Journal Information
Vol. 8. Issue 5.
Pages 4843-4848 (September - October 2019)
Download PDF
More article options
Vol. 8. Issue 5.
Pages 4843-4848 (September - October 2019)
Original Article
DOI: 10.1016/j.jmrt.2019.08.031
Open Access
Correlation of eddy current signals obtained from EDM notches and fatigue cracks
Cesar G. Camerinia,b, Lucas B. Camposb, Vitor M.A. Silvaa,b, Daniel S.V. Castroa,b,
Corresponding author

Corresponding author.
, Rafael W.F. Santosc, João M.A. Rebellob, Gabriela R. Pereiraa,b
a Department of Metallurgical and Materials Engineering(COPPE-UFRJ), Federal University of Rio de Janeiro, Rio de Janeiro, RJ, Brazil
b Laboratory of Nondestructive Testing, Corrosion and Welding (LNDC), Federal University of Rio de Janeiro (UFRJ), Rio de Janeiro, RJ, Brazil
c CENPES – PETROBRAS Research Center, Rio de Janeiro, RJ, Brazil
Article information
Full Text
Download PDF
Figures (9)
Show moreShow less
Tables (4)
Table 1. Information about the probe and the inspection procedure.
Table 2. Values for widths and signal amplitude for ECT measurements on fatigue crack.
Table 3. Values for widths and signal amplitude for ECT measurements on EDM notches.
Table 4. Comparison between the amplitude of signals from original fatigue crack (0,02 mm) and EDM notches.
Show moreShow less

For calibration of eddy current testing, artificial defects are commonly used by virtue of its simple production and ease control of geometric parameters. However, such defects are significantly different in terms of geometric aspects from real ones, even with a precise control of the manufacturing process. To ensure reliable calibration results, it is mandatory to establish a relationship between the signals from artificial (machined) and real defects. This work exposes a detailed study correlating eddy current signals from electrical discharge machine notch (EDM notch) and fatigue cracks. To perform the tests an absolute pencil probe was used. Although there is a significant difference between the signals of EDM notch and fatigue cracks, the results show that it is possible to use artificial defects to represent fatigue cracks reliably if a proper correction is implemented.

Eddy current testing
EDM notches
Fatigue cracks
Full Text

The non-destructive eddy current test is widely used to detect and characterize cracks and defects in metallic materials [1]. This process requires a prior calibration procedure, in similar conditions to the real test, to obtain databases for comparing and interpreting the signals response seeing on the part being tested and assures the test system sensitivity is sufficient to detect the required crack size. In order to improve accuracy of results, it is important that defects in the calibration standards have physical, electrical, magnetic and morphological properties similar to defects identified in field situations, producing theoretically a resembling and uniform response. However, fabrication of standards with fatigue cracks with exact size can be an arduous task due to the difficulty in controlling the dimensions. Meanwhile, the high cost, the requirement of a specific infrastructure and qualified personal greatly increases the costs involved. According to these obstacles, the use of artificial defects is a prevalent practice to represent real defects in calibration procedure and creation of a database [2,3].

An alternative is the insertion of laser notches, which correlate geometrically well with real cracks. Although, some issues such as the influence of heat input on material around the notch and the arduousness in removing all the material from the notch may produce inaccurate results. Thus, the use of laser notches as simulations of real cracks should be further investigated [5]. A popular and alternative approach is creating notches by electrical discharge machining (EDM) method [6]. The low cost, the fast obtaining and the easy defects’ dimensional control are some of its advantages. Nevertheless, it is difficult to insert deeper EDM notches with very small width, analogous to ordinary fatigue cracks [7]. It can be a meaningful problem as long as the response signal depends on the distribution of the eddy currents in the inspected material. Therefore, changes in factors that modify the currents path such as crack dimensions, tilt and skew, orientation, branching, roughness and surface quality, also modifies the response obtained. Hence, EDM notches are supposed not to be always a reasonable representation of a real fatigue crack since some of these factors have different conditions [8]. There is much discussion in the non-destructive testing community about the accuracy of using signal responses on EDM notches as references for eddy current test (ECT) inspections and some relevant works addressing this topic were developed. Mostly, the literature reports a significant effect of defects width on increasing eddy current signal amplitude, either for fatigue cracks [14] or for EDM notches [7,15–17], revealing an important relation of the amplitude to the volume of absent material. However, a comparison between these behaviors for both defect types is mandatory to legitimate the use of a standard calibration block containing EDM notches for inspections seeking fatigue cracks.

Rummel et al. [9] reported the lack of reliability of EDM slot using it an as a tool for calibration for crack sizing. It was verified a growing linear relationship between signal amplitude and EDM slots width, but different from the same analysis for fatigue cracks. An equivalent linear variation of EDM response related to width was also found by Larson et al. [7]. The authors determined that the signals from the original fatigue crack specimens correlate well to the predicted zero-width notch signals calculated by extrapolating the trend observed in experimental results for EDM notches with distinct widths. Comparing directly induced fatigue cracks (around 10–20 µm width) and EDM notches having the same depth and length, some investigations [3,8,11] noticed a considerable difference in signal amplitude and, consequently, hampering the cracks detection and underestimating its dimensions. This discrepancy was found to be larger the higher the test frequency [12,13]. Simultaneously, Ibrahim et al. [10] noticed the signal phase difference in this situation is truly slight.

On the other hand, Kurokawa [18] found that the modification in impedance due to the change in crack width was not significant in plates of Inconel 600 [19] with inserted fatigue cracks, except for values below a certain threshold, from which there is a sudden signal decrease. A corresponding conclusion was found by Nakagawa et al. [13]. For flaws parallel to current propagation, Ross and Lord [4] noticed a similarity between the fatigue crack and EDM notch signals. The complex and jagged crack shape could cause significant current interruption, whereas the straight path of the EDM notch, although much wider, interrupts nearly the same amount of current.

Some crack characteristics also has the potential to change remarkably the signal obtained. The crack closure effect may create electrical contacts inside the cracks as a result of flaw geometry and presence of oxides and asperities. It generates a short circuit of eddy current flow around the crack and may change the response considerably [3,13]. Some authors [4,20] also highlighted that cracks have their interface affected by compressive stresses, presenting a reduction in signal amplitude as the probability to create electrical contacts is higher. Additional branches in cracks surface that may boost the eddy current indications were also pointed by Lahdenperä [8] as a possible reason of the underestimation in sizing of thermal fatigue cracks. Other authors warned about the possibility of response distortions due to the influence of deformations and of the resolidification region [4], for narrower defects, and also about the inspected surface roughness [21]. Finally, Esquivel and Kim [22] noticed that for wider cracks, the signal change is closer to the lift-off change, whereas for very narrow cracks the signal modification is analogous to conductivity variation since the contact points induces a sparse variation in crack conductivity.

Summarizing the results of several researchers, a lack of consensus was verified, what keeps open the discussion to understand more clearly the aspects about a possible similarity of the signal provided different types of defects with the same geometric characteristics. The present work is a detailed study on the differences between signals provided by fatigue cracks and EDM notches. The effect of the discontinuities openings was analyzed exclusively, remaining all the other parameters fixed, and corrections to represent and detect fatigue cracks using calibration results based on EDM notches were discussed. The defects are through thickness cracks and through thickness notches and samples have enough thickness to not influence the behavior of the induced currents. In this way, the signals are exclusively related to the aperture and are not affected by the background geometry of the defects. The measurements details are explained in the following sections of the paper.


A real fatigue crack was fabricated by a fatigue test performed using compact tension (CT) specimens of Inconel 625 [19], propagating the crack without reaching the specimen rupture, as show in Fig. 1a. Inconel 625 is a paramagnetic material (relative magnetic permeability μr=1) and its electrical conductivity is 0,775 MS/m. The specimen was subjected to a load variation of 25 kN in a fatigue machine, oscillating at a frequency of 4 Hz. After the preliminary crack was obtained, a mechanical system was installed on the sample larger face to vary the crack opening for subsequent eddy current tests measurement. The system consists basically of two pins that are fixed in the same through-holes of the CT sample, which fit into the fatigue machine grips. A hole with an internal thread was machined in one of the pins whereas the other was mounted in the piece serving as a stop. A screw has been inserted and adjusted until it connects to the stop, allowing the control of the accumulated torque applied to the screw. According to the applied torque, a corresponding fatigue crack width is obtained, where the larger the torque the greater the tensile force to open the crack, as shown in Fig. 1b. The signal acquisition was performed continuously on a path that passes over the crack, as demonstrated by Fig. 1c. Fig. 2 shows an image of the CT specimen with the fatigue crack and, to the right, it is demonstrated the same specimen with the crack opening control system.

Fig. 1.

Schematic of the mechanical system created for the variation of the fatigue crack.

Fig. 2.

CT specimen with system to control the fatigue crack width.


For a comparison purpose, a specimen of the same material containing electrical discharge machine (EDM) notches with 10 different widths was also fabricated, as represented in Fig. 3. Fig. 4 shows the real specimen containing 10 EDM notches with several widths between 0,17 and 0,40 mm. The width values were measured with the support of the optical microscope. It should be noted that it is very difficult to achieve EDM notches with widths smaller than 0,17 mm, especially for great depths. The eddy current inspections were executed in the same manner as the CT specimens. From the results obtained from both, a correlation between the value of the impedance variation and the discontinuity width was made. Focusing on the validation of the use of EDM notches to represent the behavior of eddy currents in fatigue cracks zones, a comparative study was performed utilizing the signals obtained in the inspection of both types of discontinuities.

Fig. 3.

Illustration of the specimen with EDM notches with different widths.

Fig. 4.

Specimen made of Inconel 625 containing 10 EDM notches with width values from 0.16 to 0.40 mm.


For each torque value applied, the corresponding crack width and morphological details were verified by optical microscopy. Ten crack width measurements were executed for each torque (opening) condition and the mean value was considered for interrelationship with the eddy current signals. Torques were applied until the fatigue crack reached widths similar to the EDM notches so that it was possible to analyze the influence of flaws morphology on the eddy current signals. To ensure a data set acceptable for statistical analysis, five inspections were carried out for each notch and for each crack opening. All measurements were performed on a line perpendicular to the defect at half of its length. The inspection sweep velocity was 1 mm/s, the probe drive equal to 4 V and the data acquisition rate was 2 KHz, providing a plentiful amount of data. The absolute probe details and inspection information are exposed in Table 1.

Table 1.

Information about the probe and the inspection procedure.

Resonance frequency (MHz)  Coil diameter (mm)  Ferrite core diameter (mm)  Lift-off (mm)  Gain (dB) 
Plase,  45 

Fig. 5 exemplifies the relation between calculated standard penetration depths of the induced currents and the test frequency for Inconel 625 [23]. The used frequency was 260 kHz, which results in eddy currents penetration depth of around 1,1 mm. Since the thickness of the specimens used for the present test is 5 mm, almost five times greater than the penetration depth, and considering that the reduction of the current density follows an exponential function with respect to depth, it is concluded that the specimens’ thickness does not influence the collected signals. The average sample reference signal amplitude, without defect, is 11,563 V and its standard deviation is 0,233 V.

Fig. 5.

Penetration depth of eddy currents as a function of the test frequency for Inconel 625.

3Results and discussion

The mean widths values and their standards deviations (SD) for cracks and EDM notches are exposed in Tables 2 and 3.

Table 2.

Values for widths and signal amplitude for ECT measurements on fatigue crack.

Mean crack width (µm)  22,13  29,10  38,44  46,42  64,46  81,56  96,10  109,54  130,59  149,24 
Crack width SD (µm)  8,48  3,09  4,94  3,85  6,83  8,81  6,31  9,77  7,14  8,17 
Amp. (V)  3,568  4,130  4,300  4,380  4,412  4,466  4,514  4,520  4,570  4,668 
Mean crack width (µm)  168,28  185,90  196,44  228,20  260,09  270,51  288,29  310,41  359,44 
Crack width SD (µm)  7,96  8,68  8,46  11,47  12,79  14,93  14,15  15,39  18,27  – 
Amp. (V)  4,698  4,754  4,792  4,902  4,888  4,912  4,922  4,986  5,124  – 
Table 3.

Values for widths and signal amplitude for ECT measurements on EDM notches.

Mean EDM width (µm)  178,74  184,60  197,58  212,36  258,13  261,82  278,09  319,74  320,39  403,90 
EDM width SD (µm)  11,07  7,12  9,26  5,46  5,01  4,35  8,42  12,70  5,42  6,12 
Amp. (V)  4,605  4,731  4,802  4,800  4,974  4,965  4,848  4,988  5,038  5,384 

By the time the specimen is removed from the fatigue test, the crack is remarkably narrow, around 10–20 μm width, as exposed in Fig. 6. This characteristic, in accordance to the difficulty in inserting EDM notches narrower than 0,17 mm, establish a considerably large difference between the width values of the EDM defects and the real crack. For this reason, the mechanical system of Fig. 2 was developed and successfully generated flaws with intermediate widths, between the actual crack and the smallest EDM notch.

Fig. 6.

Detailed view, obtained by optical microscopy, of the fatigue crack produced in this work.


Fig. 7a and b shows some images of the crack and EDM notches with different openings, respectively. The number presented below each image is the approximate value of measured widths mean. Fig. 8 represents the relationship between the signal mean amplitude as a function of the discontinuities widths. All defects’ amplitude measurements take as starting reference (balance point) the average sample reference signal amplitude, so the values exposed are the difference between amplitude values on defects and the average amplitude value on the non-defected area. The red dots indicate the measurements performed on the fatigue specimen, while the blue dots are related to EDM notches. The error bar has a range of ±1 standard deviation from the mean and indicates the data variability. The irregular crack path, distinctly from EDM notches with flat flaw faces, makes the width measure more imprecise, as demonstrated by the higher value for width standard deviation compared to mean value.

Fig. 7.

a) Variation of fatigue crack with the increase of the accumulated torque in the screw of the mechanical system. b) Different widths achieved with the electrical discharge machine.

Fig. 8.

Signals amplitude as a function of the discontinuities width.


From the collected data it is possible to verify that flaw width has great influence on the inspection results, where, as the aperture is increased, the signal amplitude is higher. Comparing signals from original fatigue crack (0,02 mm width) signal and all the EDM notches, the difference between amplitude values are shown in Table 4. At the same time, amplitude values for defect signals between 3,56 V and 5,38V guarantee amplitudes changes between 31% and 47% of the reference signal, highlighting the good sensitivity accomplished due to a satisfactory frequency selection and to the small radius of applied probe.

Table 4.

Comparison between the amplitude of signals from original fatigue crack (0,02 mm) and EDM notches.

Mean EDM width (µm)  178,74  184,60  197,58  212,36  258,13  261,82  278,09  319,74  320,39  403,90 
Signal difference (V)  1,037  1,163  1,234  1,232  1,406  1,397  1,280  1,420  1,470  1,816 
Signal difference (% EDM signal)  22,52  24,58  25,69  25,66  28,26  28,14  26,40  28,47  29,18  33,73 

The signal difference between EDM notches and the original fatigue crack varies from 2,22 dB (22,52%) to 3,57 dB (33,73%) and are exposed in Fig. 9 for all notches. Therefore, it is essential correct the signals from EDM notches to make them representative for a real field situation, where the objective is the detection of tiny fatigue cracks. The percentage to be corrected must be calculated through the calibration curve plotted with inspection results from several discontinuities openings. Taking into account the statistical aspects of the data and the uncertainty of the measurements, it can be presumed that the general tendency is that the correction factor to be applied is larger the wider the notch.

Fig. 9.

Signal difference (dB) between measurements on EDM notches and on the original fatigue crack.


Moreover, it is also possible to identify some interesting aspects such as, for example, the compatibility of width values. Signals from EDM notches and fatigue cracks having the same width provided closely related signal amplitudes for a large range of widths, demonstrating that the influence of flaw morphology can be neglected. In this range, real cracks’ width can be predicted roughly by corresponding EDM notch width. However, in measurements related to the three smaller fatigue crack widths, the amplitude variation is quite abrupt and has a distinct decrease rate. From the fourth measure onwards, linear trend with a different slope is assumed. Hence, to estimate the flaw width in field inspections with same test conditions, the calibration linear rule to be applied will depend on the signals amplitude value.

The most likely reason for this rate change is the presence of electrical contacts between the faces of the crack for widths smaller than a given threshold, changing the configuration of current distribution around the crack and allowing the current flow between crack faces. These contact points present in the fatigue crack generated in this work are exemplified in Fig. 6. It corroborates the conclusion of some previous studies [3,13] that also gave importance to the impact of this phenomenon on the acquired signals.

In general, the obtained results are quite different from those reported by Nakagawa et al. [16] and Larson et al. [7]. Signals related to the smaller notches manufactured by the authors presented small differences for the fatigue cracks, about 12% of the amplitude, while the larger ones presented variations of 35%. However, since they were all flat notches, the depth and shape of the notch bottom influenced the sign of inspection. The signal amplitude variation behavior was in agreement with results found by Kurokawa [18] for the same material.


An alternative system to formulate calibration curves was proposed based on an induced fatigue crack and its width enlargement by applied tensions. This made it possible to delineate a calibration curve for flaws’ widths measurements with only one defect inserted.

It is noticed that the original fatigue cracks presented eddy currents signal amplitude smaller than those from the EDM notches. Once the correction of this attenuation is taken into account carefully before inspecting fatigue cracks, the EDM notches can be successfully used in the calibration process as well. In addition, it has been demonstrated that flaws’ morphology has no influence on the signals from a certain width threshold value because the amplitude increases for wider defects at approximately the same rate for both defect types. For noticeably narrow cracks, the signal decay rate in relation to the width value is more intense, indicating the presence of another concomitant effect, which is the presence of the electrical contacts between the faces of the crack. It modifies the configuration of the interruption of the currents and results in more heightened changes in the signal, even for a low conductive material such as Inconel. As a deduction, an indispensable step is to ascertain the tensions acting on the inspected material as this may cause relevant changes in crack opening, and consequently, in the collected signal. The use of a small probe also provided a good sensitivity since the amplitude variation due to defects presence was high compared to reference signal.

Conflicts of interest

The authors declare no conflicts of interest.


The authors would like to thank to CNPq and CAPES for the financial support and Petrobras for technical and financial assistance for the development of this study.

A.M. Armour.
Eddy current and electrical methods of crack detection.
J Sci Instrum, 25 (1948), pp. 209-210
A.D. Watkins, D.C. Kunerth, T.R. McJunkin.
Eddy current examination of spent nuclear fuel canister closure welds.
Proceedings of International High Level Radioactive Waste Management Conference (IHLRWM),
D. Lamtenzan, G. Washer, M. Lovez.
Detection and sizing of cracks in structural steel using the eddy current method.
U.S. Department of Transportation. Georgetown Pike, (2000),
S. Ross, W. Lord.
(1995), pp. 2043-2049
H. Wirdelius, E. Österberg.
Study of defect characteristics essential for NDT testing methods ET, UT and RT.
Swedish Nuclear Power Inspector, (2000),
M. Kunieda.
Electrical discharge machining processes.
Handbook of manufacturing engineering and technology,
B.F. Larson, C.C. Lo, N. Nakagawa.
Investigation of the effects of notch width on eddy current response and comparison of signals from notches and cracks.
Center for Nondestructive Evaluation, Iowa State University, (2010),
K. Lahdenperä.
Sizing of fatigue cracks using different eddy current techniques and simulation.
VTT Technical Research Centre of Finland, (2015),
W.D. Rummel, B.K. Christner, S.J. Mullen.
The influence of calibration and acceptance criteria on crack detection and discrimination by eddy current techniques.
pp. 939-946
M.E. Ibrahim, G.D. Surtees, G.R. Hugo, H.J. Morton, S.K. Burke.
Discrimination between fatigue cracking and mechanical damage in aircraft fastener holes by eddy-current phase rotation.
Australian Defence Science and Technology Group, (2015),
M. Kane, A. Koshti.
Comparison of signal response between edm notch and cracks in eddy-current testing.
B.A. Auld, F. Muennemann, D.K. Winslow.
Eddy current probe response to open and closed surface flaws.
J Nondestruct Eval, 2 (1981), pp. 1-21
N. Nakagawa, V. Kogan, G. Bozzolo.
Effect of crack closure on eddy current signals.
Rev Progress Quantitat Nondestruct Evaluat, 7A (1987), pp. 173-179
C.C. Lo, N. Nakagawa.
Effects of dynamic and static loading on eddy current NDE of fatigue cracks.
AIP Conf Proc, 1096 (2009),
H. Huang, N. Yusa, K. Miya.
Eddy current testing and sizing of fatigue cracks.
Proceedings of 12th A-PCNDT – Asia-Pacific Conference on NDT,
N. Nakagawa, M. Yang, B.F. Larson, E.M. Madison, D. Raulerson.
Study of the effects of EDM notch width on eddy current signal response.
AIP Conf Proc, 1096 (2009),
E. Mohseni, V. Martin, D.R. Franca.
A model-based study of the effect of semi-elliptical surface notch geometry on the Signal of a Split-D eddy current probe.
ASNT Annual Conference 2016,
M. Kurokawa.
Relationship between electric properties and width of cracks of inconel alloy.
Proceedings of the 13th International Conference on NDE in the Nuclear and Pressure Vessel Industries,
Resistivity materials.
Electrical and magnetic properties of metals, pp. 110-111
K. Newton.
The transparency of fatigue cracks to NDT methods used for the inspection of offshore structures.
Proceedings of Offshore Europe 87,
E.I. Todorov, W.C. Mohr, M.G. Lozev.
Detection and sizing of fatigue cracks in steel welds with advanced eddy current techniques.
AIP Conf Proc, 975 (2008),
O. Esquivel, Y.M. Kim.
Quantitative evaluation of flaw-detection limits of eddy current techniques for interrogating structures beneath thermal protection systems on reusable launch vehicles.
Volpe National Transportation Systems Center, (2005),
P.J. Shull.
Eddy current: basic eletromagnetic principles applied to eddy current.
Nondestrutive evaluation: theory, tecnhiques and applications, pp. 268-291
Copyright © 2019. The Authors
Journal of Materials Research and Technology

Subscribe to our newsletter

Article options
Cookies policy
To improve our services and products, we use cookies (own or third parties authorized) to show advertising related to client preferences through the analyses of navigation customer behavior. Continuing navigation will be considered as acceptance of this use. You can change the settings or obtain more information by clicking here.