Journal of Nuclear Science and Technology

ISSN: 0022-3131 (Print) 1881-1248 (Online) Journal homepage: http://www.tandfonline.com/loi/tnst20

Development of a hybrid ECT sensor for JSFR SG

double-wall tubes

Toshihiko Yamaguchi & Ovidiu Mihalache

To cite this article: Toshihiko Yamaguchi & Ovidiu Mihalache (2017): Development of a hybrid
ECT sensor for JSFR SG double-wall tubes, Journal of Nuclear Science and Technology, DOI:
10.1080/00223131.2017.1359116

To link to this article: http://dx.doi.org/10.1080/00223131.2017.1359116

Published online: 13 Sep 2017.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=tnst20

Download by: [Australian Catholic University] Date: 13 September 2017, At: 23:36
 JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY, 
https://doi.org/./..

ARTICLE

Development of a hybrid ECT sensor for JSFR SG double-wall tubes

Toshihiko Yamaguchi and Ovidiu Mihalache
Japan Atomic Energy Agency, Section of Fast Reactor Reserch and Development, Tsuruga-shi, Japan

ABSTRACT ARTICLE HISTORY

A new design has been adopted for the steam generator (SG) tubes of the Japan Sodium-cooled Received  September 
Fast Reactor (JSFR) using double-wall tubes. This paper estimates and assesses the effectiveness Accepted  June 
of detecting defects in SG double-wall tubes of the JSFR by using combined high-frequency eddy
KEYWORDS
current testing (ECT) and low-frequency remote field eddy current sensors. We confirm that the Inspection; numerical
proposed hybrid ECT sensor is highly sensitive to small defects, fatigue cracks, and other defects simulation; experiment; fast
even when located under support plates of tubes. The parameters of the hybrid ECT sensor are breeder reactor; steam
designed and optimized to detect small defects using accurate numerical simulations based on generator; Cr-Mo steel;
Downloaded by [Australian Catholic University] at 23:36 13 September 2017

the finite element method, using an in-house developed code. The sensitivity and high perfor- remote field ECT; fatigue
mance of the hybrid ECT sensor was validated with experimental measurements. crack; support plate

1. Introduction small defects on the outer surface of SG double-wall

tubes, even for larger air-gap between walls. Also, one of
The Japan Atomic Energy Agency has designed and
the main inspection targets is the detection of defects on
conducted intensive research to develop an advanced
the outer surface of tubes with depths 20% of the tube-
loop-type fast breeder reactor (FBR) known as the Japan
wall thickness.
Sodium-cooled Fast Reactor (JSFR) [1].
Until now, for single-wall tubes, conventional high-
New and innovative technologies will be used in
frequency ECT has been used successfully to detect
the JSFR to support economic competitiveness in the
defects located on the inner/outer surface of non-
coming decades by generating carbon-free energy while
magnetic tubes or the inner surface of magnetic
continuing to enhance nuclear reactor reliability and
tubes, while low-frequency remote field ECT (RF-ECT)
safety. One of the key concepts in the JSFR design
proved to be sensitive for both inner and outer sur-
is to reduce the amount of plant commodities by
faces of ferromagnetic tubes with high relative mag-
decreasing the number of cooling loops. This is facili-
netic permeability [7–10]. By using both probes in the
tated by using high-chromium steels with low thermal-
same time (ECT and RF-ECT) for detection of defects
expansion coefficients and high-temperature strength,
in ferromagnetic tubes, it is possible to make the dis-
therefore shortening the piping length.
tinction between detection of an inner or an outer tube
The steam generator (SG) unit consists of thousands
defect. However, in the ISI of ferromagnetic SG tubes,
of long, straight tubes made of 9Cr-1Mo steel, bundled
these sensor units (RF-ECT and ECT) are unconnected,
in a cylindrical steel vessel with high-pressure water
spatially separated, and scan the SG tube surfaces inde-
and steam running through them and low-pressure hot
pendently of each other. This results in different noise
sodium flowing inside the vessel. The two walls (inner
distributions for each sensor unit (for example due to
and outer) of the SG double-wall tube are separated by
lift-off, sensors’ speed, and others), and therefore makes
a very small gap of less than 10 μm [1–2].
it difficult to use combinations of sensors’ signals to fur-
Ferromagnetic SG double-wall tubes, composed of
ther suppress the noise levels or to correlate their com-
inner and outer tubes in very close proximity, have
bined signals to the same location of a defect.
undergone continuous development in recent decades
In the new hybrid ECT approach, these sensors are
[1] in order to ensure highly reliable SGs in FBRs. How-
combined in the same location to improve the ISI of
ever, the gap between the walls should interfere with
FBR SG double-wall tubes. The two signals are acquired
the use of other nondestructive testing (NDT) tech-
simultaneously at the same location and with similar
niques as well as electromagnetic NDT ones. There-
noise distributions, creating new possibilities for fur-
fore, for ferromagnetic SG double-wall tubes, it is nec-
ther signal processing to enhance defect detection. The
essary to investigate and develop new methods and sen-
hybrid ECT sensor operation point is adjusted in a
sors for in-service inspection (ISI) technology based
way to better detect and differentiate the defect type
on eddy current testing (ECT) [3–6]. These methods
(if it is located on inner or outer tube surface) while
should confirm the ECT sensors’ sensitivity to detect

CONTACT Toshihiko Yamaguchi [email protected]

©  Atomic Energy Society of Japan. All rights reserved.
 2 T. YAMAGUCHI AND O. MIHALACHE

maximizing also sensors’ sensitivity by reducing signal

interferences between sensors.
The geometry of the RF-ECT coil system was pre-
viously optimized for FBR SG tubes with a single-wall
[11–13]. This was done using either two-dimensional
(2D) or three-dimensional (3D) finite element method
(FEM) numerical simulations to enhance the detection
of small defects located near or under the tube support
plates (SPs).
In the present paper, the authors investigate the feasi-
bility of applying a combination of high-frequency ECT
and low-frequency RF-ECT to the ISI of ferromagnetic
SG double-wall tubes. The hybrid ECT sensor has both
a direct near field and an indirect remote field ECT sys-
tem integrated into a single unit. Both systems are opti-
mized specifically for the ferromagnetic SG double-wall
tubes of the JSFR.
Downloaded by [Australian Catholic University] at 23:36 13 September 2017

The paper shows that the current ECT technology

used in ISI for FBR SG tubes is also adequate for detect-
ing defects located on both surfaces of inner or outer
tube walls of an SG double-wall tube. Such defects
should be detected with almost the same sensitivity,
regardless of any variation in the air gap between the Figure . Schematics of steam generator and double-wall tube.
tube walls or other noise effects.
SG double-wall tubes must be confirmed periodically
to ensure their integrity and the high reliability of the
2. New approach sensor for JSFR SG tubes heat-transfer mechanism.
SG double-wall tube integrity can be assured in two
2.1. Steam generator with double-wall tube ways. The first method is based on the continuous mon-
In JSFR, the SGs must be mechanically and structurally itoring of inner or outer tube leaks during SG opera-
designed to take into account the sodium–water reac- tion. The second method requires periodic, time-based
tion in the event of a tube failure. Also, they must main- inspection of the tubes during plant shutdown.
tain a high and reliable heat transfer through their tube The main causes of SG tube failure are considered to
walls from the hot liquid sodium, that is outside the be (a) the cracks initiation caused by a departure from
tubes and that is heated by the reactor core, to the nucleate boiling temperature fluctuations, (b) the tube
steam-water that is inside the tubes and that drives fretting wearing against the tube SPs, and (c) the tube
the electrical turbine. The main geometrical character- fatigue due to flow-induced vibrations [14]. Therefore,
istics and electromagnetic properties of SG tubes are any tube inspection system must be able to confirm the
presented in Table 1. Numerous SG tubes are bun- tube soundness by means of volumetric testing for arti-
dled inside a stainless-steel cylindrical vessel contain- ficial and fatigue cracks located on the outer surface
ing a large amount of hot sodium at temperatures above of outer-wall tube of the double-wall tubes and defects
500 °C. In the double-wall tube structure design, even located near or under the tube SP.
if one side of a tube wall is pitted, cracked, or even The newly implemented JSFR design of an SG
fails, the other tube wall prevents water from leaking with double-wall tubes minimizes the consequence of
into the sodium, as shown in Figure 1. However, the sodium–water reactions if the heat-transfer tube is pen-
soundness of both the inner and outer tube walls of etrated by the cracks, and also reduces the probability of
such tube failures occurring.
Table . Design specifications of SG double-wall tubes.
Items Values 2.2. RF-ECT inspection technique
SG tube type Double-wall (straight Volumetric inspection techniques have been widely
tube type)
Outer-wall tube: outer-diameter/thickness ./. (mm)
developed for the austenitic stainless-steel heat-transfer
Inner-wall tube: diameter/thickness ./. (mm) tubes of SGs. High-frequency ECT is the most conven-
Double-wall outer/inner diameter ./. (mm) tional technique used in the ISI of SG tubes of pres-
Tube material Cr-Mo steel
Tube SP Cr-Mo/stainless steel surized water reactors [15,16]. However, standard high-
Electrical conductivity . (MS/m) frequency ECT is not adequate to detect defects located
Relative magnetic permeability 
(estimated) on the outer surface of ferromagnetic material tubes.
This is because the high relative magnetic permeability
 JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY 3

of the tubes results in a lower skin depth at high current

frequencies. However, low-frequency RF-ECT [17] can
be used for ferromagnetic tubes made of 2.25Cr-1Mo
steel. This technology has been developed and widely
validated by both 2D and 3D FEM numerical simu-
lations and experimental measurements at the Monju
FBR for their SG single-wall tubes [18,19].
In the JSFR, the SG double-wall tubes are thinner
and made of 9Cr-1Mo steel that also has ferromagnetic
properties but with a different relative magnetic per-
meability than that of 2.25Cr-1Mo steel. However, for
future ISI of SG double-wall tubes using ECT, defects
located on both inner and outer tube wall should be
detected with the same or better accuracy than defect
detection in SG single-wall tubes of an FBR. It is well
known that ECT sensors are more sensitive to the detec-
tion of defects located on the inner surface than on the
Downloaded by [Australian Catholic University] at 23:36 13 September 2017

outer surface of magnetic SG tubes. Therefore, by focus-

ing in designing an RF-ECT sensor to address mainly
the issues of high sensitivity detection of defects located
on outer surface of magnetic tubes will implicitly assure Figure . Examples of multi-coil configurations: vertical type
at least the same sensitivity of ECT sensors for defects (upper), horizontal type (lower).
on inner surface of tubes.

Each channel represents a differentially connected two-

coil system for eddy current detection in the multi-coil
2.3. Hybrid ECT sensor: integration of multiple
approach. Straight SG double-wall tubes were used as
RF-ECT and conventional ECT sensors
the test specimens, with similar geometrical and elec-
An RF-ECT bobbin-type probe has been developed for tromagnetic characteristics to those of the JSFR SG
the ferromagnetic SG tubes of the Monju reactor [11– tubes. Artificial defects were machined into each test
13]. This probe is sensitive to defects located on the specimen using electrical discharge machining (EDM).
outer surface of single-wall tube such as fully circumfer- The RF-ECT signal depends on the geometry and
ential grooves that resemble the fretting wear near the size of the detection coils. The authors designed the
tube SP. However, it is less good at detecting small par- detection coils using various configurations, shown in
tial cracks. Figure 2, based on multiple vertical and horizontal coils
For the ISI of SG double-wall tubes, the authors by using an FEM simulation code. These multiple ver-
have investigated a new sensor that is also based on a tical/horizontal geometries were optimized to detect
multiple-coil RF-ECT detection system. This uses the smaller defects situated on the outer surface of double-
indirect field to enhance signal detection when the sen- wall tubes, while detection of the same defects located
sor is located in the remote field area. By comparison, a on the inner surface of tubes has at least the same
conventional ECT sensor is located in the direct path of sensitivity. Correspondingly, the diameter and number
the electromagnetic field transmission. of turns of the copper wire in the multiple coils are
In the multi-coil approach, the sensor is more sen- proportional to their size. Increasing the size of the
sitive to defects located between the coils as the num- multiple coils means that fewer coils could be phys-
ber of channels (coils) increases. However, smaller coils ically arranged in the limited space available inside
weaken the raw threshold-detection signal for defects. the tubes.
Therefore, its signal becomes more susceptible to elec- The new sensor (which is referred to as the hybrid
tronic noise when amplified. The appropriate coil size ECT one) is based on both indirect electromagnetic
and the number of channels were obtained by FEM field of the remote field ECT technique and direct elec-
numerical simulations and after calibration with the tromagnetic field in the standard ECT, i.e. using both
experimental minimum signal threshold. The latter was fields simultaneously. A 3D view of the hybrid ECT sen-
obtained from previously designed and experimentally sor, including the ferrite cores located before and after
measured [19,20] multiple RF-ECT coil systems for SG each coil in both excitation and detection unit, is shown
tubes made of 2.25Cr-1Mo steel, as in the Monju FBR. in Figure 3. The outer diameter of the sensor is 11.2 mm,
To investigate the indirect field sensitivity, a basic which leaves a gap for insertion in the double-wall tube
test was performed using a double-wall tube to opti- whose inner diameter is 12.8 mm, as listed in Table 1.
mize the coil arrangement and the number of channels. In the hybrid ECT sensor, the standard ECT sensor is
 4 T. YAMAGUCHI AND O. MIHALACHE

FEM simulations, using meshes with up to 650,000

triangles, were conducted to assess the performance of
the multi-coil system operating in the RF-ECT mode.
Various FEM meshes were employed, with sizes vary-
ing between 450,000 to 650,000 triangles for each wall
gap size. Without this approach, FEM meshes could
easily increase above several million triangles after
including all possible wall gap values, greatly increas-
ing the simulation time in further extensive FEM
investigations.
Compared with SG single-wall tubes in the Monju
Figure . D view of the hybrid ECT sensor. FBR, the electromagnetic properties of the JSFR double-
wall tubes are different because of different tube mate-
rials. The electrical conductivity decreases from 3.5
located in the middle part of the multi-coil RF-ECT to 1.0 MS/m, while the relative magnetic permeabil-
sensors. ity increases from 95 to 232. These values were deter-
The direct field sensor is used to detect defects mined using a separated calibration method based on
Downloaded by [Australian Catholic University] at 23:36 13 September 2017

located under the double-wall tube SP, and also to direct comparison between 2D FEM axisymmetric sim-
classify/discriminate between inner and outer tube ulations and experimental measurements using simple
defects. The indirect field is used to detect small defects bobbin-type coils without ferrite cores. The RF-ECT
(defects could be initiated due to SG tube failure modes signals from bobbin-coils for various inner and outer
described previously in Section 2.1) on the outer sur- tube defects were computed at frequencies between 100
face of the SG double-wall tube. The sensor operation and 1000 Hz and compared with experimental mea-
point is chosen in a way to minimize the interference surements in double-wall SG tubes, in which the peak
between the direct and indirect fields. The optimized amplitude occurred between 500 and 600 Hz. While in
configuration was determined by FEM numerical sim- literature [25], for 9Cr-1Mo alloys, previous reported
ulations presented in the next section. values for electrical conductivities were from 1.69 to
1.83 MS/m and relative magnetic permeability were
from 37 to 133, the authors founded different values for
electromagnetic properties of double-wall SG tubes that
3. FEM numerical simulation and optimization could best fit the measured defects signals in the double-
of hybrid ECT sensor for JSFR SG tubes wall SG tubes used in the paper. The electromagnetic
characteristics of the ferrite cores used in the design of
3.1. Procedure of numerical simulations
the hybrid ECT sensor are 1.7 MS/m (electrical con-
The numerical simulations were conducted using an ductivity) and 17,000 (relative magnetic permeability).
electromagnetic FEM code that was developed at the The current density in the FEM simulations of the entire
Japan Atomic Energy Agency. It was benchmarked [21– excitation system was fixed to 1 A/mm2.
23] for the specific conditions of FBR SG tubes with
theoretical solutions, experimental measurements, and
3.2. FEM simulations for hybrid ECT sensor
also with other commercial FEM software [24].
The FEM simulation code is based on a 2D approx- A schematic of the RF-ECT system is presented in
imation using an axisymmetric geometry that is well Figure 4. The coil parameters P = (A, B, C, D, E, F)
suited for the long, straight geometry of SG tubes. In were chosen based on various FEM simulations and
the FEM simulations, the excitation coil system and parameterizations, using a brute-force attack in which
the tube-defect geometry were assumed to obey the 2D the parameters’ size was changed in steps dP = (1, 2,
axisymmetric requirements, while the detection excita- 2, 0.5, 0.5, 0.5) mm but subjected to the following con-
tion system was 3D. The latter does not interfere with straints (listed in Table 2) resulting from either manu-
2D FEM solution because it is assumed that the detec- facturer requirements or other considerations based on
tion coils do not represent a source for eddy currents. practical experimental measurements. Sensor dimen-
By neglecting eddy currents inside wires of detection sions of smaller than 1 mm are difficult to manufacture
coils, with a zero electrical conductivity and air relative while those above 80 mm result in over-sized configu-
magnetic permeability, there are no additional source- rations.
current terms in the electromagnetic equations to The goal of the FEM simulations was to find the
violate the 2D approximation. Therefore, the electro- proper sensor configuration that provides a large signal-
magnetic force signal induced in the detection coils is to-noise (S/N) ratio for detecting small defects when
calculated in the 3D geometry by integrating the mag- subjected to various effects. In the first analysis, based
netic field potentials around the coils. on experimental measurements (as shown in later
 JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY 5

Figure . Schematic of the sensors and tube parameters to be optimized for increased S/N ratio.

Table . Constraints on the sensor parameters (A, B, C, D, E, F) in minimum depending on the initial starting iteration
the optimization procedure. point for sensor parameters.
Parameter Value (mm) Reasons Examples of various configurations and their S/N
ratios are listed in Table 3, while the design parameters
A  < A <  lower, higher limit due to manufacturer
requirements (A, B, C, D, E, F) were chosen based on the medium
B  < B <  lower limit to ensure stronger values obtained in the FEM simulations. The S/N ratio
excitation field
higher limit to avoid over-sized sensor of the designed sensor is shown in Figure 5 for the real
C  < C <  lower limit to ensure RF-ECT operation part of the signal, in which the signal is rotated in such
Downloaded by [Australian Catholic University] at 23:36 13 September 2017

higher limit to avoid over-sized sensor

D <D< lower limit due to manufacturer a way in the Lissajous figure that the imaginary part of
requirements the signal is minimized and discarded as noise while its
E <E< higher limit to avoid over-sized sensor
F . < F < .
real part is maximized and used in further analysis.
The present FEM simulations of the hybrid ECT
sensor were concerned only with the design of the
chapters in Figures 13, 17), it was recognized that the multi-coil RF-ECT sensors, while the bobbin-coil ECT
signal from the inner-tube ripple is the one that limits operating point was assumed to not interfere with RF-
the performance of the sensor. The inner-tube ripples ECT signals. This assumption was proved to be cor-
are due to the manufacturing process of tubes result- rect (by decreasing the amplitude of ECT signal to
ing in a regular helical pattern variation of tubes-wall the point at which its interference with RF-ECT signal
thickness. When the cylindrical surface of the tube is could be neglected) based on experimental measure-
unfolded to a flat geometry, the tube-wall thickness dis- ments and practice, shown later in the next section in
tribution will have a stripe pattern-shape with constant Figure 15.
distance between stripes and aligned at a constant angle The newly designed RF-ECT sensor was next inves-
to the axial direction of tubes (the pattern is similar with tigated specifically using FEM in order to ensure its
the shape of signal showing later in Figure 15). proper operation point in varying conditions as in
In numerical simulations, the S-signal represents the experimental measurements.
amplitude of the signal from a defect in an SG tube
(slit 0.5 mm wide and 20%tw), while the N-noise is
the amplitude signal from the inner-tube ripples (depth
3.3. Effect of inner-tube ripple size
approximately 50 μm). In the FEM simulations, the
inner-tube ripples were modelled by approximating the In FEM simulations, the maximum variation of the
changes in the inner radius of tube with a square-wave SG inner-tube diameter was considered to be 50 μm,
shape (as in Figure 4) in the axial direction of tube and approximating the inner-tube ripple depth. After cali-
constant depth along circumferential direction. Since brating the FEM simulations with measurements, the
the length of the inner-tube ripples was not known, the relative amplitude of the signals due to inner-tube rip-
FEM simulations took into consideration the following ples could be estimated.
lengths: 12.5, 15, 17.5, and 20 mm. They also tried to The maximum influence of various ripple configu-
find the sensor parameters (A, B, C, D, E, F) that result rations was assessed by considering its geometry as in
in a high and uniform S/N ratio, mostly independently Figure 6, with ripple lengths ranging from 5 to 30 mm
of ripple size. The FEM simulations showed that it is with a 5 mm step size. Experimental visual confirma-
difficult to obtain a regular pattern of S/N behaviour, tion with various SG tubes showed several variations in
with a single-sensor configuration giving the best S/N the ripple length. Therefore, the hybrid ECT sensor is
ratio. Instead, it was found that there are many possible expected to exhibit a high S/N ratio not only for a spe-
combinations of the parameters that show similar S/N cific ripple length but for a larger range variation.
ratios. Finding these cases requires either a brute-force Figure 6 shows the FEM numerical simulations
attack (that can be accelerated by using clusters of cen- of S/N ratio for an outer defect (OD) narrow full-
tral processing units (CPUs) for calculations) to inves- circumferential slit that is 0.25 mm wide and 20%tw. A
tigate all possible sensor parameters combinations (A, high and approximately uniform S/N ratio is obtained
B, C, D, E, F), or gradient step algorithms that do not in all cases at an operating frequency of 1 kHz for the
converge to the global minimum but instead to a local RF-ECT sensors.
 6 T. YAMAGUCHI AND O. MIHALACHE

Table . Examples of various sensor configurations and designed sensor

parameters.
Configuration Ripple size (mm) Configuration A—B–C: D—E–F (mm) S/N ratio

 . . – . – .:. – . – . .

 . . – . – .:. – . – . .
 . . – . – .:. – . – . .
 . . – . – .:. – . – . .
 . . – . – .:. – . – . .
 . . – . – .:. – . – . .
 . . – . – .:. – . – . .
 . . – . – .:. – . – . .
Designed sensor . . – . – .:. – . – . .
Downloaded by [Australian Catholic University] at 23:36 13 September 2017

Figure . FEM signal from OD %tw slit and noise due to inner-tube ripples for the designed sensor (S/N = .).

Figure . Influence of the inner-tube ripple size on the S/N ratio for OD %tw full-circumferential slit.
 JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY 7
Downloaded by [Australian Catholic University] at 23:36 13 September 2017

Figure . Influence of double-wall gap on the S/N ratio for OD %tw full-circumferential slit.

3.4. Effect of SG double-wall tube gap size amplification factor of 0.54 × 106 was required, while
the shape of the signal in the Lissajous representa-
In the next step, the influence of the SG double-wall
tion had to be rotated to 45°. In Table 4, the S/N
tube gap size was assessed by FEM simulations. This
ratio validated at two different sizes of inner-tube rip-
effect is important since the wall gap varies in the actual
ple (15 and 17.5 mm) is given for three outer-tube
experimental case, and therefore a uniform or low-
defects. Based on FEM simulations and experimental
variation S/N ratio is desired. While a maximum value
measurements for small defects (as shown in Figure
of 10 μm is expected for the wall gap in the actual SG
13) and fatigue cracks (as shown in Figure 17), the
double-wall tube, even larger values were considered in
confirmation of ECT signal variation and its S/N ratio
the FEM simulations to obtain a better understanding
showed that a ripple size of 15–17.5 mm in the sim-
of the sensor operation.
plified geometrical model can be considered as giv-
In the simulations, a fixed 15 mm inner ripple size
ing the closest results between simulations and mea-
was assumed, while a variable wall gap ranging from
surements as in Figure 9. FEM numerical simula-
0 to 15 μm was used. The proposed sensor operat-
tions of the S/N ratio showed very good agreement
ing at 1 kHz exhibits a large S/N ratio, as shown in
with experimental measurements, confirming the RF-
Figure 7.
ECT sensor sensitivity. Furthermore, it also confirmed
the appropriateness of the chosen parameters of the
3.5. FEM simulations for defects and validation inner-tube ripple size, i.e. 15–17.5 mm and 50 μm
with experimental measurements depth.
The designed multiple RF-ECT coils were manufac-
The sensitivity of the hybrid ECT sensor for various
tured based on the analysis results and are described
outer-tube defects with narrow and broad openings was
in Table 5 and Figure 10. By taking into consideration
simulated by FEM at 1 kHz and a 5 μm wall gap. The
the design specifications of the tubes for actual use in
results are illustrated in Figure 8 for a 15 mm inner-
the JSFR, a core adjustment mechanism was installed,
tube ripple size. For larger open defects, the noise due
whereby each channel of the multiple RF-ECT coils
to inner-tube ripples decreased, confirming the sensor
was connected to a 50 m long cable inserted through
performance.
the SG tube. The structure and an external view of the
In order to calibrate the FEM simulations with
manufactured hybrid ECT sensor (including multiple
experimental measurements, the simulations had to
RF-ECT coils and a conventional ECT one) are shown
be changed according to the measurements. An
in Figure 11.
 8 T. YAMAGUCHI AND O. MIHALACHE

Table . Validation of S/N ratio computed using FEM with experimental measurements.
Defect Simulated S/N ratio (ripple  mm) Simulated S/N ratio (ripple . mm) Experiment S/N ratio

OD %tw,  mm . . .

OD %tw, . mm . . .
OD %tw, . mm . . .
Downloaded by [Australian Catholic University] at 23:36 13 September 2017

Figure . FEM simulations of S/N ratio for OD %tw with various widths.

Figure . Validation of FEM simulations (ripple =  mm, . mm) with experimental data.

Figure . Parameters of the hybrid ECT sensor.

 JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY 9

Figure . Defect amplitude (pinhole ϕmm and OD %tw) and

relative angle between detection coil and defect, for channels ,
Figure . Hybrid ECT sensor for SG double-wall tubes. , and  located near the detect position.

Table . Parameters of the hybrid ECT sensor.

Items Values on FEM numerical simulations. The defect detectabil-
RF-ECT ity between the coils was improved to -3.8 dB compared
Downloaded by [Australian Catholic University] at 23:36 13 September 2017

Exciter coil number of wires:  turns to the maximum peak amplitude of one sensor channel
Detector coil number of wires:  turns
ECT
when located just above the defect.
Exciter and detector coil number of wires:  turns Experimental measurements showed that multiple
Hybrid ECT exciter and detector core material: PC Permalloy vertical coils have a better S/N ratio compared to that
of multiple horizontal ones, which was in agreement
Table . Parameters of the defects in the double-wall tubes. with the FEM numerical simulations. All grooves and
Items (defects located on
circumferential slits were well detected with a high S/N
the outer surface) Values ratio. Figure 13 presents a C-scan of the measured sig-
Circumferential slit W.mm × %tw (. mm)
nal at 1 kHz for the outer-tube surface defects: a full-
W.mm × %tw (. mm) circumferential groove, pinhole, axial and circumferen-
W.mm × %tw (. mm) tial notches.
W.mm × %tw (. mm)
Circumferential notch W.mm × Lmm × %tw (. mm) Figure 14 shows the results when the gap between
Axial notch W.mm × Lmm × %tw (. mm) the two walls of the tube is less than 10 μm, in the case
Pinhole ϕmm × %tw (. mm)
Partial groove Wmm × %tw (. mm) of a circumferential notch. The signal amplitude level
Fatigue crack L.mm × %tw (. mm) (0.89 V) for external defects decreases by up to 10%
L.mm × %tw (. mm)
compared with the signal amplitude (0.98 V) obtained
from the same notch but machined in a single-wall tube
without a gap.
4. Experimental measurements In the hybrid ECT sensor, both low-frequency multi-
coil RF-ECT sensors and high-frequency bobbin-type
4.1. Experiments to detect artificial defects (EDM)
ECT sensor operate simultaneously. In order to avoid
Artificial defects were machined in the SG double-wall signal interference between the two sensors, the ampli-
tube using EDM, with their parameters listed in Table tude of the ECT signal based on direct field coupling is
6. The performance of the multiple-coil RF-ECT sys- decreased up to the point that minimizes its effect on
tem was assessed experimentally using these defects in the RF-ECT sensors. Experimental results from known
various configurations. outer-tube defects show that this operation point can
Figure 12 shows the relationship between the chan- be acquired and that little interference is observed in
nel sensitivity, when measuring the defect signal ampli- the RF-ECT signal when the ECT bobbin-coil signal is
tude from a pinhole ϕ1mm and 20%tw (from tube-wall switched ON/OFF (see Figure 15).
thickness) on the outer surface of tube, and the circum-
ferential relative angle (measured in the circumferential
direction of tube) between the middle of the detection
4.2. Experimental measurements to detect fatigue
coil and the middle of the defect. The signal from one
crack
channel is represented by the signal from a pair of coils
that are separated spatially by a small gap in the axial A fatigue crack was created on the outer surface of
direction of the tube and that are connected differen- a double-wall tube using a fatigue testing machine in
tially. order to clarify the detectability of a crack defect with
In order to increase defect detectability in the inter- a narrow opening. Table 7 shows the parameters of two
mediate area between adjacent detection channels, a 12- fatigue cracks 30%tw and 50%tw, while the fractogra-
channel sensor was selected and manufactured based phy of the fatigue cracks is shown in Figure 16.
 10 T. YAMAGUCHI AND O. MIHALACHE

Figure . C-scan experimental measurements using multiple RF-ECT sensors.

Downloaded by [Australian Catholic University] at 23:36 13 September 2017

outer surface of double-wall tube and used to manu-

facture the fatigue crack, could not be detected by the
hybrid ECT sensor. In the second step, the fatigue crack
was manufactured on the outer surface of a double-
wall tube using a fatigue testing machine. The raw data
confirmed that a 30%tw circumferential fatigue crack
is detectable, validating the performance of the hybrid
ECT sensor.

4.3. Experiments for detection of defects located

under tube support plate
In SG tubes, impact fretting is regarded as a very impor-
tant tube-failure mechanism. The mechanism of tube
impact fretting wear damage is the result of tubes repet-
itively impacting and rubbing against their SPs.
Two types of material have been under consideration
for the SPs of JSFR SG tubes: austenitic stainless steel
and 9Cr-1Mo steel. Considering the RF-ECT sensor, the
austenitic stainless-steel SP has a small influence on the
amplitude signal from a defect hidden beneath the SP.
Therefore, the sensor detectability is high, as was also
Figure . Effect of wall gap between inner and outer tubes on confirmed by FEM numerical simulations. If the tube
the signal amplitude of outer circumferential notch. SP is made of 9Cr-1Mo steel, the RF-ECT signal has low
amplitude because of the high relative magnetic perme-
Table . Parameters of fatigue crack. ability of the SP, and detection of defects located under
the SP is obstructed by the large amplitude of the SP sig-
Items Tension Counter
nal. In order to overcome this difficulty, a new method
Fatigue crack %tw Max . kN– Min . kN , was adapted that uses both fields (the direct and indirect
Fatigue crack %tw Max . kN– Min . kN ,
electromagnetic fields) simultaneously to detect defects
under the tube SP. The optimized excitation frequency
for the ECT coils was 20 kHz.
Experimental results (Lissajous diagram and real and The following fretting wear defects, using EDM, were
imaginary components) for detection of the circumfer- created on the outer surface of double-wall tube, located
ential fatigue crack with only one channel of the multi- under the tube SP (see Figure 18): a full circumferential
ple coil RF-ECT system, operating at 1 kHz, are shown slit 0.3 mm wide and 20%tw and a partial circumferen-
in Figure 17. In the first step, it was confirmed exper- tial groove 5 mm wide and 20%tw that is spanned 90°
imentally that the initial starter notch, located on the around tube circumference.
 JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY 11
Downloaded by [Australian Catholic University] at 23:36 13 September 2017

Figure . C-scan using RF-ECT signal from partial and full circumferential outer-tube defects using (a) only the indirect remote field,
and (b) both indirect remote and direct ECT fields.

Figure . Fatigue testing machine and the fractography of fatigue crack.
 12 T. YAMAGUCHI AND O. MIHALACHE
Downloaded by [Australian Catholic University] at 23:36 13 September 2017

Figure . Experimental measurements using multiple RF-ECT sensors inspection for fatigue cracks in SG double-wall tube (results
for one channel of the RF-ECT sensor).

high and uniform S/N ratio in detecting very narrow

(0.3 mm) outer-surface tube defects. Based on valida-
tion of FEM simulations with experimental measure-
ments, the length and depth of inner-tube ripples and
their impact on the S/N ratio were confirmed.
Development of the proposed hybrid ECT sensor to
be used in JSFR double-wall tubes, which includes a
combined system made of multiple coils based on RF-
ECT and a bobbin-type coil unit based on standard
ECT, was confirmed through experimental measure-
Figure . View of SG tube, SP, defect and their schematics. ments. Its operation point was adjusted so that the RF-
ECT unit was not influenced by the ECT one. It was
confirmed that multiple RF-ECT coils were better at
Figure 19 shows the performance of the bobbin-coil detecting smaller defects (all-round thinning, a circum-
ECT sensor for a slit and a 90° circumferential groove, ferential slit, and a circumferential fatigue crack with a
both of them located on the outside double-wall tube depth of 30% of the tube-wall thickness) compared with
surface and under tube SP. The measurements result the existing ECT bobbin-coil. However, a circumferen-
shows that this approach is feasible, resulting in a high tial slit (on the outer surface of double-wall tube) with
S/N ratio in detecting both defects. a depth of 20%tw under a ferromagnetic SP was found
to be detected better using a conventional ECT coil.
The experimental results showed a decrease of 10%
5. Conclusion
in the signal amplitude from an all-round tube thinning
Using FEM simulations, multiple RF-ECT sensors were when the double-wall gap increased from zero (corre-
designed to be sensitive to the detection of very small sponding to the single-wall case) to less than 10 μm.
outer-tube defects, specially optimized for JSFR SG In spite of initial concerns about a deterioration of
tubes. The influence of the double-wall gap on the sen- defect detectability because of the wall gap between
sor signal was shown to be minimal, assuring a relatively inner and outer tubes in double-wall tube configuration
 JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY 13
Downloaded by [Australian Catholic University] at 23:36 13 September 2017

Figure . Experimental measurements of detection of defects under tube SP (Cr-Mo steel) using direct ECT field effect at  kHz.

or inner tube ripples, their detectability (all-round thin- Energy Society Japan; 2010 Mar 26-28; Mito, Japan.
ning, circumferential and axial notches) was confirmed Japanese.
even for defects located on the outer surface of tube with [4] Mihalache O, Yamaguchi T, Ueda M, et al. FEM simu-
lations for ISI of double wall of SG tubes in JSFR using
depths of 20% of the tube-wall thickness, which was one eddy currents. Annual Meeting of the Atomic Energy
of the main inspection targets. Society Japan; 2010 Sep 15-17; Sapporo, Japan.
The newly developed hybrid ECT sensor system was [5] Yamaguchi T, Mihalache O. [Study on inspection and
shown to be more than adequate for future use in ISI of repair technologies in FBR toward development of
JSFR SG tubes after its S/N sensitivity was adjusted by inspection device for JSFR components – multi-coil RF-
ECT]. Annual Meeting of the Atomic Energy Society
extensive FEM simulations and experimental measure-
Japan; 2011 Sep 19-21; Kokura, Japan. Japanese.
ments. [6] Yamaguchi T, Mihalache O. [Development of multi-
coil RF-ECT sensor for JSFR SG tubes]. Annual Meet-
ing of Japan Society Maintenology; 2012 Jul 25-27;
Disclosure statement Tokyo, Japan, p. 273–275. Japanese.
[7] Kumano S, Kawase N, Kawata K, et al. Signal process-
No potential conflict of interest was reported by the authors. ing of rotating pancake eddy current signal for steam
generator tubes. Proceedings of the 13th International
References Conference on NDE in the Nuclear and Pressure Vessel
Industries; 1995 May 22–25; Kyoto, Japan. Ohio, USA:
[1] Aoto K, Uto N, Sakamoto Y, et al. Design study and ASM international. p. 413–420.
R&D progress on Japan sodium-cooled fast reactor. [8] Fukutomi H, Takagi T, Nishikawa M. Remote field eddy
J Nucl Sci Technol. 2011;48:463–471. current technique applied to non-magnetic steam gen-
[2] Kurome K, Kawamura M, Enuma Y, et al. Steam gen- erator tubes. NDT E Int. 2001;34:17–23.
erator with straight double-walled tube – Development [9] Chen Z, Chigusa N, Isaka H, et al. Data process-
of fabrication technologies of main structures made of ing of corrosion noise polluted ECT signals for heat
high chrome steel-made. Proceedings of the Interna- exchanger tubes of Cu-Ni alloy. Studies Appl Elec-
tional Conference on Fast Reactors and Related Fuel tromagnet Mech, Electromagnet Nondestruct Eval.
Cycles: Challenges and Opportunities; 2009 Dec 7–11; 2004;24:128–135.
Kyoto, Japan. [10] Ganugapati V, Arunchalam K, Ramuhalli P, et al. A
[3] Yamaguchi T, Mihalache O. [Conceptual design study polynomial mixing algorithm for suppressing TSP sig-
and related R&Ds toward the demonstration reactor nals in bobbin coil eddy current data. Studies Appl
of JSFR (9) development of inspection device for FBR Electromagnet Mech, Electromagnet Nondestruct Eval.
structure components]. Annual Meeting of the Atomic 2006;26:213–219.
 14 T. YAMAGUCHI AND O. MIHALACHE

[11] Mihalache O, Yamaguchi T, Ueda M, et al. 3D RFEC ferromagnetic SG tubes covered by a sodium layer.
simulations for the in-service inspection of steam gen- Studies Appl Electromagnet Mech, Electromagnet
erator tubes in fast breeder reactors. Int J Appl Electro- Nondestruct Eval. 2014;39:144–154.
magnet Mech. 2010;33:1165–1171. [19] Yamaguchi T, Mihalache O, Ueda M. Multi-coils ECT
[12] Mihalache O, Yamaguchi T, Ueda M. Computational for SG tubes of FBR: 3D numerical FEM simulations
challenges in numerical simulations of ISI of ferritic vs. experimental measurements. 17th ISEM; 2015 Sept;
steam generator tubes in fast breeder reactors using Kobe, Japan.
eddy currents and multi-frequency algorithms. E-J Adv [20] Mihalache O, Yamaguchi T, Ueda M. Multi-frequency
Maintenance. 2011;3–2:54–77. ECT for a multi-coils system using 3D FEM simula-
[13] Mihalache O, Yamaguchi T, Ueda M. Large scale tions of FBR SG tubes covered by sodium. Studies Appl
3D FEM electromagnetic simulations and validations Electromagnet Mech, Electromagnet Nondestruct Eval.
for FBR steam generator tubes. Studies Appl Elec- 2016;41:9–17.
tromagnet Mech, Electromagnet Nondestruct Eval. [21] Mihalache O. Advanced remote field computational
2012;38:94–102. analysis of steam generators tubes. Studies Appl Elec-
[14] Ando M, Kubo S, Kamishima Y, et al. Study on in- tromagnet Mech, Electromagnet Nondestruct Eval.
service inspection program and inspection technolo- 2006;26:220–227.
gies for commercialized sodium-cooled fast reactor. [22] Mihalache O, Ueda M, Yamashita T. 2D axisymmet-
ICONE14-89558; 2006 Jul 17-20; Miami, USA. p. 635– ric ECT simulation of the World Federation’s first
642. eddy current benchmark problem. Studies Appl Elec-
[15] Takagi T, Hashimoto M, Fukutomi F, et al. Benchmark tromagnet Mech, Electromagnet Nondestruct Eval.
Downloaded by [Australian Catholic University] at 23:36 13 September 2017

models of eddy current testing for steam generator tube: 2007;26:275–282.

experiment and numerical analysis. Int. J Appl Electro- [23] Garcia DR, Mihalache O. FEM comparisons between
magnet Mat. 1994;5:l49–162. RF-ECT signals in quasi-static or transient and linear
[16] Pigeon M, Levy R. In-service inspection of PWR or nonlinear regimes. Int J Appl Electromagnet Mech.
steam generators by eddy currents; French achieve- 2012;39:297–303.
ments in defect detection and evaluation. Nucl Eng Des. [24] FEMLAB 3.0 user’s manual. Comsol Inc. 2004.
1984;81:17–25. [25] Liu J, Hao XJ, Zhou L, et al. Measurement of
[17] Atherton DL, Sullivan S, Daly M. A remote field eddy microstructure changes in 9Cr-1Mo and 2.25Cr-1Mo
current tool for inspecting nuclear reactor pressure steels using an electromagnetic sensor. Scr Mater.
tubes. Br J Nondestruct Testing. 1988;30:22–28. 2012;66:367–370.
[18] Yamaguchi T, Mihalache O, Ueda M. Experimen-
tal measurements and simulations of ECT signal for