Download 1489_1.pdf

LASER ULTRASONIC STUDY OF CRACK TIP DIFFRACTION
M. Ochiai1, D. Levesque2, R. Talbot2, A. Blouin2, A. Fukumoto1 and J.-P. Monchalin2
1
Power and Industrial Systems R&D Center, Toshiba Corp., Yokohama 235-8523 Japan
2
Industrial Materials Institute, National Research Council Canada, Boucherville, PQ,
J4B6Y4 Canada
ABSTRACT. Ultrasonic diffraction pattern at a crack tip and crack depth sizing based on the
diffraction are studied by using laser-generated bulk waves. Directivity patterns oriented essentially
along the normal for the diffracted longitudinal wave and essentially at 45 (deg) for the diffracted
shear wave for any incident angles is obtained. A laser-ultrasonic system combined with time-offlight (TOFD) analysis is produced and demonstrates accurate crack sizing on an artificial slot
having a variable depth from 0 mm to 10 mm. An improvement of signal-to-noise ratio by split
spectrum processing (SSP) is also suggested.
INTRODUCTION
Regarding defect inspection, not only detecting but also sizing of cracks is
desirable because a depth of the crack is one of the most important parameter to evaluate
deteriorations of the materials. Ultrasonic testing (UT) is an inspection technique widely
used in a field of non-destructive evaluation (NDE) and some UT methods based on
detection of diffraction waves from a crack tip are considered as the most reliable
technique to measure a depth of the crack in a material. The time-of-flight diffraction
(TOFD) is a technique sensing the crack tip diffraction, in which the depth and the
location of the crack are analyzed by the time-of-flight analysis [1].
On the other hand, laser technique offers non-contact remote maintenance
and is suitable to be used for components having complicated shape installed in narrow
space. Compact and flexible optical fibres easily deliver laser beams into such narrow
space and also simple optical mechanisms easily scan small laser spots following onto
such contour surfaces.
In general, TOFD is a technique based on two-probes pitch-catch
measurement and piezoelectric transducers (PZTs) are utilized for both the generation
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/$20.00
1489
and the detection of ultrasonic waves. One technique that enables laser-based TOFD is to
use a laser-ultrasonics, which means an ultrasonic wave generation by a laser pulse and
its detection by an optical interferometer [2]. Recently, a few techniques to size surfacebreaking cracks have been proposed by using laser-generated surface wave [3-5],
although these can be useful only for the shallower cracks, say a depth of less than a few
mm, in principle. TOFD technique combined with laser-ultrasonics has been also
attempted by using surface-skimming wave [6] and bulk waves [7]. These successes are
however limited by relatively poor sensitivity of the laser-based detection systems.
This paper will describe a performance of complete laser ultrasonic TOFD
system on realistic tight cracks. Details of diffraction patterns at a crack tip measured by
a highly sensitive laser-based detection system will be presented. Then sizing capability
of laser ultrasonic TOFD system will be demonstrated by using a surface-breaking
artificial slot. Finally, stress corrosion cracking (SCC), which is one of the most severe
and potentially dangerous deteriorations in nuclear plants, in a stainless steel plate are
examined by using the laser ultrasonic TOFD method. A typical B-scan containing
diffraction signals from SCC and a result will be presented. A further signal processing to
improve a signal-to-noise ration (SNR) of diffraction indication will be proposed.
EXPERIMETAL METHODS FOR LASER ULTRASONIC TOFD
An experimental configuration of laser ultrasonic TOFD is outlined in
Fig.l. Ultrasonic waves were generated by Q-switched Nd:YAG laser pulses having a
wavelength of 532nm, a pulse duration of about 10ns and a repetition rate of lOHz. Pulse
energies of 75mJ were delivered to the sample surface. These energies are quite intense
but are still below the threshold for optical fibre delivery [8]. A lens mounted on a 1dimensional scanning stage (x'- direction) was used to focus the laser pulses onto the
sample surface with a diameter of about 100jj,m. The fluence at the surface was estimated
about IkJ/cm2 and this produced ablation plasma at each irradiated spots. This method of
excitation simultaneously generated a various ultrasonic modes; surface-skimming
longitudinal waves (P), Rayleigh waves (R), bulk longitudinal waves (L), bulk shear
waves (S) in the sample medium and shock waves (A) originated from the expansion of
ablation plasma in air. These waves except for the shock wave had frequencies of,
typically, 0.5MHz to 20MHz and respective particular directivities.
Ultrasonic
waves
generated by the incident laser pulse
and travelling in the medium were
detected using an optical detection
system based on a confocal FabryPerot interferometer [9]. Two
multimode optical fibres having a
core diameter of 400|Lim were
employed; one fibre was used to
deliver light from the detection laser
to the sample surface, while the other
FIGURE 1. Experimental set up for laser ultrasonic TOFD.
1490
one was used to bring scattered light back to the interferometer. The detection laser was a
fundamental NdiYAG laser having a wavelength of l,064nm. This detection laser
emitted long pulses (typically, 70|us) with highly stabilized frequency and high peak
power (typically, IkW). The confocal Fabry-Perot is 1m long and has an optical
bandwidth of from 2MHz to 12MHz, approximately. A PIN photodiode detected the light
demodulated by the Fabry-Perot. The detected signal was high-pass-filtered with a cut off
frequency of 1MHz and amplified by 40dB. The signals were sampled with a sampling
rate of lOOMsample/s and stored for the signal processing in a personal computer. This
computer also controlled the synchronization between the emissions of two laser sources
and the scan of the sample.
A sample to be tested was fixed on a 3-dimensional scanning stage (x-, y-, zdirections as shown in Fig.l). A combination of these stages (x-, x'-, y- and z-direction)
enabled to adjust focus (z-direction), to change scanning location (y-direction) and to
scan the detection and generation spots independently (x- and x'-directions).
EXPERIMENTAL RESULTS
Amplitude Profiles of Crack Tip Diffraction
Theoretical models have been already constructed to describe the diffraction
of ultrasound, by considering infinite plane elastic waves incident on a semi-infinite
crack having no width [10, 11]. In the case of laser-generated ultrasound, the incident
waves can be usually assumed as spherical waves because the generation source, in other
words, a laser spot for generation, is very small compared to the generated ultrasonic
wavelengths. However, the incident energies of ultrasound to the tip greatly depend upon
the incident angle due to the particular emission pattern of laser-generated ultrasound
[12]. As it is well known, in ablation regime, directivity of the laser-generated
longitudinal wave is essentially normal to the surface and of the laser-generated shear
wave is angled typically from 30° to 60°.
Diffraction patterns of a tip were interrogated by using the setup. A stainless
steel plate having a thickness of 15mm was used. A surface-breaking artificial slot (EDM
notch) having a dimension of 15mm(L) x 0.3mm(W) x 10mm(D) was machined in a
surface. The generation and detection laser beams were irradiated onto the surface
opposite to the slot. Each laser beam was scanned on the surface independently; the
generation spot was scanned to cover the incident angle, /?, to the tip from 0° to 76° by
400 points while the detection spot was scanned to cover the receiving angle, 0, from the
tip from -63° to +63° by 100 points. Typically, step angles were from 0.6° at ft =0° to
0.03° at /?=76° for the generation and from 2.3° at 0 =0° to 0.47° at 0 =±63° for the
detection. Because of the same geometrical reason, it is noted that the difference of
travelling distance in both generation and detection should be considered in order to
calibrate the amplitude of diffracted.
1491
(a) LdfL-wave
(b) SdfS-wave
FIGURE 2. Diffraction pattern at a crack tip.
A total of 100 waveforms (100 different angles of detection) was used to
reconstruct a 2-dimensional image of B-scan for each incident angle (generation position).
The amplitude profiles of the diffraction signals were extracted from the B-scans. Typical
patterns are plotted as polar graphs in Fig.2. In each graph, the tip is located at the origin
and the direction of the incident wave is shown as a black arrow. Diffraction patterns of
the LdfL-waves with incident angles from 0° to 60° are shown in Fig.2(a) and diffraction
patterns of the SdfS-waves with the same range of incident angles are shown in Fig.2(b).
The maximum amplitude of the LdfL-waves is observed at the reception angle of 0° and
the amplitude becomes smaller with the increase of the incident angle. As to the SdfSwaves, the maximum amplitude is observed at the reception angle of about 45° in any
incident angles and the amplitude becomes the largest with the incident angle of 45°.
These patterns should involve the effects of the emission pattern of laser-generated
ultrasound, the diffraction pattern of the tip and the difference of travelling distance.
It is noted that the notation of "df' means that the signal was produced by
the diffraction at the tip, for example, "LdfS" means a diffraction signal from the tip
where the wave was incoming to as L-wave, diffracted and mode-converted at and
outgoing from as S-wave.
Calibration with Artificial Slots
For crack orientation in direction vertical to the surface, the depth and the
location of the crack can be evaluated through simple geometrical analysis in TOFD.
Although there can be some variations in TOFD, the basic configuration is the scanning
transversal to the crack as shown in Fig.3. The generation and detection spots launched
onto the same surface of the tested sample with a separation of 2S. Assuming there is a
diffraction source, such as a crack tip, at a location of x from the centre of the separation
and the depth of d in the medium, the travelling time of the diffraction signal, £diff, is
given by
,—————————
,—:——————-
' - x)2 + J(d
f
diff
~
Generation
Detection
Separation 25
(D
^
where v is an ultrasonic velocity in the medium.
The time ?diff should be always longer than the
travelling time of P-wave, tp, and shorter than one
of the bottom echo, t\» here b=LL or SS. As
1492
£———1 —— ^.JL^
^\^
4
'I
$
Depth d
.............V.....................
Crackl
FIGURE 3. TOFD configuration for crack.
scanning the generation and detection spots, the time ^iff becomes minimum at the point
of jc=0. The center position of the generation and detection shows the location of the
crack. In this case, the crack depth, d, is determined by
--S2
d=
(2)
As previously discussed, the maximum amplitude of diffraction signal is given by the
LdfL-wave with both the incident and the reception angle of 0°. This however means that
the generation and detection spots should be superimposed at the same position on the
sample surface. In this case, the R-wave and A-wave, both of which have much larger
amplitude than any bulk waves, will arrive during the time to be observed and the
relatively small diffraction signals will be hidden by these waves. It is preferable that the
arrival of R-wave, £R, which is usually earlier than that of A-wave, should be later than
one of the bottom echo of interest.
To assess the performance of laser ultrasonic TOFD system, a series of
experiments was carried out by using a stainless steel plate (15mm-thick) containing an
EDM notch. The depth of the artificial slot varies linearly from Omm to 10mm along the
direction of the slot length of 40mm (Fig.4). Some points ranging in depth from Omm to
10mm on the slot were examined by generating each B-scan image from a series of Ascan waveforms. The separation between the generation and the detection spots was fixed
in 20mm and the sample, which is equivalent to scanning the laser spots simultaneously,
Initial Position
Generation Detection
Separation: 2Q ran
Final Position
Generation Detection
120mm
m
i
Position (mm)
FIGURE 5. Typical B-scan containing the
FIGURE 4. Experimental setting for calibration.
diffraction trace.
— 1- — _ _| _ |_ _j _
— -\" ~ ~ 1~ — - t- -t - 1- - -t- - — 1- — - h -t - 1- -'•- : - -1 - t- -t - -1 - 1- -1 - 4 - 1- 4 - - 4 - — 4- — _ |_ 4 _ |_ .
- 4- -
- -t - 1- -t- - -t - 1- -t -
p
§
_ J. _
_ _l _
. 4 _ l_ 4 _ _ J. _ _ 1_ _ _ L _L _ l_ .
~
-'
1
1
I
. I I I
I
I I I .
. I
|
|
I
1
1
1
1
I
1
1
1
1
1
A
1
1
1
s
M-fch4
*r
*r-[
*
- T~
~~ r ~ - r -T - r -
- -t - 1- 4 - - 4- _ 4 _ 1- 4 -
- +- - - 1- 4 - 1- - t- - _ i_ _f _ i_ _
- i- - — U 4 — 1— -
- -\ - r ~r -
J
'i 4
- 4 - h- 4 -
S1
U 4 _
I
I
I I
1
1
1
1
1
1
^1If ^4^
- + -
:
- 4- -
- H - 1- H - H - h- -1 - -1 - U 4 -
I
.
ffi
2
n
_ J _ L _L _ _ 1 _ _ L _ _ L J _ L _
1
1
1
1
I I I .
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
2
4
-
. J _L
_ 1_
1
1
1
1
1
1
I I
1
6
8
J _
1
1
I
-
- - -t- -\
*•>
- — K -/-f ,\|
4I
1
A?
1
\- -t- H - —
1
- - ~
Q
1
fc
1
1
* 0
i^H
!* 1
I
if
-\
h 1 t- -
~7 ^
\A
/\
( i
i
t- 1- -
-
h/i
}
1
-
- ---
*4
00
0
1
o
1
1- -t 1
1
L»
/
L i.
L J. .
1
1
|
2
t-
4
e
a
10
Nominal Depth (mm)
FIGURE 6. Propagation times of the
Nominal Depth (mm)
FIGURE 7. Relation between measured
diffraction the wave at the peak.
and designed depths.
1493
was scanned by 40mm along the direction perpendicular to the slot length. A-scan
waveforms were obtained with an interval of 0.05mm so that a total of 800 waveforms
were used to reconstruct a 2-dimensional image of B-scan.
A typical B-scan, which x-axis corresponds to the position of the generation
spot, y-axis corresponds to the time and plotted colour corresponds to the signal
amplitude, are shown in Fig.5. Each image contains an arc-shape tracing the LdfL
diffraction signal. The trace should appear between the arrival of P-wave and LL-wave
and the peak of it gives the slot depth according to the equation (2) as well as the slot
location. The minimum arrival time of each arc was measured in each B-scan and was
plotted as a function of the designed slot depth in Fig.6. By using a proper ultrasonic
velocity (in this case v=5,850m/s is used as the velocity of longitudinal wave), the slot
depth was estimated for each measurement with enough accuracy as shown in Fig.7.
Results on SCC Samples
The laser-ultrasonic TOFD technique was used to inspect a stainless steel
plate (100mm(L) x 50mm(W) x 10mm(D)) that contained a few stress corrosion cracks.
These SCCs have less than O.lmm wide. The generation and detection spots were
separated by 15mm and were irradiated onto the surface opposite to the opening of SCCs.
The sample was scanned for 40mm with a step size of 0.05mm. The unprocessed B-scan,
Fig.8, shows the presence of the SCC as an arc-shape indication, even though it is not as
clearly distinguished as the artificial slots as shown in Fig.5. The crack depth was
estimated as 4mm as it is expected by the microscopic observation from the sidewall.
To improve SNR and to enhance the crack image in the B-scan, split
spectrum processing (SSP) [13] was attempted. SSP is a method where one wideband
ultrasonic pulse is filtered through a digitally implemented filter bank. The filter outputs
are fed through some nonlinear operation to create a final signal where the coherent
noises have been eliminated. The filter bank consists of a certain number of band pass
filters of different centre frequencies but constant bandwidth. The output signals are
normalized and compounded using algorithms such as minimization or polarity
thresholding. The result is shown in Fig.9.
»
li
Position (mm)
FIGURE 8. A result of TOFD analysis
FIGURE 9. A result of TOFD analysis
on SCC sample (unprocessed).
on SCC sample (SSP).
1494
DISCUSSION
As discussed, although the best SNR is obtained when the detection and
generation spots are superimposed, the detection should be separated from the generation
by a certain distance. On the contrary, the depth evaluation becomes very difficult if on
uses too wide separation (too large incident and detection angle). In fact, even the
artificial slot, it was impossible to find an indication and to evaluate the slot depth on the
slot of 9mm and 10mm deep, corresponding to the incident angles of more than 60°, by
using LdfL-wave mode. From this point of view, the shear wave, which has an angled
emission pattern, seems to be an alternative. It should be however noted that the
diffraction patterns of shear wave (dfS-wave) have null amplitude at some diffraction
angles. Therefore, the separation should be determined based on the assumed crack depth
when the shear wave is used on TOFD.
In the present approach, the dominant noise is originated from ultrasonic
propagation, such as backscattering from grains; it means sensitivity of the detection
system is not essential for the detectability. To eliminate this kind of coherent noise, we
focused on the frequency dependence of the noises. SSP and laser ultrasonics are good
combination because laser ultrasonics is a broadband generation and detection technique
that what SSP requires. As shown in Fig.8 and 9, SSP is an effective method to improve
SNR; however good results can be obtained only if all parameters, such as number of
filters and bandwidth of individual filters, are carefully tuned. There will be another
approach that uses averaging techniques. Although both temporal and spatial averaging
techniques may improve the SNR, spatial averaging with different propagation paths is
superior because temporal averaging cannot eliminate coherent noises. The synthetic
aperture focusing technique (SAFT) [14- 16], which collects many contributions with
proper calibrations from different propagation paths, offers an attractive alternative,
although this technique usually requires relatively much time to acquire many data and to
execute complex signal processing.
CONCLUSION
TOFD linked with laser-ultrasonics offers a well understood method based
on relatively simple data acquisition and processing to size a crack. We have reported
that the technique is capable to size slot depths from 0.2mm to 8.5mm in a type 304
stainless steel plate having a thickness of 15mm without any averaging and/or filtering.
However, in the cases when the ultrasonic noises in the material to be inspected are large
or the crack is small, SNR becomes poor and this causes a problem on TOFD evaluation.
The particular emission pattern of laser-generated ultrasounds is one of the
reasons of this poor SNR. The interrogated diffraction patterns of a slot tip have shown
that the diffracted L-wave appears the direction normal to the sample surface while the
diffracted S-wave appears at the diffraction angle of about 45°. The best SNR was
obtained with the incident and diffraction angles of 0°; however, this configuration was
not convenient for TOFD processing. The use of narrower separation or the use of shear
1495
wave can be alternatives; these however sometimes make the measurements unreliable
because the signal can be overlapped by the other waves or disappeared.
To overcome SNR problem, the use of SSP was suggested. Although results
of SSP were very sensitive to the parameters in the processing, it has been shown that the
SNR is improved by using carefully tuned parameters. SSP is a method essentially based
on the frequency domain analysis; the results should be therefore similar to one obtained
by a band pass filter. The other possibility is to use temporal or spatial averaging
technique, such as SAFT.
REFERENCES
1. Silk, M.G., Research Techniques in Non-destructive Testing, Vol.III, Academic Press,
New York, 1977
2. Scruby, C.B., et al., Laser Ultrasonics, Technique and Applications, Hilger, Bristol,
1990
3. Cooper, J.A., et al., "Surface acoustic wave interactions with cracks and slots: a
noncontacting study using lasers" IEEE Trans. Ultrason. Ferroelectr. Frequency
Contr., UFFC-33, 5, (1986) 462
4. Scala, C.M., et al., "Laser ultrasonics for surface-crack depth measurement using
transmitted near-field Rayleigh waves," Rev. Prog, in QNDE (QNDE'99), Montreal,
PQ, 1999,19A. (1999) 327
5. Ochiai, M., et al., "Sizing of micro cracks using laser-induced broad-band surface
waves," J. At. Energy Soc. Japan, 43, 3 (2001) 275 (in Japanese)
6. Shan, Q., et al., "Surface-breaking fatigue crack detection using laser ultrasound," /.
Appl Phys. Lett., 62, 21, (1993) 2649
7. Mihara, T., et al., "Laser ultrasonic TOFD method," Rev. Prog, in QNDE (QNDE'OO),
Ames, IA, 2000, 20A. 255
8. Schmidt-Uhlig, T. et al., "Laser shock processing with 20 MW laser pulses delivered
by optical fibers," Eur. Phys. J. AP., 9 (2000) 235
9. Monchalin, J.P., et al., "Optical detection of ultrasound," IEEE Trans. Ultrason.
Ferroelectr. Frequency Contr., UFFC-33, (1986) 485
10. Ogilvy, J.A. et al., "Diffraction of elastic waves by cracks: application to time-offlight inspection," Ultrasonics, 21 (1983) 259
11. Temple, J.A.G., "Time-of-flight inspection: theory," Nucl. Energy, 22 (1983) 335
12. Hutchins, D.A., "Directivity patterns of laser-generated ultrasound in aluminum," /.
Acoust. Soc. Am. 70 (1981) 1362
13. Karpur, P., et al., "Split spectrum processing: determination of the available
bandwidth for spectral splitting," Ultrasonics, 26 (1988) 204
14. Blouin, A., et al., "SAFT data processing applied to laser-ultrasonic inspection," Rev.
Prog, in QNDE (QNDE'97), San Diego, CA, 1997,17A. 611 (1997)
15. Levesque, D., et al., "Improved performance of laser-ultrasonic F-SAFT imaging,"
Rev. Prog, in QNDE (QNDE'98), Snowbird, UT, 1998, ISA, 309 (1998)
16. Ochiai, M., et al., "Visualization of surface-breaking tight cracks by laser-ultrasonic
F-S AFT,"to be presented in QNDE (QNDE' 02)
1496