You are on page 1of 9

Journal of Nondestructive Evaluation (2018) 37:49

https://doi.org/10.1007/s10921-018-0498-9

Coherence Weighting Applied to FMC/TFM Data from Austenitic CRA


Clad Lined Pipes
Ricardo Baiotto1 · Benjamin Knight-Gregson2 · Channa Nageswaran3 · Thomas Clarke1

Received: 3 July 2017 / Accepted: 4 June 2018 / Published online: 26 June 2018
© Springer Science+Business Media, LLC, part of Springer Nature 2018

Abstract
Lined pipe girth weld regions are challenging to inspect using ultrasonic techniques due to their anisotropic and inhomogeneous
characteristics, which lead to noise and acoustic beam distortion. This paper is focused on the application of coherence
weighting methods to decrease the background noise level and false indications on total focusing method (TFM) images,
encompassing the triple boundary of mechanically lined pipes. Three coherence weighting approaches were used here; the
coherence factor (CF), the sign CF (SCF) and the phase CF (PCF). All three were found capable to improve the TFM images,
however, the CF and the SCF performed better due to their exponential sensitivity factors, unlike the PCF which has a linear
sensitivity factor.

Keywords Mechanically lined pipes · FMC · TFM · Austenitic weld · Coherence weighting

1 Introduction 1.1 Lined Pipes

Austenitic welds are known to significantly impair ultrasonic Mechanically lined pipes are comprised of an external pipe
inspection [1–3], which is the case for the girth welds found made of C–Mn steel and an inner corrosion resistant alloy
on mechanically lined pipes. This kind of pipe is typically pipe [4,5]. A girth weld is the preferred union method
employed in the transportation of oil and gas, both onshore employed to join lined pipes segments. However this union
and offshore. Due to the inherent risk associated with the process is complex due to the dissimilar materials present.
handling of such materials, it is of great importance to non- The joint preparation method consists of the removal of the
destructively assess the integrity of lined pipes. Besides the liner from the extremity of the pipe, followed by a seal weld
problems due to the weld microstructure, ultrasonic inspec- between the liner and host pipe. After the seal weld, the
tion offers a feasible set of techniques to evaluate their girth remaining exposed inner surface of the C–Mn steel pipe is
welds, given that the anisotropy and inhomogeneity issues covered by an overlay weld. Only then the girth weld is cre-
are addressed. ated [6]. Figure 1 shows an illustration of a mechanical lined
pipe weld joint. This paper will focus on the triple boundary
region where the cladding meets the liner and the external
B Ricardo Baiotto pipe.
ricardo.baiotto@ufrgs.br During the life cycle of mechanical lined pipes, the over-
Benjamin Knight-Gregson lay cladding is subjected to corrosive fluids [7]. In addition,
ben.k-g@postgrad.manchester.ac.uk they undergo cyclic loading during installation [4,6,8] and,
Channa Nageswaran sometimes, during operation [9,10]. These conditions, asso-
channa.nageswaran@twi.co.uk ciated to possible manufacturing defects, can lead to crack
Thomas Clarke nucleation in the cladding region that can compromise the
tclarke@demet.ufrgs.br corrosion resistance and structural integrity of the instal-
1 Physical Metallurgy Laboratory (LAMEF) - PPGE3M, lation. Given that, it is important to be able to detect and
UFRGS, Porto Alegre, Rio Grande do Sul, Brazil characterise these defects.
2 TWI Technology Centre (Wales), TWI, Port Talbot, UK Linear array ultrasonic non-destructive methods are used
3
in the assessment of welds on a regular basis due to the rel-
NDT Department, TWI, Cambridge, Cambridgeshire, UK

123
49 Page 2 of 9 Journal of Nondestructive Evaluation (2018) 37:49

development as a form to predict the grain orientation of


coarse grained weld joints [18]. By using information such
as the weld geometry, electrode diameter and number of
weld passes, MINA was capable to repeatedly describe the
microstructure of the welds [18–20].
Nonetheless, the imaging method applied in this paper was
the total focusing method (TFM) [21–24]. Through delay
Fig. 1 Illustration of a typical girth weld on lined pipes, all the welds
are made using with corrosion resistant alloy consumables
and sum operations, TFM utilizes full matrix capture (FMC)
data in order to synthetically focus on every point of a previ-
ously defined grid of pixels inside the material [21–23]. The
atively high inspection productivity achieved [11]. In the FMC data set is the collection of the A-scans representing all
case of lined pipes, the consumables employed the welds the possible combinations of transmit and receive elements
need to match the corrosion resistance of the inner pipe, in an array [21,22]. Equation 1 can be used to calculate a
thus generally leading to an austenitic and coarse-grained TFM image in a configuration which involves a linear array
weld microstructure. These microstructure characteristics are transducer, coupled to the flat surface of a homogeneous and
known to have a detrimental effect upon ultrasonic non- isotropic test piece (and equally isotropic coupling medium
destructive methods [1,3,12], since coarse-grain structures such as a wedge or water layer) [21].
lead to increases in backscattered noise, attenuation and beam
distortion [1]. The first two effects are caused by acoustic  N N
   
energy being reflected back by the grain boundaries, whereas  
IT F M (x, y) =  Am,n tx,y,m + tx,y,n  . (1)
the third arises from beam skewing due to grain interfaces  
m=1 n=1
[1]. Since overlay claddings typically have a thickness of
only a few millimeters, it is thought that errors due to beam
skewing are less relevant for the work presented in this In Eq. 1, IT F M (x, y) is the intensity of the pixel located at
paper. (x, y) coordinate, Am,n is the A-scan taken from the FMC for
transmit m and receive n, N is the number of active elements
1.2 Imaging Method in the array, tx,y,m and tx,y,n are the time values it takes
for the sound to travel between the transmitting element m
A number of approaches have been developed to help and the focusing point and between the focusing point and
improve the ultrasonic imaging of coarse grain materials. the receiving element n, respectively. Due to the reciprocity
Among them there is the Decomposition of the Time Rever- principle, tx,y,m is independent of the propagation direction,
sal Operator, known by its French acronym DORT [13]. i.e., tx,y,m is the same regardless of whether the element is the
This method has been shown to be effective in improving transmitter or receiver. As such, the time of flight is defined
the signal to noise ratio (SNR) by isolating the contribu- by Eq. 2.
tion of single scattering from the signal [13,14] and in the
elimination of artifacts caused by surface guided waves [15]. 
Another approach proposed by Sharma et al. [16] makes use (X x,y,m
a − Xm
0 )2 + (Y a
x,y,m − Ym )
0 2
of the ensemble empirical mode decomposition to improve tx,y,m =
C1
SNR thus revealing indications from flat bottom holes in a B- 
scan obtained from an austenitic stainless steel sample. The (x − X x,y,m
a )2 + (y − Yx,y,m
a )2
main advantage of those methods is that they do not need + . (2)
C2
any sort of information about the microstructure of the test
piece.
However, some authors resort to computational modeling a
In Eq. 2, X x,y,m a
and Yx,y,m are the Cartesian co-ordinates
in order to assess how sound waves behave inside specific of the sound wave incidence point in the boundary between
coarse grain structures. CIVA is an example of commercial the two materials, X m0 and Y 0 are the Cartesian co-ordinates
m
software packages capable of simulating this kind of phe- of the center of the transducer, x and y are the co-ordinates
nomena. Using a semi-analytical beam computation feature of the focusing point, C1 and C2 are the velocities of the
of this software, Mahaut et al. [17] successfully simulated sound wave exited by the transmitting element in the coupling
propagation in a polycrystalline stainless steel by employing medium and in the test piece, respectively.
Voronoi cells with randomly distributed size and velocity. Figure 2 is intended to illustrate the TFM imaging pro-
Additionally, modelling tools such as MINA (Modelling cedure described above. In order to solve Eqs. 1 and 2
Anisotropy from Notebook of Arc Welding) have been under a
X x,y,m a
and Yx,y,m must be known beforehand. An analytical

123
Journal of Nondestructive Evaluation (2018) 37:49 Page 3 of 9 49

the PCF in (x, y) coordinates is based on the calculation of


the standard deviation of the phases and is determined by
Eq. 5. Similarly to Eq. 4, Eq. 6 provides the weighted inten-
sity map, I PC F (x, y)), for the PCF.

PC F(x, y)

σ (x, y) ∗ P σ (x, y) ∗ P
1 − p σ0 PC F , if p σ0 PC F < 1,
= (5)
Fig. 2 Illustration of the path taken by a beam in a media composed by 0, otherwise,
two materials with different sound velocities. The beam either arrives
or leaves transducer m, point (X m 0 , Y 0 ), passes through the interface I PC F (x, y) = PC F(x, y) × IT F M (x, y). (6)
m
between the two media at point (X x,y,ma , Yx,y,m
a ) and leaves or arrives
at the focusing point at (x, y). The test piece geometry in this illustration In Eq. 5, PC F(x, y) is the PCF calculated for (x, y)
is taken from the lined pipe girth weld region presented in Fig. 1 and
coordinates, while σ0 is the standard deviation of a random
the focusing point is the triple boundary
phase distribution between −π and π , the term σ p (x, y) is
an estimate of the phase diversity calculated from the phase
method to accomplish this by solving Snell’s law with Fer- standard deviation and PPC F is the PCF sensitivity. PPC F
rari’s method as described in detail by Weston et al. [25]. works in a similar way to PC F , controlling the attenuation
level of PC F(x, y). A special case of PCF, named SCF, that
1.3 Coherence Weighting takes into account only the phase polarity was also introduced
by Camacho et al. [28] and employed elsewhere [27,29].
The FMC/TFM approach is susceptible to the increased Since only those signals with the same polarity are consid-
backscattered noise and attenuation caused by the anisotropy ered to be fully coherent, the SCF is stricter than PCF and
and inhomogeneity similarly to other ultrasonic nondestruc- is expected to perform better in reducing the noise ampli-
tive methods. In order to minimize the attenuation and tude than PCF. Equation 7 is used to calculate the SCF as
backscattered noise in the TFM images, three weighting a function of the (x, y) coordinate, while Eq. 8 gives the
factors were investigated regarding their ability to suppress (I SC F (x, y) weighted intensity map likewise Eqs. 4 and 6.
noise. They are named coherence factor (CF), sign CF (SCF)
    PSC F
and phase CF (PCF).  
  1 N  N

The concept of CF was first reported by Mallart and Fink SC F(x, y) = 1 − 1 − sm,n (x, y)  ,
[26] and is defined as the ratio between the coherent intensity  N
m=1 n=1 
and the total incoherent intensity. Equation 3 adapted from
(7)
[27] calculates CF at every (x, y) point of the TFM image.
The parameter PC F is the CF sensitivity, which controls the I SC F (x, y) = SC F(x, y) × IT F M (x, y). (8)
relative attenuation between regions with different coherence
level. The multiplication of C F(x, y) by IT F M (x, y) results In Eq. 7, sm,n (x, y) is the phase sign for the A-scan rel-
in the weighted intensity map or IC F (x, y), as demonstrated ative to transmitter m and receiver n when focusing on the
by Eq. 4. pixel located in the (x, y) coordinates and PSC F is the sign
coherence sensitivity factor, which is the analogue of PC F
C F(x, y) and PPC F .
⎛  2 ⎞ PC F Since it is not possible to obtain a perfect intensity or
 N N 
Am,n (tx,y,m + tx,y,n )
⎜  m=1 n=1 ⎟ phase coherence in a situation where noise and skewing are
=⎝ N N ⎠ , (3) present, all values in the coherence matrices are below unity
N2 n=1 |Am,n (tx,y,m + tx,y,n )|
2
m=1 and they drop with the increase of sensitivity factors. The
IC F (x, y) = C F(x, y) × IT F M (x, y). (4) amplitude drop as a function of the sensitivity factor leads to a
more marked difference between the higher and lower ampli-
Camacho et al. [28] first presented SCF and PCF as tude pixels, which increases resolution. For CF and SCF the
an alternative to reduce the influence of grating and side amplitude drops exponentially towards zero and does not
lobes indications while increasing the lateral resolution in have an upper limit for PC F or PSC F above which the pixel
ultrasonic imaging. Their approach is based on the analy- amplitude is null. However, for PCF this is not the case,
sis of phase distribution and relies on the assumption that a according to Eq. 5; PCF amplitude of a given pixel drops
perfectly focused scatterer produces echoes with no phase linearly with the increase of PPC F , eventually reaching zero
σ (x, y) ∗ P
dispersion, while an unfocused scatterer tends to produce when p σ0 PC F = 1. Therefore, PCF images will lose
echoes with random phase distribution. The method to obtain pixels when PC F increases above the limiting value and fur-

123
49 Page 4 of 9 Journal of Nondestructive Evaluation (2018) 37:49

Fig. 5 Overall TFM image of the plate in the region with 0.8 mm deep
notch. The numbered arrow indicates the front wall (1), the back wall
(2), the thickness change corner (3) and the triple boundary and notch
(4)
Fig. 3 Schematic drawing of the plate used to collect the FMC data.
Dimensions are in millimeters
elements and a 0.3 mm element pitch. Data was acquired
from only 64 elements in order to keep the size of the data file
manageable. A M2M Multi++ array controller and the Multi
2000 software was used to control the probe and acquire
the data which was saved in FMC format to allow post-
processing it with a TFM algorithm. The phased array probe
was positioned above the plate, which was positioned with its
carbon steel side upwards in order to simulate an inspection
from the outside of a lined pipe. The incidence angle was
set to be approximately 11◦ . For a schematic drawing of the
inspection one can refer to Fig. 2.
Data post-processing was conducted in Matlab. The algo-
rithm first calculated the sound paths by use of Ferrari’s
Fig. 4 Overall image of the plate in the region with 0.8 mm deep notch. method as described by Weston et al. [25]. The TFM algo-
The numbered arrow indicates the front wall (1), the back wall (2), the rithm, based on Eqs. 1 and 2, was then applied with coherence
thickness change corner (3), the triple boundary (4) and the notch (5).
The scale shown at the bottom is in millimeters weighting factors calculated through Eqs. 3–8. This leads to
TFM images with and without coherence weighting; in addi-
tion to these, images of the coherence weightings themselves
ther increases in PC F will eventually leads to the complete will be presented in the next section.
loss of the image.

3 Results and discussion


2 Methodology
In Fig. 4 a macrograph of a cross section of the plate expos-
A specially manufactured plate was used in this work as ing the 0.8 mm notch below the triple boundary is shown.
mock-up in order to simulate the lined pipe triple bound- Figure 5 presents a simple TFM image of the same region
ary and cladding regions. A schematic drawing of the plate of the plate, obtained with Eqs. 1 and 2, in order to illustrate
can be found in Fig. 3. The plate was made of a 15 mm thick the standard appearance of the calculated images. Compar-
X65 steel plate half covered with a 2.5 mm thick 625 alloy ing Figs. 4 and 5 one can see that the TFM image presents
plate and half weld-overlay cladded with the same alloy. indications from the front and back wall (numbers 1 and 2),
Two notches, one approximately 0.8 mm deep and the other the thickness change corner (number 3) and the triple bound-
approximately 1.6 mm deep, were inserted via electrical dis- ary and notch (number 4). These images were obtained with
charge machining in the coated side just below the triple a conventional TFM algorithm with 0.2 mm grid resolution.
boundary. All the notches were 25 mm wide and around Figure 6 presents a close up view of the triple boundary
0.2 mm thick. In Fig. 4 one can find a macrograph of one and notch region of Fig. 4, obtained with a 0.1 mm step size
of the inserted notches. and presented with color scale in decibel (dB). In this figure
The ultrasonic inspections were conducted in an immer- the three strong indications in the center of the image refer
sion tank using a 10 MHz linear array probe with 128 to the triple boundary in the top, the notch tip in the middle

123
Journal of Nondestructive Evaluation (2018) 37:49 Page 5 of 9 49

Given that, one can observe that the indications around the
triple boundary in Fig. 6 are distorted in the vertical direction
in comparison with Fig. 4.
Figures 7, 8 and 9 display the output images generated
by Eqs. 4, 6 and 8, respectively, for varying sensitivity fac-
tors. The same data set used to obtain 5 and 6 was used
here and, similarly to Fig. 6, these images only include the
triple boundary region on a dB scale in decibels. Figure 7
shows that the multiplication of the image matrix by CF
greatly reduces the noise even for sensitivity factors as low
as one, however for higher PC F values the indication from
the tip of the notch starts to dim and is below −60 dB when
PC F = 4. Figure 8 shows a different behavior for the results
obtained from Eq. 6. In this case, for lower sensitivity val-
ues there is no significant improvement in comparison with
Fig. 6 TFM image around the region with a 0.8 mm notch. The scale
is in decibels Fig. 6 and for higher sensitivity values the indications from
the triple boundary and notch tip quickly disappear. Figure 9
shows that the SCF behave similarly to CF, although the noise
and the notch root in the bottom. The noise present in the drops faster for SCF than for CF and the three indications are
data is emphasized by the dB scale which will also be used present for a wider range of sensitivity factors. To avoid the
to evaluate the coherence weights in the following figures. It presence of low amplitude noise in the images, a threshold
is important to note that throughout this work there was no for pixel amplitude was introduced. For the images presented
attempt to correct for the distortion caused by the cladding.

Fig. 7 0.8 mm deep notch image of CF applied on the TFM image with various sensitivity factors (PC F ). The scale is in decibels

Fig. 8 0.8 mm deep notch image of PCF applied on the TFM image with various sensitivity factors (PPC F ). The scale is in decibels

123
49 Page 6 of 9 Journal of Nondestructive Evaluation (2018) 37:49

Fig. 9 0.8 mm deep notch image of SCF applied on the TFM image with various sensitivity factors (PSC F ). The scale is in decibels

Fig. 10 CF images of 0.8 mm deep notch region using various sensitivity factors (PC F ). The scale is in decibels

Fig. 11 PCF images of 0.8 mm deep notch region using various sensitivity factors (PPC F ). The scale is in decibels

in this paper, the threshold is −60 dB below the maximum the possibility of creating ultrasonic images without using
amplitude of each image. the amplitude of the signals directly.
Figures 10, 11 and 12 present the coherence weights Figures 13 and 14 present images of the different CFs
C F(x, y), PC F(x, y) and SC F(x, y) used to create Figs. investigated in the regions with a 1.6 mm deep notch and
7, 8 and 9, respectively. One can notice that both sets of without notch, respectively. Overall, the images follow the
images are similar, especially for higher sensitivity values trend presented by the previous images for the 0.8 mm deep
where the coherence weights become dominant. This raises notch. The sensitivity factors were selected in order to obtain

123
Journal of Nondestructive Evaluation (2018) 37:49 Page 7 of 9 49

Fig. 12 SCF images of 0.8 mm deep notch region using various sensitivity factors (PSC F ). The scale is in decibels

Fig. 13 TFM image (a), compared with CF (b), PCF (c) and SCF (d) for the 1.6 mm deep notch region. The scale is in decibels

Fig. 14 TFM image (a), compared with CF (b), PCF (c) and SCF (d) for a region without notches. The scale is in decibels

clear indications of the notch and triple boundary while located around the coordinates (31, 35 mm), are positioned
minimizing the amplitude of noise. However, two extra indi- around the back-wall of the sample or below. Since the
cations at the bottom of the image in the same depth position macrographs and visual inspection did not show any reflec-
as the root of the notch indication are seen in Fig. 13. Fur- tors in that region, it is assumed that those indications are
thermore, all the coherence weights images presented extra artefacts caused by the anisotropy and inhomogeneity of the
indications when there is only the triple boundary present cladding.
(Fig. 14b–d). The indications, other than the triple boundary,

123
49 Page 8 of 9 Journal of Nondestructive Evaluation (2018) 37:49

4 Conclusions 6. Yuan, L., Kyriakides, S.: Liner wrinkling and collapse of bi-
material pipe under bending. Int. J. Solids Struct. 51, 599–611
(2014). https://doi.org/10.1016/j.ijsolstr.2013.10.026
In the present paper the procedure to employ three coherence 7. Wilmot, D., Montague, P.: The suitability of CRA lined pipes for
weightings, namely CF, PCF and SCF was described. It was flowlines susceptible to lateral buckling. In: SUT Global Pipeline
shown that the three factors are capable of improving the Buckling Symposium, Perth (2011)
TFM image by reducing noise. However the three different 8. Toguyeni, G.A., Banse, J.: Mechanically lined pipe: installation
by reel-lay. In: Offshore Technology Conference, Rio de Janeiro
CFs presented here can be used only within a range of their (2012)
respective sensitivity values, i.e., too low a sensitivity value 9. Siskandarajah, T., Roberts, G., Rao, V.: Fatigue aspects of CRA
is not capable improving significantly the image, whereas lined pipe for HP/HT flowlines. In: Offshore Technology Confer-
too high a sensitivity value attenuates every indication but ence, Rio de Janeiro (2013)
10. Tkaczyk, T., Denniel, S., Pépin, A.: Fatigue and fracture of mechan-
the strongest one in the image (for PCF when PPC F is high
ically lined pipes installed by reeling. In: Offshore and Arctic
enough, all the indications are attenuated). The selection of Engineering, Rio de Janeiro (2012)
the proper sensitivity factors for an unknown specimen being 11. Schmerr, L., Song, S.J.: Ultrasonic Nondestructive Evaluation Sys-
inspected can be troublesome; in some cases, a reference tems Models and Measurements. Springer, New York (2007)
12. Limthongkul, S.: Validation of wave propagation in austenitic
testing block may be needed in order to determine an useful
welds using vibrometer. Master’s Thesis, Brunel University Lon-
range for those factors. Furthermore, the results obtained in don, London (2015)
this work exposed some limitations of the different CFs with 13. Shahjahan, S., Aubry, A., Rupin, F., Chassignole, B., Derode,
regard to their ability to suppress false indications. Besides A.: Improvement of flaw detection with ultrasonic array probes
in multiple scattering polycrystalline materials by means of a
the improvement in the noise level in all the cases, the pres- random matrix approach. In: 13th International Symposium on
ence of false indications can be misleading. Non-destructive Characterization of Materials. Le Mans (2013)
From the comparison between the images created through 14. Aubry, A., Derode, A.: Detection and imaging in a random medium:
the multiplication of the coherence weights matrices by the A matrix method to overcome multiple scattering and aberration.
J. Appl. Phys. 106, 044903 (2009)
TFM matrix and the images from the coherence weights 15. Villaverde, E.L., Robert, S.: Ultrasonic imaging of defects in
matrices themselves it is possible to conclude that, in the coarse-grained steels with the decomposition of the time reversal
present work, Eqs. 4, 6 and 8 have only limited effect upon operator. J. Acoust. Soc. Am. 140, 541–550 (2016)
the resulting images. This leads to the conclusion that, at least 16. Sharma, G.K., Kumar, A., Jayakumar, T., Rao, B.P., Mariyappa, N.:
Ensemble empirical mode decomposition based methodology for
for the present case, there is no need to calculate the TFM ultrasonic testing of coarse grain austenitic stainless steels. Ultra-
image and one could rely solely on the coherence weights sonics 57, 167–178 (2015)
images to inspect this sample. 17. Mahaut, S., Darmon, M., Chatillon, S., Jenson, F., Calmon, P.:
Recent advances and current trends of ultrasonic modelling in
Acknowledgements The authors would like to thanks the Brazilian CIVA. Insight 51, 78–81 (2009)
Governmental Agency Conselho Nacional de Desenvolvimento Cient- 18. Moysan, J., Apfel, A., Corneloup, G., Chassignole, B.: Modelling
fico e Tecnolgico (CNPq) and BG Group for partially sponsoring this the grain orientation of austenitic stainless steel multipass welds to
research through the Science Without Borders Program. improve ultrasonic assessment of structural integrity. Int. J. Press.
Vessels Pip. 80, 77–85 (2003)
19. Apfel, A., Moysan, J., Corneloup, G., Fouquet, T., Chassignole,
B.: Coupling an ultrasonic propagation code with a model of the
heterogeneity of multipass welds to simulate ultrasonic testing.
References Ultrasonics 43, 447–456 (2005)
20. Mark, A.F., Fan, Z., Azough, F., Lowe, M.J.S., Withers, P.J.: Inves-
1. Dissimilar Consortium: Phased Array Ultrasonic Inspec- tigation of the elastic/crystallographic anisotropy of welds for
tion of Dissimilar Metal Joints. Report, TSB Project No: improved ultrasonic inspections. Mater. Charact. 98, 47–53 (2014)
P11/MFE/6/IAA058J (2011) 21. Holmes, C., Drinkwater, B.W., Wilcox, P.D.: Post-processing of
2. Fan, Z., Mark, A.F., Lowe, M.J.S., Withers, P.J.: Nonintrusive esti- the full matrix of ultrasonic transmit-receive array data for non-
mation of anisotropic stiffness maps of heterogeneous steel welds destructive evaluation. NDT E Int. 38, 701–711 (2005). https://
for the improvement of ultrasonic array inspection. IEEE Trans. doi.org/10.1016/j.ndteint.2005.04.002
Ultrason. Ferroelectr. Freq. Control 62, 1530–1543 (2015). https:// 22. Nakahata, K., Tokumasu, S., Sakai, A., Iwata, Y., Ohira, K., Ogura,
doi.org/10.1109/TUFFC.2015.007018 Y.: Ultrasonic imaging using signal post-processing for a flexible
3. Ye, J., Moysan, J., Song, S., Kim, H.J., Chassignole, B., Gueudre, array transducer. NDT E Int. 82, 13–25 (2016). https://doi.org/10.
C., Dupond, O.: Influence of welding passes on grain orientation— 1016/j.ndteint.2016.04.002
the example of a multi-pass v-weld. Int. J. Press. Vessels Pip. 93–94, 23. Sutcliffe, M., Weston, M., Dutton, B., Charlton, P., Donne, K.: Real-
12–21 (2012). https://doi.org/10.1016/j.ijpvp.2012.02.007 time full matrix capture for ultrasonic non-destructive testing. NDT
4. Yuan, L., Kyriakides, S.: Plastic bifurcation buckling of lined pipe E Int. 51, 16–23 (2012). https://doi.org/10.1016/j.ndteint.2012.06.
under bending. Eur. J. Mech. A 47, 288–297 (2014). https://doi. 005
org/10.1016/j.euromechsol.2014.04.005 24. Felice, M.V., Velichko, A., Wilcox, P.D.: Accurate depth measure-
5. Denniel, S., Tkaczyk, T., Pepin, A.: Reeled mechanically lined ment of small surface-breaking cracks. NDT E Int. 68, 105–112
pipe: cost efficient solution for static and dynamic applications in (2014). https://doi.org/10.1016/j.ndteint.2014.08.004
corrosive environment. In: Deep Offshore Technology Conference,
Perth (2012)

123
Journal of Nondestructive Evaluation (2018) 37:49 Page 9 of 9 49

25. Weston, M., Mudge, P., Davis, C., Peyton, A.: Time efficient 28. Camacho, J., Parrilla, M., Fritsch, C.: Phase coherence imaging.
auto-focussing algorithms for ultrasonic inspection of dual-layered IEEE Trans. Ultrason. Ferroelectr. Freq. Control 56(5), 958–974
media using full matrix capture. NDT E Int. 47, 43–50 (2012) (2009). https://doi.org/10.1109/TUFFC.2009.1128
26. Mallart, R., Fink, M.: Adaptive focusing in scattering media 29. Zhang, J., Drinkwater, B.W., Wilcox, P.D.: Comparison of ultra-
through sound-speed inhomogeneities: the van Cittert Zernike sonic array imaging algorithms for nondestructive evaluation. IEEE
approach and focusing criterion. J. Acoust. Soc. Am. 96(6), 3721– Trans. Ultrason. Ferroelectr. Freq. Control 60, 1732–1745 (2013)
3732 (1994). https://doi.org/10.1121/1.410562
27. Knight-Gregson, B., Peyton, A., Scully, P., Weston, M., Davis, C.:
Improvement of ultrasonic cross-sectional imagery via the applica-
tion of coherence weighting and the total focussing method. In: 55th
Annual Conference of the British Institute for Non-Destructive
Testing, Nottingham (2016)

123

You might also like