You are on page 1of 7

NDT&E International 54 (2013) 56–62

Contents lists available at SciVerse ScienceDirect

NDT&E International
journal homepage: www.elsevier.com/locate/ndteint

A novel approach for measuring of thickness of induction hardened layers


based on the eddy current method and the finite element modeling
J.M. Szlagowska-Spychalska n, M.M. Spychalski, K.J. Kurzydlowski
Division of Material Design, Faculty of Material Science, Warsaw University of Technology, Woloska Street 141, 02-507, Warsaw, Poland

a r t i c l e i n f o a b s t r a c t

Article history: The thesis concerns applications of the Eddy Current (EC) for measurements of the thickness of the layers.
Received 23 August 2012 Additionally, the special finite element models have been developed which simulate eddy currents in
Received in revised form materials with surface layers. These models can be used for direct measurements of the thickness of the
30 November 2012
induction hardening layers in complex shape elements. The numerical models have been validated by the
Accepted 1 December 2012
Available online 11 December 2012
measurements carried out on specially designed test samples. The results obtained in the study allow for
building-up the reference database for experimental measurements. They also provide tools for analyses
Keywords: of the eddy current impedance diagrams collected under variable probe parameters and for different
Finite element method properties of the layers and substrates.
Non-destructive testing
& 2012 Elsevier Ltd. All rights reserved.
Eddy current
Induction hardened layers

1. Introduction conductivity difference between the layer and the core. More
complicated shapes of the samples should also be taken into
The possibility of using the Finite Element Method (FEM) for account. This issue became the subject of the submitted work.
analysis of electromagnetic phenomena and the eddy currents
calculations was investigated in the early 1980s [1]. The first works
in this field were based on simple two-dimensional models. These 2. Experimental studies
models allow to obtain maps of distribution of electromagnetic field
and current density [2]. In following years scientists began to 2.1. Materials properties
examine the distributions of the fields around defects [3–5]. There
were also the first attempts to determine the characteristics of the The procedure for measuring the thickness of layers has been
signal from the crack [6]. developed for components made of steel AMS 6414 (4340).
Current development of numerical methods and increasing AMS 6414 (4340) is a heat treatable, low alloy steel containing
computational power of computers allowed to take on more nickel, chromium and molybdenum. It is known for its tough-
complex issues, such as the modeling of eddy currents density ness and capability of developing high strength (TS ¼1793 MPa,
distribution for the surface layers and coatings [7]. There has also YP ¼1496 MPa) in the heat treated condition while retaining good
been work devoted to the analysis of the impedance probe and fatigue strength. Typical applications are for structural use, such
consequently the more accurate determination of the character- as aircraft landing gear, power transmission gears and shafts and
istics of the eddy current diagram [8]. Currently, attempts are other structural parts [14]. In order to allow proper modeling
being made for the three-dimensional modeling of the electro- of eddy current effect in the test material and appropriate
magnetic field distribution around the cracks [9,10] and the selection of eddy current frequency, the material properties of
geometry of the crack [11], modeling probe with different cores the steel, such as conductivity and permeability were evaluated.
[12], modeling ferromagnetic materials [13], etc. Measurements of these parameters were performed for samples
Analysis of the papers and the current state of knowledge of from the starting material and for samples from the induction
eddy current modeling, allowed to postulate that significant hardened material (at industrial conditions). The results are
progress would be to perform calculations and quantitative shown in Table 1.
analysis of eddy currents for materials with low electrical
2.1.1. Samples
An important stage of research was the creation of the base of
n
Corresponding author. standard samples with hardened layers. At first authors had the
E-mail address: szjustyna@inmat.pw.edu.pl (J.M. Szlagowska-Spychalska). base of reference samples containing flat samples with marked

0963-8695/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ndteint.2012.12.001
J.M. Szlagowska-Spychalska et al. / NDT&E International 54 (2013) 56–62 57

Table 1
Electromagnetic properties of the AMS 6414 steel.

Properties Not hardened steel Hardened steel

Conductivity ½S=m (2.3 70.01)  106 (1.03 7 0.01)  106


Permeability mR 16.9 7 0.1 11.1 7 0.1

Fig. 2. Microstructure of the layer of the standard sample.

Table 2
Thickness of the cylindrical samples layers.

Number of sample Thickness, 7 0.5 mm

Sample 1 0.60
Sample 2 0.75
Sample 3 1.15
Sample 4 1.45
Sample 5 1.85
Sample 6 2.55

Fig. 1. Values of H as a function of distance from the surface of the standard


from 0.6 to 2.55 mm were examined. The measurements were
sample, thickness 0.75 mm. performed for the frequencies which should allow to obtain
standard depth of penetration (selected for the thickest layer).
However, in the several papers there was suggestion that for
layers: 0.5,1,1.5 and 2 mm. These samples were used to deter- measurements of thickness of hardened layer the better solution
mine the initial parameters of the process. Additionally, it is to use frequency which gives us standard depth of penetration
was decided to create a base of cylindrical samples with different two times bigger than thickness of single layer. According to this
layers (f23  20 mm) which could allow to determine effects requirement chosen frequencies for flat samples were respec-
of sample shape on the eddy current signal. These samples were tively: 5.5 (standard depth of penetration) and 2 kHz (standard
prepared by company which produces parts for the aviation depth of penetration two times bigger than thickness of layer)
industry. and for cylindrical samples: 3 kHz (standard depth of penetration)
and 1.5 kHz (standard depth of penetration two times bigger
2.2. Characterization of the induction hardened layers than thickness of layer). The phase angle for each of the obtained
characteristics was measured. Sample lift-off curves obtained
Further studies concerned the characterization of induction during the eddy currents measurements are shown in Fig. 3.
hardened layer. In order to determine the thickness of the layers
of cylindrical samples, the measurements of hardness values, such
as a function of distance from the surface of the sample, were
conducted (hardness measurements were performed using the 3. Numerical calculations. Distribution of eddy current
Triboindenter HYSITRON) as well as microstructures observations density and lift-off curves obtained by the numerical
(microstructures observations were performed with the use of the simulation
microscope NIKON Epiphot 200). A sample diagram is shown
in Fig. 1. In modeling electromagnetic phenomena using ANSYS soft-
The results revealed a step change in the properties on the ware one of two formulas of calculation could be selected. The
boundary between layer and core. A very clear difference between first one involves the use of scalar magnetic potential formulation
!
the hardness of the core material and the layers allowed to ( T  O: electric vector potentialscalar magnetic potential; [15]).
accurately determine the thickness of the layers. It was also The second one is based on magnetic vector potential formulation
! !
confirmed by observations of microstructures (Fig. 2). ( A , V A : magnetic vector potential, electric scalar potential
The thickness of six cylindrical samples obtained during hard- magnetic vector potential; [16,17]). The selected type of the finite
ness measurements and microstructures observations are shown elements and solutions (transient, harmonic, etc.) determines the
in Table 2. method of calculation. In this study calculations were based on
Magnetic Vector Potential (MVP) formulation.
2.3. Eddy currents measurements To simulate distribution of eddy current density and charac-
teristics of signal for subsequent layers the parametric, 2D and 3D
With a base of reference samples the next stage of testing, models of probe-sample relation were created. Simulations were
involving eddy current measurements, could have been executed. performed in ANSYS 12. The movement of the lowering probe on
Eddy current measurements were performed with the use of the surface of a sample was simulated. Obtained models allow for
the defectoscope MIZ 27SI (ZETEC). The spring loaded, hand-held, examination of:
coreless probes were used. The operating scale of probes was
1–5 and 5–50 kHz. Other probe parameters were: coil diameter  Various location of the probe relative to the sample.
9.5 mm, high 25 mm, number of turns 100. Flat samples from not-  Electromagnetic field distribution depending on the electrical
hardened and hardened steel, flat samples with layer thicknesses properties of surface layer.
ranging from 0.5 to 2 mm and cylindrical samples with layers  Distribution of eddy currents density, etc.
58 J.M. Szlagowska-Spychalska et al. / NDT&E International 54 (2013) 56–62

Fig. 3. Lift-off curves obtained by the eddy current testing for: (a) flat samples with hardening layers, frequency 2 kHz and (b) cylindrical samples with hardening layers,
frequency 1.5 kHz; Defectoscope MIZ 27 SI.

Fig. 4. Lift-off curves indicating a thinning curve for the subsequent layers, frequency Fig. 6. Lift-off curves indicating a thinning curve for the subsequent layers,
5.5 kHz. frequency 2 kHz.

Fig. 5. Thinning curve, frequency 5.5 kHz. Fig. 7. Thinning curve, frequency 2 kHz.

for layers less than 1.5 mm decrease of penetration is significant


For flat samples (2D, axisymmetric model) the simulations were (Fig. 5).
performed for frequencies 2 and 5.5 kHz. Numerical calculations Then, for frequency 2 kHz a measurement range covering layer
for frequency 5.5 kHz (standard depth of penetration) are shown thicknesses up to 2 mm was achieved (Figs. 6 and 7).
in Fig. 4. In the circle the part of the characteristics obtained This was also confirmed by maps of distribution of eddy current
during eddy current measurements can be seen. density (Fig. 8). Again in the circle the part of the characteristics
The best measurement results for frequency 5.5 kHz could be obtained during eddy current measurements can be seen. It was
obtained only for layers with thickness below 1 mm. Especially confirmed that the best results could be obtained for the frequency
J.M. Szlagowska-Spychalska et al. / NDT&E International 54 (2013) 56–62 59

Fig. 8. Distribution of eddy current density ½A=m2  for frequency 2 kHz: (a) layer thickness 0.50 mm and (b) layer thickness 2.00 mm.

Fig. 11. Lift-off curves indicating a thinning curve for the subsequent layers,
frequency 1.5 kHz.
Fig. 9. Lift-off curves indicating a thinning curve for the subsequent layers,
frequency 3 kHz.

Fig. 12. Thinning curve, frequency 1.5 kHz.


Fig. 10. Thinning curve, frequency 3 kHz.

This was also confirmed by distribution of eddy current


which gives penetration two times greater than the estimated density (Figs. 13 and 14). Although this dependence for cylindrical
thickness of the layer. samples is not as visible as it was in the case of flat samples.
For cylindrical samples (3D model) simulations were again
performed for two frequencies: 3 kHz (standard depth of pene-
tration, selected for the thickest layer) and 1.5 kHz (standard 4. Verification and discussion of the results
depth of penetration two times bigger than thickness of layer).
Numerical calculations for frequency 3 kHz (standard depth of Verification was based on comparison between the phase angles
penetration) are shown in Figs. 9 and 10. obtained during numerical simulation and the phase angles
And again it can be seen that better results were achieved for obtained during eddy current measurements. In order to compare
lower frequency (Figs. 11 and 12). the numerical results with experimental results the normalization
60 J.M. Szlagowska-Spychalska et al. / NDT&E International 54 (2013) 56–62

Fig. 13. Distribution of eddy current density ½A=m2  for frequency 1.5 kHz; layer thickness 0.60 mm.

Fig. 14. Distribution of eddy current density ½A=m2  for frequency 1.5 kHz; layer thickness 2.55 mm.

of phase angles was necessary. This was realized by using presented where xN is the normalized phase angle value, xW is the phase
equation (1) angle for subsequent layers (w¼0.6–2.55) [1], xNH is the phase angle
for not hardened steel [1] and xH is the phase angle for hardened
xW xNH steel [1]. Comparison of phase angle values for flat samples is shown
xN ¼ ð1Þ
xH xNH in Table 3.
J.M. Szlagowska-Spychalska et al. / NDT&E International 54 (2013) 56–62 61

Table 3 Table 4
Comparison of experimental and numerical normalized phase angle values for Comparison of experimental and numerical normalized phase angle values for
different thickness of hardening layers, frequency 2 kHz, flat samples. different thickness of hardening layers, frequency 1.5 kHz, cylindrical samples.

Samples Numerical values xN Experimental values xN Samples Numerical values xN Experimental values xN

Not hardened steel xNH 0.00 0.00 Not hardened steel xNH 0.00 0.00
Layer 0.5 mm 0.34 0.35 Layer 0.60 mm 0.17 0.16
Layer 1.0 mm 0.66 0.70 Layer 0.75 mm 0.29 0.21
Layer 1.5 mm 0.83 0.80 Layer 1.15 mm 0.40 0.32
Layer 2.0 mm 0.93 0.95 Layer 1.45 mm 0.51 0.53
Hardened steel xH 1.00 1.00 Layer 1.85 mm 0.68 0.63
Layer 2.55 mm 0.92 0.95
Hardened steel xH 1.00 1.00

Fig. 15. Normalized phase angles, experimental values, flat samples, frequency
2 kHz.

Fig. 17. Normalized phase angles, experimental values, cylindrical samples, frequency
1.5 kHz.

Fig. 16. Normalized phase angles, numerical values, flat samples, frequency 2 kHz.

The best convergence was received for the frequency 2 kHz,


as it can be seen on graphic illustration of the results—Figs. 15
and 16. Fig. 18. Normalized phase angles, numerical values, cylindrical samples, fre-
quency 1.5 kHz.
For frequency 5.5 kHz the convergence was worse, probably
because of too small depth of penetration of eddy currents. This
had an impact on measurement accuracy. Comparison of phase that penetration of eddy current should be two times greater than
angle values for flat samples is shown in Table 4. the estimated thickness of the layer.
Similarly to the case of flat samples, for the lower frequency
1.5 kHz, which gives us the depth of penetration of eddy currents
two times greater than the estimated thickness of the layer, the 5. Conclusions
convergence of experimental results with the results of numerical
analyses was the best. The graphic illustration of the results for It has been assumed in this work that modern methods of
cylindrical samples is (are) shown in Figs. 17 and 18. numerical simulations allow for direct estimation of the layers
At this stage of research it can be concluded that the results thickness from the EC impedance measurements. To this end,
obtained in the course of experimental verification of the model special finite element models have been developed which simu-
allow to assume that models are correct and can be used for late eddy currents in materials with surface layers. These models
estimation of the layers thickness. But it must be kept in mind can be used for direct measurements of the thickness of the
62 J.M. Szlagowska-Spychalska et al. / NDT&E International 54 (2013) 56–62

induction hardening layers in complex shape elements. The [3] Burais N, Foggia A, Nicolas A, Sabonnadiere JC. Electromagnetic field for-
numerical models developed in the study have been validated by mulation for eddy current calculations in nondestructive testing system. IEEE
Trans Magn 1982;18:1058–60.
the measurements carried out on specially designed test samples. [4] Lord W. Applications of numerical field modeling to electromagnetic meth-
The correctness of the method was confirmed by the following ods of nondestructive testing. IEEE Trans Magn 1983;19:2437–42.
results which were achieved during research: [5] Rebican MI. Dependence of simulated ECT signal on defect conductivity. In:
7th international symposium on advanced topics in electrical engineering
(ATEE); 2011. p. 1–4.
1. Lift-off curves obtained by numerical simulations for the sub- [6] Palanisamy R, Lord W. Prediction of eddy current probe signal trajectories.
sequent layers are consistent to lift-off curves obtained by the IEEE Trans Magn 1980;16:1083–5.
eddy current testing (Figs. 3, 4, 7 and 12). [7] Gotoh Y, Matsuoka A, Takahashi N. Electromagnetic inspection technique of
2. Phase angles obtained by numerical simulations for the sub- thickness of nickel-layer on steel plate without influence of lift-off between
steel and inspection probe. IEEE Trans Magn 2011;47:950–3.
sequent layers are consistent with phase angles obtained by [8] Shin Y, Lee J, Song M. Preparation of eddy current impedance plane diagram
the eddy current testing (Tables 3 and 4). by finite element modeling. Key Eng Mater 2004;270–273:579–84.
[9] Marklein R, Rahman MU. Numerical modeling and inverse profiling in non-
destructive testing. In: Applied electromagnetics conference (AEMC); 2009.
The results obtained in the study allow for building-up the
p. 1–4.
reference database for experimental measurements. They also [10] Thomas V, Joubert PY, Vourc’h E, Placko D. A novel modeling of surface
provide tools for analyses of the eddy current impedance dia- breaking defects for eddy current quantitative imaging. In: Sensors applica-
grams collected under variable probe parameters and for different tions symposium (SAS); 2010. IEEE. p. 154–7.
[11] Jesenik M, Gorican V, Hamler A, Trlep M. Finding a crack in a material and
properties of the layers and substrates. Numerical simulations also
determining of depth. In: IET 8th international conference on computation in
allow for estimation of the layers thickness from the EC impedance electromagnetics (CEM 2011); 2011. p. 1–2.
measurements. [12] Zeng Z, Udpa L, Udpa SS. Finite-element model for simulation of ferrite-core
eddy-current probe. IEEE Trans Magn 2010;46:905–9.
[13] Chang D, Zuo X, He Y, Tian GY, Zhang H. Modelling and experimental
Acknowledgments investigation of ferromagnetic material for angular defect detection. In:
17th international conference on automation and computing (ICAC); 2011.
p. 246–50.
This work was in part financed by Developmental Project 0397/ [14] Corus Engineering Steels. UK; 2007.
!
P02/2006/01. This work constitutes part of Ph.D. thesis realized by [15] Ren Z. ( T O) formulation for eddy-current problems in multiply connected
J. Szlagowska-Spychalska. regions. IEEE Trans Magn 2002;38:557–60.
[16] Biro O, Preis K. On the use of the magnetic vector potential in the finite
element analysis of three dimensional eddy currents. IEEE Trans Magn
References 1989;25:3145–59.
[17] Biro O, Preis K, Richter K. On the use of the magnetic vector potential in the
nodal and edge finite element analysis of 3D magnetostatic problems. IEEE
[1] Palanisamy R, Lord W. Finite Element Modeling of Electromagnetic NDT
Trans Magn 1996;32:651–4.
Phenomena. IEEE Trans Magn 1979;15:1478–81.
[2] Nehl TW, Demerdash NA. Application of finite element eddy current analysis
to nondestructive detection of flaws in metallic structures. IEEE Trans Magn
1980;16:1080–2.

You might also like