THE OPTIMIZATION OF LAMB AND RAYLEIGH WAVE
GENERATION USING WIDEBAND-LOW-FREQUENCY EMATS
S. Dixon, C. Edwards and S.B.Palmer
University of Warwick, Department of Physics, Coventry CV4 7AL, UK
ABSTRACT. - This paper describes a non-contact ElectroMagnetic Acoustic Transducer (EMAT)
that can be used to generate both Lamb and Rayleigh waves on metal samples. The generated waves
are wideband and low frequency with a dominant frequency content centred on approximately
200kHz extending to around 500kHz. Detection of the waves is achieved using a linear coil detection
EMAT. The transducers (generator & detector) have been used on both aluminium and steel, but
operate more efficiently on aluminium due to its lower electrical resistance and density when
compared to steel. Some considerations are described for the design of the generation EMAT
including applications where the dynamic field from the coil alone is used to obtain the Lorentz
interaction with the sample surface eddy current.
INTRODUCTION
The use of EMATs [1-4] permits non-contact surface or flexural wave measurements to
be made on metallic samples. There are alternative non-contact methods to the EMAT such
as laser or air-coupled ultrasonic generation sources but there are instances where it is
preferable to use an EMAT. Some advantages of using the EMATs described in this paper
are:•
Non-contact - doesn't load the sample or change the boundary conditions.
•
Good way to generate / detect in-plane ultrasonic displacements.
•
No couplant is required.
•
Operation on rough surfaces due to non-contact nature and relatively low frequency.
•
High temperature performance.
•
Wideband, low frequency signals yield good measurement resolution and propagation
distances.
The low frequency operation not only enhances EMAT efficiency but permits us to
make non-dispersive So mode Lamb wave measurements on samples up to a few
millimetres thick. The low frequency also tends to reduce Rayleigh wave attenuation and
increase propagation distance.
CP657, Review of Quantitative Nondestructive Evaluation Vol. 22, ed. by D. O. Thompson and D. E. Chimenti
© 2003 American Institute of Physics 0-7354-0117-9/03/S20.00
297
EXPERIMENTAL DETAILS
The approach that we used in these experiments was to generate a wideband, low
frequency wavemode using a current pulse of several microseconds duration and a peak
current value of approximately SOOamps [5]. The resultant ultrasonic modes have
wideband frequency content, centred at approximately 200kHz with significant frequency
content to 500kHz, dependent on coil impedance. The 'low' frequency of operation
enhances the generation efficiency [2,3] and the wideband nature means that the resultant
wavemode is relatively temporally sharp when compared with a toneburst signal of a
similar frequency. Previously we have shown that the zero order symmetric Lamb wave
mode out-of-plane displacement is similar in shape (and thus frequency content) to the
generation current pulse. At 200kHz the electromagnetic skindepth [1,2] in aluminium is
approximately 0.18mm so that the generation source is effectively confined within this
depth.
The set up used for EMAT generation of Lamb waves generated both the So and the AQ
modes in thin sheets and is shown below in figure 1. We have used this technique [5] to
measure the velocity variation of the So mode in metal sheets for texture determination.
The set up used for EMAT generation of Rayleigh waves is shown schematically in figure
2. The crack could be detected either positioned between the EMATs (a) or in a reflectiontype geometry (b) and this is demonstrated in the results. In figures 1&2 the generation coil
shown is a spiral pancake coil but linear type coils have also been used.
generation EMAT
(spiral coil)
detection EMAT
(linear coil)
aluminium sheet
FIGURE 1. Schematic diagram of Lamb wave generation and detection on thin metal sheet.
generation EMAT
(spiral coil)
detection EMAT
(linear coil)
crack positions
aluminium bar
FIGURE 2. Schematic diagram of Rayleigh wave generation and detection on thin metal sheet.
298
Measurements were taken using a 'spiral pancake' type coil with no permanent
magnetic field and with the permanent magnetic field directed normal to and into the
sample surface and normal to and out of the sample surface. This was done on a 0.25mm
thick aluminium sheet (for Lamb wave measurements) and then on a 63.5mm thick
aluminium bar (for Rayleigh wave measurements).
Measurements were then taken on a sectioned aluminium billet with an as-cast rough
surface that contained a partially closed surface crack. The crack was positioned between
the generator and detector (location 'a') and with the crack disposed to the side of the
detector not between generator or detector (location 'b') as indicated on figure 2. A linear
coil EMAT was used to generate the Rayleigh wave in this case.
Finally an optimised spiral pancake coil EMAT was used to generate Rayleigh waves
on a steel railtrack head that contained a thermic weld in the middle of the 1m long sample.
The generator was positioned close to one end of the track section and measurements were
recorded as the detector was moved away from the generator along the head surface.
RESULTS
Lamb Waves on 250um Thick Aluminium Sheet
Using the spiral pancake coil, measurements were taken on a 0.25mm thick aluminium
sheet. The waveforms shown below in figure 3 are obtained with the permanent magnetic
field (BSN) into and normal to the surface of the sheet, out of and normal to the surface of
the sheet and with no permanent field present. In all cases the static field magnitude was
0.35T at the aluminium surface, where a static magnetic field was applied.
So
reflections
BSN out of
sheet
_Q
&
CD
BSN into
sheet
_
05
no static
field
50
100
150
200
time (JLIS)
FIGURE 3. Lamb waves detected on 250|im thick aluminium sheet with the permanent magnet in
different orientaions.
299
Note that the signal at approximately 5jis on the waveforms of figure 3 is electrical
Note that the signal at approximately 5µs on the waveforms of figure 3 is electrical
noise from the generation current pulse. The first ultrasonic arrival occurs at approximately
noise from the generation current pulse. The first ultrasonic arrival occurs at approximately
SOjis and is the zero order symmetric Lamb wave (So). The zero order antisymmetric Lamb
30µs and is the zero order symmetric Lamb wave (S0). The zero order antisymmetric Lamb
wave (Ao) starts to arrive at around lOOjis. The polarity
of the So mode is the same for the
wave (A0) starts to arrive at around 100µs. The polarity of the S0 mode is the same for the
case
of
no
applied
static
magnetic
field
(BSN)
and
with
BSN
into the sheet, opposite to the
case of no applied static magnetic field (BSN) and with BSN into the sheet, opposite to the
polarity
with
BSN
out
of
the
sheet.
The
largest
amplitude
So
mode
observedwith
withBBSN
into
polarity with BSN out of the sheet. The largest amplitude S0 mode isisobserved
SN into
the
sheet.
The
amplitude
of
the
AO
mode
is
hardly
effected
by
the
application
of
the
static
the sheet. The amplitude of the A0 mode is hardly effected by the application of the static
magnetic
field, apart from interference with the AO mode at approximately 135jis.
magnetic field, apart from interference with the A mode at approximately 135µs.
0
Rayleigh
on 63mm
63mm Thick
Thick Aluminium
Aluminium Bar
Bar
Rayleigh Waves
Waves on
The
waveforms shown
shown below
below in
in figure
figure 44 were
wereobtained
obtainedwith
withthe
thepermanent
permanent
The Rayleigh
Rayleigh waveforms
magnetic
field
(BSN)
out
of
and
normal
to
the
surface
of
the
bar,
with
no
permanent
field
magnetic field (BSN) out of and normal to the surface of the bar, with no permanent field
present
and
with
BSN
into
and
normal
to
the
surface
of
the
bar.
Note
that
in
this
case
the
present and with BSN into and normal to the surface of the bar. Note that in this case the
largest
amplitude
wave
is
obtained
with
BSN
out
of
the
bar,
which
is
the
opposite
largest amplitude wave is obtained with BSN out of the bar, which is the opposite
orientation that yielded
yielded the
the largest
largest SSo
Lamb wave mode. The Rayleigh wave with the
0 Lamb wave mode. The Rayleigh wave with the
smallest amplitude was obtained
obtained with
with BBSN
directed into
into the
the bar,
bar, all
all other
other conditions
conditions
SN directed
remaining
The shape
shape and
and polarity
polarity ofof each
each ofof the
the Rayleigh
Rayleigh waves
waves isis
remaining the same. The
approximately the same, the only
only significant
significant difference
difference being
beingtheir
theirrelative
relativeamplitudes.
amplitudes.
amplitude (arb.)
0.60.6
(U
•D
_
CO
BBSN
outofofbar
bar
SN out
no
no static
static field
field
0.40.4
BBSN
intobar
bar
SN into
0.2
0.20.0
0.0-0.2
-0.2-0.4
-0.44.0
4.0
4.4
4.4
4.8
4.8
time
time (µs)
(jis)
5.2
5.2
5.6x10
5.6
-3
FIGURE 4. Rayleigh waves detected on a 63.5mm thick aluminium block with different static magnetic
FIGURE
4. Rayleigh waves detected on a 63.5mm thick aluminium block with different static magnetic
field orientations.
field orientations.
300
Rayleigh Waves on a Rough Surface Aluminium Billet
Figure 5 shows the 'single-shot' waveform {a} obtained on the rough surface of the
billet with a real partially closed crack positioned midway between the linear coil generator
and detector. There is some transmission through the partially closed crack of around 25%
of the signal amplitude that was detected on a defect free region. Assuming that energy is
proportional to the square of amplitude this gives a figure of 6% of Rayleigh wave energy
transmitted through the crack. Waveform {b} of figure 5 was obtained over a defect free
region with the detection EMAT 30mm from the crack that reflected a Rayleigh wave back
towards the detector (observed at 74jis in figure 5).
Ravleigh Waves on Steel Railtrack Head
Figure 6 shows the Rayleigh wave detected on a length of railtrack head, generated by
the spiral coil EMAT with a static magnetic field applied out-of and normal to the plane of
the top of the rail to provide enhanced generation compared to no applied field. The
distance of propagation in this case was 400mm. The wave observed on the steel railhead
is not as wideband as that observed on the aluminium samples. This may be due to
magneto-elastic effects or the presence of a work-hardened surface layer on the rail head.
o.o
3.0-
/ through crack
1
(a)
*Hvv*^^vHhvvw^|f^fi^^
2.5-
03
direct
\
2.0- Rayleigh wave\ |
|
1.5i
1.0i
0
20
40
reflected fronr
/ crack
(b)
60
80
100
time i
FIGURE 5. Rayleigh waves detected on an aluminium billet containing a crack
2
o>
o.4H
0.0-
Q.
-0.4-
0
20
40
60
time
FIGURE 6. Rayleigh waves detected on a steel railtrack head.
301
80
100
DISCUSSION
DISCUSSION AND
AND CONCLUSIONS
CONCLUSIONS
The
that
itit is
is that
that itit is
is possible
possible to
to generate
generate various
various
The first
first result
result
thatCONCLUSIONS
is important
important to
to note
note is
DISCUSSION
AND
ultrasonic
eddy current
current generated
generated in
in the
the sample
sample surface
surface
ultrasonic modes
modes via
via an
an interaction
interaction of
of the
the eddy
with
the
dynamic
magnetic
field
from
the
generation
coil.
This
interaction
is
usually
too
first resultmagnetic
that it isfield
important
to generate
various
with The
the dynamic
from to note is that it is possible
interaction
is usually
too
small
to
observe
with
conventional
EMAT
drive
currents
which
are
much
smaller
in
ultrasonic
modes
via
an
interaction
of
the
eddy
current
generated
in
the
sample
surface
small to observe with conventional
currents which are much smaller in
amplitude
than
the
one
used
here.
We
need
to
consider
the
various
forces
that
are
generated
with
the
dynamic
magnetic
field
from
the
generation
coil.
This
interaction
is
usually
too
amplitude than the one used here.
various forces that are generated
to observe
with conventional
EMATthe
drive
currents which are much smaller in
ininsmall
the
samples
to
the aluminium
aluminium
samples
to begin
begin to explain
results.
amplitude
than thediagram
one usedofhere.
We7need
to consider
the various forces
that
areeddy
generated
The
figure
cross-sectional
view of
of
the
eddy
current
The schematic
schematic
shows
the cross-sectional
view
the
current
in the aluminium
to begin
to explain
theno
results.
generated
in
aluminium
skindepth
with
permanent static
static magnetic
magnetic field.
field. The
The
generated
in the
the samples
aluminium
permanent
The schematic
diagram
of figureinteraction
7 shows the
view of
the eddy
current
tangential
forces
to
of cross-sectional
the eddy (or mirror)
mirror)
current
are shown
shownas
as
tangential
forces due
due
to the
the Lorentz
Lorentz
current
are
generated
in
the
aluminium
skindepth
with
no
permanent
static
magnetic
field.
The
FDT
and
FDT
and
arise
due
the
fringing
normal
dynamic
magnetic
field
components
from
FDT and FDT and
magnetic field components from
tangential
to positions.
the
LorentzThe
interaction
of the eddy
(or mirror)
are shown
as
the
external
coil
these
magnetic
fieldcurrent
radial
tangential
the
externalforces
coil at
atdue
these
positions.
largest dynamic
magnetic
field
isis aa radial
tangential
F
and
F
and
arise
due
the
fringing
normal
dynamic
magnetic
field
components
from
DT
DT
in-plane
interaction of
of this
this in-plane
in-plane dynamic
dynamic
in-plane field
field from
from the
the generation
generation coil. The Lorentz interaction
the external coil
at these
positions. The
largest dynamic magnetic
fieldinto
is a the
radial
tangential
magnetic
normal and
and
surface
of the
the
magnetic field
field and
and the
the eddy
eddy current
current generates a force normal
into the
surface
of
in-plane field from the generation coil. The Lorentz interaction of this in-plane dynamic
metal.
metal.
magnetic field and the eddy current generates a force normal and into the surface of the
The
cross-sectional view
view of
of the
the eddy
eddy current
current
The schematic
schematic diagram
diagram of figure 8 shows the cross-sectional
metal.
generated
in
the
aluminium
skindepth
with
a
permanent
applied
normal
and
into
generated
in
the
aluminium
static
field
applied
normal
and
into
The schematic diagram of figure 8 shows the cross-sectional view of the eddy current
the metal
metalinsurface.
surface.
Interactions
fromfield
the applied
coil or normal
in-plane
fringing
totogenerated
the
Interactions
with with
the dynamic
fieldstatic
in-planeand
fringing
the aluminium
skindepth
a permanent
into
fields
from
thesurface.
static field
field
are not
notwith
in figure
Thethe
tangential
(inplane)
radially
fields
from
the
static
are
considered
tangential
(inplane)
radially
to the
metal
Interactions
the dynamic
field8.from
coil or in-plane
fringing
polarised
forces
due
to
the
interaction
of
the
eddy
(or
mirror)
current
are
shown
as
polarised
forces
due
to
the
interaction
of
the
eddy
(or
mirror)
current
are
shown
as FFST.
ST.
fields from the static field are not considered in figure 8. The tangential (inplane) radially
This
shear force
force arises
arises
due
the
Lorentzofinteraction
interaction
of
normal
to
This
shear
Lorentz
of the
the
normal
to and
and
into-plane
static
polarised
forces
due todue
the the
interaction
the eddy (or
mirror)
current
are into-plane
shown as Fstatic
ST.
magnetic
field
witharises
the eddy
eddy
(or
mirror)
magnetic
field
with
the
current.
This shear
force
due (or
the mirror)
Lorentzcurrent.
interaction of the normal to and into-plane static
magnetic field with the eddy (or mirror) current.
BSN
BSN
BSN
B
SN
BSN
BSN
uuuuuuuu
sample surface
surface
sample
sample surface
FST
FST
FST
eddy
eddy (or
(or mirror)
mirror)
eddy
(or
current
current mirror)
current
FST
FST
FIGURE 7.
7. In-plane
In-plane tangential
tangential force
force (FST)
(FST) generated
FIGURE
generated via
via the
the Lorentz
Lorentz interaction
interaction of
of the
the eddy
eddy current
current with
with
FIGURE
7. In-plane
tangential force (FST) generated via the Lorentz interaction of the eddy current with
the
static
magnetic
field.
the static magnetic field.
the static magnetic field.
sample
sample
sample
surface
surface
surface
BDN
BDN
FFDN
DN
BDN
BDN
DT
BBDT
B
BDT
DT
FFDT
DT
FDT
eddy
(or
mirror)
eddy(or
(ormirror)
mirror)
eddy
current
current
current
BDN
BDN
FDT
F
DT
DN
FFDN
DT
FFFDT
DT
FIGURE 8. Out-of plane normal force (F ) and in-plane tangential forces (F
& F’ ) generated via
DT& F’DTDT
FIGURE8.8.Out-of
Out-ofplane
planenormal
normal force
force (F
(FDN
) generated via
DN) and in-plane tangential forces (FDT
FIGURE
DN) and in-plane tangential forces (FDT & F'DT) generated via
the Lorentz
interaction
of the
eddy
current
with
the coil’s
coil’s dynamic
dynamicmagnetic
magnetic field.
Lorentz
interaction
eddycurrent
currentwith
withthe
the
thethe
Lorentz
interaction
ofof
thetheeddy
coil's dynamic
magnetic field.
field.
302
Firstly consider the Lamb wave generation mechanisms, where a symmetric mode is
only generated by a symmetric force and an anti-symmetric mode can only be generated by
an anti-symmetric force. The dominant force for the Lorentz interaction of the eddy current
with the self-field of the coil is normal to and into the plane of the sheet. The
electromagnetic skindepth of aluminium at 200kHz is approximately 0.2mm, so FDN
extends through the sheet thickness with exponentially decreasing amplitude and varying
phase. Any force can be broken down into a purely symmetric and a purely anti-symmetric
part. Thus FDN is capable of generating the So mode or the AO mode and it is observed to do
so (figure 3). The particle displacement associated with the AO mode is predominantly
normal to the plane, in the same direction that FDN acts. The particle displacement
associated with the So mode is predominantly in-plane and so whilst FDN can generate both
modes it generates the AO mode more efficiently than the So mode as the driving force and
particle displacement are in the same direction for the AO mode. The same argument
applies to the case of the in-plane tangential force (FST) generated by the Lorentz
interaction of the eddy current with the static magnetic field. In that case the SO mode
dominant particle displacement and the driving force FST act in the same direction and
thus FST predominantly generates the So mode with very little contribution to the AO mode.
The explanation for the enhancement of the Rayleigh wave using an applied magnetic
field is not straightforward. For Rayleigh wave generation there are no components of
force that are symmetric throughout the thickness of the sample - a Rayleigh wave would
not be generated if this was the case. The Rayleigh wave has both in plane and out-of plane
displacements when it propagates along a surface. The spatial profile of the in-plane and
out-of-plane forces are complex and either force is capable of generating a Rayleigh wave
mode with significantly different characteristics. The superposition of each of these
spatially and time dependant forces would yield the resultant Rayleigh wave. In this paper
we are limited to stating that the simplified forces shown in figure 9 appear to generate
larger amplitude Rayleigh waves than those shown in figure 10.
FN
FIGURE 9. In-plane tangential force (FT) and
out-of plane normal force (FN) that generated the
largest amplitude Rayleigh wave.
FN
FIGURE 10. In-plane tangential force (FT) and
out-of plane normal force (FN) that generated the
smallest amplitude Rayleigh wave.
ACKNOWLEDGEMENTS
We would like to acknowledge the EPSRC (UK) for funding this work through an
Advanced Fellowship.
REFERENCES
1.
2.
3.
4.
5.
Frost H.M., Electromagnetic ultrasonic transducer: principles, practice and
applications, m:Physteal Acoustics XIV, ed. W P Mason and R N Thurston (New
York Academic), 1985,ppl79-275
Dobbs E.R., Electromagnetic generation of ultrasonic waves, in: in:Physical
Acoustics X, op. cit, 1980, pp!27-189
Kawashima K., Experiments with two types of electromagnetic acoustic transducers,
J. Acous. Soc. Am, 60, pp365-373, (1976)
Maxfield B W and Fortunko C M, The design and use of electromagnetic-acoustic
wave transducers (EMATs) Mater. Eval 41 1399-1408, (1983)
Dixon S, Edwards C and Palmer S.B, Texture measurements of metal sheets using
wideband Electromagnetic Acoustic Transducers (EMATs) J. Phys D:Appl Phys.,
35 pp816-824, (2002)
304