Stress corrosion cracking detection using non-contact ultrasonic techniques

Full text

Stress corrosion cracking detection using non-contact ultrasonic
techniques
F. Hernandez-Valle, A.R. Clough, R.S. Edwards∗
Department of Physics, University of Warwick, Coventry, CV4 7AL, UK
Abstract
In this work a method is presented for detecting and locating stress corrosion cracking (SCC)
in stainless steel pipe samples. The method combines laser generation and either laser or
electromagnetic acoustic transducer (EMAT) detection, scanning the generation point across
the sample surface. Using laser-generated ultrasonic waves that interact with the cracks,
and performing time-frequency analysis techniques to examine changes in the generated
wavemodes, surface plots that clearly resolve the spatial extent and geometric alignment of
the cracks are created and presented here. The method is demonstrated using components
removed from service after exhibiting SCC.
Key words: Stress corrosion, stainless steel, Ultrasonics, Nondestructive testing,
Time-frequency analysis
1. Introduction
The interaction of a corrosive environment and tensile stress (e.g. directly applied stresses
or in the form of residual stresses) can produce failure in the form of stress corrosion cracking
(SCC) in susceptible metallic components [1]. The damage produced by SCC is not always
obvious to casual inspection, so failure can be both unexpected and catastrophic. Thus,
early detection of such defects is important in order to have sufficient time for adequate
measures to be taken, for example repair or replacement of the damaged component.
∗Corresponding author
Email address: r.s.edwards@warwick.ac.uk (R.S. Edwards)
Preprint submitted to Corrosion Science August 15, 2013

Various nondestructive evaluation (NDE) techniques have been trialled to detect and
locate this type of cracking, all with certain advantages and disadvantages. For example,
dye penetrant testing is a widely used and relatively simple method which can be easily per-
formed at remote test sites. However, besides requiring surface preparation, dye penetrant
can only detect defects that are open at the sample surface, and it performs poorly on hot,
dirty, and rough surfaces as well as on porous materials [2, 3]. Eddy current and pulsed
eddy current techniques can be used for rapid inspection, and have a relatively small probe
size, and there is no need to physically contact the test samples. However, eddy current
inspection only detects surface or near-surface defects, it is very sensitive to a wide range of
parameters related to the conductivity and magnetic permeability of the test sample, and
is very sensitive to lift-off variations [4–6].
Radiographic inspection has the advantage over many other NDE techniques in that
analysis and interpretation is almost intuitive. Amongst radiographic methods, X-ray to-
mography excels where information is needed in three spatial dimensions [7–9]. However,
despite the advantages, radiation techniques have serious safety concerns due to possible
overexposure to large amounts of radiation. In addition, a closed crack will generally only
be detectable in a radiograph at certain orientations, ideally when the long dimension of the
crack is parallel to the direction of radiation propagation [7–9].
Various ultrasonic techniques have been used for SCC detection, in particular time of
flight diffraction (TOFD). This technique relies on the detection of weak diffracted waves
arising at the edges or tip of a crack, and can locate and size defects either within the bulk
of a sample, or on the surface [10, 11]. Although TOFD is well understood and widely used
it has some limitations, such as the assumption that there is no interference from other
wavemodes. In some geometries this will limit its applicability to cases where there is only
a single defect, thus research to add new features and remove some limitations is still being
performed [10, 11].
Standard ultrasonic measurements, particularly those looking for reflections of bulk waves
from defects, have difficulties due to the low reflection and transmission coefficients for
closed and partially closed defects. However, recent work looking at the interaction of
2

surface waves with surface-breaking defects in the near-field have shown several enhancement
mechanisms which can be used for identification of cracking [12–20]. For Rayleigh wave
propagation on thick samples, the signal enhancement observed when scanning the detector
across the crack is due to the constructive interference of incident, reflected and mode-
converted waves [12, 13, 15]. For defects propagating at an angle to the surface, where the
local thickness changes throughout the defect, further large enhancements are also seen in
the time and frequency domains [21–23]. Similar effects are seen when scanning thin samples
using Lamb waves, with enhancements observed as an increase in magnitude of the signal
at certain frequencies [24]. These distinctive features can be used to identify the defect and
give some information about its geometry [23].
For laser generation, further enhancement effects have been observed due to the changes
in generation conditions when the generation spot is over a defect [18–20, 25]. As the
laser spot passes over the defect, the boundary conditions of generation on the surface will
change, and this has been shown to give an increase in the magnitude of the signal at certain
frequencies. If the defect is partially closed and the laser spot source is directly illuminating
the defect, then the material will also undergo thermo-optic crack closure, which has been
shown to produce higher order frequency components [20].
In this paper we examine the near-field interactions of laser generated ultrasonic waves
with stress corrosion cracking in stainless steel pipe samples removed from service, and use
the ultrasonic signal enhancement to resolve the spatial extent and geometric alignment of
those cracks. Ultrasonic waves generated in bounded media, such as pipes, take the form
of guided waves. These travel along the pipe with different propagation and displacement
behaviour depending on the particular wavemode (e.g. longitudinal, torsional or flexural
modes in pipes, and higher-orders of each of these) [26]. The particular wavemodes generated
depend on the pipe geometry, the material, the generation source used and also the testing
frequency. Quantification of the enhancement effect for guided waves in the time domain
is complicated due to the presence in most cases of more than one wavemode, thus time-
frequency analysis is performed in order to highlight which modes were generated and/or
enhanced [19].
3

The generation of ultrasound here was performed using a thermoelastic laser source,
described in section 2. For detection two different approaches were used; laser interferometry
techniques (section 3.1) or electromagnetic acoustic transducers (EMATs, section 3.2). Both
approaches are non-contact, and hence their influence on the sample properties and on the
wavemode being measured is negligible [27]. Both can easily be scanned across the sample
surface, and can work on rough surfaces and in hostile environments. Laser detection has
the advantage of high spatial resolution, which adds to the accuracy of defect location.
However, laser detection systems are still a costly option when compared to the EMAT
detection approach.
The paper is organised as follows. In section 2, a description of the experimental setup
and time-frequency analysis are given. Results from scans of damaged pipe samples and
discussion of each detection approach are given in sections 3.1 and 3.2. Finally, a summary
of the uses of these approaches is presented in section 4.
2. Experimental method
Experiments were performed using a pulsed Nd:YAG laser (1064 nm wavelength and
10 ns pulse duration) to generate ultrasound, focused into a point of approximately 500 µm
diameter. The laser was filtered such that it acted in the thermoelastic regime, minimising
damage to the sample, to generate broadband ultrasonic waves [28]. In this regime, the
area impinged by the laser beam is heated rapidly, expanding and generating stress as the
surrounding cooler material constrains its expansion. The associated pulse of material ex-
pansion and contraction can generate a range of ultrasonic modes, such as bulk longitudinal
and shear waves, or Rayleigh waves and other guided wave modes, depending on the sample
geometry. For detection, to investigate the near-field interactions of surface acoustic waves
with SCC, two different approaches were used; laser detection, or EMATs [27].
For laser detection, a two-wave mixer laser interferometer system from Intelligent Optical
Systems [29] was used (setup shown in figure 1a). The interferometer is sensitive to the out-
of-plane component of the surface displacement, and has a bandwidth of 125 MHz, allowing
4

SCC
S
c
a
n
d
i
r
Scan step = 0.5
Pipe sample
°
R
o
t
a
t
i
o
n
s
t
a
g
e
Detection point
(Laser or EMAT)
Generation laser
(a) Set-up (b) Pipe samples
Figure 1: Experimental setup used for inspection of stainless steel pipes containing SCC for
laser or EMAT detection (a). Dimensions and scanned regions of both samples are shown
in (b), with a schematic of the orientation of the measured defects relative to the pipe.
measurements over a wide range of frequencies. Its continuous wave laser (200 µm spot-
size) operates at 1550 nm, with a power variable up to 2 W; this is varied depending on the
sample surface quality. The interferometer works on rough surfaces without the need for
surface preparation. [29]
For EMAT measurements, linear detection coils were used, produced by wrapping 10
turns of 0.08 mm diameter insulated copper wire around NdFeB magnets of field approx-
imately 0.5 T, with the field aligned either into the sample, for measuring predominantly
in-plane particle velocity, or along the sample surface for measuring predominantly out-
of-plane velocity. The transducer active measurement area was 1 mm (width) by 35 mm
(length). For a description of the principle of operation and more detailed description of the
EMATs used, the reader is referred to reference [23].
Measurements presented here were done on two different AISI 304 stainless steel sections
of pipes removed from service. Both pipe sections had an inner diameter of 152.4 mm and
an outer diameter of 160.2 mm, and hence a wall thickness of 3.9 mm, with different axial
5

lengths; 270 mm (sample A) and 50 mm (sample B, see figure 1b). Sample A contained
two cracks (SCC 1 & SCC 2) and some pitting damage in the region of interest (figure 4a),
whereas sample B contained several features which looked like surface-breaking cracks to
visual inspection (figure 5a).
Regardless of the detection technique, the scanning process consisted of placing each
sample on a rotational stage to perform a circumferential scan, such that the generation laser
source passed over the region of interest with constant steps; these increments correspond
to the x-axis on the surface plots shown in figures 4, 5, and 10. The vertical position of the
generation and detection points was then varied, starting from the lowest part of the region
of interest and moving up until reaching the upper part with increments of approximately
2 mm; these increments correspond to the y-axis on the surface plots.
Guided ultrasonic wave modes were generated, propagated along the sample, and were
detected by the laser or EMAT. Since guided waves in pipes can consist of several different
wavemodes, with those present depending on the pipe geometry, the frequency of the ultra-
sonic wave and the generation source [26], mode identification using just the signals in the
time domain is complicated. For this reason, a time-frequency analysis was performed on
each A-Scan at each scan point to identify the arrival time of each frequency component, and
to highlight which modes were generated and/or enhanced (see figures 2, 6 and 7) [19, 24, 30].
Each wavemode has a frequency-dependent velocity which can be calculated, and hence the
sonogram analysis used here is able to identify modes based on their arrival times, and allow
identification of modes which show sensitivity to the presence of a surface-breaking defect.
An increase in the magnitudes of these modes at a defect can then be measured through
windowing the correct region of the sonogram, and used for the construction of surface plots
that resolve the spatial extent and geometric alignment of SCC in the pipe’s surface.
3. Results and discussion
3.1. Laser-Laser SCC measurements
For measurements using the laser-laser system, the laser spot source and laser detector
were scanned circumferentially around the pipe section in increments of 0.5◦(scan step of
6

approximately 0.67 mm) with a fixed circumferential separation, such that the spot source
passed over the defect region, as shown in figure 1a. To identify the arrival time of a specific
mode at a chosen frequency a sonogram was produced at each scan position [19, 24, 30]. For
positions at which the source was located far away from the defect a low frequency surface
acoustic wave was observed, and was used as representative data for a defect free region of
the pipe; an example of this is shown in figure 2a. The bottom part of this figure shows the
A-Scan, with the top part of the figure the frequency content at each time. Primarily low
frequency modes are generated, and the measurement is sensitive to out-of-plane motion,
hence will only be sensitive to certain wavemodes [26]. For positions when the source is fully
illuminating the defect a high frequency enhancement can be observed, in the form of the
extra wavemode shown in figure 2b, which is accompanied by a visible change in the A-Scan
at the source location. The sharp wave at about 18-20 µs is the wavemode shown by the
near-vertical line on the sonogram.
As can be seen on the A-Scans in figure 2, analysis in the time domain of these sig-
nals is complicated, and the amplitude measurements used when studying enhancements of
Rayleigh waves are not easily applicable [15]. However, this high frequency enhancement
is very similar to that observed when Lamb waves are generated at defects on thin, flat
plates, which is due to the generation of new mode converted waves at the defect, and the
change in the boundary conditions of generation that occurs as the laser source passes over
the defect [18, 19, 24, 25]. This is analysed by studying magnitude variations in selected
regions of the sonogram. To quantify the enhancement for this data, the peak magnitude
in the sonogram data (maximum value of the z-component) of two regions of the sonogram
were recorded across the duration of the scan, chosen based on knowledge of the relevant
dispersion curves [26], and on the regions which were observed to show the most significant
changes during the scan. The first region was bounded by the frequency range between
1-2 MHz and arrival times of 17.14-20.14 µs, with the variation in the magnitude as a func-
tion of scan position shown in figure 3a. This shows an increase at the defect but also some
variation at other positions. The second region was at a higher frequency range of 2 - 3 MHz,
arriving between 18.54 and 20.54 µs, with the variation in the magnitude as a function of
7

(a) Reference
(b) Enhanced
Figure 2: Sonograms with and without defect present for the laser source passing over defect
region.
8

scan position shown in figure 3b.
Both of these analysis regions show a clear increase in peak magnitude in the chosen
regions at the position of a known defect, along with variations at other positions. In order
to improve the probability of detection (PoD) of the SCC and have a clear image of the true
extent of the defects, the product of the two sets of peak magnitude data was used. This
is shown in figure 3c, and clear peaks in the magnitude are observed for positions at which
the source illuminates the defect, with the magnitude returning to a steady lower level in
regions free from surface defects.
The generation and detection points were scanned across the sample, as described in sec-
tion 2, producing a figure similar to figure 3c for each vertical position. The magnitude data
for each scan were then stacked next to each another to form a surface plot of the enhance-
ment behaviour across the sample, shown in figure 4b. Here, the colour scale represents the
extent of the enhancement in each scan as a function of the source position, with the lighter
areas corresponding to larger magnitudes, and dark areas to magnitudes below a chosen
threshold based on the ‘no-defect’ magnitude. A photograph of the scanned area (figure 4a)
shows two stress corrosion defects, labelled SCC1 and SCC2, and several instances of smaller
patches of pitting damage. In the enhancement surface plot in figure 4b, the two defects
can be clearly resolved with their spatial extent and geometric alignment identifiable from
the image. Two instances of the pitting damage can also be observed in the upper left hand
corner of the surface image as bright spots. The spatial extent of these defects is smaller
than the two stress corrosion cracks and this is reflected in their small spatial extent in the
surface plot; resolution could be improved significantly by using smaller scan steps.
The image in figure 4b highlights the advantages of the near-field inspection method
for the resolution of multiple partially closed surface defects in close proximity, with sev-
eral defects successfully resolved. A far-field inspection technique based on reflection or
transmission from the defects would show little evidence of their presence due to the low
reflection and high transmission of the incident wave at the defects, illustrated by the lack of
any significant change in the waveform when the defect lies between the source and detector.
The low reflectivity prevents a pulse-echo investigative approach, and the high transmission
9

(a) Low frequency
(b) High frequency
(c) Product
Figure 3: Peak sonogram magnitudes for low (a) and high (b) frequency regions, allowing
the production of an improved PoD by taking the product of the two (c).
10

makes it challenging to resolve the defect through changes in the transmitted waves.
The photograph of sample B (figure 5a) shows several marks, each of which could be
indicative of cracking. However, the image produced from the signal enhancements in fig-
ure 5b shows clearly that there are two partially closed defects with a small separation, close
to the sample edge. These enhancements from defects are distinct from the enhancement
arising as the spot passes over the edge of the sample (right hand side of the figure), which
also shows enhancement similar to a full thickness defect across the width of the sample.
Closer inspection of the defect area in figure 5a showed that the apparent defect located in
the upper left of the image was in fact dirt on the sample surface. The fact that no signal
enhancement was seen as the source passed over this region illustrates the effectiveness of
the technique for distinguishing real defects from damage which may falsely be identified
by visual testing. The surface plot in figure 5b highlights the advantages of the laser-laser
near-field approach used here.
3.2. Laser-EMAT SCC measurements
Whilst the laser-laser system has many advantages, including the excellent spatial and
frequency resolution, the high cost of such a system means that it is not always practical for
use. Furthermore, certain wavemodes will have only a small amplitude in the out-of-plane
direction, and a significant in-plane motion [26]. EMATs may prove to be a suitable inexpen-
sive detection alternative, with the capability to detect in-plane or out-of-plane motion [23].
In a similar fashion to the laser-laser approach, the laser spot source was scanned circum-
ferentially in increments of 0.8◦(scan step of approximately 1 mm) such that the spot source
passed over the defect region, but with detection using an EMAT. To simplify the experi-
mental procedure the EMAT was held fixed in the same position on the sample, and hence
the circumferential separation between generation and detection points increased when the
rotational stage was moved. Two different EMAT configurations were tested, one of which
was sensitive to the out-of-plane (OP) particle velocity, and other sensitive to the in-plane
(IP) particle velocity [23, 27], allowing sensitivity to different wavemodes [26, 31].
In order to identify the arrival times and frequency components of surface acoustic waves
11

SCC 1
SCC 2
Pitting
Scan position
Vertical position
(a) Sample A
(b) Surface map
Figure 4: Close-up of the region containing SCC in sample A (a), and surface plot showing
enhancement of the surface acoustic wave generated through interaction of the laser spot
source with defects (b).
12

SCC 3
Scan position
Vertical position
SCC 4
(a) Sample B
(b) Surface map
Figure 5: Close-up of the region containing SCC in sample B, and 2D surface plot showing
the surface wave enhancements with multiple defects and edge enhancements visible.
13

(a) Reference - OP
(b) Enhanced - OP
Figure 6: Sonograms for scan positions where the laser is far away (a) and fully illuminating
(b) the defect region; detecting with an EMAT sensitive to out-of-plane velocity.
a sonogram was again produced for each A-scan obtained using either the in-plane or out-of-
plane EMAT. Positions at which the laser source was located far away from the defect were
again considered as representative of a defect-free region of the pipe, and are shown for the
out-of-plane and in-plane cases in figures 6a and 7a respectively. For positions where the
laser source was fully illuminating the defect, extra signals at higher frequencies accompanied
by a visible change in the A-scan can also be observed in both cases (see figures 6b and 7b).
There are some differences between the data shown in figures 6 and 7, and that in
figure 2 for laser-laser measurements. The EMAT measurements are made with a varying
14

(a) Reference - IP
(b) Enhanced - IP
Figure 7: Sonograms for scan positions where the laser is far away (a) and fully illuminating
(b) the defect region; detecting with an EMAT sensitive to in-plane velocity.
15

separation between generation and detection points, and hence the wavemodes will arrive
at different times during the scan. The EMATs also measure particle velocity rather than
surface displacement. Finally, the expected differences between the wavemodes detected in
the in-plane and out-of-plane velocity components is clear [26].
The determination of the frequency ranges for analysis is more complicated for the EMAT
results, given the varying arrival time and the lower signal to noise ratio. The choice of
frequency range must consider the particle velocity sensor nature of the EMATs, and the
bandwidth limitations due to the coil width [32, 33]. The coil used in both EMATs has
a narrow but finite width, in this case approximately 1 mm. When the width of the coil
becomes comparable to the wavelength of the wave passing underneath the coil, the acquired
signal will not be a true representation of the particle velocity in the sample associated with
the wave propagation [31]. This effect is known as the spatial impulse response, and it has
been shown that the EMAT response for wavelengths less than one half of the coil width
will fall below 10% of the amplitude for the long wavelength limit [31], as shown in figure 8
(figure produced following the model presented in [31] for a sinusoidally varying wave). This
means that the EMATs used here will not detect significant signal amplitude for signals of
less than 1 mm wavelength (corresponding to a frequency of 3 MHz), and for wavelengths
around this value there will be resonant effects.
Once the regions of enhancement have been identified, avoiding regions where the fre-
quency is close to a minimum in the EMAT response, the same procedure to quantify the
enhancements as for the laser-laser measurements was used. The peak magnitudes in two
regions on the sonograms were tracked; the first of which was at 0.5-1 MHz and the second
at 1-2 MHz, with a window length of 4µs, with the window moved in time to match the
mode arrival time due to the increasing separation between generation and detection points.
The peak magnitudes for the low and high frequency regions are shown in figures 9a and
9b, for out-of-plane and in-plane velocities respectively.
As with the laser-laser measurements, both figures show some peaks and troughs, but
the exact enhancement region is not completely clear from each plot individually. Hence
the product of the peak magnitudes for both regions was again taken to improve the PoD.
16

1.0
0.8
0.6
0.4
0.2
0.0
Normalised EMAT response (arb.)
0.1
2 3 4 5 6 7 8 9
1
2 3 4 5 6 7 8 9
10
Log(wavelength (mm))
Coil width = 1 mm
Figure 8: EMAT response as a function of wavelength for a sinusoidally varying wave as the
wavelength is varied. The first minima occurs when the coil width is equal to the wavelength.
In this case, the signals were still noisy with a low PoD, and the out-of-plane and in-plane
products were also combined, as the presence of a defect should have some effect on both.
By obtaining the product of all four peak magnitudes a surface map was constructed. This
is shown in figure 10, and again the two defects were resolved (note that the scanned region
was not exactly the same as for the laser-laser measurements, and hence the pitting damage
is not shown).
4. Conclusions
When defects are partially closed and a laser is used to generate ultrasound, for the laser
spot source directly illuminating the defect the thermo-optic crack closure changes the gener-
ation conditions for the ultrasound and also alters the reflection and transmission behaviour
of the defect. This produces the higher order frequency components observed here which
have been used to track enhancements of the signal [20, 25]. The surface images produced
from the scans of the samples can be used to identify defects, either by visual inspection
of the data produced by the time-frequency analysis, or through automated identification
using suitable software [34].
Laser generation combined with detection using either laser interferometry or EMATs
17

15
10
5
0
Peak sonogram magnitude (arb.)
2520151050
Scan position (mm)
Low frequency
High frequency
(a) OP
50
40
30
20
10
0
Peak sonogram magnitude (arb.)
2520151050
Scan position (mm)
Low frequency
High frequency
(b) IP
Figure 9: Peak sonogram magnitudes for low and high frequency regions, detecting with an
EMAT sensitive to OP particle velocity (a), or IP particle velocity (b).
18

120x10
3
100
80
60
40
20
0
Peak sonogram magnitude (arb.)
2520151050
Scan position(mm)
(a) Product - OPIP
(b) Surface map - OPIP
Figure 10: Product of the OP and IP magnitude data, showing a notable improvement of
the S/N ratio (a), plus the surface plot that shows the presence of both SCCs on sample A.
19

has been shown to be capable of detecting and locating real stress corrosion cracks contained
in stainless steel pipe sections. The laser detection approach has the advantage that both
detector and source have a very small spatial footprint on the sample surface, thereby
allowing for the resolution of defects that are close together, as in sample A (figure 4b), where
defects are spaced by a few mm, or close to the sample edges, as in sample B (figure 5b).
This highlights the advantages of a laser-laser near-field approach, as the defects imaged
would be very difficult to resolve from the backwall signal for reflection based techniques.
The EMAT detection approach has the advantage of reduced cost when compared to the
laser interferometer, and EMATs are more easily implemented than the laser detection ap-
proach. With the information from the in-plane and out-of-plane data, access to information
about the interaction of extra wavemodes with the SCC is available. The EMAT approach
has some disadvantages, such as the reduced sensitivity, large spatial footprint and less clear
interpretation of the data. However, the laser-laser measurements can be used to inform the
results of the EMAT measurements, with EMATs used for practical implementation.
This technique shows promise with regards to characterising the defect geometry. Results
presented here measure the spatial extent of the defect at the surface of the sample, with
the level of enhancement observed related to the sample geometry [23, 24]. Measurements
of machined simulated defects on plates using the analysis presented here has shown that,
as the defect depth increases, so does the level of enhancement [24], and this has also
been observed recently using SCC-type defects grown to different depths. When using
Rayleigh waves, it has been shown that the enhancement is also sensitive to the angle of
propagation of the defect into the sample, with the in-plane and out-of-plane enhancement
used for characterisation [23]. Furthermore, once a defect is detected, the arrival times of
the reflected wavemodes can be used to identify geometry, and potentially detect branching
of the defect [35]. Combined with this detection using the signal enhancement, full defect
characterisation is feasible.
The technique has the limitation of requiring access to the pipe surface, and is sensitive
primarily to surface-breaking defects. We have presented measurements of the outer surface
of pipes, however, measurement of defects on the inner surface is possible through using
20

pipeline pigging. A full measurement of the pipe, including internal defects, is then possible
through a combination of this and other NDE techniques.
5. Acknowledgments
This work was funded by the European Research Council under Starting Independent
Researcher Grant 202735, NonContactUltrasonic.
References
[1] R. Cottis, Stress corrosion cracking, Tech. rep., The National Physical Laboratory (2000).
[2] B. Larson, Study of the factors affecting the sensitivity of liquid penetrant inspections: Review of
literature published from 1970 to 1998, Tech. rep., Center for aviation systems reliability (2002).
[3] P. Shull, Nondestructive evaluation. Theory, techniques and applications, CRC Press, 2002.
[4] H. Shaikh, N. Sivaibharasi, B. Sasi, T. Anita, R. Armirthalingam, B. Rao, T. Jayakumar, H. Khatak,
B. Raj, Use of eddy current testing method in detection and evaluation of sensitisation and intergranular
corrosion in austenitic stainless steels, Corrosion Science 48 (2006) 1462–1482.
[5] N. Yusa, S. Perrin, K. Mizuno, Z. Chen, K. Miya, Eddy current inspection of closed fatigue and stress
corrosion cracks, Measurement Science and Technology 18 (2007) 3403–3408.
[6] T. Chen, G. Tian, A. Sophian, P. Que, Feature extraction and selection for defect classification of
pulsed eddy current NDT, NDT&E International 41 (6) (2008) 467–476.
[7] S. Casalta, G. Daquino, L. Metten, J. Oudaert, A. V. de Sande, Digital image analysis of X-ray and
neutron radiography for the inspection and the monitoring of nuclear materials, NDT&E International
36 (5) (2003) 349–355.
[8] L. Babout, T. Marrow, D. Engelberg, P. Withers, X-ray microtomographic observation of intergranular
stress corrosion cracking in sensitised austenitic stainless steel, Materials Science and Technology 22
(2006) 1068–1075.
[9] D. Horner, B. Connolly, S. Zhou, L. Crocker, A. Turnbull, Novel images of the evolution of stress
corrosion cracks from corrosion pits, Corrosion Science 53 (2011) 3466–3485.
[10] P. Petcher, S. Dixon, Wideband ultrasonic time of flight diffraction combining b scans and cross sectional
imaging, in: AIP Conference proceedings 1096, 2009, pp. 650–657.
[11] P. Petcher, S. Dixon, Parabola detection using matched filtering for ultrasound b-scans, Ultrasonics 52
(2012) 138–144.
21

[12] A. K. Kromine, P. A. Fomitchov, S. Krishnaswamy, J. D. Achenbach, Laser ultrasonic detection of
surface breaking discontinuities: Scanning laser source technique, Materials Evaluation 58 (2) (2000)
173–177.
[13] J. Blackshire, S. Sathish, Near-field ultrasonic scattering from surface-breaking cracks, Applied Physics
Letters 80 (18) (2002) 3442–3444.
[14] I. Arias, J. Achenbach, A model for the ultrasonic detection of surface-breaking cracks by the scanning
laser source technique, Wave Motion 39 (1) (2004) 61–75.
[15] R. S. Edwards, S. Dixon, X. Jian, Depth gauging of defects using low frequency wideband Rayleigh
waves, Ultrasonics 44 (2006) 93–98.
[16] Y. Fan, S. Dixon, R. S. Edwards, X. Jian, Ultrasonic surface wave propagation and interaction with
surface defects on rail track head, NDT&E International 40 (2007) 471–477.
[17] X. Jian, S. Dixon, N. Guo, R. Edwards, Rayleigh wave interaction with surface-breaking cracks, Journal
of Applied Physics 101 (2007) 064906.
[18] S. Dixon, B. Cann, D. Carroll, Y. Fan, R. Edwards, Non-linear enhancement of laser generated rayleigh
waves by cracks, Nondestrutive Testing and Evaluation 23 (2008) 25–34.
[19] A. Clough, R. Edwards, Scanning laser source lamb wave enhancements for defect characterisation.,
NDT&E International Under review (2013) NA.
[20] H. Xiao, P. Nagy, Enhanced ultrasonic detection of fatigue cracks by laser-induced crack closure, Journal
of Applied Physics 83 (1998) 7453–7460.
[21] B. Dutton, A. Clough, M. Rosli, R. Edwards, Non-contact ultrasonic detection of angled surface defects,
NDT&E International 44 (2011) 353–360.
[22] B. Dutton, A. Clough, R. Edwards, Near field enhancements from angled surface defects; a comparison
of scanning laser source and scanning laser detection techniques, Journal of Nondestructive Evaluation
30 (2011) 64–70.
[23] M. Rosli, R. Edwards, Y. Fan, In-plane and out-of-plane measurements of rayleigh waves using emats
for characterising surface cracks, NDT&E International 49 (2012) 1–9.
[24] A. Clough, R. Edwards, Lamb wave near field enhancements for surface breaking defects in plates,
Journal of Applied Physics 111 (2012) 104906.
[25] S. E. Burrows, B. Dutton, S. Dixon, Laser generation of lamb waves for defect detection: Experimental
methods and finite element modelling, IEEE Transactions on ultrasonics, ferroelectrics and frequency
control 59 (2012) 82–89.
[26] J. Rose, Ultrasonic waves in solid media, Cambridge University Press, 1999.
[27] S. Palmer, S. Dixon, Industrially viable non-contact ultrasound, Insight 45 (2003) 211–217.
[28] C. B. Scruby, L. E. Drain, Laser ultrasonics. Techniques and applications, Adam Hilger, 1990.
22

[29] M. Klein, G. Bacher, A. Grunnet-Jepsen, D. Wright, W. E. Moerner, Homodyne detection of ultrasonic
surface displacements using two-wave mixing in photorefractive polymers, Optics Communications 162
(1999) 79 – 84.
[30] M. Niethammer, L. Jacobs, J. Qu, J. Jarzynski, Time-frequency representations of lamb waves, Journal
of the Acoustical Society of America 109 (5) (2001) 1841–1847.
[31] S. Dixon, S. Burrows, B. Dutton, Y. Fan, Detection of cracks in metal sheets using pulsed laser generated
ultrasound and emat detection, Ultrasonics 51 (1) (2011) 7–16.
[32] K. Kawashima, Quantitative calculation and measurement of longitudinal and transverse ultrasonic
wave pulses in solid, IEEE Transactions on sonics and ultrasonics 31 (2) (1984) 83–93.
[33] S. Dixon, C. Edwards, S. Palmer, High-accuracy non-contact ultrasonic thickness gauging of aluminium
sheet using electromagnetic acoustic transducer, Ultrasonics 39 (2001) 445–453.
[34] M. Rosli, R. Edwards, B. Dutton, C. Johnson, P. Cattani, Identifying surface angled cracks on alu-
minium bar using emats and automated computer system, in: D. Thompson, D. Chimenti (Eds.),
Review of progress in quantitative nondestructive evaluation, Vol. 29, Vol. AIP Conf. Proc. 1211, 2009,
pp. 1593–1600.
[35] F. Hernandez-Vale, A. Clough, R. Edwards, to be published.
23