Anda di halaman 1dari 8

Home Search Collections Journals About Contact us My IOPscience

Mode Conversion and Total Reflection of Torsional Waves for Pipe Inspection

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2013 Jpn. J. Appl. Phys. 52 07HC14


View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address:
This content was downloaded on 12/04/2016 at 14:07

Please note that terms and conditions apply.

Japanese Journal of Applied Physics 52 (2013) 07HC14 REGULAR PAPER

Mode Conversion and Total Reflection of Torsional Waves for Pipe Inspection
Nurmalia , Nobutomo Nakamura, Hirotsugu Ogi, and Masahiko Hirao
Graduate School of Engineering Science, Osaka University, Toyonaka, Osaka 560-8531, Japan
Received November 22, 2012; accepted March 21, 2013; published online July 22, 2013

The group velocity of each higher torsional mode in a pipe depends on the wall thickness and it will convert to a lower mode when the thickness is
smaller than a critical value, the so-called cut-off thickness. The fundamental torsional mode, T0; 1, and the first higher mode, T0; 2, are
generated in pipes by an electromagnetic acoustic transducer (EMAT) and their mode conversion behaviors are investigated by changing the
shape of a defect on the pipe. It is confirmed that the conversion behavior is sensitive to the shape of thickness transition and the total reflection of
the T0; 2 mode occurs in a pipe with a tapered defect. On the basis of the mode conversion, defects with a cross-sectional loss larger than 35%
are detectable. # 2013 The Japan Society of Applied Physics

objects. No preparation of specimen surface before inspec-

1. Introduction tion is required. Therefore, the use of an EMAT reduces
The extensive usage of pipe structures in many high-risk the total inspection time. An EMAT was used to measure
industrial applications encourages continuous studies on the resonant frequencies of a torsional wave in one of the
pipe inspection techniques. Many terrible disasters caused pioneer works,30) and some recent works have used an
by pipe ruptures have occurred leading to serious economic EMAT for pipe inspection.31,32) In the present study, a
disadvantages, environmental impact, and even the loss of particular conguration of an EMAT called a periodic
human life.1,2) One of the causes of such rupture is wall permanent magnet (PPM) is used for torsional wave
thickness reduction,3,4) and wall-thinning inspection is an generation.33,34) This EMAT is characterized by the ability
important research subject. to generate an acoustic wave with a xed wavelength, which
The currently available methods for this purpose include is determined by the periodicity of the magnet arrangement.
the magnetic ux leakage,57) eddy current810) and ultra- Taking advantage of the simple dispersion characteristic,
sonic. Among these inspection methods, the ultrasonic we have proposed a novel method for wall-thinning
method oers the ability to inspect the whole-wall thickness, inspection based on the group velocity change of the
not only the surface or near surface as in the case of other torsional wave induced by mode conversion. Group velocity
methods. In particular, ultrasonic nondestructive testing measurement is less dependent on the strength of the
(NDT) with guided waves1117) has been attractive for the received signal compared with the common amplitude
following reasons. The wave propagation follows the shape based method. We particularly focus on the group velocity
of a waveguide providing the capability to detect a hidden change induced by mode conversion. In each higher mode,
defect, which is not possible with conventional ultrasonic there is a certain thickness, the so called cut-o thickness,
testing. Also, a large distance can be inspected with a single below which that mode cannot exist. This mode is then
measurement. Among guided waves, the torsional wave expected to convert into lower mode(s) with a higher group
oers more benets because the component of particle velocity. As a result, the dependence of group velocity on
displacement is only in the circumferential direction. This thickness can be a basis for wall-thinning inspection. This
feature makes torsional modes insensitive to liquid loading method has been successfully applied to the case of a shear
and insulation material, which often covers a pipe in many horizontal (SH) guided wave propagating in a plate speci-
applications. The modes consequently propagate over a men.35,36) This study is then carried out to extend the
longer distance. Furthermore, this wave has a simple application of our method to the measurement of torsional
dispersion characteristic leading to easier interpretation. waves in a pipe.
Many studies on torsional waves have been carried out. In the previous work,37) pitch-catch measurement using
The nite element method was used to predict the reection two EMATs was carried out to measure the traveling time of
of the fundamental torsional mode from various de- torsional modes propagating in pipes with a thinning region.
fects.1820) Some studies on scattering were also re- It was found that the traveling time of a higher torsional
ported.2123) However, most reported studies were mainly mode changed depending on the depth of the thinning
based on analytical and numerical approaches. This is and a result indicating the mode conversion was obtained.
because the torsional wave is not as easy to generate as On the basis of this nding, further experimentation is
the other guided waves, although a piezoelectric array24,25) performed to investigate the mechanism of mode conver-
and a magnetostrictive transducer2628) were used. In the sion, and the inuence of the defect shape on mode
present study, we use an electromagnetic acoustic transducer conversion is examined. Experimentation to detect circum-
(EMAT). An EMAT29) is a totally noncontact transducer. ferential defects is carried out as a basic feasibility test of the
It generates acoustic waves directly inside the inspected proposed method.
material and the generation mechanism does not involve a
couplant or any mechanical bonding. This couplant-free 2. Torsional Waves Generated by PPM-EMAT
nature makes an EMAT applicable to inspection in harsh A torsional wave is a type of guided wave that exists in a
environments such as in high-temperature and fast-moving rod-shaped or cylindrical waveguide. It is characterized by
07HC14-1 # 2013 The Japan Society of Applied Physics
Jpn. J. Appl. Phys. 52 (2013) 07HC14 Nurmalia et al.

2.5 =5.22 mm
Lorentz force current
Frequency (MHz)

1.5 T(0,3)
1 T(0,1)
0.5 magnets

0 1 2 3 4 Fig. 2. Arrangement of four PPM-EMATs placed around a pipe and
Wave number (1/mm) schematic illustration of the involved elds directions. Pipe thickness is
(a) exaggerated in the left gure.

higher mode is highly inuenced by the frequency  thick-

Group velocity (mm/s)

3 T(0,1)
ness and this eect is even more pronounced in the lower
T(0,2) frequency range. When the thickness reaches the cut-o
2 tc of T(0,2) T(0,3) thickness tc , the group velocity is zero and this mode is
expected to convert into lower mode(s). Thus, when the
tc of T(0,3) frequency is xed, the group velocity shows a dependence
on the thickness of the waveguide.
An EMAT basically has two components: a coil, which
0 provides a dynamic magnetic eld, and permanent magnet(s)
0 1 2 3 4 5 6 7 or an electromagnet as a static eld source. In aluminum and
Frequency Thickness (MHz mm) other nonferromagnetic metals, an EMAT generates elastic
(b) waves by way of the Lorentz force mechanism. When an
alternating current drives the coil, which is located close to
Fig. 1. Dispersion curves for frequency-wave number (a) and group the specimen, an eddy current is generated on the surface of
velocity-frequency  thickness (b) of torsional wave propagating in
the specimen. The interaction of this dynamic eld with the
aluminum pipe with inner/outer diameters of 19/25 mm.
static eld induces the Lorentz force, which creates acoustic
waves. The relative position between these two elds
determines the type of acoustic waves to be generated (and
having a vibration only in the circumferential direction while detected).
the wave propagates in the axial direction. As a convention, We built a transducer consisting of four PPM-EMATs
innite numbers of modes of torsional waves are designated to generate a specic mode of torsional waves. In the PPM-
as T0; m, where m denes the radial/thickness mode EMAT, dynamic and static elds are provided by an
parameter. Equation (1) presents the dispersion equation of elongated spiral coil and permanent magnets, respectively.
a torsional wave propagating in a pipe with a and b as the The permanent magnets lie atop of the coil to form a
inner and outer radii, respectively. perpendicular conguration of the dynamic and static elds.
    The permanent magnets are arranged in such a way to
J1 r 0 Y1 r 0
A1  J1 r A2  Y1 r accommodate a periodically alternating magnetization in the
r ra;b r ra;b direction normal to the specimens surface. The periodicity
0 1 is equal to the wavelength of the generated waves in this
0 0
Here, J1 , Y1 , J1 , and Y1 are Bessel functions of the rst experimental case,  5:22 mm. Four PPM-EMATs were
and second kinds and their derivatives, respectively. With arranged to form a meandering pattern and placed around the
f as the frequency,  as the wavelength, cs as the shear pipe specimen. One EMAT consists of a wound coil and two
wave velocity (3.1 mm/s for aluminum), ! 2f , and pairs of magnet arrangements with opposite magnetizations.
 2=,  is dened as One set of magnet arrangements provides a static magnetic
eld to two neighboring coils. There are seven magnets in
   2: 2 each arrangement such that the total number of magnets is
c2s twenty eight. The four PPM-EMATs placed around the pipe
The aluminum pipes used as specimens in this study have and the details of their arrangements, showing the direction
the inner and outer radii of 9.5 and 12.5 mm, respectively. of each eld, are presented in Fig. 2. The T0; 1 and T0; 2
Figure 1 shows the dispersion curves derived from Eq. (1) modes are excited by this EMAT conguration, whose
for the current specimen. The broken line indicates the xed displacement is uniform in the circumferential direction and
wavenumber used in this study (1.2 mm1 ). Group velocity periodic in the axis direction, they are generated at the
is dened as cg d!=d and its corresponding dispersion respective frequencies of 0.615 and 0.77 MHz.
curves are also presented in Fig. 1(b). The group velocity of
the fundamental T0; 1 mode is constant in the entire range 3. Experimental Setup
of frequency  thickness. However, the velocity of each A pipe made of aluminum alloy A5052 is chosen as the
07HC14-2 # 2013 The Japan Society of Applied Physics
Jpn. J. Appl. Phys. 52 (2013) 07HC14 Nurmalia et al.

0.3 T(0,1)

Amplitude (arb. unit)


0.3 T(0,2)

Fig. 3. Typical experimental setup with EMAT placement around the

0 50 100 150 200 250
Time (s)
specimen for inspection. The pipe is 1000 mm in length with
inner and outer diameters of 19 and 25 mm, respectively Fig. 4. Typical received waveforms of two modes propagating in a
(3 mm wall thickness). Twelve specimens are prepared, specimen with a steplike defect.
including the original defect-free pipe as a reference.
Thinning regions of dierent shapes and sizes are introduced
at the center of each pipe specimen as per the objective of region between two EMATs.37) To understand the detailed
each experimental investigation. This region lies 200 mm mechanism of mode conversion, a further experiment is
along the axial direction and has a remaining thickness of performed by preparing a pipe with a region of wall thinning
1 mm in all specimens. In ve specimens, the thinning of 1 mm remaining thickness. This thickness is smaller than
regions cover the whole circumference and dier by the the cut-o thickness of the T0; 2 mode at a frequency of
sloping degree of the defect edges. In addition to the 0.77 MHz. Furthermore, the mode conversion behavior in
specimen with 90 slope of the defect edge (steplike edge), dierent shapes of defect edges is also investigated. The
four specimens with dierent diametrical transitions are T0; 1 and T0; 2 modes are generated and the receiver is
prepared: 30, 20, 10, and 5 slopes (from horizontal); that moved in increments of 5 mm along the 400 mm length of
is, the specimen with the 5 sloping edge has the longest the pipe to detect and record the propagating waves. The
tapered area. The other six specimens contain partially measurement site includes the defect-free regions before and
circumferential defects; that is, the defects do not cover the after the thinning region and the defect itself.
entire circumference. The defect in each specimen has a Figures 5 and 6 are assortments constructed from the
dierent length in the circumferential direction but the same received waveforms, similar to those shown in Fig. 4,
length in the axial direction (200 mm). obtained at many scanning points of each mode generation
Figure 3 shows the experimental setup. The EMAT array in dierent specimens. The defects in all specimens cover
is used to generate the waves and is driven by tone bursts of the whole pipe circumference with a remaining thickness
eight cycles with a specic frequency via the diplexer, which of 1 mm. The edges of the defect in each specimen have
ensures a unidirectional signal. This driving frequency dierent sloping degrees. From top to bottom, the gures are
determines the intended torsional mode to be generated in specimens with edges of 90 (steplike edge), 30, 20, 10, and
the specimen. The propagating mode is picked up by a point- 5 inclinations. The shape of the edge is illustrated on the
contacting transducer. Figure 4 shows the received wave- top of each gure for easier understanding. The defect is
forms displayed by the oscilloscope for driving frequencies conned between the dashed lines. These gures relate the
of 0.615 and 0.77 MHz, which are used to generate the distance from the transmitting EMAT and the propagating
T0; 1 and T0; 2 modes, respectively. The wave is time such that the slope corresponds to the inverse of the
generated in the specimen with a steplike defect and group velocity. The intended signal being focused is the
detected at 100 mm from the EMAT. The direct signals direct (the rst arrival) signal. The traveling time of this
are focused and indicated by oval marks in these gures. The signal at each location is deduced by gating it out from the
traveling time is obtained by gating out this signal and received signals38) in Figs. 5 and 6. The results are shown in
sending it to the heterodyne-spectrometer after being pre Figs. 7(a) and 7(b) for the T0; 1 and T0; 2 modes,
amplied. All the experimental parameters are set by a respectively, in the case of a steplike defect.
personal computer. The T0; 1 mode is excited at a frequency of 0.615 MHz.
Figure 5 reveals that this mode is less reected when the
4. Results and Discussion thickness transition at defects edge is smoother. The
4.1 Mechanism of mode conversion traveling time of the direct signal in every distance in the
Our previous investigation experimentally conrmed the specimen with a steplike edge [Fig. 5(a)] is investigated to
conversion of the T0; 2 mode into the T0; 1 mode by obtain the group velocity. The group velocity in the thinning
measuring the traveling time in pipes with a wall-thinning region (1 mm thick) is 3.11 mm/s, which is almost the
07HC14-3 # 2013 The Japan Society of Applied Physics
Jpn. J. Appl. Phys. 52 (2013) 07HC14 Nurmalia et al.

(a) (a)
250 250

150 150

50 50

(b) (b)
250 250

150 150

50 50

(c) (c)
250 250

Time (s)
Time (s)

150 150

50 50

(d) (d)
250 250

150 150

50 50

(e) (e)
250 250

150 150

50 50

100 200 300 400 500 100 200 300 400 500
Distance (mm) Distance (mm)

Fig. 5. Wave propagation of T0; 1 mode generated in specimens with Fig. 6. Wave propagation of T0; 2 mode generated in the same pipes
edge angles of (a) 90 , (b) 30 , (c) 20 , (d) 10 , and (e) 5 . Edges of the used to obtain Fig. 5. Arrows show the total reection in (d) and (e).
thinning region are indicated by broken lines.

same as the velocity of 3.17 mm/s for the original thickness explained by the displacement redistribution as in the case
(3 mm), as shown in Fig. 7(a). The theoretical velocity of of SH guided waves.35) This conversion into a higher mode
this fundamental mode is equal to the shear wave velocity, is not observed in the case of specimens with a smoother
which is 3.1 mm/s for aluminum. This observation thickness change at the defects edge. Comparison between
indicates that the T0; 1 mode continues to travel through this experimental result and the theoretical dispersion
this region. However, when the thickness returns to the curves is presented in Fig. 8(a). The frequency is xed at
original at the far end, the observed traveling time is 0.615 MHz, which is the same as that in the experiment, such
1.8 mm/s, which is smaller than that before the defect. In that the horizontal axis is the thickness of the specimen. The
Fig. 5(a), it appears that most of the T0; 1 mode converts numbered arrows indicate the sequence of change in the
into the T0; 2 mode with a lower group velocity, while a measured group velocity in relation to the change in
small portion persists as the T0; 1 mode. The calculated thickness.
group velocity of the T0; 2 mode at frequency 0.615 MHz The T0; 2 mode is generated at a frequency of
is 1.7 mm/s. The mechanism of this conversion can be 0.77 MHz. The group velocity of the direct signal in the
07HC14-4 # 2013 The Japan Society of Applied Physics
Jpn. J. Appl. Phys. 52 (2013) 07HC14 Nurmalia et al.

200 4

Group velocity (mm/s)

(i) T(0,1)
150 (ii) T(0,2)

1.8 mm/ s 2
100 3.17 mm/ s
Traveling time (s)

3.11 mm/ s (a) frequency=0.615 MHz

200 0 2 4 6 8
Thickness (mm)

150 4

Group velocity (mm/s)

2.06 mm/ s 2.49 mm/ s 3 (ii)
3.11 mm/ s (i)

100 200 300 400 500 1

(b) frequency=0.77 MHz
Distance (mm)
0 2 4 6 8
Fig. 7. Traveling time for (a) T0; 1 mode and (b) T0; 2 mode
Thickness (mm)
generation. The slope of time versus distance gives the group velocity. The
thinning region is located between the broken lines.
Fig. 8. Experimental group velocities plotted against the theoretical
dispersion curves with a xed frequency at (a) 0.615 MHz for T0; 1 mode
generation, and (b) 0.77 MHz for T0; 2 mode generation. The numbered
specimen with a steplike defect [Fig. 6(a)] is investigated arrows indicate the sequence of mode conversions following thickness
and the result is presented in Fig. 7(b). The observed group
velocity before the defect is 2.06 mm/s. It becomes
3.11 mm/s in the thinning region, which indicates a con-
version to the T0; 1 mode. When the thickness returns The shape of the edge transition aects the propagation
to 3 mm after the defect, the group velocity becomes behavior of the T0; 2 mode in an opposite manner. As the
2.49 mm/s. This observed group velocity is between those thickness change becomes smoother, the signal reection
for the T0; 1 and T0; 2 modes (3.11 and 2.06 mm/s). becomes clearer, as can be observed in Fig. 6. Furthermore,
Therefore, this velocity change indicates that the T0; 1 a total reection of the generated signal is observed in the
mode converts back to the T0; 2 mode while a part of last two specimens with very gentle thickness transitions
the T0; 1 mode remains, resulting in interference between (sloped by 10 and 5 ), as shown in the Figs. 6(d) and
the two signals, as shown in the far edge of Fig. 6(a). 6(e). The entire wave is reected back and there is no
Figure 8(b) shows a comparison of this experimental result transmission at all through the defect. The arrows in these
with the theoretical dispersion curves for a xed frequency gures indicate the location where the total reection
of 0.77 MHz, which shows group velocity dependence on occurs, which corresponds to a thickness of 1.87 mm. This
thickness. The rst conversion from the T0; 2 mode to the point is located within the sloping region and close to
T0; 1 mode takes place when the thickness reduces from 3 the theoretical cut-o thickness of the T0; 2 mode,
to 1 mm, indicated by arrow (i). Conversion back from the which is 2.1 mm. This behavior is also observed in the rst
T0; 1 to the T0; 2 mode occurs when the thickness returns higher mode of the SH guided wave propagating in a plate
to 3 mm [arrow (ii)]. specimen.36)
Most actual wall-thickness reduction areas have a rather
smooth edge, such as a dent or depression.39) As mentioned 4.2 Detection of partially circumferential defects with
above, it is observed in Fig. 5 that the T0; 1 mode travels mode conversion
through the defect without reection when the slope angle The above study conrmed the thickness dependence of
is small while reection occurs with a stepped edge. In the group velocity of the T0; 2 mode and the existence of
the specimen with a gradual transition, the wave cleanly a cut-o thickness, below which this mode converts to the
propagates into the region with the defect. This experimental T0; 1 mode. When traveling time is measured for a known
nding agrees with the simulation performed by Caradente distance with a xed frequency, this behavior can be a
et al.23) They predicted the reection coecient of the basis for wall-thinning detection. Thickness reductions in
T0; 1 mode when it is propagated in a tapered defect. all the specimens used so far were introduced over the
They used the ratio of the half length of the tapered region to entire circumference of the pipes. As a basic feasibility
the wavelength as the parameters. With these parameters of test of our idea with mode conversion, a further experiment
our experiment, it was reported that the reection coecient was performed with several specimens containing partially
decreases with the slope angle. circumferential defects.
07HC14-5 # 2013 The Japan Society of Applied Physics
Jpn. J. Appl. Phys. 52 (2013) 07HC14 Nurmalia et al.

250 6% of 0.4
circumference 100% of 25% circumference
Traveling time (s) circumference

100 T(0,1) -0.4 T(0,2)


Amplitude (arb. unit)

50 60% circumference
0 20 40 60 80
Cross-sectional loss (%) 0.2

Fig. 9. Traveling time of the two modes propagating in awed specimens
with dierent grades of cross-sectional loss. -0.2

-0.4 * T(0,2)

These defects dier in circumferential length but are 0.4 100% circumference
equal in axial length (200 mm). Thickness reduction is
introduced on the outer surface of the pipe specimens and 0.2
the remaining thickness is 1 mm in each specimen. This
thickness is smaller than the cut-o thickness of the T0; 2
mode; therefore, this mode should convert to the T0; 1 -0.2
mode. The conversion is observed as a group velocity
change, which causes a decrease in the measured traveling -0.4 *
time. In this experiment, a quartz transducer is replaced by 0 100 200
another identical EMAT. In the practical application based Time (s)
on the proposed method, both the transmitter and receiver
sides should be EMATs. In addition, by using an EMAT as a Fig. 10. Received waveforms when T0; 2 mode is generated in
receiver, it is possible to detect only the intended mode. The specimens containing dierent percentages of circumferential defect.
two EMATs are separated by a xed distance with the defect Signals with an asterisk are aected by conversion to T0; 1 mode at the
located between them. The rest of the experimental setup is defect.
the same as shown in Fig. 3, but a matching impedance
circuit is inserted between the receiving EMAT and the pre
amplier. The T0; 1 and T0; 2 modes are generated in wave is the original T0; 2 mode. In the last specimen with
each specimen and the traveling times are measured. a full-circumference defect, the entire wave converts to
The measured traveling time is presented in Fig. 9 as a the T0; 1 mode at the defect, resulting in a shorter traveling
function of the ratio of the cross-sectional area loss with time. It then can be concluded that with the present
respect to the original defect-free pipe. A full-circumfer- conguration, the phenomenon of conversion to a higher
ential defect reduces the cross-sectional area by 69.7%. As a mode can be a basis to detect a loss in pipe thickness that
reference, the traveling time for each mode propagating in a covers approximately one-half of its circumference.
defect-free specimen (0% cross sectional loss) for the same
distance is also included. 5. Conclusions
As expected, the traveling time of the T0; 1 mode does Torsional waves are successfully generated by the current
not show any signicant dependence on the cross-sectional EMAT and the propagation behavior of the T0; 1 and
loss. However that of the T0; 2 mode varies depending on T0; 2 modes is explored. The group velocity dependence
the cross-sectional loss. When the cross-sectional loss is on thickness induced by the mode conversion for each
close to 35%, which corresponds to approximately one-half mode is observed by traveling time measurements in pipe
of the circumference or more, the traveling time suddenly specimens containing thinning regions. The inuence of a
decreases. This is explained by the mode conversion from defects edge shape on mode conversion is also investigated.
the T0; 2 mode to the T0; 1 mode; the T0; 1 mode has It is conrmed that the group velocity of the T0; 1 mode
a higher group velocity at the defect resulting in a shorter in the wall-thinning region is independent of thickness
traveling time. Figure 10 shows the waveforms when T0; 2 owing to its nondispersive characteristic. However, the
mode is generated in specimens with approximately 25%, T0; 1 mode converts to T0; 2 mode at the far edge of
60% and full-circumference defects. In the rst specimen, the defect, in response to an abrupt thickness increase. This
the narrow defect does not cause mode conversion and the conversion does not occur when the thickness transition at
detected signal consists of only the T0; 2 mode. In the the edge is gradual. It is also observed that the T0; 1 is
second specimen, the generated wave apparently splits into hardly reected at the defects edge of this type.
two waves. The rst wave, marked with an asterisk, arrives When the thickness is smaller than the cut-o thickness,
earlier because it experiences mode conversion to the T0; 1 the T0; 2 mode converts to the T0; 1 mode. Conversion
mode, which has a higher velocity at the defect. The second back to the T0; 2 mode is observed when the thickness
07HC14-6 # 2013 The Japan Society of Applied Physics
Jpn. J. Appl. Phys. 52 (2013) 07HC14 Nurmalia et al.

increases to the original value. A more gradual thickness 16) X. Wang, P. W. Tse, C. K. Mechefske, and M. Hua: NDT&E Int. 43 (2010)
change at a defect enhances the reection of the T0; 2 17) H. Sato and H. Ogiso: Jpn. J. Appl. Phys. 49 (2010) 07HC08.
mode. When the thickness change is very slow, total 18) A. Demma, P. Cawley, M. Lowe, and A. G. Roosenbrand: J. Acoust. Soc.
reection occurs at the cut-o thickness, which is located Am. 114 (2003) 611.
within the slope. This conversion to the T0; 1 mode occurs 19) A. Lvstad and P. Cawley: NDT&E Int. 44 (2011) 553.
20) A. Lvstad and P. Cawley: NDT&E Int. 46 (2012) 83.
even when the thickness reduction exists only in approxi- 21) H. E. Engan: J. Acoust. Soc. Am. 104 (1998) 2015.
mately one-half of the circumference. From these results, it 22) H. E. Engan: IEEE Trans. Ultrason. Ferroelectr. Freq. Control 46 (1999)
can be concluded that mode conversion can be the basis of 1035.
wall-thinning detection. Combining this behavior and total 23) R. Carandente, J. Ma, and P. Cawley: J. Acoust. Soc. Am. 127 (2010)
reection phenomena can provide a potential method for 24) H. Nishino, S. Masuda, Y. Mizobuchi, and K. Yoshida: Jpn. J. Appl. Phys.
pipeline inspection. A further attempt to implement the 49 (2010) 116602.
proposed method for application is required. 25) H. Nishino, T. Tanaka, S. Katashima, and K. Yoshida: Jpn. J. Appl. Phys.
50 (2011) 046601.
26) H. Kwun, S. Y. Kim, and M. S. Choi: J. Korean Phys. Soc. 46 (2005) 1352.
27) Y. Y. Kim, C. I. Park, S. H. Cho, and S. W. Han: J. Acoust. Soc. Am. 117
1) Pipeline Accident Report: Rupture of hazardous liquid pipeline with (2005) 3459.
release and ignition of propane, Carmichael, Mississippi, November 1, 28) S. H. Cho, S. W. Han, C. I. Park, and Y. Y. Kim: J. Appl. Phys. 100 (2006)
2007. National Transportation Safety Board (2009). 104903.
2) Pipeline Accident Report: Pacic gas and electric company natural gas 29) R. B. Thompson: Physical Acoustics (Academic Press, New York, 1990)
transmission pipeline rupture and re, Sun Bruno, California, September 9, p. 157.
2010. National Transportation Safety Board (2011). 30) W. Johnson, B. A. Auld, and G. A. Alers: J. Acoust. Soc. Am. 95 (1994)
3) F. Hasan, J. Iqbal, and F. Ahmed: Eng. Fail. Anal. 14 (2007) 801. 1413.
4) Z. A. Majid, R. Mohsin, Z. Yaacob, and Z. Hassan: Eng. Fail. Anal. 17 31) M. Hirao and H. Ogi: NDT&E Int. 32 (1999) 127.
(2010) 818. 32) Y. Okawa, R. Murayama, H. Morooka, and Y. Yamashita: Jpn. J. Appl.
5) X. B. Li, X. Li, L. Chen, P. F. Feng, H. D. Wang, and Z. Y. Huang: Phys. 48 (2009) 07GD07.
J. Mech. Sci. Technol. 23 (2009) 109. 33) C. F. Vasile and R. B. Thompson: J. Appl. Phys. 50 (1979) 2583.
6) Y. Sun and Y. Kang: NDT&E Int. 43 (2010) 348. 34) M. Hirao and H. Ogi: EMATs for Science and Industry Noncontacting
7) J. Atzlesberger and B. Zagar: Procedia Eng. 5 (2010) 1401. Ultrasonic Measurements (Kluwer Academic, Boston, MA, 2003) p. 47.
8) G. Y. Tian and A. Sopian: NDT&E Int. 38 (2005) 77. 35) Nurmalia, N. Nakamura, H. Ogi, and M. Hirao: Jpn. J. Appl. Phys. 50
9) J. B. Nestleroth and R. J. Davis: NDT&E Int. 40 (2007) 77. (2011) 07HC17.
10) Z. Xu, X. Wu, J. Li, and Y. Kang: NDT&E Int. 51 (2012) 24. 36) Nurmalia, N. Nakamura, H. Ogi, M. Hirao, and K. Nakahata: NDT&E Int.
11) J. L. Rose, J. J. Ditri, A. Pilarski, K. Rajana, and F. Carr: NDT&E Int. 27 45 (2012) 156.
(1994) 307. 37) Nurmalia, N. Nakamura, H. Ogi, and M. Hirao: AIP Conf. Proc. 1511
12) M. J. S. Lowe, D. N. Alleyne, and P. Cawley: Ultrasonics 36 (1998) 147. (2013) 909.
13) J. L. Rose: Ultrasonic Waves in Solid Media (Cambridge University Press, 38) S. A. Uribe, N. Nakamura, H. Ogi, and M. Hirao: AIP Conf. Proc. 1096
Cambridge, U.K., 1999) Chap. 12. (2008) 1550.
14) H. J. Shin and J. L. Rose: Ultrasonics 37 (1999) 355. 39) K. A. Macdonald, A. Cosham, C. R. Alexander, and P. Hopkins: Eng. Fail.
15) J. L. Rose: Key Eng. Mater. 270273 (2004) 14. Anal. 14 (2007) 1667.

07HC14-7 # 2013 The Japan Society of Applied Physics