ji
Surface D
Fig. 16.2
196
The holes used for beam spread diagrams are shown in figure 16.2 and
numbered 1 to 4. Holes 1 and 3 are drilled into one face of the block and
holes 2 and 4 into the opposite face. The holes are located thus: -
Hole 1 is located 19mm below surface A 48 mm from surface E
Hole 2 is located 25 mm below surface A 83 mm from surface E
Hole 3 is located 13 mm below surface A 50 mm from surface F
Hole 4 is located 43 mm below surface A 35 mm from surface F
Since the block is 75 mm high, you can deduce: -
Hole 1 is 56 mm below surface В
Hole 2 is 50 mm below surface В
Hole 3 is 62 mm below surface В
Hole 4 is 32 mm below surface В
Therefore, without going beyond half skip, testing from surfaces A and В we
have targets at depths of 13,19, 25, 32,43, 50, 56 and 62 mm.
PLOTTING THE BEAM (VERTICAL PLANE)
We first take the slide from the beam plotter shown in the last chapter (figure
15.19). Onto the probe card, we carefully draw in the beam angle for the
probe that we are using. Figure 16.3 uses a 45° nominal angle for a shear
wave probe with an actual angle of 44°. The beam centre line is drawn at 44°
from both corners of the card and the probe serial number and actual angle
is written onto the card. The card is now related to that particular probe and
no other.
Probe № 123
44°
Fig. 16.3
197
The next thing to do is to draw horizontal lines across the beam centre at
depths of 13,19,25, 32, 43, 50, 56 and 62 mm, across both the beam centres
that we drew (See figure 16.4). Feint pencil lines should be used for this.
Next, you calibrate the timebase accurately for an appropriate range at Shear
Wave Velocity, check the probe Index and probe Angle. In this example
100mm range will do. For this example, we will assume that you are going to
plot the 20dB beam spread for this probe. To do this we must first establish
what 20dB looks like on our display and draw a line across the screen at the
20 dB level. To do this you place the probe on the IOW block and obtain an
echo from one of the holes. The gain is adjusted so that the echo height is full
screen. At this point, you note the gain setting in dB and then, with the probe
in the same position, reduce the gain by 20dB. The signal will have reduced
to about 10% of full screen height and you note the actual screen height.
You then draw a line across the screen at this height using a wax pencil as
shown in figure 16.5.
Fig. 16.5
198
t)nce you have established the 20 dB level, plot the beam spread using the
following procedure: -
a)	Maximise the signal for the hole that is 13mm below the scanning
surface as before, and bring it to full screen height.
b)	Mark the scanning surface with a fine pencil line alongside the beam
index (See figure 16.6). We will call this mark ‘a’.
c)	Scan forward until the signal drops by 20dB. Make another mark
alongside the beam index (figure 16.7). We will call this mark ‘b’.
d)	Scan back, past the maximum until the signal drops by 20 dB to the
other edge of the beam and again mark the block (figure 16.8). We will
call this mark ‘c’.
e)	You now have three marks on the block, ‘a’ beam centre, ‘b’ bottom
edge, and ‘c’ top edge. (See figure 16.9). Measure a-b and a-c carefully.
f)	This method assumes that the beam centre has been drawn in
properly (i.e., you have measured probe angle accurately).
So on your slide you plot position ‘a’ at the intersection between the
beam centre line, and the 13mm horizontal line (i.e. at the hole
depth) (See figure 16.10).
199
g)	The bottom edge of the beam at that depth is then plotted at a distance
equal to ‘a-b’mm along the horizontal line from ‘a’ as in figure 16.10.
h)	Then you plot the top edge of the beam at a distance equal to ‘a-c’mm
along the horizontal line as shown in figure 16.10
You now have the beam width at the 13mm depth marked on the beam
plotter card and you continue to repeat each of the steps a - h above for
the remaining depths. Once this is complete, you can draw along the beam
edges from your marks as shown in figure 16.11
200
It is usual only to plot to a depth that is sufficient for the thickness of weld
to be tested, in this case to 50mm. Note that the beam edges have been
[projected back to the origin. You must also remember to plot the same
[values for the other side of the plotting card so that it can be used to evaluate
[defects identified from either side of the weld centre line.
The method described above has several possible inaccuracies built-in.
Marking the oily surface of the beam spread block with a pencil can be
difficult and so it has become more common to use a wax pencil. However,
the thickness of line produced by wax pencils is thicker than an ordinary
pencil line and you have to decide which part of the tick line is your measuring
‘edge’. If the line thickness varies, which it will, there will be an inaccuracy
in your measurements. This might be further exaggerated by your reading
of the ruler and by your transferring that measurement to the plotting card.
As an alternative to measuring and plotting probe movements, the following
procedure can be used; we will just follow through the plotting of the beam
centre and two edges at one depth, in this case for the 13mm deep hole: -
a)	Having established the 20 dB level below full screen height, you again
move the probe to maximise the echo from the 13mm hole and use
the attenuator to return the signal to full screen height.
b)	Measure the range of the signal on the timebase, in this case 18mm.
You plot this point by drawing an arc, 18 mm radius, from the ‘0’ on
the plotting card. This should intersect the horizontal line where it
crosses the beam centre line, if you have measured your probe
angle properly (See figure 16.12).
c)	Go back to the block and re-establish the maximum echo at full screen
height. Now move the probe towards end surface F, and note that
j	the echo height falls, as	the range changes. As the falling signal
[	reaches your 20 dB line,	carefully note the range, in this case we
i	perhaps find it is 14 mm.	We plot this, which is the bottom edge of
the beam by drawing an arc 14 mm radius from ‘0’ on the plotter, and
[ noting where it intersects the horizontal line (See figure 16.12).
i
201
d)	Move the probe again to a obtain a maximum echo and then scan
away from surface F to find the top edge of the beam by taking the
range as the signal reaches the 20 dB line, in this case we will say at
24 mm. This point is plotted in the same way so that we now have
three arcs crossing the horizontal line corresponding to the beam
centre, bottom edge, and top edge as shown in figure 16.12.
Repeat steps ‘a - e’ for each of the other holes and again draw the beam
edges through the intersections of the arcs with the horizontal lines (See
figure 16.13). The beam plotted using this method will not be identical to the
beam plotted by the previous method, but is still valid provided you apply the
method that was used to plot the beam to evaluate defect size. Never mix
the two methods.
In practice, rather than stopping to plot each hole as you scan it, it is more
usual simply to jot down the ranges in table form, and plot the whole series
of points at one sitting.
Great care is needed during the plotting stage whichever method you use
Generally, you will find beam spread diagrams for probes between 35° and
50° fairly straightforward to plot. The points can be fitted into the classic
202
beam shape. However, more difficulty will be experienced with a 60° or 70°
probes Firstly, the exact maximum is more difficult to identify because of
the relatively long rise and decay. Secondly, the beam edges are difficult
to define because of the relatively long decay in signal amplitude; a large
probe movement produces a small signal change. During this process you
are much more likely to twist the probe and prematurely cut the signal down
to the 20dB line. The probable result is a series of points, defining the beam
jedge, which cannot be joined by a straight line. You will need to check the
^points carefully, but with the subjective influence of knowing that each point
is too far out, or not far enough out, from the beam centre. Finally, you will
have to draw beam edges that give you the ‘best fit’ for the plotted points.
It is difficult to generalise when talking about beam characteristics, but
the tendency is for beams to be less well defined in shape for rectangular
crystals, than for circular crystals, and for small diameter high frequency
probes, than for large diameter low frequency ones.
PLOTTING THE BEAM (HORIZONTAL PLANE)
The beam profile in the horizontal plane can be determined using the same
holes in the IOW block. The procedure, which is repeated for each depth, is
as follows: -
a)	Obtain and maximise the amplitude of an echo for the hole chosen.
b)	Position a guide strip across the block in line with the heel of the probe
(See figure 16.14).
c)	Scan the probe along the guide until the signal has reduced by 20 dB.
d) Mark the centre of the probe (figure 16.15).
e)	Measure the distance ‘X’ between the edge of the block and the probe
centre line that you just drew.
f)	Measure the distance ‘Y’ between the probe index and the side of the
hole.
g)	Subtract the drilled depth (22mm) from the measurement ‘X’ and the
result is half the beam width at a horizontal distance ‘Y’.
h)	Repeat steps a - g from the other side of the hole to find the width of
203
the other half of the beam (figure 16.16).
i)	Repeat steps a - h for each of the remaining holes.
j)	Draw the horizontal beam spread diagram from your recorded value1
(figure 16.17)
Fig. 16.14
Surfac e distance ----1----1----1----1 - P^EztEzE-------------1--- 1
_____________________—--------------------- 0
Beam
Index
Fig. 16 17
SIZE ESTIMATION (VERTICAL PLANE)
We will consider the case of a defect near the sidewall in a 20 mm thick
single ‘V’ weld (See figure 16.18). At this stage, we will assume that the weld
cap is dressed.
204
Fig. 16.18
Firstly, we find the maximum echo for the defect, and plot its position (figure
16.19 position ’a’) from the surface distance and beam path length, just as
we did in section chapter 15. We then move the probe towards the weld
centre line until the signal has reduced by 20 dB. At this point, measure the
surface distance and timebase range. This information is now plotted on the
slide, but this time the beam path length is plotted along the bottom edge of
the beam (figure 16.19 position ‘b’).
Fig. 16 19
The probe is then moved away from the weld centre line, through the
niaximum, back to the 20 dB drop position where the surface distance and
beam path length are again measured and the third position (‘c’ in figure
16.19) along the top edge of the beam is plotted this is the current position in
the diagram. The three marks on your plotting slide in figure 16.19 show the
defect size and orientation.
205
If the defect has a relatively large dimension in this plane, you may experience
a large amount of probe movement before the signal begins to drop towards
the 20 dB position. In such cases it is useful to plot the orientation of the
defect by plotting some intermediate points using the beam centre range,
provided those points are plotted while the signal amplitude is somewhere
near its maximum value.
SIZE ESTIMATION (HORIZONTAL PLANE)
Commonly known as the length of the defect, this dimension can bt
measured by finding the maximum echo position and scanning the probe
left and right parallel to the weld centre line to establish the left and righ
20 dB points. The total probe movement is then measured, and the surface
distance between the beam index and weld centre line is measured. The
length of the defect is then obtained by subtracting the beam width in mm
at that surface distance on your beam profile diagram, from the total probe
movement. Figure 16.20 shows the probe movement between the 20 dt
limits for a defect near the centre line of a weld.
Fig. 16.20
Suppose the probe movement is measured as 27mm and the surface
distance as 40mm. To find the length of the defect we refer to our horizonta
beam plot (figure 16.21). At the measured surface distance, (40mm) we set
that the beam width is 9mm. We subtract the 9mm from our 27mm probe
movement giving a result of 18mm for the length of the defect.
206
9mm beam width
Surface distance ------1-----1----1-----1----1-
0
Beam
Index
Fig. 16.21
There are some reservations to be considered. You must be alert to the
factors that influence amplitude when using the dB drop method. It is not a
5 safe assumption that amplitude is dropping because the beam is scanning
past the end of the defect. The other factors that you must remember are: -
। a) Area of defect surface - the defect may taper in section giving a
 reduction in cross sectional area within the beam; if this is enough
to drop the signal by 20 dB, you could well plot the cut off point several
:	millimetres before the true end of the defect.
b)	Orientation - the defect may twist and this may cause a premature cut
 off point to be chosen. Sizing the defect by using more than one probe
angle will give you the clue if twisting is present.
c)	Range - there may be a dogleg in the run of the defect, putting part of
its length in an unfavourable position.
d)	Probe rotation - inadvertently twisting the probe as you scan may also
lead to a false result.
e)	Change in surface roughness or couplant may also lead to poor
results.
MAXIMUM AMPLITUDE TECHNIQUE
This technique has at least as much merit as the dB drop system, but does
hot seem to have had the same degree of publicity and acceptance. It is a
Valuable method of crosschecking results obtained by the dB drop system,
and for certain defects, it can be clearly shown to be superior. It takes into
Account the fact that most defects that occur do not present a single, polished
Reflecting surface, but in fact take a rather tortuous path through the material,
Vvith some facets of the defect surface suitably orientated to the beam, and
207
some unfavourably orientated. Figure 16.22 illustrates this, showing a crack
propagating in the weld. The facets that are boldly outlined are those that are
Each of the reflecting facets will be at a slightly different range, and although
they may be too close together to resolve as separate signals, the signal
envelope can nevertheless be regarded as a series of overlapping separate
signals. In fact the envelope may look like figure 16.23 a, b or c, depending
on the degree of range variation from the different facets, and on the
resolution of the equipment.
As the beam is scanned across the surface of the defect, the beam centre
will sweep each facet in turn. As it does, the signal from that facet will reach
a maximum and then begin to fall, even though the main envelope may
be, at that instant, rising or falling in amplitude. As each signal reaches it
maximum, you stop, measure the surface distance to the weld centre, and
the timebase range for that facet, and plot the reflecting point on your beam
plotter. You increase the gain to follow the series of maximum echoes unti
the beam sweeps the last facet, which you plot. After that the signal only
falls, no further maximum is observed. On your beam plotter you wilt now
have a series of points marked, which trace out the extent of the defect.
Facets poorly	Better
resolved	resolution
Clearly
resolved
Fig. 16.23
208
I TIP DIFFRACTION SIGNAL
i! The maximum echo technique can exploit a phenomenon reported by
|! sproule some years ago. This is the diffraction of the sound beam at the
I tip of a defect. The diffraction signal then radiates from the tip of the defect
; as a circular wave front, rather like the ripples on a pond from the point at
i which a stone enters the water. The signal will eventually reach the material
: surface over a wide range of the surface. It can be detected by the probe
I even though the defect orientation may appear most unsuitable (See figure
16.24). The amplitude of this signal is rather weak, about 30 dB lower than
a corner reflector at the same depth.
Fig. 16.24
This tip diffraction signal will generally be the last maximum echo that you
see. If the defect is very smooth, there will probably only be three maxima,
the main echo and the tip diffraction signal from each end. Let us take a
practical example that illustrates one limitation of the dB drop method and
the use of tip diffraction to advantage. Figure 16.25 shows the A2 block,
looking at the 300mm edge. The machined slits marking the 100mm radius
can be seen to be 4mm deep, and form a corner reflector in the block.
Using a 70° probe positioned 69mm back from the slot; you should get
El bottom corner reflector signal at 73mm range. Maximise this by probe
hnovement and set it to full screen height. Plot the maximum point on your
209
beam plotter - it should come on the 25mm depth line. Now go through the
20 dB scanning procedure to determine the vertical extent of the ‘defect’
plotting top and bottom beam edge reflection points. If you do it carefully,
all three points (top edge, bottom edge and beam centre) should coincide
suggesting that the defect has no vertical extent. In other words, the system
hasn’t worked. This will always be true if the defect is as smooth as our slit
(e.g. lack of penetration), and you will only get a positive result if the top of
the slit has a slag inclusion associated with it.
If you now go back to the 69 mm surface distance and again maximise the
echo, we will try the maximum amplitude method with tip diffraction giving us
a second maximum to plot. Firstly, measure the surface distance (69mm) foi
the bottom corner maximum, and the timebase range (73mm). Plot these out
and you should get a mark at the 25 mm depth line on your plotter, just a:
before. Now adjust the gain so that the signal amplitude is at about one fifth
full screen height, note the amplitude and the gain setting. Then increase thi
gain by 30 dB and scan towards the slot. As the main signal is going down
in amplitude and moving to the left on the CRT, watch the leading edge c
the signal closely, and you will see a new signal creep up the leading edge
and maximise somewhere near one fifth screen height. As this diffractio
signal reaches its maximum, note the surface distance and the range (thes<
should be 58mm and 61 mm respectively). Plot this point out, and you shou1
get a mark on your plotter 4 mm above the bottom corner mark, showing th
vertical extent of the slot.
Another practical exercise which you can do to practice the maximui
amplitude method is to use the group of five holes in the IOW block. Thet
give a signal pattern rather like that from a jagged crack. If you scan this ar
plot each maximum, you should get a series of 5 points corresponding to tl
holes, and the row of points lie at 10° to the vertical. If you use the 20 dB dr< 1
method, you should also be able to construct a line whose length is eqi
to the distance between the first and fifth hole, and which lies at 10° to tl
vertical. For this sort of reflector, both techniques work.
210
Both methods we have discussed attempt to establish the true dimensions of
the defect. An experienced and conscientious technician will not follow either
technique blindly, but will vary the procedure according to the characteristics
of the reflecting surface. Both techniques may be used in order to confirm
a critical defect size. For example, when using the dB drop technique to
find the length of a smooth sidewall non-fusion defect, it can be shown that,
using the crystal diameter as beam width (i.e., assume a parallel beam),
or using a 6 dB drop technique produces a more accurate result than the
20 dB technique. It is useful to note which type of defect each method is
relatively good at measuring, and which type each method is relatively poor
at measuring.
Nature of Defect	dB Drop	Maximum Amplitude
Smooth without volume	Good if tip	Poor
(Lack of penetration	diffraction is	
lack of side wait fusion)	used	
. Smooth with volume (Porosity, pipes, etc)	Very good	Good
Planar, irregular profile (Cracks, lamellar tearing)	Good	Very good
Volume, irregular profile (Porosity clusters, slag)	Good	Very good
(Because of the differences in interpretation and personal choice of technique
(variations, there is likely to be a wide variation in the size estimation of the
same defect by several technicians. This fact has caused much concern in
.the past and stimulated research into improved methods of defect sizing.
(The latest of these, and the most successful if the TOFD system.
i
PEFECT SIZING USING TOFD
B"heTOFD technique, first used by Silk in 1977, uses tip diffraction to identify
211
the top, bottom and ends of a discontinuity in one pass. Silk chose to use an
angled compression wave for the TOFD technique rather than a shear wave,
for two reasons. Firstly, the tip diffraction signal is stronger than a shear wave
diffraction signal, and secondly, a lateral wave is produced which can be used
to measure the horizontal distance between the transmitter and receiver.
The tip diffraction signal is generated at the tip of the discontinuity - effectively
a ‘Point’ source. According to Huyghens, a point source produces a spherical
beam. Figure 16.26 shows a typical TOFD transducer set-up on a component
with a vertical discontinuity. There are four sound paths from the transmitter
to the receiver. Path ‘A’ is the lateral wave path travelling just below the
surface. Path ‘B’ is the tip diffraction path from the top of the discontinuity.
Path ‘C’ is the tip diffraction path from the bottom of the discontinuity and
path ‘D’ is the backwall echo path.
Figure 16.27 shows a typical unrectified trace for the four signals. Note the
phase relationships, A and C are in opposite phase to В and D. The important
difference to note is between В and C - the top and bottom diffraction signals
are in opposite phase. This phase difference allows the practitioner to identify
those points.
Backwall ‘O'
Fig. 16.26
212
Assuming that the diffracting tip is centred between the two transducers, the
depth of the tip below the surface can be calculated from: -
Depth -
( BPL Y ( HD Y

2
Where: BPL - Beam path length for the signal in question
HD	= Beam path length for the lateral wave.
The distance measurements taken from the ultrasonic trace must be made
from the same part of each waveform. In the trace shown in Figure 16.27,
the largest half cycle would be selected. For signals A & C this is negative
and for signal В positive. The advances in computer technology have
made it possible to carry out all the calculations and plotting to be handled
automatically and stored for subsequent evaluation. The method that has
been chosen to display this TOFD data presents the information in a special
‘В-scan’ form that is easy to assimilate. The way in which the positive and
negative half cycles are displayed needs explaining.
In a conventional В-scan image, the ‘slice’ is taken across the weld
perpendicular to the centre line. In the TOFD display, the ‘slice’ is taken
along the weld (figure 16.28). However, whereas the conventional В-scan is
a relatively thin slice, the TOFD image represents the volume between the
probes as they scan along the weld. The presentation is known as a ‘D-scan’.
213
Cap
Weld length
Root
Fig. 16.28
An echo arriving at the receiver is a pulse of a certain pulse width and
amplitude. In conventional В-scan displays, this pulse is displayed as a bright
spot whose diameter is proportional to the pulse width and whose brightness
is proportional to the signal amplitude. In some ways, it is like a broad pencil
tip that can be used to draw pictures in light or bold broad strokes. The pulse
is really a short burst of a few cycles of alternating waveform. In the TOFD
system, the waveform is depicted in greyscale with positive going half cycles
tending towards white, and negative going half cycles tending towards black
(see figure 16.29). This type of display will allow us to identify phase change
so that we can discriminate between he lateral wave, top and bottom defect
signals and backwall.
Fig. 16.29
This allows particular half cycles to be identified for measurement purposes,
and phase changes to be recognized for determination of top or bottom
echo. Figure 16.30 shows a typical computer screen for a TOFD inspection.
The image shows details of the component (in this case, a weld) as well
214
as the TOFD D-scan image and an А-scan trace In this image, left to right
represents the component thickness, and the vertical dimension represents
scan length.
The А-scan trace shown corresponds to a slice through the weld at the
location indicated by the ‘cross hairs’ of the cursor. The striped band on the
left of the TOFD image represents the lateral wave, and the bold striped
band to the right of the image represents the backwall echo. The difference
ih boldness is due to the different signal amplitudes. Following the horizontal
‘cross hairs’ and about half way between the lateral wave and backwall
‘stripes’, a series of feint ‘horse shoe’ shaped stripes can be seen. These
are diffraction signals from a small discontinuity. The А-scan trace shows the
signal clearly.
Fig. 16.30
In this example, the discontinuity has a very small dimension in the through-
thickness dimension, but close study of the А-scan shows a small phase shift
in the last half-cycle of the discontinuity signal. This tells the practitioner that
the distance from top to bottom of the discontinuity is about the same as the
pulse length for this particular discontinuity.
215
Fig. 16.31
hag'Deisfy

A much bolder indication can be seen towards the top of the lateral wave
line suggesting a discontinuity at, or just below the surface. In figure 16.31,
the cursor has been moved to this location. The lateral wave signal can be
seen to be longer and stronger than at the previous location. The fact that
the wave shape stays in phase suggests that the diffraction echo, which
is extending the signal, has the same phase as the lateral wave. In other
words, it is a bottom tip signal. However, it is not possible in this case to
see where the lateral wave ends and the bottom tip begins, and so it is not
possible to say how deep the discontinuity extends below the surface. The
TOFD method is limited in its ability to size near surface discontinuities when
the arrival time difference between the lateral wave and the diffraction signal
is similar to pulse length. Near surface resolution when using TOFD can be
a bit confusing if you look at it from a conventional ultrasonics point of view.
Imagine a top surface crack 4mm deep. At 5MHz, it represents more than
two wavelengths at compression wave velocity and with a reasonably short
pulse of two cycles; you might expect to resolve the bottom of the defect.
However, the path difference between the lateral wave and the tip diffraction
signals for a probe separation of 80mm is only 0.4mm and this is about the
same as the wavelength for 15MHz (See figure 16.32). You would need a
15MHz transmitter with only one cycle in the pulse to resolve the crack.
Fig. 16.32
216
The transducers used in TOFD techniques are angled compression wave
transducers. The common angles used are 60° and 70°, although other
angles may be used if the component thickness makes it necessary. The
design and construction of the transducer is important in order to promote a
good lateral wave. Previous theory has suggested that a shear wave should
also exist in the component and this is true, it does. Figure 16.33 shows a
little more of the trace for the above example. On the extreme right of both
the А-scan and TOFD D-scan, the shear wave can be seen. Since it arrives
well after the other signals, it does not present a problem in this application.
Shear wave
Fig. 16.33
Scanning with the TOFD system is fast and many scanning systems are
motorized. They all require distance encoders so that the D-scan image can
be constructed. The vertical extent of those defects that can be resolved is
many times more accurate than other sizing systems.
DISTANCE, GAIN, SIZE (DGS) SYSTEM
The DGS system was first introduced by Krautkramer in 1958 with the
intention of standardising evaluation techniques and thus reducing variation
in reporting the size of a particular defect from one practitioner to another.
We will look at the system step by step to see how it works.
pefore we start, let us quickly fix in our minds some of the positive benefits
bf the DGS system. It enables us to: -
I.
| a) Choose a sensible gain level to use for a given defect size and range.
i b) It tells us the smallest defect we could possibly detect at a given range
217
c)	It tells us the useful gain available from a given probe/flaw detector
combination.
d)	It gives us the basis for a go/no go acceptance or rejection system.
e)	It can, under certain circumstances, give us an indication of the order
of size of a particular flaw, provided that flaw does not exceed the
beam width in its largest dimension.
In chapter 7, we saw how the intensity of the beam decreases with distance
from the probe (See figure 7.10). In the far field, the intensity follows a law
of the e~2" type, where « is the attenuation coefficient for the material being
tested. We can show using area-depth reference block, that for reflectors
that are flat and at right angles to the beam, at a given scanning depth, there
is a relationship between the area of the reflecting surface and the amplitude
of the signal. The DGS system makes use of the laws of sound distribution
and reflection to relate the amplitude of signal from various sizes of ‘perfect
disc reflectors’ placed at various depths in a material.
Figure 7.10 in chapter 7 showed Intensity against Distance. Suppose we
plotted a graph showing the amplitude of the echo from a given sized flat
bottomed hole at various distances. For example, if we had suitable probes
and test blocks we might do the following experiment: -
a)	Select 5MHz 12 mm diameter compression wave probe.
b)	Select flat-bottomed hole targets in steel, 2 mm diameter, at depths o'
4, 8, 12, 16, 20, 30,40, 60, 80,100,150, and 200 mm.
c)	Set up a maximum echo for the hole depth nearest to two near field
distances (in this case the hole 40 mm down) and adjust the gain tc
give a signal amplitude of 60% full screen height. Leave the gain se1
at that value.
d)	Scan each of the holes in turn starting at the 4 mm deep hole, notinc
the amplitude of the signal as it is maximised {i.e., beam centre direct!'
over hole centre).
e)	Plot a graph of amplitude against hole depth.
218
A table with typical results is shown below
Hole depth (mm)	Screen Height (Full screen = 5 divisions)
4	1.6
8	1.35
12	1.6
16	3.75
20	4.75
30	4.25
40	3.0
60	1.6
80	0.95
100	0.6
150	0.3
200	0.15
These results are plotted in figure 16.34.
Fig. 16.34
We could imagine then, for this 5 MHz probe, making a whole series of
^rnplitude readings, at a fixed gain setting, and different targets from a full
(backwall, down to say a 1 mm flat bottomed hole, at a series of depths for
219
each reflector. We could then plot these on one graph. (See figure 16.35;
The series of curves shown in figure 16.35 are purely diagrammatic, but 
they were a true set of curves, the gain used would be such as to produc
a signal amplitude just on full screen height for a back wall echo on a bloc!
20 mm thick. At this gain set, an unknown flat-bottomed hole reflector г
55mm depth might give an echo amplitude of 50% full screen height. Draw <
horizontal line drawn across the graph at 50% full screen height. Then dra>
a vertical line up the graph from 55mm depth. The two lines can be seen t(
intersect close to the curve for the 8 mm diameter target. We could say tha
the unknown target is approximately 8 mm diameter if it was a disc reflectoi
or for a flaw of unknown shape, we could say it was equivalent to an 8 mr
diameter disc reflector. Note the words ‘approximate’ and ‘equivalent’. Then
are other factors to consider such as the shape and orientation of the flaw
and it is unlikely that any real flaw will be a true disc shape.
You will notice that the curves for reflectors smaller than the probe diameter
roughly follow the inverse square law in the far field. That is, if you double the
distance, the amplitude drops to one quarter of its original value. However,
in the curve for the back wall echo, which is an infinite reflector (i.e., much
bigger than the beam), the amplitude is inversely proportional to the distance;
if you double the distance the amplitude only drops to one half of its origina
220
^alue. You can check this for yourself by taking a selection of samples of the
*ame material but differing thickness, setting the gain for one sample, and
'hen checking all the others at the same gain and noting the amplitudes. If
^ou do this, make sure that the surface finish and couplant are the same for
jach reading, and that the samples are all bigger than the beam.
<rautkramer in 1958 used these principles to develop a standard set of
purves called the DGS diagram (AVG diagram in the original German Text).
fnhe scales chosen are not quite the same as figure 16.35, so we will explain
hem by looking at the limitations of the curves in figure 16.35.
I!)ISTANCE
:irstly, figure 16.35 was produced for a particular probe (diameter and
requency) and for a particular material. Therefore, the beginning of the far
ield in millimetres is the near field distance for that probe and material; it will
liffer for other probe diameters or frequencies, and for different materials. In
he general case, shown in the DGS diagram, the units used on the distance
>cale are near field units, and for convenience, a logarithmic scale is used,
n figure 16.36, the horizontal ‘distance’ scale goes from 0.1 to 100 near field
iistances. To use the diagram you need to work out the near field distance
{for your probe and material.
I
GAIN
[the next problem with figure 16.35 is that the gain is fixed. Remember, we
Iset a back wall echo from a 20 mm thick specimen at full screen height.
I/Vith this gain set, a 2 mm hole at 60 mm depth is small. To see it clearly we
would have to turn up the gain, which defeats the object of the exercise. So,
|ignal height is the wrong parameter to study. That is why we developed the
boncept of a calibrated attenuator - so that we can compare signal amplitude
Variations that are bigger than the screen height. In the DGS diagram, we
use gain, measured in decibels for the vertical axis.
221
03 04 0-5 0607060910	2	3	4 S 6 7 8 9Ю 20	30 40 SO 6070S030100
Fig. 16.36
222
(To achieve this we set our reference echo as before, but note the attenuator
(setting. We call this 0 dB gain. We then scan each disc or back wall reflector
(as before, but each time we note how much gain (dB) we need to bring each
 signal up to the reference echo height, and on our graph use this gain value
on the vertical scale. In figure 16.36, the vertical axis is measured in dB,
(0 dB (our biggest signals) being at the top, and increasing down the scale
(for smaller signals. For instance, the dip in the 0.1 curve at 0.5 near field
(distance, is 44 dB lower than the reference level.
A side effect of using the decibel scale for gain, because it is a logarithmic
I unit, is that the exponential curves in figure 16.35 become straight lines in
^figure 16.36.
| SIZE
I
{The third problem with figure 16.35 is similar to the first. The relationship
I between the curves for those particular sizes of disc reflector depends upon
\ the probe, and would change if the diameter changed. For this reason, in the
general case represented in the DGS diagram, the disc sizes are shown as
a proportion of crystal diameter. Thus, the 0.2 curve would represent a 2 mm
disc reflector for a 10 mm diameter probe, but it would also represent a 4mm
disc for a 20mrn diameter probe.
 Summarising, the DGS diagram in figure 16.36 relates Distance along the
beam in near field units, to Gain in dB compared to a particular back wall
reflector, and Size of the disc reflector as a proportion of crystal diameter. It
Ignores loses due to changes in surface roughness, couplant, or attenuation
iin the test material. Corrections for these losses will be dealt with later.
USING THE BASIC DGS DIAGRAM
We will now go through an imaginary ‘sizing’ exercise, but neglecting for
{the time being the losses mentioned in the last paragraph. We will assume
'that we are examining steel plate 100 mm thick, using a 10 mm 5 MHz
I
s	223
compression wave probe. We discover a small defect indication at a depth of
84 mm in the steel. We now want to ‘size’ that flaw using the DGS system to
find the equivalent flat-bottomed hole size
PROCEDURE
D2 100
a)	Calculate the near field distance from: - NF = —— = --—— = 21 mm
'	4k 4x1.192
b)	Choose a reference back wall on the A2 block approximately equal to
near field (25 mm range).
c)	Set the 25 mm back wall to a chosen reference amplitude (half screen
height for example).
d)	Note the attenuator reading (40 dB for example).
e)	Place the probe on work piece and locate the defect again, maximising
the signal.
f)	Bring this to the reference amplitude and note gain reading (72 dB for
example).
g)	Calculate the ‘GAIN’ (difference in the two readings, 72 - 40 = 32dB).
h)	Calculate defect ‘DISTANCE’ in near field units (84 4- 21 = 4NF).
i)	On the DGS diagram, mark the point of intersection of the calculated
GAIN and DISTANCE (32dB and 4NF).
j)	Choose the nearest ‘SIZE’ curve to this point (0.2 line for this example)
k) Calculate the equivalent flaw size by multiplying the probe diameter by
the SIZE curve found above (10mm x 0.2 = 2mm diameter)
This tells us that the defect that we found cannot be smaller than the
equivalent area of a 2mm diameter flat-bottomed hole at that depth. Of
course, it could be larger, and almost certainly is larger because: -
a)	The defect is not a perfect reflector with a smooth flat surface parallel
to the scanning surface.
b)	The surface of the specimen is probably not as smooth as the A2
block and therefore the coupling efficiency will not be as good.
c)	The attenuation in the specimen is probably not the same as the A2 block
224
TRANSFER LOSSES
Losses due to differences in couplant or surface roughness between the
reference block and the work piece are sometimes called transfer losses. In
the case of compression wave techniques, we can usually eliminate transfer
losses by using a back echo from the work piece as our reference point
instead of the A2 block. This back echo may not be at one NF distance,
but that does not matter because the DGS diagram allows for this. Let us
looks again at the example above, but this time, use the back echo from the
specimen (at 100 mm range) as our reference.
PROCEDURE
a)	Calculate back wall depth in near field units: -100 + 21 =5 NF’s approx.
b)	Set back wall echo to reference amplitude and note attenuator
reading (52dB for example).
c)	From the DGS diagram, the 5NF line crosses the backwall curve at
10dB Gain.
d)	Locate and maximise flaw echo and note attenuator reading (68dB for
example).
e)	Calculate defect depth in near fields (4 NF’s in our example).
f)	Calculate the GAIN (dB difference) 68 - 52 = 16dB.
g)	Now add the 10dB from c) above and the ‘GAIN’ becomes
16 + 10 = 26 dB
h)	On this new “GAIN” line, find the point where it crosses the 4NF line.
i)	Choose the nearest ‘SIZE’ curve to this point (0.3 line for this
example).
j)	Calculate the equivalent flaw size by multiplying the probe diameter by
the SIZE curve found above (10mm x 0.3 = 3mm diameter)
If this was a real case and we had followed both procedures to arrive at
the same answers (i.e. an equivalent disc size of 2mm both times), it would
mean that there was no transfer loss. In fact, we arrived at two different
‘GAIN’ values 32dB in the first case and 26dB in the second. The transfer
225
loss was 6dB, but the second method automatically corrected for this. In
practice, it is the second procedure that is normally followed for compression
wave probes. Later we will look at the use of DGS diagrams for shear wave
probes, and we will deal with transfer losses for shear wave probes then.
ATTENUATION LOSSES
The back wall echo (infinite reflector) curve on the DGS diagram follows a
law in which amplitude is inversely proportional to distance beyond about
three near fields. So the curve drops 6 dB each time we double the distance.
In our example, the specimen has been 100 mm thick, is well beyond three
near field distances for our probe. So, if there were no attenuation loss, the
dB difference between a 1st back echo and a 2nd back echo should be 6
dB. Suppose the actual difference measured was 13 dB, then the extra 7 dB
would be due to attenuation of the beam between the first and second back
echo (i.e. a return trip of 200mm). The attenuation of the material is then:
^ = 0.035 dB/mm (35dB/m).
In our example, we found that the “GAIN” for the flaw was on the 26dB line.
But if attenuation of 0.035 dB/mm was being experienced, then of that 26 dB,
84 x 2 x 0.035 = 5.88dB was due to attenuation, and only 26 — 5.88 = 20.12
was the real “GAIN” of the flaw. If we now go back to the DGS diagram, we
will see that the intersection of the 20dB “GAIN” level, with 4 NF “DISTANCE”
line, now lies midway between the 0.4 and 0.5 “SIZE” curves, and the
equivalent flaw size is 4.5mm instead of our previous estimate of 3mm. If we
ignore attenuation, we will underestimate our flaw size, as you can see.
USING DGS TO SET APPROPRIATE TEST SENSITIVITY
One important use of the DGS system is in the determination of an
appropriate test sensitivity for a particular inspection. However, this requires
‘someone’ to specify the smallest equivalent disc size that must be detected,
and too often, nobody takes this responsibility and the poor technician is left
to his own devices. The following example shows how to set sensitivity using
the DGS system, if a minimum disc size has been specified.
226
G
Fig. 16.37
227
Suppose for example, we were to be given the following task: -
Plate thickness	50 mm (2.5 Near fields approx.)
Probe frequency	5 MHz
Probe diameter	10 mm
Near field distance	21 mm
Attenuation	0.04 dB/mm
Smallest equivalent disc	3 mm diameter (0.3 “SIZE” curve)
Recording level	40% full screen height
In this example, the 3 mm diameter disc reflector specified is 0.3 times the
probe diameter, so we will use the 0.3 “SIZE” curve in figure 16.37. The
greatest scanning depth will be the back wall echo that is at 2.5 near fields
on the “DISTANCE” scale. If we run along the 0.3 “SIZE” curve we can see
that the lowest point of the curve is in the near field close to the specimen’s
top surface at a level of 22 dB. We can draw a line across the graph at the
22dB level and mark it ‘Recording level’. The 2.5 NF “DISTANCE” line cuts
the back wall echo line at 4 dB, that is 18 dB lower than our ‘Recording
level’. We can now go to the specimen and obtain a backwall echo which
we peak at an amplitude of 40% full screen height, and then turn up the gain
by a further 18 dB. The test sensitivity is now set and we know that any flaw
with an equivalent disc size of 3 mm will produce a signal of at least 40% full
screen height.
REPORTING LEVEL
With our sensitivity set in this way we know that we can ignore any signal
lower than two fifths full screen height, because we have set the sensitivity
for the worst possible circumstances. However, if any particular flaw reaches
or exceeds 40% full screen you cannot assume that the flaw is reportable:
you first have to correct for attenuation, and the actual depth of the flaw,
to see if it exceeds the 0.3 “SIZE” curve. We can speed up this correction
process by plotting an attenuation curve on the DCS diagram.
228
ATTENUATION CURVE
When we set our sensitivity, we used a back wall echo as our reference. The
pulse had been attenuated by 50 x 2 x 0.04 = 4dB in travelling out and back
through a 50 mm thickness of material with an attenuation of 0.04dB/mm.
Although we did not notice it, 4dB of that basic gain we used was to allow
for this attenuation. But, at the top surface there has been no attenuation, so
our test sensitivity is 4dB too high at the top surface, 2dB too high half way
through the specimen, and OdB at specimen thickness. If we plot three points
22dB + 4dB at the top surface, 22dB + 2dB at half way (1.25 NF) and 22dB
+ OdB at 2.5 NF, we can draw in the attenuation curve shown in figure 16.37.
When we now obtain a flaw echo at a particular depth that is above 40%
full screen height, we correct the amplitude by reducing the signal height by
the difference in dB between the 0.3 “SIZE” line and the attenuation curve
at the defect depth. If the signal is still above 40% screen height, then it is
reportable.
For example, let us suppose that during our examination of this component
at the described test sensitivity, we detect a flaw at 42 mm depth giving a
signal height of 40% full screen height. To find out if this a reportable defect
we would use the following procedure: -
42
a)	Calculate defect depth in near field units — = 2 NF’s
b)	At 2 NF’s on the DCS diagram note the dB difference between the
attenuation curve (22.5dB) and the 0.3 ‘SIZE’ curve (16.5dB) = 22.5dB
-16.5dB = 6dB.
c)	Reduce the gain by 6 dB and note the defect amplitude (40% full
screen).
d)	ANSWER defect is just reportable.
If you go through the same exercises for a defect at 21mm depth, you will
see that you need to reduce the gain by 11 dB, so that the signal would be
well below reporting level. Another example might give a result well above
229
the reporting level, and we may then have to work out the equivalent disc
size for the flaw, to include in our report.
ASSESSING USABLE GAIN
Another important advantage of the DGS system is that you can use it
to assess how much useful gain you have for any particular probe, flaw
detector, and specimen combination. From this you can go on to look at the
smallest disc reflector that could be found at any given depth, and lastly the
limit of penetration of the beam, beyond which even a backwall echo will not
produce a readable signal. The procedure follows the rules that we have
already used: -
a)	The first thing to do is to decide on the smallest deflection of the
timebase that you can see easily on the CRT Let us suppose this to be
a signal that is 2mm high. The next thing to decide is the poorest
signal to noise ratio for which you can be sure to distinguish an echo
from ‘grass’. Let us say that this is 2:1. In other words, if you have
2mm of grass on the screen, a defect would have to peak to 4mm
amplitude before you are sure to notice it. The smallest noticeable
signal then is 6 dB larger than grass level.
b)	Having decided this you connect the equipment and starting with the
gain at its minimum value, you increase gain until the average level
of grass across the screen is 2 mm (for this you will need a relatively
long timebase). You note how much gain you have put in. Let us say
for our probe and flaw detector, 52dB of gain produces 2mm of grass.
Your calibrated gain control may be marked up to 120dB of gain, but
you know that if you ever have to increase a signal by more than 52dB.
all you will do is produce more grass. So on our DGS diagram
everything below 52dB is wasted. Moreover, we know that a defect
showing above this grass would have to be at least 6dB bigger
Therefore, our maximum effective gain on the DGS would have to be
at least 6 dB bigger. So our maximum effective gain on the DGS
diagram becomes 52 - 6 = 46dB.
230
c)	Draw a line across the DGS diagram at 46dB everything below that
line is ‘Dead’ (See figure 16.38).
d)	Since all this has been done without a specimen, there has been no
attenuation. The next step is to measure the attenuation of the
specimen and plot an attenuation curve. Let us say that this works out
to be 1 dB for each near field distance. This means that at 2NF’s, 2dB
of our 46dB has been wasted on attenuation, 10dB at 10NF’s and
20dB at 20NF’s. We plot the attenuation curve shown in 16.38.
Everything below or to the right of this curve is dead.
e)	Lastly, we can look at the Dead zone of the probe. Suppose at
minimum gain the dead zone is just about 0.2NF’s, and at maximum
usable gain it is 0.7NF’s. We can plot the line shown on figure 16.38
and everything to the left of that is ‘Dead’.
What is left of the DGS curve in figure 16.38 is the usable range for that
probe and flaw detector for that material. If you are asked to find any defect
in that material which is too close to the probe, or too far away, or too small,
you can tell from the DGS diagram, and don’t need to waste hours scrubbing
a probe over the work piece for nothing.
We have now covered the uses of the general DGS diagram thoroughly,
although we have confined our study to’ the use of compression waves. From
what we have done already, you will see that the DGS system has nothing
to do with the actual size and shape of the defect. It is NOT a sizing system.
However, it does provide much very useful information and allows a good
measure of standardization and repeatability to be built into an inspection
system. The system for shear wave probes follows the same principles, but
with some changes in detail for reasons that will be discussed.
DGS FOR SHEAR WAVE PROBES
When we are using shear wave probes, part of the near field is contained
within the Perspex path length that will vary for different designs and sizes
of probe. It is not practical, therefore, to draw a general DGS diagram for
231
232
all probes, and so individual diagrams are drawn for each design, size and
frequency. Because these DGS curves are ‘custom built’ for a particular
probe, it is possible to simplify the scales. The ‘G’ scale is calibrated in
deciBels as before, but the ‘D’ scale is calibrated directly in mm beam path
distance (as read directly from the timebase), and the ‘S’ scales are flat bottom
hole reflector diameters in millimetres. Figure 16.39 shows a typical DGS
diagram for an angle probe of 4MHz using an 8mm x 9mm rectangular crystal.
SETTING THE SENSITIVITY
The sensitivity is not as simple to set for angle probes as it was for
compression wave probes because you do not normally expect to see a
back wall echo to use as the reference echo. Instead, we need to use a
reference block such as the A2 block as our primary reference. This means
that we must apply a transfer correction to account for coupling differences,
as well as an attenuation correction. We will discuss the assessment of these
values shortly. First, let us work through the following example of setting
sensitivity using given values for the two corrections.
Probe Sensitivity required Maximum range Transfer loss Attenuation A2 block Attenuation Specimen Reference Reflector	4MHz 8x9mm rectangular crystal. To detect 3mm diameter disc reflector 200mm 6dB 0.04dB/mm beam path length 0.08dB/mm beam path length. 100mm radius on A2 block.
PROCEDURE
a)	Set the echo from the 100mm radius on the A2 block to two screen
divisions amplitude.
b)	Read the dB difference on the DGS diagram between the back
wall ‘S’ curve to 100mm ‘D’ intersection, and the 3mm ‘S’ curve to
200mm intersection (figure 16.39). In our example, this is 28dB.
233
c)	Increase the amplitude of the 100mm radius echo by this amount
(+28dB).
d)	Increase the 100mm echo amplitude further, by the transfer loss
(+6dB).
e)	Calculate the attenuation difference between 100mm range on
the A2 block (A = 100 x 0.4 = 4dB) and 200 mm in the specimen
(A = 200 x 0.08 = 16dB) Attenuation difference =16-4=12 dB.
f)	Increase the 100mm echo amplitude by the attenuation correction
(+12dB).
The sensitivity is now correctly set with the 100mm radius echo at an
amplitude of two divisions plus (28+6+12) = 46dB. At this sensitivity, a 3mm
diameter disc reflector will give at least two divisions signal height.
If we want to estimate the equivalent flaw size of defect detected using this
sensitivity, we need to reduce the signal height by the difference between
the recording level (28dB below the 100mm to backwall intersection in our
example) and the attenuation curve. The recording level is shown at the
38dB ‘G’ line in figure 16.39 and the attenuation curve has also been drawn
in using the same principles as we did in figure 16.37) but this time the
difference at the surface is 16 dB decreasing to 0 at 200mm at the rate of
0.08dB/mm. The following example shows the procedure: -
Sensitivity	As above.
Defect range	50mm
Defect amplitude	recording level +	40 dB.
a)	Attenuation correction at 50mm = 12dB (difference between recording
level and attenuation curve)
b)	Corrected defect amplitude = recording level (40 -12) = 28dB.
c)	Read recording level plus 28 dB, at the 50mm depth line on DCS
curve, - coincides with 5mm diameter ‘S’ line.
.’. Equivalent flaw size=5mm diameter.
234
Fig. 16.39
8dB
45 I___________________________I - 1	1 1 I . I 1 I . 1, 111.111 111, 111111 .1111	1 1 1 1 1 I 1 1 1 I I 1 J I i 11
O 10	20	50 too 200	500	1000
401______________________I . i   I_I__i I.l.1 iLd-uiju.l i.j-iu.lujd.iJ-iJ_____________i J. 1 I i_ I  I I
0	10	20	50	100	200	SOO	1000	2000
701__________________1 L i I . 1_1___I__I L_1-..1 . L . . . 1 1 .-1.X_.L-1—1-..LX1.1_ ! 1 I_. 1 ! .1 . —
О	io 20	sb Ю0 200	500 lOOd2000
Flaw distance along surface from front edge of probe (mm)
MEASUREMENT OF SHEAR WAVE ATTENUATION
It is usual to express shear wave attenuation in terms of beam path length
indicated on the timebase, ignoring the fact that the sound travels out and
back. The method used involves two identical probes, one transmitting and
one receiving. Figure 16.40 illustrates the technique, the procedure being as
follows: -
a)	Calculate the skip distance, and half skip beam path length for the
probe angle chosen.
b)	Calibrate the timebase for sufficient range for at least one full skip
beam path length.
c)	Using a guide to align the transmitter and receiver, position the probes
one skip distance apart (positions T and R in figure 16.40).
d)	Adjust the sensitivity to bring the received signal to half screen height.
Note the attenuator reading (A1).
e)	Move the receiver to two skip distances (position R2).
f)	Bring the signal back to half screen height and note the new attenuator
reading (A2).
g)	Calculate the gain difference in dB (A1 - A2).
h)	Calculate the attenuation in dB mm from: -
Attenuation = A1 -A2 4- Full skip beam path length (dB/mm)
Fig. 16.40
236
MEASUREMENT OF TRANSFER LOSS
The procedure for measuring the transfer loss between the specimen and
the A2 block, for angle probes, is as follows: -
a)	Calculate full skip distance and half skip beam path length for the A2
block and the specimen, for the probe chosen.
b)	Using a guide to align transmitter and receiver set the probes one skip
distance apart on the A2 block as shown in figure 16.41.
c)	Adjust the received signal to half screen height and note the attenuator
reading (A1).
d)	Using the guide to align the probes, set the probes one skip distance
apart on the specimen, as shown in figure 16.42.
e)	Adjust the received signal to half screen height and note the new
attenuator reading (A2).
f)	Calculate the gain difference between the two signals (АЗ) = A1 - A2.
g)	On the DGS diagram note the gain difference between the intersection
of the VI half skip distance line and the backwall ‘S’ curve, and the
intersection of the specimen half skip and the backwall ‘S’ curve (A4).
h)	Calculate the difference in attenuation between half skip distance in
the A2 block, and half skip distance in the specimen (A5).
i)	Calculate the transfer loss in dB from: - Transfer loss =3-(A4+A5) dB.
Specimen
A2 Block
Fig. 16.42
237
USE OF TRANSPARENT DGS SCREEN INSERTS
Some manufacturers, notably Krautkramer, produce DGS scales on Perspex
screens that can be clipped over the screen, and DGS information read
directly from the signal peak. An example is shown in figure 16.43. Modern
digital flaw detectors allow a computer generated DGS scale to be overlaid
on the LCD panel. Transfer and attenuation corrections can be calculated in
the normal way, and applied to the gain setting to enable direct readings of
equivalent flat-bottomed hole size to be made.
In order to set the basic sensitivity to a level that corresponds to these DGS
curves, reference targets are used. R1 is the 100mm radius of the A2 block
R2 is the 25 mm radius of the A4 block C1 is the 1.5mm drilled hole in the A1
block C2 is the 1.5mm drilled hole in the A4 block
To set the sensitivity you position the peak of the appropriate signal in the
circle marked R2 and C2 (or R1 and 01) as appropriate, and then increase
the gain by the amount specified (i.e. R2 + 30dB or 02 + lOdB in figure
16.43). You then increase the gain further by the appropriate amount for
transfer loss and attenuation at the deepest range at which defects may
arise.
238
To assess equivalent flat-bottomed hole size when a defect is encountered,
you need to adjust the gain to suit the attenuation at the defect range and
then note the nearest ‘S’ curve to the peak of the signal.
DAC METHOD
The DAC method, like the DGS method is designed to produce standardisation
of inspection and reporting rather than to measure the dimensions of the
flaws. The procedure for setting test sensitivity is described in detail in the
inspection standards. In this example, we will describe a typical approach.
The procedure is related to a basic reference block made from the same
material, and of similar thickness and surface condition, to the work pieces.
Often, the standard calls for the calibration block to be made from material
produced in the same batch as the test item. An example of a DAC calibration
block has been described in chapter 8 and illustrated in figure 8.6.
The ‘primary reference’ (term used in the standards) is set by obtaining a
signal from the drilled reference target, scanned from a beam path length
just into the far field. This is roughly one-quarter skip or 2” whichever is the
less. The signal amplitude is adjusted to 75% of full screen height. The probe
position is shown as position 1 in figure 16.44 and the screen presentation is
shown in figure 16.45. The probe is then moved to other locations (positions,
2, 3, and 4 in figure 16.44) and the signal amplitude marked on the screen
(See figure 16.45), for each position. A curve is drawn joining these points,
rather like a DGS curve. This is called the Distance Amplitude Correction
(DAC) curve, (See figure 16.45). This line represents the reference level at
various depths in the specimen. Lines may also be drawn at 50% or 20% of
this reference level.
239
Fig. 16.45
Transfer loss is then calculated between the calibration block and the work
piece. The method described for the DGS system could be used, or a
reflector could be drilled into the work piece and compared to the amplitude
from an identical reflector in the basic reference block. This transfer loss
is added to the gain that has been set for the primary reference. The initial
test sensitivity is then set at twice the corrected reference level (i.e. you add
another 6dB of gain).
Using the test sensitivity described above, you may then locate discontinuity
indications. The sensitivity has to be adjusted back to the primary reference
level + transfer loss (i.e. 6dB down on the initial test sensitivity). The defect
indications are then compared with the DAO reference levels. All indications
greater than 20% of that reference level must be investigated to determine
the shape, identity and location of the discontinuity because cracks are
unacceptable regardless of the signal amplitude. In any event, all indications
greater than 50% of the reference level are to be recorded on the Report of
the inspection.
Although the method to be used to size defects is not specified in each
standard, some do describe a number of recommended procedures.
Acceptance criteria are specified in terms of amplitude (in excess of
240
reference level) and discontinuity length (as a function of specimen
thickness). For defect length and the 6dB intensity drop method of sizing is
commonly recommended. However, care is needed because any indication
from a ‘crack like’ discontinuity is unacceptable. The customer should be
made aware of the difficulties in positive identification of the nature of a
defect.
This section has covered a lot of ground; so don’t try to learn it off by heart.
Practice, instead, the various methods dealt with, at every opportunity, for
there is still a lot of ‘art’ mixed up in the science of ultrasonic flaw detection.
Get to know the limitations of each method, and get to know your own
limitations. Remember, above all, two things: -
-	It is easy to get the right answer when you know how big the defect is!
-	None of the methods described will consistently give the correct size
for all defects.
241
CHAPTER 17
ASSESSING THE PERFORMANCE CHARACTERISTICS OF ULTRASONIC
EQUIPMENT
All items of Flaw detection equipment will possess individual characteristics
and it is important that a check is made to ensure that these come within
certain limits if the equipment is to be fit for its primary purpose. Furthermore,
these characteristics can change with age or use, and it is equally important
to ensure that regular checks are made to detect any adverse changes in
performance. It must be remembered that some performance characteristics
are affected by both the flaw detector performance, and the probe
performance. Therefore, resolution and sensitivity checks are carried out on
the combination of probe and flaw detector you have chosen. If the result is
that the combination is unsatisfactory, you may have to decide whether to
change the probe, or the flaw detector. This can be done by trying the same
test using a different probe, or a different flaw detector.
The following checks can be carried out using the A2 block: -
a)	Timebase linearity
b)	Amplifier linearity
c)	Resolution
d)	Sensitivity or ‘Penetrating Power’
TIMEBASE LINEARITY
Checking the linearity of the timebase means that we are trying to show that
the electron beam travels across the Cathode Ray Tube at a constant speed.
Or, that the digital timebase on an LCD panel is swept at a constant speed.
If this is true, then signals occurring after equal time intervals (for example,
multiple echoes from the back wall) should appear with equal spacing on
the display. If the electron beam speeds up or slows down as it moves along
the timebase, depth information will not be accurate, and we would say that
the timebase is non-linear. Linearity checks should be carried out for depth
242
ranges most commonly encountered during your normal work. For most of
us, a check for the following ranges should be enough: -
a)	25 mm of Steel
b)	100mm of Steel
c)	500mm of Steel
TIMEBASE LINEARITY PROCEDURE
A lamination free sample should be chosen to give at least four back wall
echoes using a compression wave probe for the depth range being checked.
For convenience, four or five back wall echoes should be used and arranged
so that the first coincides with the ‘2.5’ position (4 echoes) or the ‘2’ position
(5 echoes) on the timebase. The position of each of the remaining echoes
is then carefully noted and plotted in the way shown in figure 17.1. The
maximum error you should tolerate is 1% deviation for the range chosen.
However, non-linearity of timebase is seldom a real problem with modern
flaw detectors and the most common cause of apparent none-linearity is the
poor calibration of the time-base zero by the operator. The procedure set out
in chapter 9 should be followed to avoid this potential error. It is important
during the assessment of timebase linearity that timebase readings are
taken as each signal is brought to a common amplitude. This is usually about
50% of full screen height.
Fig. 17.1
243
LINEARITY OF AMPLIFIER
It is helpful if the amplifier boosts weak signals in the same ratio as it boosts
stronger signals. The signals amplitude would then be related to the intensity
of sound being reflected from the defect or interface. This linearity will only
be over a limited range of screen height for most flaw detectors. It is normal
practice to check linearity at each probe frequency. In other words, if you
have 2.5 MHz and 5 MHz probes in your kit, linearity checks should be made
at both those frequencies.
The procedure to be followed, using a compression wave probe and the A2
test block, is straightforward: -
a)	Calibrate the timebase for 250mm.
b)	Place probe on test block to show ten multiples of 25mm
c)	Set the gain so that the ‘n’th echo (usually the first echo outside
the near field) is at a particular amplitude (80% screen height
normally).
d)	Note the amplitude of subsequent echoes (n+1, n+2, n+3 echoes
etc.).
e)	Reduce the ‘n’th echo to one half of its original value and check
that the each of the other echoes have reduced to half their original
value. If they have, then the amplifier is linear.
Deviation from linearity for any particular echo can be expressed as a
percentage relative deviation from the following calculation: -
A -2A
Deviation-—---------x100 % Where: A1 = Original Amplitude
A2 = Reduced Amplitude
The deviation from linearity as a function of amplitude can also be shown
graphically as in figure 17.2 below. Such a curve is only valid for particular
settings of frequency, pulse energy, timebase range, and for the gain settings
involved in the check.
244
MAXIMUM PENETRATIVE POWER
This term describes a check that allows you to compare the energy output for
a particular flaw detector and probe. Repeat checks compare the system’s
current performance with past performance, or with similar equipments. The
test, carried out with a compression wave probe, with the set at maximum
gain, simply tells you how many multiple echoes can be obtained from the
Perspex insert of the A2 block. The method is illustrated in figure 17.3. The
results are expressed as the number of echoes that can be seen, and the
amplitude of the echoes, see figure 17.4. In the example there are 4 backwall
echoes and the fourth echo amplitude is 30% of full screen height.
Fig. 17.3
						
				I		и
						
						
01	23456789 10
Fig. 17.4
245
RESOLUTION
The resolving power of an ultrasonic probe was discussed in chapter 4 where
we said that pulse length could be defined as the number of cycles in the
pulse multiplied by the wavelength. The long pulses give poorer resolution.
However, if the flaw detector is only capable of displaying smoothed rectified
pulses, it is difficult to assess the number of cycles. We can check probe
resolution without knowing how many cycles are in the pulse in two ways: -
a)	We can calibrate the timebase accurately, set the working sensitivity
and then obtain an echo from about half timebase range. We then
measure how much space (in millimetres) the echo occupies on
the timebase. Any two reflectors closer together than this value will not
be completely resolved. In figure 17.5, the timebase is calibrated for
100mm of steel and the echo at midrange occupies 4mm.
b)	We can carry out a standard test on a calibration block to see if two
known reflectors are resolved. This is the standard way of checking
resolution but it suffers from the drawback that it is not related to the
required resolution for a particular application. Figure 17.6 shows such
a standard test using the slot on the A2 block. This test can only be
used for compression wave probes. Figure 17.7 shows a typical trace
for this test for a probe with good resolution that clearly resolves the
three surfaces on the A2 block. Figure 17.8 shows an alternative test
on a block with two concentric drilled holes. The probe is positioned over
the step between the two holes to measure resolution. This type of block
can be used for compression wave probes or any angle of shear wave.
01	23456789	10
Fig. 17.5
246
Fig. 17.7
Plan view
Fig. 17.8
247
PROBE SQUINT
In the case of angle probes, we assume that in a plan view, the centre of the
beam is perpendicular to the front face of the probe. This is almost always
true for a new probe but uneven wear during use can cause the beam to
‘squint’ in either direction. To check for this we can draw a line perpendicular
to an edge of the A2 block as shown in figure 17.9. We then place the side
edge of the probe allong this line and scan forward to obtain an echo from
the bottom corner. As this signal reaches a maximum, we tist the probe,
first in one direction and then in the other. If the signal increases in either of
these manoevres, there is squint. As the signal reaches the new maximum,
we draw a pencil line along the side of the probe. The probe is then removed
and the pencil line extende to meet the perpendicular line and the angle
measured as in figure 17.9.
Fig. 17.9
248
CHAPTER 18
REPORT WRITING
This topic is as important as the ultrasonic inspection itself. Unless the
inspection is properly and fully recorded, it may just as well never have
been carried out. Many organisations have their own printed form on which
the Inspector makes his Report. These will all be different in detail and it is
not worthwhile giving an example here. However, you should ensure that
whichever form you are using, the following information is given, clearly and
concisely: -
a)	IDENTIFICATION
(i)	Date of the Inspection
(ii)	Time of the Inspection
(iii)	Place of the Inspection
(iv)	Customer for whom the work is done
(v)	Inspector carrying out the work
(vi)	Component examined. Serial Number, Description, Material,
(vii) Code, Specification or standards used.
b)	EQUIPMENT
(i)	Flaw Detector
(ii)	Probes, size, frequency, angle
(iii)	Calibration and reference blocks used
(iv)	Couplant.
c)	CALIBRATION
(i)	Sensitivity for all probes used
(ii)	Timebase for all probes used
(iii)	Attenuation and transfer corrections, where appropriate.
d)	TECHNIQUE
(i)	Scans made (limits and coverage with each probe)
(ii)	Sizing method used
(iii)	Recording and reporting level used
249
(iv)	Limitations on inspection quality imposed by shape or situation
of object, time or other factors.
e)	RESULTS
(i)	Indications found
(ii)	Scale drawing showing location and size of defects
(iii)	Relationship between defects found and acceptance standard.
STYLE OF THE REPORT
The Report should be made in plain language and should deal strictly with the
task you were asked to perform. Don’t be tempted to add superfluous detail
just to show how clever or experienced you are. Technical terms should be
used in their correct senses, and initials or abbreviations should only be used
after you have used the full term once in association with that abbreviation,
for example, ‘3mm diameter (ф)’, ‘flat-bottomed hole (fbh)’. Results that can
be shown in tabular form, or in scale drawings, are easier to follow than long
written descriptions. Remember that the person who eventually gets the
Report may not be an expert in Ultrasonics.
250