Final Report

Fault Detection in Small Diameter Pipes Using

Ultrasonic Guided Wave Technology

April 2013
This page is intentionally left blank.
Abstract

Ultrasonic guided wave technology is one of the more recent developments in the
field of non-destructive evaluation. It was introduced in 1998, mainly in the oil and
gas industry, to monitor pipelines for corrosion, defects and loss of metal. Ultrasonic
guided wave is an attractive technology as it requires only exposing the area where
the transducers will be placed, hence requiring minimal insulation removal and
excavation for a buried pipe. In this project small diameter pipes (with diameter less
than 5 cm) are considered due to a request by Scotia Gas Network Ltd. The aim
of this project was to investigate propagation of torsional mode, T(0,1), in various
different scenarios.

Experimental study was conducted on a bare steel pipe with a diameter of 3.4 cm
and a wall thickness of 0.5 cm. A defect was machined onto the pipe which occupied
8.3% cross sectional area of the pipe wall. The aim was to find the frequency at
which the wave must be transmitted, in order to detect the defect located in the
pipe. The results obtained show that if the torsional mode, T(0,1), was transmitted
at a frequency of 50 kHz the defect can be detected. Whereas, when the guided
waves were transmitted at higher frequencies there was high reverberation level that
prevented the defect echo to be clearly distinguished. This high reverberation was
due to the limited number of transducers being used in the experiment. The next
experiment involved rotating the receiving transducer around the circumference of
the pipe so that multiple readings could be taken; it was then averaged to give a
resulting waveform. This experiment demonstrated taking multiple readings around
the circumference of the pipe reduces reverberation level at high frequencies, by
eliminating non-axisymmetric modes as it has phase difference at different positions.

Next, guided waves were propagated through a bitumen coated pipe, to investig-
ate the response of guided waves. The results obtained were poor as guided waves
travelled very limited distance, due to high attenuation caused by the bitumen. Lastly,
guided waves were propagated through a buried bare pipe in order to investigate
whether guided waves can detect the defect. The results were positive as the defect
echo was clearly identifiable. Further work using more transducers, using a chirp
instead of tone burst input signal and testing on a corroded pipe is recommended.
iv

Key words : non-destructive evaluation; pipeline; guided waves; defect detec-

tion; ultrasonic.
Contents

Abstract iii

List of Figures ix

List of Symbols xiii

List of Abbreviations xv

Acknowledgements xvii

1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Aims and Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Brief outline of dissertation . . . . . . . . . . . . . . . . . . . . . . . 3

2 Pipeline Inspection 5
2.1 Ultrasonic Guided Wave Technology . . . . . . . . . . . . . . . . . . 5
2.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Guided waves in pipes . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3.1 Types of guided waves in pipes . . . . . . . . . . . . . . . . . 7
2.3.2 Naming of guided waves . . . . . . . . . . . . . . . . . . . . . 8
2.3.3 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3.4 Choosing a wave mode and frequency for inspection . . . . . 9
2.3.5 Torsional vs. Longitudinal waves . . . . . . . . . . . . . . . . 10
2.3.6 Dispersion curves . . . . . . . . . . . . . . . . . . . . . . . . 11
2.4 Ultrasonic Guided Wave Technology vs. Conventional Ultrasonics . 12
2.5 Generation of guided waves . . . . . . . . . . . . . . . . . . . . . . . 14

3 Experiments 17
3.1 Single position measurement . . . . . . . . . . . . . . . . . . . . . . 17
3.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.1.2 Experimental setup . . . . . . . . . . . . . . . . . . . . . . . 17
3.1.3 Results and analysis . . . . . . . . . . . . . . . . . . . . . . . 21
3.1.4 Discussion and experimental conclusion . . . . . . . . . . . . 26
vi CONTENTS

3.2 Multiple circumferential position measurements . . . . . . . . . . . . 28

3.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.2 Experimental setup . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.3 Results and analysis . . . . . . . . . . . . . . . . . . . . . . . 30
3.2.4 Discussion and experimental conclusion . . . . . . . . . . . . 38
3.3 Bitumen coated pipe . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.3.2 Experimental setup . . . . . . . . . . . . . . . . . . . . . . . 40
3.3.3 Results and analysis . . . . . . . . . . . . . . . . . . . . . . . 43
3.3.4 Discussion and experimental conclusion . . . . . . . . . . . . 46
3.4 Buried Pipe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.4.2 Experimental setup . . . . . . . . . . . . . . . . . . . . . . . 48
3.4.3 Results and analysis . . . . . . . . . . . . . . . . . . . . . . . 51
3.4.4 Discussion and experimental conclusion . . . . . . . . . . . . 55
3.5 Finding attenuation of the bare pipe . . . . . . . . . . . . . . . . . . 57

4 Conclusion 59
4.1 Main findings of this dissertation . . . . . . . . . . . . . . . . . . . . 59
4.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

References 61

Appendices

A Phase and group velocities I

B Calculation for number of transducers required III

C CD V

Word Count - 9998

List of Figures

2.1 Transmission of guided waves along the pipe. . . . . . . . . . . . . . 6

2.2 Flow chart showing the types of guided waves in pipes. . . . . . . . 7
2.3 Schematic representation of pipe geometry. . . . . . . . . . . . . . . 8
2.4 Different modes propagating along the pipe. . . . . . . . . . . . . . . 8
2.5 Classification of reflector kinds in a pipeline. . . . . . . . . . . . . . . 11
2.6 Propagation of waves in Ultrasonic Guided Wave Technology and
conventional ultrasonic approach. . . . . . . . . . . . . . . . . . . . . 13
2.7 Ring of transducers. . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.8 Group velocity dispersion curve for a steel pipe with an external
diameter of 75 mm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3.1 Properties of the bare steel pipe used in the experiment. . . . . . . . 18

3.2 Defect machined on the pipe. . . . . . . . . . . . . . . . . . . . . . . 18
3.3 A module with a piezoelectric transducer in place. The module was
placed on the pipe so that the transducer can produce tangential
displacement in order to generate torsional mode. . . . . . . . . . . 19
3.4 Transmitting and receiving transducers placed at zero degree config-
uration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.5 Setup for the experiment showing two transducers placed on the pipe.
The transmitting and receiving transducers here are placed 6.2 cm
apart from each other. . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.6 Group velocity dispersion curve, for torsional mode T(0,m), for a steel
pipe with an external diameter of 35 mm. . . . . . . . . . . . . . . . 21
3.7 Schematic representation of the expected path line for wave propaga-
tion along the bare pipe containing a defect. The burst signal is
transmitted in both directions of the pipe. . . . . . . . . . . . . . . . 22
3.8 Received signal, showing different echoes, for guided waves transmit-
ted at 50 kHz and the transducers placed at a zero degree configuration. 23
3.9 Amplitude of received signal, for guided waves transmitted at different
frequencies, when a single reading at zero degree configuration was
taken. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
viii LIST OF FIGURES

3.10 Variation in normalised reverberation RMS, for guided waves trans-

mitted at different frequencies, when a single reading at zero degree
configuration was taken. . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.11 Group velocity dispersion for a steel pipe with an external diameter of
34 mm and wall thickness of 5 mm. . . . . . . . . . . . . . . . . . . . 27
3.12 Rotation of receiver transducer around the circumference of the pipe. 29
3.13 Alternative method of performing the experiment by having a collar of
transducers around the circumference of the pipe. . . . . . . . . . . 29
3.14 The first graph shows the received signals for 8 evenly spaced po-
sitions of the receiving transducer around the circumference of the
pipe, for guided waves transmitted at 100 kHz, whereas the second
graph shows the average signal of these 8 waveforms. . . . . . . . . 31
3.15 Amplitude of average received signal, for guided waves transmitted
at different frequencies, when 8 evenly spaced readings are taken
circumferentially around the pipe. . . . . . . . . . . . . . . . . . . . . 32
3.16 Variation in ratio between the relative reverberation RMS, for an
average of eight evenly spaced readings around the circumference
of the pipe and a single reading at zero degree configuration, when
guided waves were transmitted at different frequencies. . . . . . . . 33
3.17 Comparing how the signal to reverberation ratio varies, for guided
waves transmitted at different frequencies, when a single reading at
zero degree configuration and when eight evenly spaced measure-
ments are taken around the circumference of the pipe. . . . . . . . . 34
3.18 Comparing how resulting waveforms vary, for guided waves transmit-
ted at 100 kHz, with variation in transducer positions. . . . . . . . . . 36
3.19 Comparing how normalised reverberation RMS varies, for guided
waves transmitted at 100 kHz, when a single reading at zero de-
gree configuration, when the receiver transducer is placed below the
transmitting transducer and when eight and sixteen evenly spaced
readings are taken around the circumference of the pipe. . . . . . . 37
3.20 Comparing how signal to reverberation ratio varies, for guided waves
transmitted at 100 kHz, when a single reading at zero degree config-
uration, when the receiver transducer is placed below the transmitting
transducer and when eight and sixteen evenly spaced readings are
taken around the circumference of the pipe. . . . . . . . . . . . . . . 37
3.21 Bitumen coated pipe along with the area for which the coating was
removed to place the transducers. . . . . . . . . . . . . . . . . . . . 41
3.22 Transducers placed on the region of pipe where bitumen coating has
been removed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.23 Dimensions and properties of the bitumen coated pipe. . . . . . . . 41
LIST OF FIGURES ix

3.24 Setup for the experiment showing two transducers placed on the
bitumen coated pipe. The transmitting and receiving transducers here
are placed 6.2 cm apart from each other and the bitumen coating is
removed so that the transducers can couple on the pipe. . . . . . . . 42
3.25 Comparing how the amplitude of received signal varies in a bitumen
coated pipe, for guided waves transmitted at different frequencies,
when a single reading at zero degree configuration was taken. . . . 44
3.26 Variation in amplitude of direct echo in non-coated and bitumen
coated pipe. The graphs shows results from normal pipe where the
input burst was transmitted at Vpk−pk of 19 V and from bitumen coated
pipe where the input tone burst was transmitted at Vpk−pk of 19 V and
4.75 V. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.27 Variation in normalised reverberation RMS in a bitumen coated pipe,
for guided waves transmitted at different frequencies, when the tone
burst was transmitted at Vpk−pk of 19 V and 4.75 V. . . . . . . . . . . 46
3.28 Dimensions of the box through which the pipe goes through. This
was then filled with 0.0324 m3 of sand. . . . . . . . . . . . . . . . . . 48
3.29 Part of pipe along with the defect buried in the sand box. The box
contained 50 kg of sand. . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.30 Transmitting and receiving transducers placed on the pipe, at 0 ◦
configuration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.31 Setup for the experiment showing two transducers placed on the pipe.
The transmitting and receiving transducers here are placed 6.2 cm
apart from each other and it has a defect buried in sand. . . . . . . . 50
3.32 Amplitude of average received signal, for guided waves transmitted
at different frequencies through the buried pipe, when eight evenly
spaced readings are taken circumferentially around the pipe. . . . . 52
3.33 Comparing how the average normalised reverberation RMS varies,
for guided waves transmitted at different frequencies, when the pipe
is buried and not buried. . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.34 Comparing how SRR varies, for guided waves transmitted at different
frequencies, when the pipe is buried and not buried. . . . . . . . . . 53
3.35 Variation in ratio between peak amplitude for a buried and non-buried
pipe, when guided waves are transmitted at different frequencies. . . 54
This page is intentionally left blank.
List of Tables

2.1 Benefits of using a torsional mode over longitudinal mode. . . . . . . 10

2.2 Typical range the guided waves propagate with a standard transducer. 12
2.3 Advantages of Ultrasonic Guided Wave Technology for pipeline in-
spection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3.1 Comparison between theoretical and experimental arrival times of

the echoes, in a received signal. . . . . . . . . . . . . . . . . . . . . 23
3.2 Analysis of signals obtained when 8 and 16 readings are taken around
the circumference of the pipe, for guided waves transmitted at 100 kHz. 39
This page is intentionally left blank.
List of Symbols

Symbol Units Description

a m Radius of pipe
c ms−1 Wave velocity
E Nm−2 Young’s modulus
f Hz Frequency
h m Thickness
k m−1 Wave-number (spatial frequency)
L m Length of pipe
m Dimensionless Counter variable / Mode
n Dimensionless Order of circumferential variation
N Dimensionless Number of . . .
r m Radial direction of pipe
SNR dB Signal to Noise Ratio
SRR dB Signal to Reverberation Ratio
U m Displacement vector
x m Distance
z m Axial direction of pipe

Greek Units Description

Symbol

ω rads−1 Angular frequency

θ rad Angle direction of pipe
ν Dimensionless Poisson’s ratio
ρ kgm−3 Material density
△ m Transducer spacing
λ m Wavelength
xiv LIST OF SYMBOLS

Subscript Description

bitumen bitumen
deg degrees
ext external
g group
i initial
int internal
mean arithmetic mean
p phase
pipe pipe
pk-pk peak to peak
r readings
rev reverberation
rms root mean square
s signal
t transducers
List of Abbreviations

Abbreviation Description

CSA Cross Sectional Area

GPIB General Purpose Interface Bus
NDT Non Destructive Testing
RMS Root Mean Square
SNR Signal to Noise Ratio
SRR Signal to Reverberation Ratio
UGWT Ultrasonic Guided Wave Technology
This page is intentionally left blank.
Acknowledgements

Firstly I would like to express gratitude to my supervisors for their enthusiasm, en-
couragement, dedication, support and helpful guidance throughout this dissertation.
I am greatly honoured to have supervisors who added great value to this dissertation
with his vast knowledge and experience of working with ultrasonic technology. His
knowledge and advice have helped me to stay on track and work at a smooth pace.

I am greatly indebted and thankful to my family and would like to dedicate this
dissertation to them. I would like to thank my parents for striving hard to provide
a good education for me. Thank you for all your understanding, love, support and
motivation that you have provided me with at every stage of my life. I would also like
to thank my twin brother, Rohit, for his support and regular humour that has helped
in relieving stress throughout the course.

I am obliged to many of my classmates and staff of the University who supported

me during tough times and provided me with constant encouragement at all times.
Special thanks to my friends Lasse and Upasna for providing me with motivation
and creative ideas that has helped me throughout the dissertation.
This page is intentionally left blank.
Chapter 1

Introduction

1.1 Motivation

Pipelines play an important role in our everyday lives. With energy companies,
spending billions in exploring and extracting fossil fuels; people tend to neglect the
role of pipelines in delivering these fuels to the final consumer.

Pipelines are similar to a human body, as it ages, it deteriorates and regular tests
must be performed to prevent further deterioration. They are similar to other modes
of transportation available in a country, except that they are more capital-intensive
and are long-term assets to a country. Regular inspection of pipelines is a must, in
order to avoid complete replacement of pipelines, which can be expensive. Testing
of pipelines is a very important issue especially in the petrochemical industry to
avoid leakages, as they transport inflammable liquids, which can cause harm to both
the environment and living organisms. These pipelines can be as long as millions
of kilometres, therefore it can be quite difficult to regularly monitor these lines. As
regular maintenance of these pipelines is required it is important to find techniques
that have high sensitivity, reliability, can perform inspections over long distances, are
quick and at the same time are not very expensive to operate.

There are many different techniques available that can be used to detect defects
and corrosion in pipes. Some of the more common techniques include radiography,
eddy-current, magnetic flux, electrochemical and the use of ultrasonic guided waves.
Ultrasonic Guided Wave Technology (UGWT) is one of the more attractive methods
of inspection as this technique allows quick, accurate and long range inspection of
pipes. It only requires excavation during initial instalment of the transducer arrays
that excite the ultrasonic waves and during the replacement of the damaged areas.
This technique was developed in NDT Lab at Imperial College [1] and most of the
research has been performed on large diameter pipes using torsional mode, T(0,1).
In this project, tests were performed using the torsional mode, T(0,1), to detect a
2 Introduction

defect located in the pipe and further tests were performed to understand how this
technique responds when the pipe is coated with bitumen and when the pipe is
buried in sand.

1.2 Aims and Objectives

This report aims to investigate how UGWT can be used to detect defects in a
small diameter pipe. It also looks at increasing the frequency at which guided waves
are transmitted, as high frequency waves are more sensitive to defects. The problem
is that when transmitted at high frequencies there is higher reverberation level, due
to the limited number of transducers used, thereby making it difficult to identify the
defect echo. Project objectives were set at the start of the project to ensure that
there was an appropriate procedure to achieve the aim. Project objectives included-

• Carry out a literature review on UGWT to develop background knowledge,

understand the research that has already taken place and how the technology
operates along with its benefits and drawbacks.

• Conduct tests on a bare pipe using two transducers by transmitting guided

waves at different frequencies to see if the defect can be detected.

• Investigate the effect of transducer separation on reverberation quantity, where

the receiver transducer is rotated around the circumference of the pipe, to see
if there is an improvement in results at higher frequencies.

• Propagate guided waves through a bitumen coated pipe to understand how

the technology responds when the pipe is coated.

• Perform investigation on a buried pipe to check if the defect can be detected

and compare the results with non-buried case.
1.3 Brief outline of dissertation 3

1.3 Brief outline of dissertation

The next chapter gives a general introduction about UGWT and how the techno-
logy operates. A brief overview about the types of guided waves in pipes and the
naming convention used for these waves is provided. It also states the advantages
of T(0,1) mode, the mode preferred for the experiment, over other modes. A com-
parison is provided between UGWT and conventional ultrasonic to understand why
UGWT is better. Lastly, it discusses how the guided waves can be generated.

Chapter 3, describes the experiments performed along with the results and
conclusion for each experiment. First experiment focuses on transmitting guided
waves at different frequencies and taking a single measurement to determine the
frequency at which the defect on the pipe is clearly detected. Next experiment
involved taking multiple readings around the circumference of the pipe to see how
it affects the results. Tests were even performed on a bitumen coated pipe and a
buried pipe to determine how the guided waves respond in this scenario.

Chapter 4, contains the main findings of the dissertation as well as potential

future work.
This page is intentionally left blank.
Chapter 2

Pipeline Inspection

2.1 Ultrasonic Guided Wave Technology

This chapter introduces the basic concepts and theories of ultrasonic guided wave
propagation in hollow cylindrical pipes; it explains how this technology operates and
detects defects in pipes. UGWT is regarded as an attractive non-destructive testing
(NDT) method available for inspection of pipelines. There are two known methods
to inspect defects in pipes, destructive and non-destructive. Destructive testing is
where the sample pipe is taken to a laboratory for the mechanical properties to be
examined. It uses hydrostatic testing to ensure that the pipe is within safe operating
conditions. Whereas, NDT is generally used to detect flaws, cracks and corrosion; it
is a preferred method as the pipeline operation is not usually disrupted [2].

Guided wave inspection, also known as long range inspection, exploits ultrasonic
waves that propagate along a medium and are guided by the geometric boundaries
of the medium [3]. These waves are generated by exciting a transducer with high
voltage electric pulses; these are converted to a mechanical vibration by a trans-
ducer. The waves then propagate through the material and are reflected back if
there are any defects, welds or pipe ends along the path of the wave (see Figure
2.1). The reflected wave is then received by the transducer and converted back to
an electrical signal to be displayed on the oscilloscope. Received signal usually
have higher amplitude than the background noise level [4, 5].

The difficulty with guided waves is that the velocity is highly dependent on
wavelength and frequency, known as dispersion. Hence, dispersion curves are used
as it shows relation between the wave mode, velocities and frequency.

If the wave velocity and the travel time are known they can be used to calculate
the location of the reflection i.e. discontinuity. There are two known approaches
to ultrasonic testing - through transmission and pulse echo. Through transmission
6 Pipeline Inspection

method is used when there is access to two opposing sides of the material being
tested, where the transmitting and receiving transducers are placed far away from
each other. Pulse echo is when only one side of the material being tested is
accessible and therefore both transmitting and receiving transducers or sometimes
one transmitter-receiver transducer is located here. Pulse echo transmission method
is adopted for this project.

Figure 2.1: Transmission of guided waves along the pipe, with a piezoelectric
transducer (green circle) placed in each module.

2.2 Background
Guided waves were first studied by Rayleigh and Lamb, early in the 19th century;
they studied elastic wave propagation in traction-free, isotropic plates; this was later
acknowledged by naming guided waves after them1 [6]. Pochamer, in 1876, and
Chree, in 1889, investigated the propagation of guided waves in a free bar [1]. In
1959, Gazis released a paper where he discussed the propagation of waves in hollow
cylinder. He found an exact solution for propagation of free harmonic waves along
the hollow cylinder, using linear theory of elasticity; he also spoke about evolution of
wave structures of different modes at different frequency [7]. This was followed by
Fitch performing an experiment using axisymmetric and non-axisymmetric modes
to confirm the result suggested by Gazis [8]. Research was also performed by Pao
and Mindilin (1960), Onoe et al, Meeker (1962) and Meitzler (1972) which developed
three dimensional problem of a circular cylinder in vacuum [1]. In 1965, Viktorov
et al. analysed the method of generating guided wave in isotropic layers using a
wedge, both theoretically and analytically [9].

Kumar in 1971 performed a theoretical study of dispersion of axially symmetric

waves in empty and fluid filled circular cylindrical shells of various wall thicknesses
1
Rayleigh guided waves on surface of semi-infinite solids and Lamb guided waves in free plates.
2.3 Guided waves in pipes 7

[10]. In 1979, Silk and Bainton researched the use of guided waves for inspection
of small heat exchange tubing; they found similarities between ultrasonic waves in
hollow cylinder and lamb waves in flat plates. A new notation for the wave modes
was also provided by them [11], discussed ahead in this report. In recent times
Alleyne and Cawley (1995) have reported on how it is possible to excite lamb waves
using dry coupled piezoelectric transducers [12]; Lowe et al. (1998) performed
corrosion screening in pipes using long range guided waves [13] and Alleyne et
al. (2001) used guided waves to detect cracks and corrosion in chemical plant
pipes [14].

2.3 Guided waves in pipes

2.3.1 Types of guided waves in pipes

Guided waves that propagate along a pipe can be generated by axisymmetric

or non-axisymmetric loading. Longitudinal and torsional modes are axisymmetric,
whereas flexural modes are non-axisymmetric. Axisymmetric waves are symmetric
around the pipe axis; this is not the case for non-axisymmetric (see Figure 2.2).

Guided waves
in pipes

Axially sym- Non axially sym-

metric modes metric modes

Longitudinal
Torsional mode Flexural mode
mode

Figure 2.2: Flow chart showing the types of guided waves in pipes.

Torsional (twisting) modes involve a shearing motion with a displacement, rep-

resented as ’U’, in the theta direction (see Figure 2.3). Longitudinal (compression)
modes involve a compressional motion with displacements in the radial and axial
direction. Flexural (bending) modes are more complicated and can travel in either
radial, axial or the theta direction (see Figure 2.4) [15].

In general axisymmetric modes are preferred over non-axisymmetric when it

comes to testing, as they are easier to excite and have relatively simple acoustic
8 Pipeline Inspection

fields [16].

Figure 2.3: Schematic representation of pipe geometry - ′ r′ is radial direction, ′ z ′ is

along the axial direction, ′ θ′ is angle position and ′ h′ is the wall thickness.

Figure 2.4: Different modes propagating along the pipe.

2.3.2 Naming of guided waves

Commonly used mode naming convention is the one used by Silk and Bainton,
which is in accordance to the mode family and wave order. Each mode is assigned
a letter according to the wave type - Longitudinal (L), Torsional (T) or Flexural (F)
- and it also has two indices that identifies the mode number. For example T(n,m)
where ’T’ shows it is a torsional mode, the first index ’n’ represents the harmonic
order of circumferential variation; axially symmetric modes have this index as zero.
The second index ’m’ represents the mode; fundamental modes are usually the first
mode [11]. Hence the three types are [17] -

• Torsional mode is represented as T (0, m). Ur = Uz = 0, Uθ 6= 0

• Longitudinal mode is represented as L(0, m). Ur = Uz 6= 0, Uθ = 0
• Flexural mode is represented as F (n, m). Ur = Uz = Uθ 6= 0
2.3 Guided waves in pipes 9

2.3.3 Background

When detecting faults in pipes the fundamental torsional mode, T(0,1), and lon-
gitudinal mode, L(0,2), are the most commonly used modes. These are preferred,
because of their non-dispersive nature over a wide frequency band. Non-dispersive
motions are those in which the wave motion does not change as it propagates; their
phase velocity is independent of wave-number.

Traditionally L(0,2) has been preferred mode by many researchers, it was used
by Silk and Bainton when they were experimenting on heat exchange tubing [11], by
Alleyne et al. while testing on chemical pipes [14] and by Lowe et al. when detecting
defects in pipes [13]. Though, recently there has been a lot of work performed
using the torsional mode T(0,1). In 2001, Alleyne and Pavalakovic used T(0,1)
along with L(0,2) to detect defects in a pipe [14]. Demma in 2003, experimented
with T(0,1) to analyse and study reflections from defects, cracks and notches in
pipes [1, 18]. Davies and Crawley in 2009, performed synthetic focusing using T(0,1)
for imaging crack-like defects [19]. In 2010, Cardente et al. assessed the scattering
of T(0,1) due to tapered notches in pipes [20] and recently in 2012, Lovstad and
Crawley studied the reflection of T(0,1) due to pitting in pipes [21]. For this project
the fundamental torsion mode, T(0,1), has been used; the reasons for choosing this
mode are discussed in the sections ahead.

2.3.4 Choosing a wave mode and frequency for inspection

As mentioned in the section above, there are three different modes to choose
from after which the frequency at which the wave is transmitted at must be decided.
There are certain modes and frequencies that are sensitive to particular defects and
therefore in order to obtain reliable results it is necessary to choose appropriate
mode for the wave. Wilcox et al. defined a list of factors which influence the mode
and frequency selection and these include dispersion, attenuation, sensitivity, excit-
ability, detectability and mode selectivity and in practical implementation speed of
single test, testing tool design and level of difficulty in analysing data must also be
taken into account [1].

The frequency at which guided waves are transmitted determines the smallest
defect that can be detected. As the frequency increases the waves become more
sensitive to small defects, but the frequency is limited by vibration levels. Whereas
when testing a long pipe, low frequency guided waves are selected due to their
ability to travel further [5]. In this experiment the torsional mode T(0,1) was used as
it is non-dispersive and it is transmitted at frequency of 50 kHz, though the aim is to
try to be able to transmit the waves at a frequency of 100 kHz, using two transducers.
10 Pipeline Inspection

Problem occurs as high frequencies tend to generate unwanted modes.

2.3.5 Torsional vs. Longitudinal waves

In torsional mode the oscillation of particles is perpendicular to the direction
of travel of motion, whereas in longitudinal the particles oscillate in the direction
of wave propagation. The torsional mode can be excited by rotating the shear
transducers used to excite the longitudinal mode by 90 degrees, so that they apply
force circumferentially rather than axially [14].

One problem with L(0,2) is that in the operating frequencies above 35 kHz
there is another mode, L(0,1), which interferes as it has lower group velocity than
L(0,2). Whereas T(0,1) is the only torsional mode in the frequency range of interest,
as T(0,2) has higher cut-off frequency [3]. Table 2.1 shows advantages of using
torsional mode.

Advantages of Torsional Mode

• Torsional mode, unlike L(0,2), does not have radial displacement therefore it
is unaffected by non-viscous liquid loading [3].

• Even though torsional modes (3.2 kms−1 ) travel along the pipe slower com-
pared to longitudinal modes (5.4 kms−1 ), they are more sensitive to discon-
tinuities [14, 22].

• Torsional modes have less mode conversion, refraction and reflection com-
pared to longitudinal modes; this is because they propagate slower along
the pipe [4, 23].

• Torsional mode has the ability to detect longitudinal defects, whereas the
longitudinal mode is less able detect defects which are narrow and parallel
to pipe axis. This is because the torsional mode has dominant shear strain
which is perpendicular to axis [14]. In one of the experiments carried out
by Zenghua Liu et al. in 2006, he shows that T(0,1) mode can detect both
longitudinal and circumferential defects [24].

Table 2.1: Benefits of using a torsional mode over longitudinal mode.

Mode conversation depends on degree of asymmetry (see Figure 2.5), hence

on the circumferential extent of defect. If a defect is axisymmetric the only mode
that is reflected is the incident T(0,1), whereas an axial defect would result in mode
conversion [18]. Reflected waves can contain different possible modes such as
F(1,2), F(2,2) . . . F(n,2) modes which can make it difficult to interpret the results [25].
Another problem with torsional modes is that due to their sensitivity to axial features
they reflect strongly from support brackets. This makes it difficult to detect corrosion
2.3 Guided waves in pipes 11

and then in that case the extent of mode conversion produced by a reflector can
be measured. It also reduces the test range, then in that case L(0,2) would be
preferable [14].

Reflectors
in pipe

Symmetrical Asymmetrical
reflectors reflectors

Flanges and Corrosion Erosion wear

Weld Branch Pipes
supports damage of pipe wall

Figure 2.5: Classification of reflector kinds in a pipeline [22].

2.3.6 Dispersion curves

One problem with guided waves is that the wave velocity depends on the fre-
quency, known as dispersion. A change in frequency of the wave mode causes a
change in mode shape, phase velocity, group velocity and the rate of attenuation [2]
(refer to Appendix A for phase and group velocities). Therefore, it is necessary to
plot a dispersion curve as it shows relation between frequency, phase and group
velocity, mode and thickness. (See figure 2.8 for a group velocity dispersion curve.)

Dispersion curves can be plotted for either phase velocity or group velocity, they
are usually calculated for pipe in vacuum as there is not much difference between
solutions of air or vacuum, due to acoustic impedance of air being negligible com-
pared to a pipe.

Phase velocity dispersion curves show the velocity of the wave cycles within a
signal; it depends on the thickness of the pipe. Group velocity dispersion curves
show velocity of finite-time wave packets, hence they are used to calculate the travel
time of wave signal and it depends on stress magnitude [3, 26].

Gazis, in 1959 derived the dispersion curve for a hollow cylinder [7]. In 1967,
Vikotrov calculated the dispersion curve for an isotropic cylinder and compared the
results with the experimental results obtained from piezoelectric transducers. Viko-
trov used a form where the wave velocity was on y-axis and thickness/wavelength
ratio on x-axis. Though, the form that is currently preferred is wave velocity on y-axis
12 Pipeline Inspection

and frequency-thickness product on x-axis, which was initially used by J. and H.

Krautkrmer as well as Floyd Firestone [27]. Pavlakovic, Lowe and Alleyne have
developed a software, disperse, that plots dispersion curve; the first version was
released in 1991 [28]. In this project a Matlab script will be used to plot a dispersion
curve, which was written by Zaghari (2012).

2.4 Ultrasonic Guided Wave Technology vs.

Conventional Ultrasonics
Conventional ultrasonic method of detection is one of the traditional NDT meth-
ods, used to detect radial cracks on a pipe. These are being replaced by other
NDT methods like UGWT. Conventional ultrasonic operate by exciting a wave that
propagates through the thickness of the material, whereas UGWT performs by
propagating a wave along the axis of the pipe which can travel long distances (see
Table 2.2 for distance range based on application). UGWT then requires analysing
of the echoes in order to calculate where the defect is located. Figure 2.6 shows
how both these technologies operate.

Application Range in each direction (m)

Ideal Conditions 80+
30 year old pipe with little internal and external 40
corrosion
30 year old pipe with general corrosion 20
Pipe wrapped in factory applied foam 15
Heavily corroded pipe 5
Bitumen coated pipe 5*
* Range in bitumen coated pipe is dependent on bitumen condition

Table 2.2: Typical range the guided waves propagate with a standard transducer [29].

Guided waves have the advantage of being able to inspect coated pipes. It only
requires exposing a small area of circumference to place the transducers, therefore
saving time and money. It also has the ability to inspect buried pipes, therefore
making it cheaper to detect faults if the pipelines are buried below a road. Con-
ventional method can result in road being excavated costing up to US$50000 [30],
whereas when using guided waves excavation is only required during installation
and replacement [30] (see Table 2.3 for more advantages).
2.4 Ultrasonic Guided Wave Technology vs. Conventional Ultrasonics 13

Figure 2.6: Propagation of waves in Ultrasonic Guided Wave Technology and

conventional ultrasonic approach.

Advantages of Ultrasonic Guided Wave Technology

• Allows in-service inspection of pipes transporting gas and oil [3].

• It is highly safe and doesn’t emit harmful radiation like radiography inspection
methods [3].

• It is possible to detect faults in complex geometry [3].

• Depth of penetration for flaw detection is better, compared to other NDT

methods [4].

• It can detect changes in wall thickness, which might be due to corrosion [4].

• It is sensitive to up to 3% changes in Cross Sectional Area (CSA) [29].

• It has the ability to detect defects in buried, insulated or elevated pipes and
also detect in areas that would be inaccessible otherwise [30].

• Propagates wave in both direction of long pipe from a single test point [31].

• Has a higher sensitivity than conventional ultrasonic and other NDT methods
[31].

• Capable of inspecting both internal and external pipe walls [31].

• Saves time and money as insulation needs to be removed only where the
transducers will be placed [31].

Table 2.3: Advantages of Ultrasonic Guided Wave Technology for pipeline inspection.

Though UGWT has quite a few benefits, one of the drawbacks is the limitation in
distance of wave propagation for heavily corroded and bitumen coated pipes [3] (see
Table 2.2). UGWT also has limitation in ability to detect small volume defects like
14 Pipeline Inspection

localized pitting [3] and it does not have the ability to distinguish between internal
and external corrosion, which means visual inspection must be performed leading
to additional costs as insulation needs to be removed at these locations [32].

The other problem is that the output waveform depends on the type of material
being inspected e.g. cast iron and other grain structures give poor results due to
low sound transmission and high signal noise [4]. Furthermore, UGWT being a
sophisticated technology requires very experienced technicians to operate it [4].

Despite these drawbacks, there are still positives about this technology and it
is one of the technologies constantly improving. One of the leading oil companies,
BAPCO, had started using this technology in 2000 [32].

2.5 Generation of guided waves

Generation of guided waves can either be via contact or non-contact methods.
The contact method uses a series of transducers (piezoelectric in this experiment)
which are in contact with the pipe. A function generator generates an electric wave-
form that is converted to mechanical vibrations by piezoelectric transducer, before
being transmitted through the pipe. The reflected wave is then converted back to
an electrical signal by piezoelectric transducer. Usually a ring of transducers are
placed around the circumference of the pipe at an axial location. Technology using
a ring of shear dry-coupled piezoelectric transducers has been patented by Imperial
College (see Figure 2.7) [30].

In order to get accurate results the spacing between

the transducers and the number of transducers must be
sufficient. The number of transducers is related with the
order of the mode, transducer spacing and the pipe dia-
meter [12].

Alleyne, 1996 and Lowe, 1998 have written about how

to calculate the number of transducers required [12]. The
Figure 2.7: Ring of trans-
example used in these papers was for mode L(0,2) which ducers [30].
propagates at 70 kHz along a 7.5 cm outer diameter pipe.
As shown in figure 2.8, the group velocity is 5.6 kms−1 therefore it requires 12
piezoelectric transducer to excite the axisymmetric mode (refer to Appendix B for
calculations).

The paper also states that the number of piezoelectric transducers required is
2.5 Generation of guided waves 15

approximately proportional to pipe diameter so if 12 transducers are required for 7.5

cm diameter pipe then 24 will be required for a 15 cm diameter pipe. It also suggests
that to excite a pure axially symmetric mode the transducer spacing (△) around
the pipe should be less than half a wavelength of the inspection mode (△ ≤ λ2 )
according to Nyquist-shanon sampling criteria [19].

In this project if inspection is to be carried out at a frequency of 100 kHz it would

require a minimum of 7 transducers (refer to Appendix B for calculations).

Figure 2.8: Group velocity dispersion curve for a steel pipe with an external diameter
of 75 mm [13].
This page is intentionally left blank.
Chapter 3

Experiments

3.1 Single position measurement

3.1.1 Introduction
As mentioned earlier, UGWT allows long range inspection of pipes in order to
detect corrosion and defects. This experiment was performed to gain a deeper
understanding of how the technology operates, what each echo represents and to
find out at what frequency the T(0,1) mode must be transmitted so that the defect
and pipe end echoes are clearly distinguishable. This is an important experiment,
called frequency tuning, and has been performed by past researchers also [23, 24].
The selection of frequency at which guided wave is transmitted is important as it
determines the ability to send the ultrasonic energy and it must be carefully selected
so that this energy can travel from position A to B [23].

As known from theory, high frequency waves are more sensitive to small defects
and the lowest frequency range used in practical testing with torsional mode T(0,1)
is 45-65 kHz for a 3 inch pipe and 10-20 kHz for 24 inch pipe [33]. It is also stated
that frequency range of 20-100 kHz can be used to detect defect as small as 5% of
the pipe’s CSA [33]. Therefore the hypothesis was that as the frequency increases
it can become difficult to detect the defect echo, due to high reverberation.

3.1.2 Experimental setup

Most of the experiments were performed on a bare steel pipe1 with an outer
diameter of 3.4 cm and a length of 280.3 cm (see Figure 3.1 for more details on
the pipe). The specimen had a notch that was machined 247 cm from the left
end of the pipe. It was 5 mm in depth (100%) with an axial length of 4 mm and a
circumferential length of 20 mm occupying 8.3% CSA of the pipe wall (see Figure
1
Bitumen coated pipe experiment was performed on a different pipe.
18 Experiments

3.2). The tests were performed with the pipe horizontally placed and the ends
resting on supports. Torsional fundamental mode, T(0,1), was transmitted along the
pipe for all the experiments. It was transmitted over a range of frequencies varying
from 30 kHz to 200 kHz, in order to determine the frequencies at which the defect
echo was identifiable.

Figure 3.1: Properties of the bare steel pipe used in the experiment.

(a) Picture of defect. (b) Schematic representa-

tion of defect.

Figure 3.2: Defect machined on the pipe.

For all the experiments two dry coupled piezoelectric transducers were used,
where one of them transmitted the guided waves along the pipe and other received
it (pulse echo transmission method was adopted) (see Figure 3.3). Piezoelectric
transducers were used because of their high sensitivity, low price and convenient
usage [34]. It produced tangential displacement, thereby generating the torsional
mode. The transducers were placed 6.2 cm apart from each other, this was the
minimum possible distance. The receiving transducer was placed 186.8 cm to
the left of the defect, while the transmitting transducer was located 54 cm to the
right of the left pipe end wall; they were placed at zero degrees configuration (see
Figure 3.4). Function generator2 was connected to the transmitting transducer that
2
Function generator HP 33120A was used.
3.1 Single position measurement 19

provided an input tone burst of three cycles at a burst rate of 10 kHz and Vpk−pk
of 19 V. This was the maximum voltage that could be applied; applying a lower
voltage would result in smaller amplitude as the transducers would get excited slowly.
Function generator was connected to the channel 1 of the oscilloscope3 to display
the input waveform. The receiver transducer was connected to a signal amplifier4
that increased the amplitude of the output. This then led to a bandpass filter that
removed the high frequency signals and low frequency noise, thereby making the
output waveforms smoother before being displayed by the oscilloscope (see Figure
3.5). A General Purpose Interface Bus (GPIB) cable was used, that enabled the
data to be transferred to a computer so that the results can be recorded by a Matlab
data capture script that also allowed controlling of the oscilloscope. The script was
edited by Zaghari (2012).

Figure 3.3: A module with a piezoelectric transducer in place. The module was
placed on the pipe so that the transducer can produce tangential displacement in
order to generate torsional mode. Two modules were used in the experiment, each
with a transducer where one transmitted the guided wave and the other received it.

Figure 3.4: Transmitting and receiving transducers placed at zero degree configura-
tion.

3
Oscilloscope LeCroy 9304C was used.
4
ORTEC Precision AC Amplifier 9452 was used.
 20

Figure 3.5: Setup for the experiment showing two transducers placed on the pipe. The transmitting and receiving transducers here are
placed 6.2 cm apart from each other.
Experiments
3.1 Single position measurement 21

3.1.3 Results and analysis

In this section, the results obtained from the experiment are looked at. The first
part of the experiment involved analysing the received signals to distinguish the
defect and the pipe end echoes.

To understand what feature each echo represents it is important to know the

speed at which the waves propagate. A group velocity dispersion curve was used
to find the group velocity of the torsional mode, T(0,1). It can be estimated, using
figure 3.6, to be 3309 ms−1 .

Figure 3.6: Group velocity dispersion curve, for torsional mode T(0,m), for a steel
pipe with an external diameter of 35 mm. This graph was produced by running a
MATLAB script which is written by Zaghari (2012).

Using the group velocity, 3309 ms−1 , for torsional mode T(0,1) and the distances
of defect and pipe end walls it was possible to estimate the arrival of each echo. A
schematic representation of the expected path line of wave propagation is shown in
figure 3.7. The diagram shows a burst signal that is transmitted in both directions by
the transmitting transducer. The signal transmitted to the right is first received by
the receiving transducer known as direct signal, whereas the signal transmitted to
the left hits the pipe end and is reflected back to be received by the transducer. The
burst signals propagate along the pipe till the energy dissipates.
22 Experiments

Figure 3.7: Schematic representation of the expected path line for wave propagation
along the bare pipe containing a defect. The burst signal is transmitted in both
directions of the pipe.

Figure 3.8, shows the signal obtained when the guided waves were transmitted
at 50 kHz. This was analysed to understand what feature each echo represents.
Table 3.1 shows the expected arrival time of each echo calculated using speed,
3309 ms−1 (0.3309 cm(µsec)−1 ) and the actual arrival time, the instance at which
the echo arrives. As shown in the table, the arrival of direct echo has the largest
difference with theoretical estimate, because the distance was short and the echo
arrived quickly making it difficult to measure the arrival time, whereas the other
echoes arrive close to the expected time. The other reasons for the difference
are discussed in the next section. The second echo due to the defect was not
distinguishable, due to it being too close to the previous echo.

The hypothesis for this experiment was that as the frequency at which guided
waves are transmitted increases, it is less likely the defect echo will be detected due
to high reverberation. As the time at which the defect echo arrives is known, it is
possible to compare how this echo varies when guided waves are transmitted at
different frequencies. Guided waves were transmitted, for a range of frequencies
varying from 30 kHz to 200 kHz, to assess how the defect echo varies (see Figure
3.9).

As shown in figure 3.9, it was possible to detect the defect echo when guided
waves were transmitted at 50 kHz. When transmitted at 100 kHz the defect echo
is present but it is difficult to distinguish and identify, whereas at frequency of 50
kHz the pipe end and defect echoes are clearly visible and distinguishable. The
3.1 Single position measurement 23

reason that the defect echo was not visible when waves were transmitted at higher
frequencies is because high frequencies produce very strong reverberation which
causes the defect echo to be lost, while at low frequencies (30 kHz) there is high
noise due to which the defect echo cannot be distinguished.

Figure 3.8: Received signal, showing different echoes, for guided waves transmitted
at 50 kHz and the transducers placed at a zero degree configuration.

Echo Type Distance trav- Theoretical Actual arrival

elled till receiv- arrival time time (µsec)
ing transducer (µsec)
(cm)
Direct Echo 6.2 18.7 30.6
Echo from left end wall 114.2 345.1 357.5
First defect echo 379.8 1147.8 1172.0
Echo from right end wall 446.4 1349.0 1350.0
Second defect echo 487.8 1474.2 Not clearly dis-
tinguishable
Second echo from left 566.8 1712.0 1706.0
end wall

Table 3.1: Comparison between theoretical and experimental arrival times of the
echoes, in a received signal.
 24

Figure 3.9: Amplitude of received signal, for guided waves transmitted at different frequencies, when a single reading at zero degree
configuration was taken. The time interval between the black dotted lines is analysed ahead for reverberation root mean square and the time
interval between the red dotted lines is used to calculate signal root mean square, which is used to calculate the signal to reverberation ratio.
Experiments
3.1 Single position measurement 25

In order to prove that the reverberation increases as the frequency of trans-

mission increases, Root Mean Square (RMS) was calculated for a common time
interval. It was calculated for the time interval between the pipe end echo and the
defect echo. This region was selected because it is the longest time interval where
no other echo occurs, thereby making it easy to calculate the reverberation RMS
(see Figure 3.9 for the time interval). The lower the reverberation RMS the better, as
it means that there is less disturbance present in the received signal. The problem
that occurred while calculating reverberation RMS is that the amplitude of each
signal was different at different frequencies. Hence, in order to normalise the data
the peak amplitude for the signal was adjusted to unity; it must be noted that the
peak could occur at any instance in the waveform.

Reverberation RMS (VRMS ) was calculated using the formula 3.1, where Nr is
the total number of readings taken, Vmean is the arithmetic mean of all the readings
and Vi is the original reading.
v
u
u 1 X Nr
VRMS = t (Vi − Vmean )2 (3.1)
Nr i=1

Figure 3.10: Variation in normalised reverberation RMS, for guided waves transmit-
ted at different frequencies, when a single reading at zero degree configuration was
taken.

As shown in figure 3.10, the reverberation RMS increases, as the guided waves
were transmitted at higher frequencies. It also shows that the reverberation RMS
was lowest for guided waves were transmitted at 50 kHz; hence the defect echo was
easily distinguishable.
26 Experiments

3.1.4 Discussion and experimental conclusion

In this section an interpretation of the results obtained is provided, using which a
conclusion is reached at. It will also be checked if the results support the hypothesis
or not. It is necessary to find an optimum frequency to transmit the waves in order
for the defect echo to be distinguishable.

The main findings of this experiment are -

• As transmission frequency increases it becomes more difficult to detect the

defect echo.

• The reverberation RMS increases as guided waves are transmitted at higher

frequencies.

The hypothesis set before the experiment was proven correct and also the two
findings are related to each other. The reason it is difficult to detect the defect echo
is due to the fact that when the piezoelectric transducers are coupled onto the pipe
though the aim is to excite only torsional wave, through tangential displacement, it
also excites some longitudinal and flexural modes at the same time. This is more
clearly seen using the group velocity dispersion curve, figure 3.11, where it is visible
that other modes are also excited at the same time. For example it can be seen
that when T(0,1) mode is excited at 50 kHz, it also excites other modes like L(0,1),
F(1,1), F(1,2), F(2,1) and F(3,1). Similarly, as the transmission frequency increases
there are more non-axially symmetric modes that get excited as a result of which the
received waveform displays signals from other modes that interfere with the defect
signal, thereby making the defect echo less easily identifiable and distinguishable.
Figure 3.9 shows, that the arrival of defect and pipe end echoes is not the same for
all frequencies. This is due to there being a mode change when the guided waves
hits the pipe end or defect, thereby resulting in variation in arrival speeds.

Reverberation occurs due to reflection of part of the transmitted wave that is then
received by the transducer, thereby making it difficult to detect the defect. It occurs
due to both axisymmetric and non-axisymmetric modes. Reverberation is different
from noise. If noise is a limiting factor then the signal can be transmitted at higher
level, thereby increasing transmission echo and not noise; but if reverberation is a
limiting factor then increasing the signal will increase reverberation. Reverberation
increases with frequency and it determines the frequency the test can be performed
at, as shown in figure 3.9, whereas frequency determines the smallest defect that
can be detected by the guided wave.
3.1 Single position measurement 27

Figure 3.11: Group velocity dispersion for a steel pipe with an external diameter of
34 mm and wall thickness of 5 mm. This graph is produced from PCDISP MATLAB
scripts written by Seco and Jimenez (2012).

It is known that guided waves transmitted at high frequencies are more sensitive
to smaller defects than waves transmitted at low frequencies. The problem occurs
that the defect echo gets lost in reverberation, when waves are transmitted at high
frequencies using one transmitting and receiving transducer. In order to be able to
clearly distinguish the defect echo this reverberation must be reduced.

It was mentioned earlier, in section 2.5, that increasing the number of transducers
enable to suppress the additional unwanted modes that get excited alongside the
torsional mode; thereby reducing the reverberation level when guided waves are
transmitted at high frequencies. Therefore the second experiment focuses on rotat-
ing the receiving transducer around the circumference of the pipe and averaging the
signal to verify if reverberation reduces at high frequencies, hence making it easier
to detect the defect echo.
28 Experiments

3.2 Multiple circumferential position measurements

3.2.1 Introduction
The aim of performing this experiment was to try to reduce the reverberation
level that occurs when guided waves are transmitted at high frequencies. Alleyne
and Cawley stated that it is possible to suppress the non-axisymmetric modes if
array of transducers are attached around the circumference of the pipe [12]. This
experiment was performed to test whether the theory is valid in practical application,
and if it reduces reverberation, thereby enabling ultrasonic signals to be transmitted
at higher frequencies. The aim was to be able to get a defect echo that can be
clearly distinguished and identified, when transmitting waves at 100 kHz.

It is known from theory that reverberation is caused by excitation of other modes.

Therefore hypothesis for this experiment was that taking multiple readings around
the circumference of the pipe and averaging the results would reduce the average
reverberation, by mitigating the effect of non-axisymmetric modes.

3.2.2 Experimental setup

The experimental setup was very similar to the first experiment, except that in
this case the receiving transducer was rotated around the circumference of the pipe
whereas previously it was positioned at zero degree configuration. The receiving
transducer was evenly spaced around the circumference of the pipe and readings
taken before averaging all the waveforms.

The spacing between the transducers must be equal to or less than half the
wavelength of mode that is supposed to be suppressed. In this case the aim was
to try to eliminate all the modes that have a lower group velocity than T(0,1) when
transmitted at 100 kHz. According to suggestions by Alleyne and Cawley, a minimum
of 7 transducers were required around the circumference of the pipe in order to be
able to detect the defect echo when guided waves are transmitted at 100 kHz [12]
(refer to Appendix B). One of the major problems is coupling between the receiving
transducer and the pipe, as it is difficult to produce same coupling each time the
receiving transducer is rotated. There is also possibility of errors in measurement of
angles, when rotating the transducer, as this was measured by finding the equivalent
distance around the circumference using tape measure. Hence the readings for
all frequencies were taken at the position of transducer, before rotating it, to keep
results consistent. The other approach to performing this experiment is to have a
collar of transmitting and receiving transducer around the circumference of the pipe,
3.2 Multiple circumferential position measurements 29

see figure 3.13, this method was not adopted due to budget constraints.

(a) Transducers placed at 45◦ (b) Transducers placed at 90◦

Figure 3.12: Rotation of receiver transducer around the circumference of the pipe.

Figure 3.13: Alternative method of performing the experiment by having a collar of

transducers around the circumference of the pipe [35].
30 Experiments

3.2.3 Results and analysis

This experiment was divided into two sub parts. The first part of the experiment
involved analysing the average output signals produced, for guided waves trans-
mitted at different frequencies, when 8 evenly spaced readings are taken around
the circumference of the pipe. The results obtained from this experiment were then
compared with the last experiment, where only a single readings was taken. The
second part of the experiment involved investigating how output signals vary with
transducer position, for guided waves transmitted at 100 kHz.

As mentioned earlier, a minimum of 7 evenly spaced transducers were required,

around the circumference of the pipe, to suppress the flexural modes F(n,m) for a
pipe of outer diameter 34 mm (refer to Appendix B). The tests were performed using
8 different receiver positions around the circumference of the pipe and finally the
average of the 8 output signals was taken. Figure 3.14 shows the 8 readings taken
around the circumference of pipe, each separated by 45◦ , and the averaged signal
of the 8 waveforms for guided waves transmitted at 100 kHz. This was similarly
performed for other frequencies ranging from 30 kHz to 200 kHz, as shown in figure
3.15.

Figure 3.15 shows the results obtained, and it is visible that the pipe end and de-
fect echoes are more clearly identifiable when multiple readings were taken around
the pipe rather than a single reading. The reason for this improvement is by taking
readings around the circumference of the pipe and averaging the results it reduces
the reverberation caused by non-axially symmetric modes, flexural modes, and it
also averages out noise.

In order to prove numerically, that the reverberation decreases as more readings

are taken around the circumference of the pipe, reverberation RMS was calculated
for the same time interval between the pipe end and defect echo using formula
3.1. In this case the relative values of reverberation RMS was calculated, and not
the not the normalised, as results were compared for waves transmitted at same
frequencies but when a single reading was taken.
 3.2 Multiple circumferential position measurements

Figure 3.14: The first graph shows the received signals for 8 evenly spaced positions of the receiving transducer around the circumference of
the pipe, for guided waves transmitted at 100 kHz, whereas the second graph shows the average signal of these 8 waveforms.
31
 32

Figure 3.15: Amplitude of average received signal, for guided waves transmitted at different frequencies, when 8 evenly spaced readings are
taken circumferentially around the pipe. The time interval between the black dotted lines is then analysed ahead for reverberation RMS and
the time interval between the red dotted lines is used to calculate signal RMS, which is used to calculate the signal to reverberation ratio.
Experiments
3.2 Multiple circumferential position measurements 33

Figure 3.16: Variation in ratio between the relative reverberation RMS, for an average
of eight evenly spaced readings around the circumference of the pipe and a single
reading at zero degree configuration, when guided waves were transmitted at
different frequencies.

Figure 3.16, shows how the ratio of relative reverberation RMS, between average
of eight evenly spaced readings around the circumference of the pipe and a single
reading at zero degree configuration, varies for guided waves transmitted at differ-
ent frequencies. As shown from the graph, reverberation RMS reduces for all the
frequencies when more readings are taken around the circumference of the pipe, as
the ratio is less than 1. The most improvement occurs for waves transmitted at 100
kHz, as expected. The improvement at 100 kHz occurs not only against noise, but
also against reverberation by elimination of the non-axisymmetric modes. If there
was improvement only against noise then the ratio is expected to follow formula
3.2, where Vrev,RMS is the reverberation RMS for waveform produced from single
reading or ’k’ that is the average waveform of the multiple readings taken, and Nr is
the number of readings.

p
Vrev,RMS,k ∗ Nr = Nr ∗ Vrev,RMS,1 (3.2)

The ratio of reverberation RMS, for an average of multiple readings by single

reading, is expected to be √18 ≈ 0.35 for 8 readings taken if noise was the only
improvement. At 100 kHz the result is lower than this, demonstrating that there is
improvement not only against noise but also against reverberation. At 50 kHz the im-
provement is majorly against noise. The figure also shows that at 150 and 200 kHz
there is improvement against reverberation, but there are more non-axisymmetric
modes present and to eliminate these modes more readings must be taken around
the circumference of the pipe.
34 Experiments

After this the signal RMS was calculated for the time interval of the defect echo.
These two RMS were used to calculate the Signal to Reverberation Ratio (SRR).
SRR is the ratio between signal RMS (Vs,RMS ) and the reverberation RMS (Vrev,RMS )
and is calculated in dB using the formula below -

Vs,RMS
SRR = 20 log10 ( ) (3.3)
Vrev,RMS
The aim is to have a resulting waveform with high SRR, as it would mean that
the received waveform has more useful information rather than reflected signal. A
waveform with high SRR would enable the user to distinguish and clearly identify
the defect, as it would not be lost in reverberation.

Figure 3.17: Comparing how the signal to reverberation ratio varies, for guided
waves transmitted at different frequencies, when a single reading at zero degree
configuration and when eight evenly spaced measurements are taken around the
circumference of the pipe.

As shown in figure 3.17, SRR improved at all frequencies when multiple readings
were performed, but the most significant improvement was at 50 kHz and 100 kHz.
This improvement at 100 kHz was due to reduction in reverberation that enabled the
defect echo to be distinguished. The other interesting point is that at frequencies
150 and 200 kHz, for single measurement, despite the SRR being better than at 50
kHz the defect was not visible. This could be because of scattering of signals due to
high energy and also because of high reverberation due to which the defect echo
gets lost and is not clearly distinguishable.
3.2 Multiple circumferential position measurements 35

In the second part of this experiment it was looked at how the output waveforms
vary with the position of receiver transducer, for the guided waves transmitted at 100
kHz. The problem with this part of the experiment was that all the readings were
performed at different times; hence there might be variation in coupling between the
transducer and the pipe. As can be seen from figure 3.18, the defect echo was not
visible when the receiving transducer was placed at zero degree configuration or
below the transmitting transducer due to high reverberation whereas it was clearly
distinguishable when 8 or 16 readings were taken around the circumference of the
pipe. It is also noticeable that the waveform produced when 16 readings were taken
around the pipe has echoes that are clearer and have larger amplitude making them
easily distinguishable compared to the waveform produced for 8 readings.

Normalised reverberation RMS was calculated for these waveforms using formula
3.1, so that it can be compared how reverberation varies with the transducer position
and the number of readings taken. It was calculated for the same time interval, as in
previous cases, between the pipe end echo and the defect echo. The waveforms
were normalised, where the peak to peak amplitude was made to unity so that the
comparison is consistent. Figure 3.19, shows the variation in reverberation RMS
with the number of readings taken around the circumference of the pipe. The figure
shows that when more readings were taken around the circumference of the pipe,
the reverberation RMS was lower. It is also noticeable that the reverberation RMS
does not improve vastly when the number of readings around the circumference of
the pipe increases from 8 to 16. This is because there is no reduction in reverbera-
tion due to non-axisymmetric modes and the slight improvement is against noise.
The reverberation RMS was higher when the receiving transducer was placed below
the transmitting transducer than at zero degree configuration, this is even visible
when looked at figure 3.18, due to which the defect echo was not identifiable.

Lastly, SRR was calculated for all waveforms following the same procedure as
in previous experiment and using formula 3.3. The higher the SRR the better the
chances of detecting the defect echo. As shown in figure 3.20, taking multiple
readings around the circumference of the pipe improves the SRR. Thereby demon-
strating that the probability of detecting defect echo is higher when more readings
are taken around the circumference of the pipe. It can be seen that the SRR doubles
from approximately 3.5 dB to 7 dB, when the number of readings increases from 8
to 16. The reason why the SRR is better when 16 readings were taken is, because
there was an improvement in signal RMS as the amplitude of the defect echo is
greater than when 8 readings were taken. The SRR was very low and almost similar
when only one reading was taken, either at zero degree configuration or when the
receiving transducer is placed below the transmitting transducer.
 36

Figure 3.18: Comparing how resulting waveforms vary, for guided waves transmitted at 100 kHz, with variation in transducer positions. It
shows, the output waveforms obtained when the transmitting and receiving transducer are placed at a zero degree configuration, when
the receiving transducer is placed below the transmitting transducer, and average of 8 (at 45 ◦ each) and 16 (at 22.5 ◦ each) readings taken
around the circumference of the pipe.
Experiments
3.2 Multiple circumferential position measurements 37

Figure 3.19: Comparing how normalised reverberation RMS varies, for guided waves
transmitted at 100 kHz, when a single reading at zero degree configuration, when
the receiver transducer is placed below the transmitting transducer and when eight
and sixteen evenly spaced readings are taken around the circumference of the pipe.

Figure 3.20: Comparing how signal to reverberation ratio varies, for guided waves
transmitted at 100 kHz, when a single reading at zero degree configuration, when
the receiver transducer is placed below the transmitting transducer and when eight
and sixteen evenly spaced readings are taken around the circumference of the pipe.
38 Experiments

3.2.4 Discussion and experimental conclusion

This main aim of performing this experiment was to try to exploit the theory
stated by Alleyne and Cawley [12], that mentioned by having a series of elements
around the circumference of the pipe, it should be possible to suppress the non-
axisymmetric F(n,m) modes (refer to Appendix B). To detect small defects ultrasonic
waves must be transmitted at higher frequencies and in order to transmit waves at
high frequencies, reverberation must be reduced by eliminating the extra modes
transmitted.

The hypothesis set before the experiment has been proven correct by the results
obtained. The results show that the reverberation decreases if more readings are
taken around the circumference of the pipe.

The main findings of this experiment are -

• Reverberation RMS reduces as the number of readings around the circumfer-

ence of the pipe increases, thereby enabling defect echo to be detected for
waves transmitted at 100 kHz, when 8 readings were taken around a pipe with
an external diameter of 3.4 cm.

• SRR improves as the number of readings increase around the circumference

of the pipe.

Reverberation is caused by both axisymmetric and non-axisymmetric modes,

that get transmitted along with the fundamental torsional mode T(0,1). Taking an
average of multiple readings works as it mitigates the effect of non-axisymmetric
modes that gets excited and it also helps in reducing the effects of background noise,
preventing any disruptions occurring in detection of the defect echo. The spacing
between transducers must be equal to half the wavelength of the mode that has to
be suppressed. In this case the aim was to suppress all modes with speed less than
T(0,1) and as T(0,1) has a constant group velocity at all frequencies the spacing
was less than half the wavelength of T(0,1) mode at 100 kHz. Non-axisymmetric
modes get eliminated because they have different phase at different position, hence
taking multiple readings and then averaging these readings results in a waveform
with a lower reverberation. At higher frequencies there are more non-axisymmetric
modes; hence more readings must be taken around the circumference of the pipe
in order to have a lower reverberation and a higher SRR for waves transmitted at
greater than 100 kHz.

The other part of the experiment involved analysing the waveforms obtained, for
waves transmitted at frequency of 100 kHz, at four different transducer positions.
3.2 Multiple circumferential position measurements 39

The results prove that it is possible to detect defect echo more clearly if there are
more transducers around the circumference of the pipe or if a transducer is reposi-
tioned physically over the circumference of the pipe and multiple readings taken.
It can be seen that when there was only one reading taken for waves transmitted
at 100 kHz, be it the case with zero degree configuration or receiver placed below
transmitter, the reverberation RMS is very high resulting in low SRR and thereby
resulting in defect echo getting lost in reverberation. Whereas, when there were
8 or 16 readings taken around the pipe then the reverberation RMS was low, but
still not zero because axisymmetric modes cannot be eliminated (see Table 3.2).
Due to this the SRR is high and the defect echo visible, when multiple readings are
taken around the circumference of the pipe. There is a possibility there might be
imperfections in coupling, when rotating the transducer, thereby resulting in some
non-axisymmetric modes still being present.

Disturbance 8 readings 16 readings

Reverberation due to non- Eliminates this No Improvement
axisymmetric modes
Reverberation due to axisymmet- Cannot get rid of this Cannot get rid of this
ric modes
Background Noise Averaged out Only improvement

Table 3.2: Analysis of signals obtained when 8 and 16 readings are taken around
the circumference of the pipe, for guided waves transmitted at 100 kHz.

In the next section another important experiment was performed to see how
guided wave propagate through bitumen coated pipes. This is an important ex-
periment as most pipes are coated externally and sometimes internally in order to
prevent corrosion, increase flow of gas and reduce energy costs in pumping and
compressor stations [36].
40 Experiments

3.3 Bitumen coated pipe

3.3.1 Introduction
The actual aim of this experiment was to perform tests on a corroded pipe to
determine if UGWT can be used to detect the presence of corrosion on pipes. The
company that was supplying the pipe, sent a pipe that was coated with bitumen
instead of corroded pipe.

Due to time constraint, the experiment was performed on the bitumen coated
pipe. The aim was to understand the propagation of waves through it and to see
if the waves attenuate as strongly, as mentioned in theory. This is also a useful
experiment as bitumen has been used to protect pipes from corrosion for over 40
years [37]. Therefore most old pipes are coated with bitumen and in order to inspect
these pipes for defects, it is important to know how guided waves perform when the
pipe is coated with bitumen.

As shown by table 2.2, guided waves propagate very limited distance in a bitumen
coated pipe. The hypothesis set for this experiment was that the bitumen coating
will strongly attenuate the waves and hence the pipe end echoes5 will be less visible
compared to a bare pipe and the amplitude of echoes is also expected to be smaller
in this case.

3.3.2 Experimental setup

The experimental setup was similar to previous experiments, with torsional mode
T(0,1) being used; except that in this case the non-coated pipe with defect is
replaced with a bitumen coated pipe that contains no defect (see Figure 3.21). In
this experiment only a single position measurement was taken with the transmitting
and receiving transducers placed at a zero degree configuration. The bitumen
coating was removed for the region where the transducers were placed6 , to enable
the transducers to couple onto the pipe and be able to transmit waves through the
pipe (see Figure 3.21, 3.22). Figure 3.23 shows the dimensions of the bitumen
coated pipe and figure 3.24 shows the setup of the experiment.

5
This pipe had no defect so there was no defect echo, as with previous cases.
6
Bitumen coating was only removed for the area where the transducers were placed and not all
the way around the pipe.
3.3 Bitumen coated pipe 41

Figure 3.21: Bitumen coated pipe along with the area for which the coating was
removed (the black region) to place the transducers.

Figure 3.22: Transducers (receiving at right and transmitting at left) placed on the
region of pipe where bitumen coating has been removed. It was only removed for
the top section of the pipe and not all the way around the pipe.

Figure 3.23: Dimensions and properties of the bitumen coated pipe.

 42

Figure 3.24: Setup for the experiment showing two transducers placed on the bitumen coated pipe. The transmitting and receiving transducers
here are placed 6.2 cm apart from each other and the bitumen coating is removed so that the transducers can couple on the pipe.
Experiments
3.3 Bitumen coated pipe 43

3.3.3 Results and analysis

This experiment involved assessing how ultrasonic guided waves propagate
through a bitumen coated pipe and to verify whether it dissipates as quickly as
stated in theory. Guided waves were transmitted at different frequencies, ranging
from 30 kHz to 200 kHz, to see if transmission frequency has an effect on wave
propagation and reverberation in a bitumen coated pipe. Figure 3.25, shows the
signals obtained for different transmission frequencies in a bitumen coated pipe
when an input tone burst was applied at Vpk−pk of 19 V7 .

As shown in figure 3.25, guided waves respond very poorly when transmitted
through bitumen coated pipe, they attenuate very highly and as a result dissipate
quickly. It is visible that at lower frequencies, i.e. 30 kHz and to an extent 50 kHz,
there is some aspects of pipe end echo visible, whereas at higher frequencies the
only visible echo is due to direct transmission. This could be because as known low
frequency waves travel further than high frequency waves. Another experiment was
performed with an input burst applied at Vpk−pk of 4.75 V to investigate if transmitting
the input signal at a lower strength has an affect on wave propagation through the
pipe. The results obtained were very similar to figure 3.25, but the only difference
was that the echo had a lower amplitude. The results show that propagation of
waves is not affected by the strength at which input burst is transmitted at.

The amplitude of direct echo produced for guided waves transmitted at different
frequencies was compared for the input tone burst transmitted at Vpk−pk of 4.75 V
and 19 V through a bitumen coated pipe and when the input burst was transmitted
at Vpk−pk of 19 V through a bare pipe. The distance between the transmitter and
receiving transducer was 6.2 cm in all cases, to ensure results are consistent.
Figure 3.26 shows how the amplitude of direct echo varies in each case. The graph
shows that there was a similar trend for all the experiments and that the amplitude
at 50 kHz is lower than all other frequencies. This is because the reverberation
is low for a single measurement, when the guided waves are transmitted at a
frequency of 50 kHz and hence there is less disturbance. It also shows that the
received signal for bitumen coated pipe has a lower amplitude compared to results
from the bare pipe, because of high attenuation caused by bitumen coating. The
results display that when the tone burst was transmitted at a Vpk−pk of 4.75 V it
produced amplitude almost 4 times smaller than the amplitude produced for tone
burst transmitted at Vpk−pk of 19 V, as it has 4 times more energy than at 4.75 V.
Therefore, the amplitude of the direct echo is proportional to the voltage the tone
burst is transmitted at through a bitumen coated pipe.

7
In the previous experiments also, the tone burst was applied at Vpk−pk of 19 V.
 44

Figure 3.25: Comparing how the amplitude of received signal varies in a bitumen coated pipe, for guided waves transmitted at different
frequencies (burst applied at Vpk−pk of 19 V), when a single reading at zero degree configuration was taken. The time interval between the
black dotted lines is then analysed ahead for reverberation RMS.
Experiments
3.3 Bitumen coated pipe 45

Figure 3.26: Variation in amplitude of direct echo in non-coated and bitumen coated
pipe. The graphs shows results from normal pipe where the input burst was trans-
mitted at Vpk−pk of 19 V and from bitumen coated pipe where the input tone burst
was transmitted at Vpk−pk of 19 V and 4.75 V.

To analyse how the reverberation varies with frequency in a bitumen coated pipe,
reverberation RMS was calculated. The signals were normalised first to ensure
comparison can be made and then reverberation RMS was calculated for the time
interval between the direct echo and the pipe end echo, using formula 3.1. Figure
3.27, displays how the reverberation RMS varies with frequency when the tone burst
was transmitted at Vpk−pk of 19 V and 4.75 V. To ensure reliable results the readings
for both scenarios were taken for the same position of transducers, so that coupling
between the pipe and transducer stays consistent, when transmitting tone burst at
both voltages.

The results demonstrate that reverberation RMS is almost same for the tone
bursts transmitted at Vpk−pk 19 V and 4.75 V. This implies that the reverberation
does not depend on the energy of tone bursts, but rather on the frequency at which
waves are transmitted at. It is visible that the results are consistent with the previous
experiments, as the reverberation RMS was lowest at 50 kHz for a single point
measurement. This suggests that the trend in which reverberation varies across
frequencies does not depend on whether the pipe is coated with bitumen or not,
though the relative values could be different.
46 Experiments

Figure 3.27: Variation in normalised reverberation RMS in a bitumen coated pipe,

for guided waves transmitted at different frequencies, when the tone burst was
transmitted at Vpk−pk of 19 V and 4.75 V.

3.3.4 Discussion and experimental conclusion

The main aim of performing this experiment was to assess how guided waves
propagate through a bitumen coated pipe. The hypothesis for this experiment was
that the guided waves attenuate highly when transmitted through a bitumen coated
pipe, due to its strong bonding with the pipe. From the results it is proven that this
hypothesis set before the experiment was correct.

The main findings of this experiment are -

• Ultrasonic guided waves attenuate strongly through bitumen coated pipe,

thereby reducing the examination range significantly.

• Reverberation RMS does not depend on the voltage of the tone burst and only
depends on the transmission frequency of the guided waves.

• Amplitude of echo is lower when guided waves are transmitted through a

bitumen coated pipe than a bare pipe.

The first finding is one of the reasons why UGWT is considered poor for detecting
defects in bitumen coated pipes. This is a disadvantage because high wave attenu-
ation limits the defect detection sensitivity and monitoring range of the technology.
When pipes are coated with a viscoelastic material, it causes the energy of guided
waves to leak into the coated material from the pipe wall. The rate of leakage
is determined by the strength of the coating. Bitumen being a strongly adhesive
material binds firmly to the pipe, thereby resulting in leakage into the coating and
3.3 Bitumen coated pipe 47

causing significantly high attenuation of guided wave [38]. This high attenuation,
thereby results in limitation to the distance the guided waves can propagate. Bitu-
men coated pipes attenuate guided waves as the pipe is a bi-layer system and any
energy carried in bitumen layer is attenuated quickly, whereas in a corroded pipe the
test ranges is less as energy is scattered by the corroded surface [30]. It is known
from previous research that insulation coatings that have strong adhesive strength,
such as bitumen, cause high attenuation of guided waves whereas mineral wool
coating almost has no effect as it does not bond strongly with the pipe [39]. It must
also be remembered that the thickness and age of coating also have an effect. The
other factor that influences attenuation is the frequency at which guided waves are
transmitted and as the frequency increases the attenuation increases (see Figure
3.25). This is because the wavelength of the wave is shorter, thereby dissipating
easily into the bitumen coating [38].

The next findings are related to the amplitude and reverberation of waveforms in
a bitumen coated pipe. This experiment showed that the amplitude of the direct echo
varies in a similar manner across the frequencies and it is lower for bitumen coated
pipe, due to high reverberation and absorption by the viscous coating when com-
pared with the bare pipe for the input burst transmitted at Vpk−pk of 19 V. Whereas,
when input burst was transmitted at Vpk−pk of 4.75 V the amplitude is lower, because
the input burst was transmitted at lower energy, but the trend in which the amplitude
varies across frequencies is similar.

To investigate how the reverberation varies with frequency, reverberation RMS

was calculated. This experiment showed that the reverberation is lowest at 50 kHz
for a single point measurement similar to when a single point measurement was
performed on a non-coated pipe. Figure 3.27, shows that reverberation does not
depend at the voltage the input tone burst is transmitted at, as the reverberation is
same for both Vpk−pk 19 V and 4.75 V. This implies that introducing more or less
energy into the pipe wall does not help to improve the reverberation.

Most pipelines are coated with bitumen or with other viscoelastic material for
protection purposes, and in practice most pipelines are buried underground. Similar
to insulation coating on a pipe, soil also has an influence on the propagation of
waves along the pipe. In the next experiment performed propagation of guided
waves through a buried pipe was looked at.
48 Experiments

3.4 Buried Pipe

3.4.1 Introduction
The aim of performing this experiment was to assess how guided waves perform,
in a practical scenario, where pipes are buried in soil. UGWT is preferred over
convectional ultrasonic when inspecting buried pipes, because it allows for screen-
ing of pipe by placing transducers at a single convenient location thereby requiring
excavation in only one area.

There have been papers released by researchers discussing the propagation of

guided waves through a buried pipe, where the performance varied depending on
the type of soil [40–42]. The hypothesis for this experiment was that sand around
the pipe would result in high attenuation of waves hence, making it difficult to detect
the defect when the pipe is buried.

3.4.2 Experimental setup

The experimental setup was similar to the first two experiment, where a non-
coated pipe with a defect was used. A box was made of plywood, through which the
pipe went through and the box was filled with sand so that the defect was buried
in sand. The dimensions of the box are shown in figure 3.28, and it was filled up
with 50 kg of sharp sand (see Figure 3.29). Figure 3.31 shows the set up for this
experiment. Eight readings were taken around the circumference of the pipe to
detect the defect in this experiment.

Figure 3.28: Dimensions of the box through which the pipe goes through. This was
then filled with 0.0324 m3 of sand.
3.4 Buried Pipe 49

Figure 3.29: Part of pipe along with the defect buried in the sand box. The box
contained 50 kg of sand.

Figure 3.30: Transmitting and receiving transducers placed on the pipe, at 0 ◦

configuration.
 50

Figure 3.31: Setup for the experiment showing two transducers placed on the pipe. The transmitting and receiving transducers here are
placed 6.2 cm apart from each other and it has a defect buried in sand.
Experiments
3.4 Buried Pipe 51

3.4.3 Results and analysis

In this section it will be analysed how guided waves propagate through the buried
pipe and if it is able to detect the defect in the pipe when it is buried. Eight evenly
spaced readings were taken around the circumference of the pipe, for guided waves
transmitted at frequencies ranging from 30 kHz to 200 kHz. It was used to assess
at what frequency the guided wave must be transmitted to detect defect in a buried
pipe. The results obtained were then compared with the results from non-buried pipe.

Figure 3.32, shows the waveforms obtained when the guided wave were trans-
mitted through a buried pipe. As shown in the figure, guided waves propagate better
through a buried pipe than through a bitumen coated pipe. It also shows that the
defect echo is visible at high frequencies as it can be identified clearly at 50 kHz
and 100 kHz, whereas at 150 and 200 kHz it is present but not clearly identifiable
due to reverberation. The waveforms produced for the buried pipe were very similar
to the waveforms obtained for the non-buried pipe (see Figure 3.15). This was a
surprising result as it was expected the buried pipe to have a higher reverberation
than non-buried pipe.

To determine how reverberation varies, reverberation RMS was calculated for

normalised signals, using the same procedure as previous cases. It was calculated
using formula 3.1 for the time interval between the pipe end echo and the defect
echo, so that comparisons could be made between the buried and non-buried
pipe. Figure 3.33, shows how reverberation RMS varies in buried and non-buried
pipe. From this figure, it can see that reverberation for a buried pipe is higher than
non-buried when guided waves were transmitted at low frequencies i.e. 30 and 50
kHz, whereas at 100 kHz it is similar in both cases. For waves transmitted at higher
frequencies, the buried pipe performs better than the non-buried pipe i.e. 200 kHz.

After calculating reverberation RMS, signal RMS was calculated for the time
interval of the defect echo. The two RMS was then used to calculate SRR, using
formula 3.3. Figure 3.34 shows how the SRR varies with transmission frequency
of guided waves for buried and non-buried pipe. As shown in figure, when guided
waves were transmitted at low frequencies the SRR was high for non-buried pipe;
this is linked to low reverberation RMS for non-buried pipe. When transmitted at 100
kHz, the results were similar for both scenarios and for high frequency waves the
buried pipe has a better SRR than non-buried pipe, because at high frequencies the
reverberation is lower for the buried pipe.
 52

Figure 3.32: Amplitude of average received signal, for guided waves transmitted at different frequencies through the buried pipe, when eight
evenly spaced readings are taken circumferentially around the pipe. The area between the black dotted lines is then analysed ahead for
reverberation root mean square and the area between the red dotted line is used to calculate signal root mean square, which is used to
calculate the signal to reverberation ratio.
Experiments
3.4 Buried Pipe 53

Figure 3.33: Comparing how the average normalised reverberation RMS varies, for
guided waves transmitted at different frequencies, when the pipe is buried and not
buried.

Figure 3.34: Comparing how SRR varies, for guided waves transmitted at different
frequencies, when the pipe is buried and not buried.
54 Experiments

Figure 3.35: Variation in ratio between peak amplitude for a buried and non-buried
pipe, when guided waves are transmitted at different frequencies.

In order to compare how the peak amplitude varies in the two scenarios, ratio
between the peak amplitude of buried pipe and non-buried pipe was calculated for
the average waveforms of eight readings taken around the circumference of the pipe.
Figure 3.35, shows how the ratio of peak amplitude varies with frequency. As can be
seen from the figure, the amplitude ratio is almost one when the guided waves were
transmitted at 100 kHz. This result was expected as the reverberation RMS and
SRR were almost the same for both cases at that frequency. As 8 readings were
taken around the pipe it mitigates all the non-axisymmetric modes that get trans-
mitted at 100 kHz. At lower and higher frequencies the amplitude ratio is less than
one, as the peak amplitude for a buried pipe is lower than when the pipe is not buried.
3.4 Buried Pipe 55

3.4.4 Discussion and experimental conclusion

The main aim of performing this experiment was to assess how the guided waves
propagate when the pipe is buried. The hypothesis for this experiment was that the
guided waves attenuate highly when transmitted through a buried pipe and thereby,
the amplitude of the guided wave would be lower than when pipe is not buried. This
experiment proves that the amplitude is lower, but the guided waves do not attenu-
ate as highly as initially expected and it attenuates less than the bitumen coated pipe.

The main findings of the experiment are -

• Guided waves do not attenuate strongly in buried pipe as compared to a

bitumen coated pipe.

• When guided waves are transmitted at high frequencies the reverberation

RMS and SRR are better for buried pipes than non-buried pipes.

• Transmission of guided waves at 100 kHz produces similar results for both bur-
ied and non-buried pipes, when 8 readings are taken around the circumference
of the pipe.

The first finding is a plus point of UGWT and one of the reasons it is used by
petrochemical industries, to detect defects, as most pipes are buried underground.
The results produced show that sand does not bind with the pipe as strongly as thick
bitumen coating does; therefore there is less leakage of signal. Though it must be
noted attenuation of guided wave depends on soil type - clay, silt, sharp sand, soft
sand etc. and the depth of soil. Previous researchers have shown that some soil
types cause very high attenuation [40–42].

This experiment demonstrated that reverberation is less for a buried pipe than
a non-buried pipe, when guided waves are transmitted at high frequencies. This
results in high SRR at higher frequencies, as more useful signal is obtained. Re-
verberation for buried pipe is low at high frequencies, because the extra modes are
suppressed by the sand. The effect of low reverberation at high frequencies is the
reasons why at high frequencies the peak amplitude is lower, than the non-buried
pipe case, as there is less disturbance in the received signal. It can also be seen
that when the signal is transmitted at 100 kHz, for 8 readings taken around the cir-
cumference of the pipe, the reverberation RMS and SRR is same for both scenarios
and the amplitude of peak waveform is almost equal in both cases. The reason
for similar result at 100 kHz is because when eight readings are taken around the
circumference of the pipe it eliminates the reverberation, due to non-axisymmetric
modes and thereby gives a similar waveform in both cases where the defect echo
can be identified. It shows that taking multiple readings around the circumference of
56 Experiments

pipe benefits even when the pipe is buried.

From this experiment it can be concluded that UGWT performs well when testing
for a defect in buried pipes and it does not produce a high attenuation. Though
there are other factors also that determine how well the guided waves perform when
testing through a buried pipe and these must be taken into account, such as the
type of soil the pipe is buried in and how adhesive the soil is to the pipe. If the soil
is strongly adhesive then guided waves would perform poorly, because the waves
would get leaked to the soil.
3.5 Finding attenuation of the bare pipe 57

3.5 Finding attenuation of the bare pipe

In order to compare attenuation, attempts were made to estimate the attenu-
ation of the bare pipe. There were two approaches adopted for this. The first one
was where the guided waves were transmitted through the uncoated pipe and the
reduction of signal strength was measured over short distances by moving the
receiver transducer along the pipe length and taking readings of the peak to peak
amplitude. This is similar to reverberation measurements in a reverberant field,
example reverberation chamber. The results implied that attenuation was very low
for the pipe and the results were unreliable as the differences were small and hence
a much longer pipe is required. These results were also unreliable, because the
readings were corrupted by other additional modes that were transmitted.

The second approach involved taking a single measurement at zero degree

configuration, where the guided waves were transmitted at a range of frequencies,
to look at decay of the signal for each frequency. The signal were looked at a longer
time compared to previous cases. The RMS of the signal was calculated for 10
sets of 0.005 seconds, to see how the signal decays. The results produced in this
case suggested the attenuation is very high. The problem with this experiment was
that there was an assumption made that energy is trapped in one mode and it dies
exponentially, but in reality every time the waves hit boundary or defect there is a
mode change which probably dissipate quicker and therefore energy gets absorbed
by other mode. Therefore even this approach did not work.
This page is intentionally left blank.
Chapter 4

Conclusion

4.1 Main findings of this dissertation

In this dissertation, ultrasonic guided wave technology was studied and the
propagation of these waves was assessed in different scenarios. Experiments were
performed to investigate at what frequency the guided waves must be transmitted
to detect the defect located in the pipe, how rotating the receiver around the pipe
affects the received signal and defect detection. Tests were even performed to
understand how the technology responds, in industrial scenarios, when the pipe is
coated with bitumen and when the pipe is buried in sand.

The experimental tests were conducted on a steel pipe with a diameter of 3.4 cm
a wall thickness of 5 mm. A defect of 8.3% CSA in size was machined onto the pipe
to investigate at what frequency the guided waves must be transmitted to detect
the defect, it was detected at 50 kHz. At higher frequencies the defect echo was
not visible, as the echo was lost in the reverberation when two transducers were
used. In order to decrease the reverberation level at higher frequencies, it required
increasing the number of transmitting and receiving transducers to suppress the
non-axisymmetric modes. Due to budget limitation it was not feasible to purchase
many transducers. Therefore receiver transducer was physically rotated around
the circumference of the pipe and readings taken, that allowed the defect echo to
be detected at higher frequencies. In averaging the signal, the non-axisymmetric
modes were eliminated, due to phase difference when the receiving transducer is
rotated; hence leaving a resultant signal with low reverberation level.

The next experiment involved testing the technology on a bitumen coated pipe
where results were very poor as the guided waves attenuated highly when transmit-
ted through a bitumen coated pipe; hence the distance the guided waves propagated
was very limited. The other experiment was performed on a buried pipe where
results were encouraging as the defect that was buried in sand was detected by the
60 Conclusion

technology. This was possible because the sand on the outside does not bind to the
pipe as strongly as bitumen coating, thereby there is less leakage of energy.

4.2 Future work

As ultrasonic guided wave is still a relatively new technology there are a variety
of different experiments that can be performed, barring there is less time constraint
and a higher budget available.

If there is a higher budget available then an experiment should be carried out with
a transducer that is capable of transmitting and receiving signals, rather than using
two transducers. Also, instead of the transmitting guided waves in both directions of
the pipe it must be looked if it is possible to transmit guided wave in one direction so
that the transducer can be placed close to pipe end. This would reduce the number
of echoes received, making it easier to identify the defect echo. The other change is
to increase the number of transducers used around the circumference of the pipe.
This would help to reduce reverberation and allow the defect echo to be detected at
high frequencies as high frequency waves are more sensitive to small defects.

The other experiment that can be performed is using a chirp as an input signal,
instead of tone burst. A tone burst is a sinusoidal wave with limited cycles, whereas
chirp is a signal where frequency increases or decreases with time. A chirp signal is
preferred as it might help to reduce reverberation, due to it producing a frequency
sweep. In order to get a good resolution, short acoustic pulses must be transmitted;
the signal energy is proportional to the signal pulse duration and to get a good SNR
the pulses must be longer [43]. The solution to achieving the balance between single
echo resolution and SNR performance is to use a chirp instead of conventional tone
burst [43].

Lastly, guided waves can also be transmitted through a corroded pipe to investig-
ate how guided waves respond and whether they can detect the corrosion on the
pipe. The corrosion can also be introduced electrochemically onto the pipe. Guided
waves can even be propagated through pipes that are coated with other materials
such as epoxy, wax, polyethylene, etc. to see how the technology respond in those
cases.
References

[1] A. Demma, The interaction of guided waves with discontinuities in structures.

PhD thesis, Imperial College, London, 2003.

[2] V. Varma, “State of the art natural gas pipe inspection,” u. s. department of
energy report, Oak Ridge National Laboratory, Oak Ridge, Oak Ridge, TN
37831, Sept. 2002.

[3] S. Izadpanah, G. Rashed, and S. Sodagar, “Using ultrasonic guided waves

in evaluation of pipes,” in The 2nd International Conference on Technical
Inspection and NDT, (Tehran, Iran), Oct. 2008.

[4] The Hashemite University (N.D), “Ultrasonic testing, 2012.” Non-destructive

Testing Center. http://www.eis.hu.edu.jo/ACUploads/10526/Ultrasonic%
20Testing.pdf, [PDF], [Accessed 15 Oct.2012].

[5] T.H.Gan, “The application of guided wave ultrasonic technique for the inspection
of corrosion under insulation (cui) of pipes and pipelines.” Petromin Pipeliner,
Sept. 2010.

[6] J.W.S. Rayleigh, “The theory of sound.” Dover publications, Inc., Newyork, N. Y.
10014, 1945.

[7] D. Gazis, “Three dimensional investigation of propogation of waves in hollow

circular cylinders. i. analytical foundation.,” Journal of the Acoustical Society of
America, vol. 31, pp. 568–578, 1959.

[8] H. Fitch, “Obseravation of elastic-pulse propogation in axially symmetric and

nonaxially symmetric longitudinal modes of hollow cylinders,” Journal of the
Acoustical Society of America, vol. 35, pp. 706–708, 1963.

[9] K. Shivaraj, K. Balasubramaniam, C.V. Krishnamurthy, and R. Wadhwan, “Ul-

trasonic circumferential guided wave for pitting-type corrosion imaging at in-
accessible pipe-support locations,” Journal of Pressure Vessel Technology,
vol. 130, 2008.

[10] R. Kumar, “Dispersion of axially symmetric waves in empty and fluid-filled

cylindrical shells,” Acoustica, vol. 17, 1972.
62 REFERENCES

[11] M. Silk and K. Bainton, “The propagation in metal tubing of ultrasonic wave
modes equivalent to lamb waves,” Ultrasonic International, vol. 17, pp. 11–19,
1979.

[12] D. N. Alleyne and P. Cawley, “The excitation of lamb waves in pipes using
dry-coupled piezoelectric transducers,” Journal of Nondestructive Evaluation,
vol. 15, pp. 11–20, 1996.

[13] M. Lowe, D. Alleyne, and P. Cawley, “Defect detection in pipes using guided
waves,” Ultrasonic International, vol. 36, pp. 147–154, 1998.

[14] D. Alleyne, B. Pavlakovic, M. Lowe, and P. Cawley, “Rapid, long range inspection
of chemical plant pipework using guided waves,” Review of progress in quantitve
nondestructive evaluation, vol. 20, pp. 180–187, 2000.

[15] J.Davies, Inspection Of Pipes Using Low Frequency Focused Guided Waves.
PhD thesis, Imperial College, London, 2008.

[16] D.Sonia and B.Fouad, “Propagation of guided waves in a hollow circular cylinder
application to non destructive testing,” in 14th International Conference on
Experimental Mechanics, (Poitiers, France), June. 2010.

[17] J. L. Rose, Ultrasonic waves in solid media. Cambridge CB22RU: Cambridge

University Press, 1999.

[18] A. Demma, P. Cawley, M. Lowe, and A. G. Roosenbrand, “The reflection of the

fundamental torsional mode from cracks and notches in pipes,” Journal of the
Acoustical Society of America, vol. 114, pp. 553–555, 2003.

[19] J. Davies and P. Cawley, “The application of synthetic focusing for imaging
crack-like defects in pipelines using guided waves,” IEEE Transactions on
Ultrasonics, Ferroelectrics, and Frequency Control, vol. 56, p. 760, 2009.

[20] R. Carandente, J. Ma, and P. Cawley, “The scattering of the fundamental

torsional mode from axi-symmetric defects with varying depth profile in pipes,”
Journal of the Acoustical Society of America, vol. 127, pp. 3440–3448, 2010.

[21] A. Lvstad and P. Cawley, “The reflection of the fundamental torsional guided
wave from multiple circular holes in pipes,” NDT&E International, vol. 44,
pp. 553–555, 2011.

[22] B.E. Paton, V.A. Troitskii and A.I. Bondarenko (2010), “Method and technology
of diagonstics of pipeline systems by guided low-frequency ultrasonic waves.”
Guided NDE LLC. http://www.ndt.net/article/ecndt2010/reports/1_03_
38.pdf, [PDF], [Accessed 10 Oct. 2012].
REFERENCES 63

[23] J. L. Rose, “Ultrasonic guided waves in structural health monitoring,” Key

Engineering Materials, vol. 270-273, pp. 14–21, 2004.

[24] Z. Liu, C. He, B.Wu, X.Wang, and S. Yang, “Circumferential and longitudinal de-
fect detection using t(0,1) mode excited by thickness shear mode piezoelectric
elements,” Ultrasonics, vol. 44, pp. e1135–e1138, 2006.

[25] S. Fletcher, M. Lowe, M. Ratassepp, and C. Brett, “Detection of axial cracks

in pipes using focused guided waves,” Review of progress in quantitve
nondestructive evaluation, vol. 29, pp. 56–59, 2009.

[26] J. L. Rose, “Dispersion curves in guided wave testing,” Materials Evaluation,

vol. 61, pp. 20–22, 2003.

[27] G. K. Sohi, Study of Rayleigh Lamb Wave Propagation in elastic and micropolar
solids. PhD thesis, Thapar University, Punjab India, 2012.

[28] M. Lowe (2011), “The guided wave testing method for screening of pipelines.”
V Pan American Conference for Non Destructive Testing Cancun. http:
//www.ndt.net/article/panndt2011/presentations/143_Lowe.pdf, [PDF],
[Accessed 18 Nov. 2012].

[29] T.M.Webb (2011), “Benefits and challenges of ultrasonic guided wave inspec-
tion.” 2011 API Inspection Summit and Expo. http://new.api.org/meetings/
proceedings/2011inspectionsummit.cfm?renderforprint=1, [PPT], [Ac-
cessed 18 Nov. 2012].

[30] M. Lowe and P.Crawley, “Long range guided wave inspection usage current
commercial capabilities and research directions,” tech. rep., Department of
Mechanical Engineering Imperial College London, London UK, 2006.

[31] J. L. Rose, “A baseline and vision of ultrasonic guided wave inspection potential,”
Journal of Pressure Vessel Technology, vol. 124, pp. 273–277, 2002.

[32] M. I. Al-Ammandi, “Long range ultrasonic inspection an end user’s view,” in

3rd MENDT - Middle East Nondestructive Testing Conference & Exhibition,
(Manama, Bahrain), Nov. 2005.

[33] A. Demma, P. Cawley, M. Lowe, and A. G. Roosenbrand, “The reflection of the

fundamental torsional mode from cracks and notches in pipes,” Journal of the
Acoustical Society of America, vol. 114(2), pp. 611–625, 2003.

[34] S. Li-ying, Y. Xiao-dong, and L. Yi-bo, “Research on transducer and frequency

of ultrasonic guided waves in urban pipe inspection,” Industrial Electronics and
Applications, 2009. ICIEA 2009. 4th IEEE Conference on, vol. -, pp. 2708–2711,
2009.
64 REFERENCES

[35] Envirocoustics (N.D), “Guided ultrasonic waves pipe inspection.” Envirocous-

tics - Non Destructive Testing. http://www.envirocoustics.gr/services/
ut_guided/GUL.pdf, [PDF], [Accessed 21 March 2013].

[36] Pipelines International (2010), “Inside and out: benefits of pipeline coating.”
Pipelines International - December 2010. http://pipelinesinternational.
com/news/inside_and_out_benefits_of_pipeline_coating/053589/, [On-
line], [Accessed 21 March 2013].

[37] F. Simonetti and P. Cawley, “On the nature of shear horizontal wave propagation
in elastic plates coated with viscoelastic materials,” The Royal Society.

[38] V. Patel, Petrochemicals. -: InTech, 2012.

[39] J. wen Cheng, The Attenuation of Guided Wave Propagation on the Pipelines.
PhD thesis, National Sun Yat-Sen University, China - Taiwan, 2006.

[40] A. Demma, D. Alleyne, and B. Pavlakovic, “Testing of buried pipelines using

guided waves,” in 3rd MENDT - Middle East Nondestructive Testing Conference
& Exhibition, (Manama, Bahrain), Nov. 2005.

[41] J. Rose, J. Mu, J. Hua, R. Royer, and S. Kenefick, “Ultrasonic guided wave
phased array inspection of coated and buried pipe (part1: theory),” in JRC-NDE
2009 - Developments in Ultrasonic Guided Wave Inspection, (-), June 2010.

[42] J. Rose, J. Mu, J. Hua, R. Royer, and S. Kenefick, “Ultrasonic guided wave
phased array inspection of coated and buried pipe (part2: experiment),” in
JRC-NDE 2009 - Developments in Ultrasonic Guided Wave Inspection, (-),
June 2010.

[43] J. E. Ehrenberg and T. C. Torkelson, “Fm slide (chirp) signals: a technique

for significantly improving the signal-to-noise performance in hydroacoustic
assessment systems,” Fisheries Research, vol. 47(2-3), pp. 193–199, 2000.

[44] D. J. Chinn, M. J. Quarry, and J. L. Rose, “Group velocity and phase velocity,”
tech. rep., Lawrence Livermore National Laboratory, Livermore, CA, 2005.

[45] M. S. Santhanam (2009), “The guided wave testing method for screening
of pipelines.” IISER, Pune/PRL, Ahmedabad. http://www.iiserpune.ac.in/
~santh/gvel.pdf, [PDF], [Accessed 27 Sept. 2012].
Appendices
Appendix A

Phase and group velocities

It is important to understand the concept and importance of phase and group

velocity in order to analyse dispersion curves. Phase velocity is described as the
speed of the phase of the guided wave, whereas the group velocity is the speed at
which guided wave packet propagates [44].

A simple analogy was used by a lecturer to describe the difference between

phase and group velocity. If there is race, where group of people, start at the same
time. Initially all appear to be running at same speed. After a while, the group will
disperse and each runner has a different speed. Phase velocity now can be thought
of as speed of individual runner and group velocity as the speed of entire group.
Phase velocity of waves is usually higher than group velocity [45].

Phase velocity is noted as Cp , whereas group velocity is noted as Cg they are

calculated using [17] -
ω
Cp = (A.1)
k
∂ω
Cg = (A.2)
∂k
’ω’ and ’k’ in this equations are natural frequency and wave number respectively.

Cg can be also written as -

∂(kCp ) ∂Cp
Cg = = Cp + k (A.3)
∂k ∂k

Group velocity is used in identifying and locating a defect, whereas phase velocity
is used in design of guided wave sensor for mode excitation. More details about the
concept and equations can be found in Ultrasonic waves in solid media by Rose [17].
This page is intentionally left blank.
Appendix B

Calculation for number of

transducers required

The calculation performed here is for T(0,1) wave mode that is transmitted at 100
kHz along a pipe with an outer diameter of 34 mm. First the group velocity is found
using the dispersion curve. The group velocity for T(0,1) remains constant through-
out the frequency range at 3309 ms−1 . Using the group velocity and frequency the
wavelength of the wave is calculated using -

Cg = f × λ (B.1)

This results in a wavelength of 0.033 m. As mentioned earlier, the transducer

spacing (△) for an axially symmetric wave must be -

λ
△≤ (B.2)
2
The minimum number of transducers (Nt ) required is then calculated using
the formula below, where ’aext ’ is the external pipe radius which is 17 mm and the
spacing being 16.5 mm.

2πaext
Nt = (B.3)
△
Therefore the minimum number of piezoelectric transducers required to
propagate T(0,1) at 100 kHz in a 34 mm diameter pipe is 7. Similarly to propagate it
at 50 kHz we would require 4 piezoelectric transducers.
This page is intentionally left blank.
Appendix C

CD

This CD contains all files as detailed below:

• PDF version of the report

• Latex files of the report
• Excel files with all the laboratory readings
• Poster