Open Crack Depth Evaluation Using Eddy Current

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/269274219

Open crack depth evaluation using eddy current methods and GMR detection

Conference Paper · May 2014


DOI: 10.1109/MetroAeroSpace.2014.6865905

CITATIONS READS

16 747

4 authors:

Dario Pasadas Artur Ribeiro


Technical University of Lisbon Institute of Telecommunications
61 PUBLICATIONS 450 CITATIONS 148 PUBLICATIONS 1,570 CITATIONS

SEE PROFILE SEE PROFILE

Tiago Jorge Rocha Helena G. Ramos


Technical University of Lisbon University of Lisbon
43 PUBLICATIONS 481 CITATIONS 198 PUBLICATIONS 2,857 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Helena G. Ramos on 09 February 2015.

The user has requested enhancement of the downloaded file.


Open Crack Depth Evaluation Using Eddy Current
Methods and GMR Detection

D. J. Pasadas, A. Lopes Ribeiro, T. J. Rocha, Helena Geirinhas Ramos


Instituto de Telecomunicações
Instituto Superior Técnico, Universidade de Lisboa
Lisboa, Portugal

Abstract—In this paper the eddy current nondestructive solution of the Maxwell’s equations, to establish a way to
method is used to determine the depth of linear cracks that were determine the excitation coil impedance. They reconstructed
machined on aluminum plates. To improve this method a the conductivity distribution in different situations. In 1999
constant field probe was used to produce linear eddy currents Yang and others [14] used the finite element method to study
launched across the crack. The tangential to plate magnetic field the velocity effect on the detection of cracks in tubes using
components perpendicular and parallel to the applied field were magnetic flux leakage processes. In 2000 Yoshida and Bowler
measured by a giant magneto-resistor sensor at different [15] used an integral vector potential formulation to tackle the
frequencies to devise one method for crack’s depth measurement. eddy-current interaction with cracks inside conductors. In the
same year Yang and others [16] applied perturbation methods
Keywords—eddy current;non-destructive evaluation; inverse
problem; fatigue cracks;
to finite element analysis to study stress corrosion cracking
when the magnetic flux leakage inspection method is applied in
tubes. In 2004 Gotoh and Takahashi [17] proposed a method
I. INTRODUCTION using magnetic flux leakage with alternating excitation at
In this paper we shall use the eddy current method to different frequencies to characterize multiple cracks in
characterize cracks that have been machined on aluminum ferromagnetic steel materials. In 2006 Rebican and others [18]
plates of the kind usually found in aerospace vehicles. The applied stochastic methods to characterize multiple cracks by
eddy current method has been widely used for a long time. inversion of eddy current signals. In the same year Joshi, Udpa
Nowadays, with a large number of aged aircrafts in use, the and others [19] used adaptive wavelets to characterize cracks in
nondestructive methods related to airplane maintenance the inspection of pipes using the magnetic flux leakage method.
continue being one of the most important ways to detect and In 2008 Amineh, Nikolova, Reilly and Hare [20] used the
characterize flaws in metallic non-ferromagnetic materials. magnetic flux leakage method to characterize rectangular
Other methods are used to detect open aging fatigue cracks surface-breaking cracks in steel measuring one tangential
such as dye penetrants [1,2], thermography [3,4], radiographic component of the leaking field.
methods [5], ultrasonic testing [6,7], acoustic emission [8] or
In recent years a lot of work on eddy current methods,
magnetic flux leakage [9].
employing different probe topologies continue being published.
The eddy current method has been tackled by many authors Some authors use pulsed or sinusoidal eddy currents and giant
both theoretically and experimentally. The seminal analytic magneto-resistor sensors [21-26], others prefer to develop
work of C. V. Dodd and W. E. Deeds calculating analytically numerical or analytical models [27,28]. In this paper we aim at
the impedance of one coil over a metal stratified material [10] characterizing rectangular straight cracks in aluminum plates,
is still used as a reference. In 1991, J. Bowler, S. Jenkins, with a special attention to the measurement of depth. The
L. Sabbagh and H. Sabbagh studied the eddy current flaw general procedure consists in scanning the area including
inspection problem [11] and expressed the electromagnetic cracks to determine their orientation. This is done when the
field scattered by a flaw by a volume integral with a dyadic crack orientation is not known. Then the scanning procedure is
kernel. They used the numerical solutions to preview the repeated using a constant field excitation probe [29] oriented in
excitation coil impedance to be compared with the measured a way that the induced currents are launched perpendicularly to
values. In 1992 Haywood and Bowler studied the measured the crack's line. Two components of the tangential magnetic
magnetic field inversion problem [12] with a subsurface slot field are measured with two giant magneto-resistor sensors,
manufactured into an aluminum plate. The magnetic dipole one parallel and the other perpendicular to the cracks line.
density on the slot faces was estimated using an integral Thus, we obtain two maps, depicting those two components as
operator and the inversion was performed with singular value measured on the scanned area. These data contain useful
decomposition in an iterative process with least squares error information about the crack's parameters, namely their length
minimization. In 1993 Norton and Bowler presented a theory and depth. In section two it is described the experimental
of eddy current inversion [13] where they used a forward apparatus in use. In section three some results are presented
problem established in terms of an integral form for the and commented and in section four we draw some conclusions.
II. THE EXPERIMENTAL SETUP III. EXPERIMENTAL RESULTS
The experimental setup is depicted in Fig.1. Fig. 3 depicts the output voltage obtained when the plate
was scanned with the probe #1 with the GMR sensor aligned in
order to measure the magnetic field perpendicular to the strong
excitation. The probe excitation current is always oriented
perpendicularly to the crack's lines. With such orientation we
expect to obtain eddy currents that must overcome the barrier
imposed by the cracks presence. Those induced currents may
pass the crack in two ways. The can go around the crack with
larger density at the surface and very high density on the crack
tops, or pass under the crack. The choice of these two modes
must depend on several parameters such as the crack length,
the depth and those parameters related to the standard depth of
penetration i.e. the operation frequency and the electrical
conductivity.

Fig. 1. Experimental setup with the planar excitation coil over an aluminum
plate. Data acquisition, positioning system and excitation current are
controlled by a desktop computer running a matlab program.

Fig. 1 represents one planar excitation probe over an


aluminum plate. The coil probe is planar. It has been Fig. 3. Amplitude of the sinusoidal output of the GMR sensor when the plate
manufactured on a printed board with parallel lines in the was scanned. In this case, the measured component of the magnetic field
is perpendicular to the primary excitation field. The excitation current, in
central region. The primary excitation field produced by this all experiencies, is always perpendicular to the cracks lines.
probe is spatially uniform in the central region. Thus, when the
coil is excited with a time-varying sinusoidal current, the eddy The data represented in Fig. 3 were obtained using an
current induced under the region below the middle coil zone is excitation frequency fex  500 Hz . After scanning the region
spatially uniform as well. One giant magneto-resistor (GMR)
sensor is positioned on the upper probe surface (probe including the cracks, the complex amplitudes representing the
number 1) and measures the magnetic field component parallel sinusoidal voltage at points far from defects were subtracted
to the imposed current, and perpendicular to the excitation from the data. Thus Fig. 3 represents the perturbation on the
primary field. Two probes were manufactured, the only sinusoidal amplitude at each measured point. At this frequency
difference between them being the orientation of the GMR the depth of penetration in the aluminum is   3.78 mm . It is
sensor. In the probe number 2 the GMR sensor reads the obvious that the amplitude of the peaks decrease when the
magnetic field component in the direction of the applied field. crack's depth also decreases. The peak voltages result from the
high current density on the crack's tops. Fig. 4 represents the
Fig. 2 depicts the aluminum plate under test. It presents GMR output voltage taken on the line passing over all defects.
seven artificial cracks. Their only difference is the depth, equal
to 3.5 mm for the deepest and 0.5 mm for the shallowest. A 0.5
rectangular surface containing all the cracks was scanned with
the two probes. 0.4
Amplitude / V

0.3

8mm 65mm 0.2


70mm
1mm

0.1

0
0 100 200 300 400 500 600
4mm x / mm
3.5mm 3mm 2.5mm 2mm 1.5mm 1mm 0.5mm
Fig. 4. Voltage peaks measured when the probe passed over the cracks.
Fig. 2. Seven rectangular surface cracks machined on one aluminum plate There is a clear correlation between the crack's depth and the peak's
with 4 mm of thickness. The crack depth decreases from left (3.5 mm) to amplitudes.
right.(0.5 mm).
Considering the results presented in Fig. 4 it is obvious that 0.45

the voltage peaks observed in the GMR sensor output by 0.4

scanning the defects are closely correlated with the crack’s 0.35

depths. This fact could be used to establish the method for the 0.3

Amplitude / V
determination of the depths. 0.25

0.2

It is important to show the evolution of the signals that 0.15

were obtained with the probe #2 that measured the field in the 0.1

same direction of the primary excitation field. This 0.05

measurement is more difficult, due to the superposition of the 0


0.5 1 1.5 2 2.5 3 3.5
strong excitation. The data represented in Fig. 5 were obtained Depth of Crack / mm
after subtraction of that field. There is one important difference Fig. 7. GMR output peak voltage for the seven cracks, with the measured
between the two measurements. The perturbation field field perpendicular to the crack.
component perpendicular to the excitation presents four peaks,
two on each crack end, and the perturbation field component 1.4

parallel to the excitation presents just one peak on the crack 1.2

center, the two measurements being performed in the same


1
conditions, with the eddy currents always directed

Amplitude / V
perpendicularly across the crack. The results obtained for the 0.8

two probes were depicted on a 2D representation as shown in 0.6

Fig. 6.
0.4

0.2

0
0.5 1 1.5 2 2.5 3 3.5
Depth of defect / mm

Fig. 8. GMR output peak voltage for the seven cracks, with the measured
field parallel to the crack.

The aluminum plate was scanned at different frequencies.


Fig.s 9 and 10 represent the data obtained with one linear scan
over cracks with 0.5 and 3.5 mm of depth using the probe
Fig. 5. Output of the GMR sensor when the plate was scanned. In this case, measuring the field component perpendicular to the crack’s
the measured component of the magnetic field is parallel to the primary line. The frequencies ranged from 500 Hz to 10 kHz.
excitation field. The excitation current, in all experiencies, is always
perpendicular to the cracks lines. 0.1
Defect with 0.5 mm of depth

500 Hz
0.09 1000 Hz
2000 Hz
60 0.08
3000 Hz
0.4
Amplitude / V

4000 Hz
0.07
y / mm

40 5000 Hz
Amplitude / V

0.06 10000 Hz
0.2
20 0.05

0 0 0.04
0 50 100 150 200 250 300 350 400 450 500
x / mm 0.03

(a) 0.02

1 0.01
Amplitude / V

20 0.8 0
0 10 20 30 40 50 60 70 80
y / mm

0.6 Distance / mm
40
0.4
Fig. 9. GMR output peak voltage for a linear scan over the crack with
60 0.2
0.5 mm of depth and measuring the field perpendicular to the crack.
50 100 150 200 250 300 350 400 450
x / mm Defect with 3.5 mm of depth
(b) 0.7
500 Hz
1000 Hz
0.6
2000 Hz
Fig. 6. 2D representation of the GMR voltage output when the measured 3000 Hz
component of the magnetic field is (a) perpendicular or (b) parallel to 0.5 4000 Hz
5000 Hz
the primary excitation field. The excitation current, in all experiencies, is
Amplitude / V

10000 Hz
0.4
always perpendicular to the cracks lines.
0.3

Fig. 7 and Fig. 8 depict the amplitude of the peaks


0.2
measured with the two probes. The peak values represented in
Fig. 7 were extracted from the complete matrix values on a line 0.1

parallel to the cracks axes, and the peak values on Fig. 8 are 0
0 10 20 30 40 50 60 70 80
taken exactly over the cracks line. It is obvious that the result Distance / mm

obtained on the second case is more precise because the data Fig. 10. GMR output peak voltage for a linear scan over the crack with
presents one single peak on a smoother curve. 3.5 mm of depth and measuring the field perpendicular to the crack.
The results presented in the last two figures were obtained must be taken not to saturate the high sensitivity magneto-
with the probes passing over the crack’s tips were the resistive sensor and to post-process the acquired data to
maximum output sensor voltage is obtained. However, it was subtract the excitation component from the collected data.
verified that the sensor signal presented one single maximum As represented in Fig.s 9 and 10 it was decided to scan the
when the component parallel to the crack was measured. The deepest and the shallowest cracks. Note that deep cracks are
maximum appears when the sensor is positioned over the those with depth equal or larger than the standard penetration
crack’s middle point, as depicted in Fig. 6(b). It was also depth for the frequency and material in use or through cracks
verified that the values of those maxima are better correlated independently of material thickness. When the cracks were
with the crack’s depth as Fig. 8 suggests. scanned with different frequencies it was obvious that, for deep
Being the results obtained by measurement of the field cracks, the signals were almost independent from frequency.
parallel to the crack’s line of better quality, it was decided to This fact results from the currents that are constrained to
repeat those measurements as a function of frequency in the circulate around the crack. They are not able to pass beneath,
same way as those represented in Fig. 9 and in Fig. 10. These because the crack is a deep one.
new results are depicted in Fig. 11 and Fig. 12. Another important result was obtained with the probe #2,
Defect with 0.5 mm of depth measuring the field component parallel to the crack with one
0.35
500 Hz pass perpendicularly over the crack, with variable excitation
1000 Hz
0.3
2000 Hz frequency. The system response as a function of frequency is
0.25
3000 Hz
4000 Hz correlated to the crack’s depth. This fact shall be used in future
5000 Hz
work.
Amplitude / V

10000 Hz
0.2

0.15 ACKNOWLEDGMENT
0.1
This work was developed under the Instituto de
0.05 Telecomunicações projects KeMANDE and OMEGA and
supported in part by the Portuguese Science and Technology
0
0 10 20 30 40
Distance / mm
50 60 70 80 Foundation (FCT) projects: PEst-OE/EEI/LA0008/2013,
SFRH/BD/81856/2011 and SFRH/BD/81857/2011. This
Fig. 11. GMR output peak voltage for a linear scan over the crack with support is gratefully acknowledged.
0.5 mm of depth and measuring the field parallel to the crack.

Defect with 3.5 mm of depth REFERENCES


2

1.8
500 Hz
1000 Hz
[1] A. J. McEvily, “Failures in inspection procedures: case studies”,
1.6
2000 Hz Engineering Failure Analysis, Vol.11, (2004) pp.167-176.
3000 Hz

1.4
4000 Hz [2] T. G. Santos, R. M. Miranda, C. de Carvalho, “A new NDT technique
5000 Hz
based on bacterial cells to detect micro surface defects”, NDT&E
Amplitude / V

1.2 10000 Hz
International, Vol. 63, (2014), pp.43-49.
1

0.8
[3] P. Broberg, “Surface crack detection in welds using thermography”,
NDT&E International, Vol. 57, (2013), pp.69-73.
0.6
[4] S. Bagavathiappan, B. Lahiri, T. Saravanan, J. Philip, T. Jayakumar,
0.4
“Infrared thermography for condition monitoring – A review”, Infrared
0.2
Physics & Technology, Vol. 60 (2013), pp.35-55.
0
0 10 20 30 40 50 60 70 80 [5] C-Y Yeh, R. Zoughi, “A Novel Microwave Method for Detection of
Distance / mm
Long Surface Cracks in Metals”, IEEE Trans. on Instrumentation and
Fig. 12. GMR output peak voltage for a linear scan over the crack with Measurement,Vol. 43. No. 5, Oct. 1994, pp.719-724.
3.5 mm of depth and measuring the field parallel to the crack. [6] M. Morbidini, P. Cawley, “The detectability of cracks using sonic IR”, J.
Applied Physics, Vol. 105, 093530 (2009), pp.1-9.
These two pictures present new interesting phenomena. For [7] J. Zhang, B. Drinkwater, P. Wilcox, “Longitudinal Wave Scattering
shallow cracks the output sensor signals are stronger for higher From Rough Crack-Like Defects”, IEEE Transactions on Ultrasonics,
Ferroelectrics, and Frequency Control, Vol. 58, no. 10, October 2011,
frequencies, and for deep cracks the signals are stronger for the pp. 2171-2180.
lowest frequencies. [8] A. Nair, C. Cai, “Acoustic emission monitoring of bridges: Review and
case studies”, Engineering Structures, Vol. 32 (2010), pp. 1704-1714.
IV. DISCUSSION AND CONCLUSION [9] H. Kim, G. Park, “A Study on the Estimation of the Shapes of Axially
Oriented Cracks in CMFL Type NDT System”, IEEE Transactions on
In this paper we introduced a method to assess the depth of Magnetics, Vol. 50, No. 2, Feb. 2014.
fatigue cracks in aircraft structures. The results presented in the [10] C. V. Dodd, W. E. Deeds, “Analytical Solutions to Eddy-Current Probe-
previous section show that, when uniform field excitation and Coil Problems”, Journal of Applied Physics, Vol. 39, No. 6, 1968,
GMR detection are used, the better field component to be pp. 2829-2837.
measured is in the same direction of the excitation field. The [11] J. Bowler, S. Jenkins, L. Sabbagh, H. Sabbagh, “Eddy-current probe
experiment must be performed in a way that the induced impedance due to a volumetric flaw”, Journal of Applied Physics,
Vol. 70, No. 3, 1991, pp.1107-1114.
currents are directed perpendicularly to the crack. Thus, it is
[12] N. Haywood, J. Bowler, “Eddy-Current Imaging of Buried Cracks by
assumed that detection, location and 2D surface Inverting Field Data”, IEEE Transactions on Magnetics, Vol. 28, No. 2,
characterization were already successfully accomplished. Care 1992, pp.1336-1339.
[13] S. Norton, J. Bowler, “Theory of eddy current inversion”, Journal of Inspection of Aircraft Structures”, IEEE Transactions on Magnetics,
Applied Physics, Vol. 73, No. 2, 1993, pp.501-512. Vol. 46, No. 3, 2010, pp.910-917.
[14] S. Yang, Y. Sun, L. Udpa, S. Udpa, W. Lord, “3D Simulation of [22] Z. Zeng, Y. Deng, X. Liu, L. Udpa, S. Udpa, B. Koltenbah, R. Bossi G.
Velocity Induced Fields for Nondestructive Evaluation Application”, Steffes, “EC-GMR Data Analysis for Inspection of Multilayer Airframe
IEEE Transactions on Magnetics, Vol. 35, No. 3, 1999, pp.1754-1756. Structures”, IEEE Transactions on Magnetics, Vol. 47, No. 12, 2011,
[15] Y. Yoshida, J. Bowler, “Vector Potential Integral Formulation for Eddy- pp.4745-4752.
Current Probe Response to Cracks”, IEEE Transactions on Magnetics, [23] H. Ramos, O. Postolache, F. C. Alegria, A. L. Ribeiro, “Using the skin
Vol. 36, No. 2, 2000, pp.461-469. effect to estimate cracks depths in metallic structures”, in Proc. IEEE
[16] S. Yang, Y. Sun, L. Udpa, S. Udpa, W. Lord, “Application of Instrum. and Meas. Technol. Conf., May 2009, pp. 1361-1366.
Perturbation Methods in Finite Element Analysis of Stress Corrosion [24] A. L. Ribeiro, H. G. Ramos and J. C. Couto, “Liftoff insensitive
Cracking”, IEEE Transactions on Magnetics, Vol. 36, No. 4, 2000, thickness measurement of aluminum plates using harmonic eddy current
pp.1714-1718. excitation and a GMR sensor”, Measurement, Vol. 45, pp.2246-2253,
[17] Y. Gotoh, N. Takahashi, “Proposal of Detecting Method of Plural Sept. 2012.
Cracks and Their Depth by Alternating Flux Leakage Testing: 3-D [25] A. Bernieri, G. Betta, L. Ferrigno, M. Laracca, “Crack Depth Estimation
Nonlinear Eddy Current Analysis and Experiment”, IEEE Transactions by Using a Multi-Frequency ECT Method”, IEEE Transactions on
on Magnetics, Vol. 40, No. 2, 2004, pp.655-658. Instrumentation and Measurement, Vol. 62, No. 3, 2013, pp. 544-552.
[18] M. Rebican, Z. Chen, N. Yusa, L. Janousek, K. Miya, “Shape [26] A. Bernieri, G. Betta, L. Ferrigno, M. Laracca, “Multifrequency
Reconstruction of Multiple Cracks from ECT Signals by Means of Excitation and Support Vector Machine Regressor for ECT Defect
Stochastic Method”, IEEE Transactions on Magnetics, Vol. 42, No. 4, Characterization”, to be published in a future issue of IEEE Transactions
2006, pp.1079-1082. on Instrumentation and Measurement.
[19] A. Joshi, L. Udpa, S. Udpa, A. Tamburrino, “Adaptive Wavelets for [27] J. Bowler, T. Theodoulidis, H. Xie, Y. Ji, “Evaluation of Eddy-Current
Characterizing Magnetic Flux Leakage Signals From Pipeline Probe Signals Due to Cracks in Fastener Holes”, IEEE Transactions on
Inspection”, IEEE Transactions on Magnetics, Vol. 42, No. 10, 2006, Magnetics, Vol. 48, No. 3, 2012, pp.1159-1170.
pp.3168-3170. [28] R. Miorelli, C. Reboud, D. Lesselier, T. Theodoulidis, “Eddy Current
[20] R. K. Amineh, N. K. Nikolova, J. P. Reilly, J. R. Hare, “Characterization Modeling of Narrow Cracks in Planar-Layered Metal Structures”, IEEE
of Surface-Breaking Cracks Using One Tangential Component of Transactions on Magnetics, Vol. 48, No. 10, 2012, pp.2551-2559.
Magnetic Leakage Field Measurements”, IEEE Transactions on [29] A. Lopes Ribeiro, H. G. Ramos, D. Pasadas, T. Rocha, “Regularization
Magnetics, Vol. 44, No. 4, 2008, pp.516-524. Methods to Assess the Eddy Current Density inside Conductive non-
[21] G. Yang, A. Tamburrino, L. Udpa, S. Udpa, Z. Zeng, Y. Deng, P. Que, Ferromagnetic Media”, Review of Progress in Quantitative
“Pulsed Eddy-Current Based Giant Magnetoresistive System for the Nondestructive Evaluation, Baltimore-Maryland, Jul. 2013.

View publication stats

You might also like