ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 52, no.
4, april 2005 653
Omnidirectional Guided Wave Inspection of
Large Metallic Plate Structures Using an
EMAT Array
Paul Wilcox, Mike Lowe, and Peter Cawley, Member, IEEE
Abstract—The design of an electromagnetic acoustic
transducer (EMAT) array device for the inspection of large
areas of metallic plate-like structures using the S0 guided
wave mode is described. The reasons for using the S0 mode
are discussed and it is shown how the choice of mode
determines the nature of the EMAT array elements. A
novel array construction technique is shown to be neces-
sary whereby the EMAT coils for adjacent elements are
overlapped in order to achieve the required element den-
sity. Results are presented that illustrate the operation of
the device on steel and aluminum plate specimens in the
thickness range from 5 to 10 mm. An area of at least 10 m2
can be inspected from a single location. Spurious signals
in the results are caused both by the unwanted A0 mode
and by S0 sidelobes, the latter occurring at the same radial
distance from the array as the genuine S0 signal from a re-
flector, but in the wrong direction. The signal-to-coherent
noise performance of the complete system is determined
by the amplitude ratio of the largest genuine S0 signal to
the largest spurious signal. This is typically around 30 dB.
The sensitivity of the device to artificial defects and gen-
uine corrosion patches is demonstrated and the limitations
of its operation are discussed. The feasibility of using the
device with the S1 guided wave mode to inspect a 20 mm
thick plate is also demonstrated.
I. Introduction
his paper describes an omnidirectional guided wave
T inspection device, comprising a circular array of elec-
tromagnetic acoustic transducer (EMAT) elements, that
enables a large area of a metallic plate-like structure to
be inspected from a single location. An earlier publication
[1] describes the mathematical basis of the algorithm that
has been developed to process the data from omnidirec-
tional guided wave arrays such as this. The current paper
presents the physical design and construction of the EMAT
array itself, together with results that have been obtained
from a variety of plate specimens.
The use of guided waves in nondestructive evaluation
(NDE) applications is widespread [2] and can be grouped
into three categories according to the propagation dis-
Manuscript received March 2, 2004; accepted September 16, 2004.
This work was supported by the UK Engineering and Physical
Sciences Research Council (grant number GR/M04839/01), Shell,
Chevron-Texaco, BP, and Exxon-Mobil.
P. Wilcox is with the Department of Mechanical Engineer-
ing, University of Bristol, Bristol, BS8 1TR, UK (e-mail:
p.wilcox@bris.ac.uk).
M. Lowe and P. Cawley are with the Department of Mechanical
Engineering, Imperial College, London SW7 2AZ, UK.
0885–3010/$20.00

c 2005 IEEE
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 06,2020 at 07:24:59 UTC from IEEE Xplore. Restrictions apply.
tances involved. In short-range applications, guided waves
are used to obtain localized data about a specimen in situ-
ations where conventional ultrasonic techniques cannot be
applied. Examples include air-coupled inspection of com-
posites [3], Rayleigh wave detection of surface discontinu-
ities in artillery shells [4], and acoustic microscopy [5]. In
medium- and long-range applications, the reason for us-
ing guided waves is to increase the amount of a structure
that can be inspected from one location, hence reducing in-
spection costs and sometimes enabling inaccessible regions
in a structure to be tested. Long-range applications typ-
ically involve one-dimensional (1D) waveguide structures
such as pipes [6] and rails [7] where the propagation dis-
tances are in the order of tens or even hundreds of meters.
Conversely, plate-like structures such as storage tanks and
pressure vessels behave as two-dimensional (2D) waveg-
uides, and their inspection using guided waves falls into
the medium-range category, with propagation distances of
a few meters. Guided wave plate inspection in the form of
continuous line scanning between two transducers is ide-
ally suited to the continuous monitoring of, for example,
steel plate in rolling mills [8] and also forms the basis of
Lamb wave tomography [9]. The guided wave device de-
scribed in [10] employs a wavelength tunable meander coil
EMAT to excite and detect the fundamental SH mode in
a pulse-echo configuration for the purpose of inspecting
laser welds. This device is mechanically moved along the
plate, parallel to the weld under inspection, to generate a
B-scan image of the area of the plate on either side of the
weld and the weld itself. However, there has been very lit-
tle commercial exploitation of guided waves for inspecting
large areas of plate-like structures from a single location
in a manner analogous to that employed in pipe and rail
inspection.
The practical obstacles that must be overcome for the
successful guided wave inspection of structures that behave
as 2D waveguides are somewhat different from those for
1D waveguides. In the latter, the number of possible prop-
agation directions is only two, but the number of guided
wave modes that exist at typical operating frequencies is
high. Conversely, in the case of 2D plate-like waveguides,
the number of possible guided wave modes may be as low
as three (these being the fundamental symmetric and an-
tisymmetric Lamb wave modes and the fundamental SH
mode) whereas the number of propagation directions is
infinite. In some cases, the problem of multiple propaga-
tion directions is surmountable by the design of the ex-
 654 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 52, no. 4, april 2005
periment. For instance, in a laboratory guided wave test,
the size of the sample can be made large enough so that
any spurious reflected signals from the edges of the sam-
ple arrive later than the reflection from the feature of in-
terest [11]–[13]. Similarly, in some industrial applications,
useful data can be obtained on the state of the structure
between two guided wave transducers operating in pitch-
catch mode by examining the amplitude or arrival time [9]
of the direct transit signal. In these configurations there is
only one wave propagation direction of interest, and the
2D waveguide is effectively reduced to a 1D one that can
be represented by the classical Rayleigh-Lamb model [14].
Much work founded on this model has been published for
transducers [15], including array devices [16] and modal
selectivity [17].
However, in order to develop a device for inspecting the
surrounding area of a plate-like structure in all directions,
full account needs to be taken of the fact that the struc-
ture acts as a 2D waveguide. For this reason, the issue of
guided wave transducer directionality becomes at least as
important as the issue of modal selectivity. Sidelobes and
grating lobes of the main beam from a guided wave trans-
ducer are of crucial importance when reflectors can lie in
any direction. For example, a defect-free rectangular plate
contains four major reflectors due to the edges (and, de-
pending on the guided wave mode, a further four due to
the corners). If the directionality of a guided wave trans-
duction device is not understood, then spurious sidelobe
reflections from any of these large reflectors can easily be
misinterpreted as signals from defects in other directions.
The design and construction of the plate-testing device
is described in Section II. Results from tests on a vari-
ety of specimens that indicate the potential of the device
and its operating limits in terms of plate thickness, defect
detectability, etc., are presented in Section III
II. System Design
A. Overview
The plate-testing device uses an array of EMAT ele-
ments arranged in a circular pattern within a protective
housing that is approximately 200 mm in diameter. The
array is connected via an umbilical to the instrumentation,
which is in turn controlled from a laptop computer where
the bulk of the signal processing is performed. The output
from the signal processing is a B-scan, which is referred
to as an omnidirectional B-scan as it covers the surround-
ing area of the plate under inspection up to a distance
of several meters and over a full 360◦ . The grayscale of
the B-scan image indicates the amount of reflected guided
wave energy as a function of position. The following sub-
sections summarize the key points in the design route that
led to this configuration and describe in detail the physical
construction of the final prototype system.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 06,2020 at 07:24:59 UTC from IEEE Xplore. Restrictions apply.
B. Selection of Guided Wave Mode
The guided wave mode and the frequency at which it is
used is referred to as the operating point of the inspection
system. It has been noted [17] that the primary criteria
that should be used to select an operating point are those
criteria determined by the structure under inspection. Of
interest in this work are the attenuation of waves in the
structure and the sensitivity of waves to surface corrosion.
For two reasons, it was initially decided to restrict consid-
eration to the fundamental guided wave modes, A0 , S0 , and
SH0 (the fundamental SH mode). First, a reasonable body
of research already exists that shows how these modes in-
teract with corrosion-like defects [18], [19]. Second, it was
anticipated that the problems of achieving modal selectiv-
ity as well as directionality would be alleviated somewhat
if there were only a limited number of modes present. The
phase and group velocity dispersion curves for these modes
in steel plate are shown in Figs. 1(a) and (b), respectively.
A major potential application area of the work de-
scribed here is the inspection of liquid-filled storage tanks.
The effect of liquid loading on one or both sides of a 2D
waveguide is to provide a mechanism for energy radiation,
and hence guided wave attenuation. If the phase velocity of
guided waves in the plate is higher than the bulk longitudi-
nal wave velocity in the surrounding liquid, energy may be
radiated into the liquid as bulk waves, even if the liquid is
inviscid. The degree of attenuation depends on the amount
of out-of-plane displacement due to the guided wave at the
surface of the waveguide. The attenuation curves for a steel
plate with water on one side are shown in Fig. 1(c). The
requirement to be able to inspect plates with liquid on
one side rules out many operating regions on the disper-
sion curves where the attenuation is too high, including
the entire A0 mode. The remaining options are to utilize
either the S0 mode at relatively low frequencies (before its
attenuation becomes too large) or the SH0 mode.
In terms of sensitivity to corrosion defects there is little
to choose between the S0 and SH0 modes. Both have mode
shapes dominated by in-plane particle displacements that
are almost constant through the thickness of the plate, dis-
tinguished by the fact that in the SH0 case this motion is
perpendicular to the direction of propagation rather than
parallel as in the case of S0 . The ultimate choice of oper-
ating on the S0 mode was due to the transduction consid-
erations described in the following subsection.
C. Selection of Transducer Element
Because the overall system is required to be omni-
directional, it is logical that this omnidirectionality ex-
tends down to the individual array elements. Each element
should behave as either a point transmitter or a point re-
ceiver of the chosen guided wave mode, with uniform sensi-
tivity to that mode in all directions. For a mode with a sig-
nificant out-of-plane surface displacement component this
is straightforward to achieve using a point transducer that
is sensitive to out-of-plane motion, as shown in Fig. 2(a).
 wilcox et al.: use of an emat array for omnidirectional guided wave inspection 655

Fig. 1. Dispersion curves showing (a) phase velocity and (b) group velocity for a steel plate as functions of frequency thickness product. The
attenuation curves for the same guided wave modes in a steel plate with water on one side are shown in (c).

Fig. 2. Schematic diagrams illustrating the excited guided wave field by (a) a point source in the out-of-plane direction, (b) a point source
in the in-plane direction, (c) distributed surface tractions polarized in the radial direction, and (d) distributed surface tractions polarized
in the tangential direction.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 06,2020 at 07:24:59 UTC from IEEE Xplore. Restrictions apply.
 656 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 52, no. 4, april 2005
This approach has been used to build omnidirectional
guided wave arrays of piezoelectric disk elements that use
the A0 mode [20], [21]. However, the low attenuation prop-
erties of the S0 and SH0 modes mean that they cannot
be efficiently excited or detected by transducers sensitive
to out-of-plane surface motion. A point transducer that is
sensitive to in-plane motion will excite and detect both the
S0 and SH0 modes [22] but will not do so in an omnidirec-
tional manner since, in transmission, the excitation force
must be in a particular direction. The resulting field from
such an element is shown schematically in Fig. 2(b). The
requirement for exciting and detecting either the S0 mode
or the SH0 mode omnidirectionally requires a transducer
element that, when transmitting, applies an axisymmetric
distribution of surface traction to the structure over a fi-
nite area. The required surface traction distributions for
exciting the S0 and SH0 modes are illustrated schemati-
cally in Figs. 2(c) and (d), respectively.
The problem is to devise a transducer element that can
apply one or the other of these surface traction distribu-
tions. Dry contact piezoelectric devices have been used to
excite guided waves using in-plane surface tractions in a
number of applications [6], [7], although existing devices
are not omnidirectional. It is conceivable that a disk- or
annular-shaped piezoelectric element could be poled in a
such a way that its surface motion was axisymmetric. How-
ever, in order for such an element to transmit guided waves
in an omnidirectional manner, the coupling between the
element and the structure would need to be uniform over
the area of contact. This would be very hard to achieve in
practice on the structures of interest, which are unlikely to
have sufficiently flat and smooth surfaces. A further dis-
advantage of using dry contact piezoelectric elements is
that they require loading onto the surface of the struc-
ture, and hence require something to react against. In the
case of pipe inspection, this is relatively straightforward to
achieve since the transducer elements are typically held in
a ring around the circumference of the pipe. In the case of
an array of piezoelectric elements applied to a flat plate-
like structure, the generation of a transducer loading force
requires either an external force to be applied to the ar-
ray or some attachment made to the structure (e.g., using
magnets or suction).
To overcome the coupling and loading problems associ-
ated with piezoelectric elements, an electromagnetic solu-
tion was sought. A pancake coil EMAT for guided wave
excitation comprises a planar spiral coil and a perma-
nent magnet above the plate in the configuration shown
schematically in Fig. 3. When an alternating current is
passed through the coil, alternating eddy currents are gen-
erated in the surface of the plate below it. These eddy
currents mirror the current in the coil and thus flow in
a circular pattern. The permanent magnet provides a DC
magnetic field that in the immediate vicinity of the coil
is in the out-of-plane direction with respect to the plate
surface. The interaction of the DC out-of-plane magnetic
field with the alternating in-plane eddy currents flowing
in the angular direction produces an alternating Lorenz
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 06,2020 at 07:24:59 UTC from IEEE Xplore. Restrictions apply.
Fig. 3. (a) Pancake coil EMAT and (b) interaction of eddy currents
with a magnetic field to produce radially polarized surface traction
forces when the EMAT is transmitting.
force in the in-plane radial direction. This provides pre-
cisely the surface traction distribution required to generate
S0 guided waves in an omnidirectional manner, and it was
for this reason that the S0 mode was ultimately utilized
rather than the SH0 mode.
A test program to investigate the performance of pan-
cake coil EMATs was undertaken, an important aspect
of which was to study the effect of coil geometry. Pairs
of EMATs with identical geometries were tested in pitch-
catch configuration, 500 mm apart on a 2.5-m by 1.25-
m by 5-mm-thick aluminum plate using 5-cycle Hanning-
windowed tonebursts with center frequencies of 50, 75, 100,
125, 150, 175, 200, 225, and 250 kHz. The separation of the
EMATs and the large size of the plate enable the first ar-
rival signals of the S0 and A0 modes to be resolved in time,
both from each other and from the first reflected S0 signals
from the plate edges. The amplitude of the peak current
passing through the transmitting EMAT was monitored
and the results were normalized to this to allow compar-
ison between different EMATs and different frequencies.
A graph showing the typical amplitudes of the A0 and S0
modes for a particular EMAT as a function of frequency is
shown in Fig. 4. A model has also been developed, the re-
sults from which show good agreement with experimental
data. This work will be described in detail in a forthcom-
ing publication and here only the aspects relevant to the
design of the EMAT array are discussed.
The sensitivity of an EMAT to a particular guided wave
mode is dependent on a number of factors that may be di-
vided into three interlinked categories: geometric, modal,
and electrical. The geometric factors are the inner and
outer diameters of the EMAT coil and the number of
turns. The inner and outer diameters govern the spatial
distribution of sensitivity to in-plane radial surface mo-
tion and introduce a wavelength dependence to the sensi-
 wilcox et al.: use of an emat array for omnidirectional guided wave inspection
Fig. 4. Modal sensitivity of a typical EMAT (24 turns; outer coil
diameter = 28 mm; inner coil diameter = 14 mm) as a function of
frequency.
tivity. The excitability of different modes by in-plane sur-
face traction is governed by their mode shapes and intro-
duces frequency-dependent terms into the sensitivity that
are different for each mode. The combination of the geo-
metric and modal factors governs the relative sensitivity of
an EMAT to different modes, and the only way in which
this can be altered is by the geometry of the EMAT coil.
The electrical impedance of an EMAT, which is also partly
determined by the coil geometry, and the impedance of the
associated instrumentation introduce further frequency de-
pendence of the EMAT sensitivity to all modes. Hence,
electrical effects do not alter the relative sensitivity of the
EMAT to different modes at a particular frequency and,
although absolute signal amplitude is important, the re-
sults presented in Section III show that signal-to-coherent
noise is the factor that ultimately limits the performance
of EMAT array system. For this reason, the most impor-
tant part of the EMAT element design is to maintain as
high a ratio of S0 to A0 sensitivity as possible, because sen-
sitivity to A0 leads to coherent noise. The coil geometry is
therefore the aspect that must receive attention. The key
results from the investigation into EMAT coil geometry
are summarized as follows:
1. If the inner diameter is less than half of the outer di-
ameter of the EMAT, the modal sensitivity is governed
by the outer diameter-to-wavelength ratio alone.
2. The peak sensitivity of a typical EMAT to the S0 mode
occurs at a frequency where the coil outer diameter
is approximately two-thirds of the wavelength of the
S0 mode. This is indicated by the right-hand vertical
dotted line in Fig. 4.
3. The sensitivity of an EMAT to the A0 mode is typi-
cally an order of magnitude lower at the point of peak
sensitivity to the S0 mode. However, as the frequency
is reduced, the sensitivity to the A0 rises (and ulti-
mately peaks close the frequency where the outer coil
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 06,2020 at 07:24:59 UTC from IEEE Xplore. Restrictions apply.
657
diameter is two-thirds of the wavelength of the A0
mode).
4. As a receiver, the sensitivity of the EMAT is increased
by increasing the number of turns (either by reducing
the inner diameter of the coil or by increased the den-
sity of the turns).
5. As a transmitter, the EMAT coil has an optimum
number of turns at which the maximum amplitude of
transmitted signal occurs. This optimum is a function
of the output impedance of the device used to drive
the EMAT.
It should be stressed that results 2 and 3 above are
applicable only for the specific instrumentation and oper-
ating frequency range used by the array described here.
The actual ratio of outer diameter to wavelength at which
the peak sensitivity to a mode occurs is heavily influenced
by both the modal and instrumentation characteristics.
The EMAT testing program also examined the omni-
directionality of EMATs by using a reference EMAT to
measure the amplitude of the S0 signal at evenly dis-
tributed positions around a circle centered on a transmit-
ting EMAT. Typical results for a single EMAT in isolation
are shown in Fig. 5(a). The uniformity of the field indicates
that the transmitting EMAT is behaving as an omnidirec-
tional source and also that the coupling of the movable
reference EMAT is highly consistent. The results for mul-
tilayered EMATs shown in Fig. 5(b) are discussed in the
following subsection.
D. Assembly of EMAT Elements to Form an Array
In parallel with the development of omnidirectional
EMAT elements, the layout of elements in an array was
investigated. The mathematical basis of the algorithm to
process data from a circular guided wave array is described
in detail elsewhere [1], and here it is only necessary to
summarize the key points in order to understand the con-
straints that the algorithm imposes on array layout. The
algorithm processes data from the array to simulate the
operation of a monolithic guided wave transducer that is
tuned to the S0 mode in pulse-echo configuration and ro-
tated through 360◦ to produce the final omnidirectional
B-scan. By using simulated data sets, it was found that
the angular resolution improved as the ratio of overall ar-
ray diameter to wavelength was increased. This is con-
sistent with the beam divergence angle of a conventional
ultrasonic transducer being governed by the aperture to
wavelength ratio [23]. From the point of view of physically
constructing an array, the more important results were re-
lated to the maximum interelement spacing. It was found
that for any array configuration, there was an upper limit
on acceptable inter-element spacing above which spurious
grating lobe signals appeared. There did not appear to be
a precise relationship between this maximum inter-element
spacing and the appearance of grating lobes as in the case
of a linear ultrasonic array [24]. However, it was empiri-
cally determined that an inter-element spacing that was
less than one-third of the wavelength of the S0 mode at
 658 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 52, no. 4, april 2005
Fig. 5. Sensitivity of S0 mode as a function of angle for (a) a single EMAT and (b) an EMAT with coils printed on each of four overlapping
layers in a multilayer PCB.
the center frequency of the transmitted signal was always
sufficient to eliminate grating lobes [1].
Unfortunately, this means that there are two contra-
dictory requirements for the size of EMAT elements. On
the one hand, in order to avoid grating lobes there should
be at least three elements per wavelength, which implies
that the elements must be less than one-third of the wave-
length in diameter. The upper frequency limit that this
constraint places on a particular EMAT is indicated by
the left-hand vertical dotted line in Fig. 4. On the other
hand, the results from the EMAT testing program indi-
cate that the element outer diameter needs to be of the
order of two-thirds of a wavelength in order to maximize
the sensitivity to the S0 mode and minimize the sensitiv-
ity to the A0 mode. One possible means of satisfying these
conflicting requirements is to physically shift the elements
in the array during a test, in order to simulate the effect of
the elements being located closer to each other than their
physical dimensions allow. This approach is not attractive,
since the need to use moving parts negates one of the key
advantages of using an array.
An alternative approach, and that which was ultimately
adopted, is to overlap the elements. This concept is illus-
trated schematically in Fig. 6(a). The overlapping coils
for different elements share a common permanent magnet
that provides the magnetic field for all of them. In order
for this to be realized in practice, the coils were printed
onto double-sided flexible polyamide printed circuit board
(PCB) material, 0.1 mm thick. To avoid having to cross
conductive tracks, each coil consists of two spiral layers
shown schematically in Fig. 6(b). On one side of the PCB,
the conductor follows a spiral path into the center at which
point it passes through the PCB via a plated-through hole.
It then follows a path that spirals outward on the other
side of the PCB. The spirals have a common center point
and their orientation is such that the current flows in the
same sense on both sides of the PCB.
The final array comprised four overlapping layers of
EMAT coils printed on four separate sheets of 0.1-mm-
thick polyamide PCB material in the configuration shown
in Fig. 6(c). These four printed layers were separated by
insulator layers of the same PCB material, and the lower
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 06,2020 at 07:24:59 UTC from IEEE Xplore. Restrictions apply.
surface of the array was protected with a layer of 0.38-mm
(0.015 )-thick polypropylene shim.
There was a concern that the overlapping of PCBs con-
taining EMAT coils in offset positions might affect the
omnidirectionality of the elements. The omnidirectional-
ity test described in the preceding subsection was repeated
for EMAT coils on each of the layers in the four-layer con-
figuration and the results are shown in Fig. 5(b). It can
be seen that there is no discernible loss of omnidirection-
ality in the multilayer configuration, although there is a
uniform decrease in amplitude for the coils in the layers
farthest from the plate surface. Conceivably, this variation
in amplitude could be compensated for at the data pro-
cessing stage, although this has not been implemented to
date.
Several reasons for not using omnidirectional piezoelec-
tric elements have already been discussed in Section II-C.
Even with suitable poling to make them sensitive to radial
in-plane motion, piezoelectric elements would suffer the
same conflicting requirements of diameter and spacing as
the EMAT elements. However, while the EMAT coils can
be overlapped to overcome this problem, the same solution
cannot be applied to piezoelectric elements.
E. Physical Construction of Array
Numerical simulations indicate that the optimum lay-
out of elements in an array is for them to be uniformly
distributed around a single circular ring. In this configura-
tion, all elements must be used as both transmitters and
receivers. However, although simple pitch-catch measure-
ments between separate EMATs in close proximity have
been successfully performed, the initial attempts to as-
semble an array of EMATs on a multilayered PCB were
not successful. Large amplitude reverberations dominated
the signals and rendered them unusable. This effect was
worst when the transmitter and receiver elements were on
the same PCB layer; hence it was concluded that the re-
verberations were occurring in the PCB material. Such
waves are necessarily excited by the same mechanism as
that which enables an EMAT to function, since Lorenz
forces also act on the current carrying conductors in the
 wilcox et al.: use of an emat array for omnidirectional guided wave inspection 659

Fig. 6. Schematic diagrams showing (a) concept of overlapped coils sharing a common magnet and (b) double-sided coil pattern on PCB.
Note that in (b) only a few turns are shown for clarity; the real coils have many more turns. The layout of the coils in the final prototype
array is shown in (c).

EMAT coils themselves. Furthermore, it was observed that
even if physically separate PCBs were used for transmit-
ter and receiver, reverberations were still apparent if the
two shared a common magnet. This is thought to occur
because the transmitter excites waves in the magnet due
to Lorenz forces acting on eddy currents induced in it.
For these reasons, it was decided to use separate trans-
mitting and receiving elements in the final array device
and to attempt to mechanically isolate them from each
other. The configuration chosen was an inner ring of 16
transmitting elements, surrounded by a concentric outer
ring of 32 receiving elements as shown in Fig. 6(c). For
both rings, the same four-layer overlapping PCB construc-
tion was employed. Ideally, two separate annular, mono-
lithic permanent magnets should have been used to provide
the magnetic fields for the transmitter and receiver rings.
However, for the prototype device this was too costly, so
instead rings containing a number of cylindrical perma-
nent neodymium-iron-boron magnets (Eclipse Magnetics,
Sheffield, UK, Part No. N35A200610) were assembled on
two concentric annular steel pole pieces. The polarity of
the magnets is reversed in one ring and an isolating layer
of rubber sheet above the magnets is used to separate the
two rings from a disk-shaped steel upper pole piece, com-
mon to both rings. The purpose of the upper pole piece is
to close the magnetic circuit made between the two rings
and hence to increase the magnetic field strength in the
specimen under inspection. The complete EMAT array as-
sembly is shown in the exploded picture in Fig. 7.
F. Instrumentation and Operation
The objective of the data acquisition stage when the
EMAT array is in operation is to capture a complete set
of 512 time-traces, one for each possible pair of transmit-
ter and receiver elements. The optimum configuration for
achieving this is to simultaneously record signals from all
receiving elements as each transmitting element is acti-
vated in turn. The physics of wave propagation in the
structure limit the time between successive transmissions,
since the reverberating signals due to excitation of one
transmitter must decay before the next transmitter can be
activated. Therefore, it is desirable to minimize the num-
ber of transmitting elements, hence the choice of using 16
transmitters and 32 receivers, rather than the other way
around. The minimum time between successive transmis-
sions is typically of the order of 0.1 s; hence the shortest
possible time for data acquisition with 16 transmitters is
1.6 s. If averaging is performed, then the shortest possible
acquisition time increases in proportion to the number of
averages taken.
A block diagram of the instrumentation for the pro-
totype system is shown in Fig. 8. With the exception
of the controlling computer, all of the instrumentation
is custom-made. The signal generator and power ampli-
fier deliver a high-voltage (of the order of 100 V p-p),
high-current (of the order of 100 A p-p) signal through
one of 16 relays to the currently selected transmitting el-
ement. The received signals from all 32 receiving elements

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 06,2020 at 07:24:59 UTC from IEEE Xplore. Restrictions apply.
 660 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 52, no. 4, april 2005

Fig. 9. A typical raw time-domain signal obtained from one EMAT

transmitter-receiver pair in the array during a test on a 6-mm-thick
steel plate at 170 kHz.

are recorded on 32 parallel amplification and digitization

channels. The analogue amplification stage contains a 30-
to-300 kHz band-pass filter, and the digitization is per-
formed using 12-bit digitizers that have selectable sample
rates of 0.5, 1, and 2 MHz. The maximum number of points
that can be recorded in each time-trace is 4000. Each of
the channels contains a microprocessor to allow between 1
and 16 averages to be performed over successive digitized
signals. The recorded, averaged signals from each channel
are uploaded to the controlling computer via the dedicated
10BaseT network connection and this operation can be
performed while the next set of data is being recorded and
averaged. Data acquisition times are typically 20–30 s, the
Fig. 7. Schematic exploded diagram showing the construction of the
array. rate-determining step being the upload time if less than
10 averages are used. A typical raw time-trace is shown in
Fig. 9.

III. Results

A. Presentation of Results and Limitations of Operation

The omnidirectional B-scan images produced by the

prototype device are plotted using a decibel (dB)
grayscale, with 0 dB corresponding to the amplitude of
the largest signal in the image. For printing clarity, the
gray levels are separated into 5-dB bands. This process
necessarily makes small noise signals appear much more
significant than if the data were plotted on a continuous
grayscale.
A distance amplitude correction (DAC) proportional to
the square root of the radial distance from the array is
applied to the images in order to compensate for beam
Fig. 8. Block diagram of instrumentation used in the prototype sys- spreading. With this correction, the reflected signals from
tem.
straight reflectors (e.g., the edges of the plate) should be

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 06,2020 at 07:24:59 UTC from IEEE Xplore. Restrictions apply.
 wilcox et al.: use of an emat array for omnidirectional guided wave inspection
of equal amplitude in the absence of any other attenuation
mechanisms such as material damping, leakage, or scatter-
ing. However, the reflected signal amplitude of point reflec-
tors decreases with the distance of the reflector from the
array, due to secondary spreading of the reflected waves.
The theoretical dynamic range of the prototype array
is in excess of 40 dB and is ultimately limited by the ap-
pearance of spurious signals due to artefacts of the angular
deconvolution algorithm [1]. The reason why the dynamic
range achievable in practice is only 30 dB is thought to be
a result of variability in coupling between the elements in
the array due to the multilayered PCB structure and un-
evenness of the magnetic field over the area of the array.
The manifestation of the dynamic noise limit is the appear-
ance of coherent signals in the processed omnidirectional
B-scan image that do not correspond to the position of a
physical reflector. One source of these spurious signals is
the presence of the unwanted A0 mode. Due to its lower
velocity, A0 mode signals are most problematic when the
array is located close to a large reflector such as a plate
edge. Any A0 mode signals from that reflector that remain
after processing will be treated as S0 mode signals and will
hence appear at incorrect angular positions farther away
from the array. Hence they could be mistaken for a defect
reflection elsewhere in the plate. A second source of spu-
rious signals is due to S0 mode sidelobes in the angular
direction around genuine reflectors. Although the angu-
lar deconvolution algorithm should theoretically suppress
these to more than 40 dB below the genuine signal, vari-
ability in the construction of the array can make them
appear at higher levels. Sidelobe signals can also be mis-
taken for defect signals as they can appear in any angular
direction, not just directly adjacent to the genuine signal.
A further practical limitation of the device caused by its
physical construction is its inability to operate effectively
below about 150 kHz. Below this frequency, mechanical
resonances in the array cause a decaying reverberation that
obscures an increasingly large part of the near field in the
final omnidirectional B-scan.
B. Ten- and 20 mm Thick Steel Plates
It can be seen from Fig. 1(c) that above a frequency
thickness product of about 1.5 MHz mm, the attenuation
of the S0 mode rises sharply in the case when a plate is
liquid-loaded. Hence, as the plate thickness is increased,
the maximum operating frequency on the S0 mode de-
creases. Because the prototype array has a low frequency
resonance that precludes testing below 150 kHz, the great-
est thickness of plate that can be inspected with it is
10 mm.
Fig. 10(a) shows the results obtained when a
3 m × 1.5 m × 10 mm thick steel plate was tested at
150 kHz. The condition of the plate was generally good de-
spite some superficial corrosion. It can be seen that with a
30 dB dynamic range there is significant noise present, par-
ticularly at a radial distance of more than around 0.75 m
from the array. This is thought to be random (rather than
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 06,2020 at 07:24:59 UTC from IEEE Xplore. Restrictions apply.
661
coherent) noise, as the DAC used to generate the plot am-
plifies background noise more as the radial distance in-
creases. For this particular data set, a dynamic range of
25 dB is more appropriate, as shown in Fig. 10(b).
The usable frequency range of the S0 mode in a 20 mm
thick plate is below 75 kHz and this is too low for the
current prototype to be able to exploit. Even if the array
could perform below 75 kHz it is questionable whether the
radial resolution would be acceptable, since this is approx-
imately inversely proportional to the operating frequency.
An alternative strategy is to use a suitable higher-order
mode at a higher frequency, such as the S1 mode. This
mode also has a zero attenuation point for the case when
a plate is liquid-loaded and this occurs close to its peak
group velocity. In the case of a 20 mm thick steel plate this
point is at a frequency of approximately 220 kHz, which is
within the operating range of the current prototype. Re-
sults obtained using the S1 mode at 220 kHz to inspect
a 3 m × 1.5 m × 20 mm thick steel plate are shown in
Fig. 11(a). The signal-to-noise ratio in this case is only
around 20 dB, as shown by the results in Fig. 11(b). Al-
though the performance is poorer than in other cases, this
is an encouraging result that indicates the potential of us-
ing a higher order mode. It should be noted that the use
of a higher-order mode brings its own set of problems, in
particular the problem of suppressing signals from a larger
number of unwanted modes, which the current prototype
array and signal processing was not specifically designed
to do.
C. Six mm Thick Steel Plate Containing Simulated Defects
A 2 m × 1.5 m × 6 mm thick steel plate containing 44
documented artificial defects with a variety of morpholo-
gies was tested at a number of locations and using a variety
of test frequencies. The condition of the plate (apart from
the defects) was as-new, unpainted with a good surface
finish and square-cut edges. A typical result obtained at
a test frequency of 170 kHz is shown in Fig. 12(a) and
the detailed description of the defects is indicated on the
diagram shown in Fig. 12(b).
The result in Fig. 12(a) demonstrates a dynamic range
in excess of 30 dB, with the specular reflections from
the four edges of the plate sample clearly visible. The
plate sample was not specifically designed for testing with
guided waves and many of the defects are too small to be
visible and too close to be resolved. The key defects that
are visible are the three deepest flat-bottomed holes in the
upper right quadrant of the plate, the two deepest slots in
the upper left quadrant, and the larger-diameter holes in
the lower left quadrant. The type of feature that is most
representative of an isolated corrosion patch is arguably a
flat-bottomed hole, and the shallowest flat-bottomed hole
that is visible on the 30 dB grayscale is 3 mm (i.e., half the
plate thickness) deep and 20 mm (i.e., just over three times
the plate thickness) in diameter. This gives an indication
of the sensitivity of the EMAT array in its current form,
although the precise relationship between the amplitude
 662 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 52, no. 4, april 2005

Fig. 10. Results obtained at 150 kHz on a 10 mm thick steel plate presented with a dynamic range of (a) 30 dB and (b) 25 dB.

Fig. 11. Results obtained at 220 kHz on a 20 mm thick steel plate using the S1 mode presented with a dynamic range of (a) 30 dB and
(b) 20 dB.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 06,2020 at 07:24:59 UTC from IEEE Xplore. Restrictions apply.
 wilcox et al.: use of an emat array for omnidirectional guided wave inspection 663

Fig. 12. (a) Experimental data obtained at 170 kHz on a 6 mm thick steel plate containing 44 artificial defects, the locations of which
are indicated by open circles. The schematic diagram in (b) describes the geometry of the artificial defects. The defect dimensions are in
millimeters.

of the reflection and the defect shape, size, and distance is
the subject of ongoing research [25].
D. Generally Corroded Plate 8 mm Thick Steel Plate
With the exception of the artificial defects, the plate in
the previous example was in perfect condition. The EMAT
array has also been tested on a severely corroded steel
plate. The plate in question was nominally 8 mm thick
and had been roughly flame cut from the floor of an oil
storage tank. The results obtained from the EMAT array
at a center frequency of 170 kHz are shown in Fig. 13(a).
The thickness of the entire area of this plate has also been
measured using a semi-automated conventional ultrasonic
technique, and the results from these measurements are
shown in Fig. 13(b).
From the ultrasonic thickness measurements, it can be
seen that the plate is in very poor condition with many
deep corrosion pits extending through more than half of
the plate thickness. The EMAT array clearly cannot re-
solve each pit separately and at best can give only a very
approximate estimate of individual defect severity. On the
other hand, it does provide a rapid method of gaining an
indication of the general condition of the plate. It should
also be noted that the poor surface condition of the plate
does not prevent the EMAT array from being used to ob-
tain data.
IV. Conclusion
The feasibility of using a guided wave EMAT array for
rapidly inspecting large areas of steel plate structures has
been demonstrated. The current prototype can typically
achieve a dynamic range of around 30 dB. Improvements
to the build quality of the array and the use of a monolithic
permanent magnet to make the magnetic field more uni-
form should enable the dynamic range to be increased to
nearer the theoretical limit of 40 dB. It would also be de-
sirable to increase the damping within the array to reduce
low-frequency resonances, as this would enable the lower
limit of the operating frequency range to be reduced.
The dynamic range (and therefore the sensitivity) of the
array is determined by the ratio of the amplitude of the
genuine signal from the largest reflector to the amplitude
of the largest spurious signal due to that reflector. In the
case of free plates, the largest reflectors are the edges with
a reflection coefficient approaching unity. However, in the
case of a plate welded into a real structure, the absolute
size of the largest reflector is likely to be typically around
14 dB [26] smaller. Hence, although the dynamic range
relative to the largest reflector that can be achieved is the
same as in the free-plate case, the absolute sensitivity that
can be achieved on a welded plate will be 14 dB greater.
The presence of liquid loading on one or both sides of
the plate under test will cause slight attenuation of the
S0 mode. For example, at 150 kHz in a 5 mm thick steel
plate with water on one side, the S0 mode attenuation is
approximately 1 dB m−1 whereas the A0 mode attenua-
tion is approximately 57 dB m−1 . The S0 signal from a
single reflector 1 m away from the array will therefore ap-
pear 2 dB smaller than in the unloaded case. This will
have a negligible effect on the dynamic range that can be
achieved by the array since spurious S0 sidelobe signals due
to that reflector will be attenuated by the same amount

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 06,2020 at 07:24:59 UTC from IEEE Xplore. Restrictions apply.
 664 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 52, no. 4, april 2005
Fig. 13. (a) Experimental data obtained at 170 kHz on a generally corroded steel plate sample nominally 8 mm thick, and (b) map of
ultrasonic thickness measurements made over the entire plate.
and hence the signal-to-coherent noise ratio will remain
the constant. Even with multiple reflectors at different dis-
tances, the signal-to-noise ratio could be preserved by ap-
plying a suitable distance amplitude correction to the pro-
cessed B-scan image to compensate for the small level of
S0 attenuation. The presence of liquid loading may actu-
ally improve the signal-to-noise ratio that can be obtained
because any spurious signals caused by the A0 mode will
be suppressed much more than S0 mode signals due to the
higher attenuation of the A0 mode.
Acknowledgment
The 6 mm thick steel plate sample containing artificial
defects and the generally corroded plate sample were pro-
vided by Röntgen Technische Dienst in Rotterdam, The
Netherlands. The 10- and 20 mm thick steel plates were
provided by Shell Global Solutions International in Ams-
terdam, The Netherlands.
References
[1] P. D. Wilcox, “Omni-directional guided wave transducer arrays
for the rapid inspection of large areas of plate structures,” IEEE
Trans. Ultrason., Ferroelect., Freq. Contr., vol. 50, no. 6, pp.
699–709, 2003.
[2] D. E. Chimenti, “Guided waves in plates and their use in ma-
terials characterization,” Appl. Mech. Rev., vol. 50, no. 5, pp.
247–284, 1997.
[3] A. Safaeinili, O. I. Lobkis, and D. E. Chimenti, “Quantita-
tive materials characterization using air-coupled leaky Lamb
waves,” Ultrasonics, vol. 34, no. 2–5, pp. 393–396, 1996.
[4] R. B. Thompson, “Experiences in the use of guided ultrasonic
waves to scan structures,” in Review of Progress in Quantitative
NDE. vol. 16A, D. O. Thompson and D. E. Chimenti, Eds. New
York: Plenum Press, 1997, pp. 121–128.
[5] G. A. D. Briggs, Acoustic Microscopy. Oxford, UK: Clarendon
Press, 1992.
[6] D. N. Alleyne, B. Pavlakovic, M. J. S. Lowe, and P. Cawley,
“Rapid long-range inspection of chemical plant pipework using
guided waves,” Insight, vol. 43, no. 2, pp. 93–96 and 101, 2001.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 06,2020 at 07:24:59 UTC from IEEE Xplore. Restrictions apply.
[7] P. Wilcox, B. Pavlakovic, M. Evans, K. Vine, P. Cawley, M.
Lowe, and D. Alleyne, “Long range inspection of rail using
guided waves,” in Review of Progress in Quantitative Nonde-
structive Evaluation. vol. 22A, D. O. Thompson and D. E. Chi-
menti, Eds. New York: AIP Conference Proceedings, 2003, pp.
236–243.
[8] D. F. Ball and D. Shewring, “Some problems in the use of Lamb
waves for the inspection of cold-rolled steel sheet and coil,” Non-
destr. Test., vol. 6, no. 3, pp. 138–145, 1973.
[9] E. V. Malyarenko and M. K. Hinders, “Fan beam and double
crosshole Lamb wave tomography for mapping flaws in aging
aircraft structures,” J. Acoust. Soc. Amer., vol. 108, no. 4, pp.
1631–1639, 2000.
[10] H. J. Salzburger, G. Dobmann, and H. Mohrbacher, “Quality
control of laser welds of tailored blanks using guided waves and
EMATs,” IEE Proc. Sci. Meas. Technol., vol. 148, no. 4, pp.
143–148, 2001.
[11] B. C. Lee and W. J. Staszewski, “Modelling of Lamb waves for
damage detection in metallic structures: Part II. Wave interac-
tions with damage,” Smart Mater. Struct., vol. 12, no. 5, pp.
815–824, 2003.
[12] C. Valle and J. W. Littles, “Flaw localization using the re-
assigned spectrogram on laser generated and detected Lamb
modes,” Ultrasonics, vol. 39, no. 8, pp. 535–542, 2002.
[13] M. J. S. Lowe and O. Diligent, “Low-frequency reflection char-
acteristics of the S0 Lamb wave from a rectangular notch in a
plate,” J. Acoust. Soc. Amer., vol. 111, no. 1, pp. 64–74, 2002.
[14] I. A. Viktorov, Rayleigh and Lamb Waves: Physical Theory and
Applications. New York: Plenum, 1967.
[15] J. L. Rose, S. P. Pelts, and M. J. Quarry, “A comb transducer
model for guided wave NDE,” Ultrasonics, vol. 36, no. 1–5, pp.
163–169, 1998.
[16] E. Moulin, J. Assaad, C. Delebarre, S. Grondel, and D. Balageas,
“Modeling of integrated Lamb waves generation systems using
a coupling finite element-normal modes expansion method,” Ul-
trasonics, vol. 38, no. 1–8, pp. 522–526, 2000.
[17] P. D. Wilcox, M. J. S. Lowe, and P. Cawley, “Mode and trans-
ducer selection for long range Lamb wave inspection,” J. Intell.
Mater. Syst. Struct., vol. 12, no. 8, pp. 553–565, 2001.
[18] M. J. S. Lowe, P. Cawley, J. Y. Kao, and O. Diligent, “The low
frequency reflection characteristics of the fundamental antisym-
metric Lamb wave A0 from a rectangular notch in a plate,” J.
Acoust. Soc. Amer., vol. 112, no. 6, pp. 2612–2622, 2002.
[19] O. Diligent, T. Grahn, A. Bostrom, P. Cawley, and M. J. S.
Lowe, “The low-frequency reflection and scattering of the S-0
Lamb mode from a circular through-thickness hole in a plate:
Finite element, analytical and experimental studies,” J. Acoust.
Soc. Amer., vol. 112, no. 6, pp. 2589–2601, 2002.
[20] P. Wilcox, M. Lowe, and P. Cawley, “Lamb and SH wave trans-
ducer arrays for the inspection of large areas of thick plates,” in
Review of Progress in Quantitative Nondestructive Evaluation.
 wilcox et al.: use of an emat array for omnidirectional guided wave inspection
vol. 19B, D. O. Thompson and D. E. Chimenti, Eds. New York:
AIP Conference Proceedings, 2000, pp. 1049–1056.
[21] P. Fromme, P. Wilcox, and P. Cawley, “Remote monitoring of
plate-like structures using guided wave arrays,” in Review of
Progress in Quantitative Nondestructive Evaluation. vol. 22A,
D. O. Thompson and D. E. Chimenti, Eds. New York: AIP Con-
ference Proceedings, 2003, pp. 157–164.
[22] J. D. Achenbach and Y. Xu, “Wave motion in an isotropic elas-
tic layer generated by a time-harmonic point load of arbitrary
direction,” J. Acoust. Soc. Amer., vol. 106, no. 1, pp. 83–90,
1999.
[23] J. Krautkramer and H. Krautkramer, Ultrasonic Testing of Ma-
terials. 4th ed. Berlin, Germany: Springer-Verlag, 1990.
[24] B. D. Steinberg, Principles of Aperture and Array System De-
sign. New York: John Wiley and Sons, 1976.
[25] O. Diligent, M. J. S. Lowe, P. Cawley, and P. Wilcox, “Reflection
of the S0 Lamb mode from a part-depth circular defect in a
plate, when the incident wave is created by a small source,” in
Review of Progress in Quantitative Nondestructive Evaluation.
vol. 22A, D. O. Thompson and D. E. Chimenti, Eds. New York:
AIP Conference Proceedings, 2003, pp. 197–204.
[26] P. Cawley, M. J. S. Lowe, D. N. Alleyne, B. Pavlakovic, and P.
Wilcox, “Practical long range guided wave testing: Applications
to pipes and rail,” Mater. Eval., vol. 61, no. 1, pp. 66–74, 2003.
Paul D. Wilcox was born in Nottingham,
England, in 1971. He received an M.Eng. de-
gree in engineering science from the Univer-
sity of Oxford, Oxford, England, in 1994 and
a Ph.D. degree from Imperial College, Lon-
don, England, in 1998.
From 1998 to 2002, Dr. Wilcox was a re-
search associate in the nondestructive testing
research group at Imperial College where he
worked on the development of guided wave
array transducers for large area inspection.
Since 2000 he also has acted as a consultant
to Guided Ultrasonics Ltd., Nottingham, England a manufacturer of
guided wave test equipment.
Dr. Wilcox now works as a lecturer in dynamics at the University
of Bristol, Bristol, UK. His current research interests include long-
range guided wave inspection, array transducers, and signal process-
ing.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY MADRAS. Downloaded on April 06,2020 at 07:24:59 UTC from IEEE Xplore. Restrictions apply.
665
Michael J. S. Lowe was born in Belfast,
UK, in 1957. He received a B.Sc. degree in
Civil Engineering from the University of Ed-
inburgh, Edinburgh, Scotland, in 1979, and
M.Sc. and Ph.D. degrees in Mechanical Engi-
neering from Imperial College, University of
London, London, England, in 1987 and 1993,
respectively. He has been a chartered mem-
ber of the Institution of Mechanical Engineers
since 1986. He is also a member of the British
Institute of NDT, the Institute of Physics, and
the French Society of Acoustics.
From 1979 to 1989 he worked in engineering consultancy, special-
ising in the application and development of numerical methods for
the solution of problems in solid mechanics. His principal clients were
in the nuclear power and offshore oil industries.
Since 1989 he has worked in the Department of Mechanical Engi-
neering at Imperial College, University of London, and is currently a
Reader. His research interests are in the use of ultrasonic waves for
nondestructive evaluation (NDE). His specialist areas of expertise
are guided waves, the interaction of waves with defects and struc-
tural features, and numerical modeling.
Peter Cawley (M’93) was born in Sheffield,
England, in 1953. He received B.Sc. and Ph.D.
degrees in Mechanical Engineering at Univer-
ity of Bristol, Bristol, England, in 1975 and
1979, respectively.
He worked in industry from 1979–1981 and
then joined the Mechanical Engineering staff
at Imperial College, London, England, ini-
tially as a lecturer and then successively senior
lecturer, reader, and professor. He has worked
on a wide variety of projects using sonic and
ultrasonic methods applied to the inspection
of adhesive joints, composite materials, mass produced components,
aerospace structure, pipework, dental implants, and other structures.
He has published over 120 refereed journal papers and a similar num-
ber of conference papers in this field and holds 4 current patents.
Professor Cawley is a director of three spin-out companies set
up to exploit technology developed in his reasearch group (Guided
Ultrasonics, Ltd., Guided Ultrasonics (Rail) Ltd., and Integration
Diagnostics Ltd.), and he is a consultant to a variety of industries.