Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Abstract
A robust feature in multi-frequency eddy current (MEC) testing has been found that can be directly linked to the thickness of the plate
under test. It is shown mathematically that the peak frequency of the imaginary part of the inductance change when an air-cored coil is
placed next to a non-magnetic metallic plate is inversely proportional to the thickness of the plate for a given material. Experimental
results indicate that this relationship also holds for a ferrite-cored U-shaped coil. In addition, this peak frequency has been shown to be
relatively independent of lift-off variations. Use of this new feature provides a fast and accurate method to gauge the thickness of plates.
Measurements made for a sample air-cored and ferrite U-cored coil next to copper and aluminium plates of various thicknesses verified
the proposed method.
r 2006 Elsevier Ltd. All rights reserved.
1. Introduction tions has been explored. For example, the use of multi-
frequency methods in abstracting conductivity and perme-
Eddy current methods have been used in a range of ability profiles along the depth in metal structures has been
technological applications such as thickness measurement, shown [12–14]. The potential of using MEC sensors in non-
quality inspection, coating and surface treatment [1–5]. contact microstructure monitoring in steel production has
Both single-frequency eddy current techniques, where the also been demonstrated [15,16]. Analytical and numerical
coil is excited with a sinusoidal signal, and pulsed eddy solutions for a variety of different sensor geometries have
current (PEC) techniques [6–10], where the coil is excited been published. From these models, it is possible to
with a rectangular stimulus have been developed. The PEC determine and significantly reduce the lift-off effects [17].
technique, as a time domain method, contains frequency- In most of these methods, theoretical impedance values
rich information and shows promising results in detecting were computed using numerical or analytical solutions for
flaws in greater depth. Three features, namely the time to a variety of possible combination of the parameter values,
peak, the peak height and more recently reported the rising a set of parameters for which the theoretical data were
time [11] of the PEC signal, are used in PEC testing to close as possible to the experimental data were chosen. This
identify and characterise defects in metallic targets. usually involves a simplex search method. For a large
However, both single frequency and pulse eddy current number of parameter values, the search process was much
testing can suffer from errors due to variations in the greater than the time required for the measurements. A
distance between the sensor and the test piece, known as feature-based inversion method for conductivity and
lift-off effects. thickness measurement greatly reduced the time [18].
Recently, the potential of using multi-frequency eddy However, these features are still coupled together and
current (MEC) sensors for non-destructive testing applica- therefore a multi-dimensional look-up table has to be
constructed to inverse the parameters.
Corresponding author. Tel.: +44 161 3064808; fax: +44 161 3064789. In this paper, a robust feature has been found in MEC
E-mail address: wuliang.yin@manchester.ac.uk (W. Yin). testing, which has clear physical explanations; can be
0963-8695/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ndteint.2006.07.009
ARTICLE IN PRESS
44 W. Yin, A.J. Peyton / NDT&E International 40 (2007) 43–48
Fig. 2. The imaginary parts of DL0 for copper plates with thickness 22, 44, y, 22 6 mm (simulation results).
Fig. 3. The imaginary parts of DL0 for aluminium plates with thickness 22, 44, y, 22 6 mm (simulation results).
of the integral: Further, it was found that Eq. (11) can be approximated
by
DLðoÞ ¼ fða0 ÞDL0 . (8)
rffiffiffiffiffiffi
o o o
Letting o0 ¼ ð2a20 c þ 2a0 Þ=sm0 c, Eq. (7) can be expressed fða0 Þ ¼ j 1þj þ 0:15j . (12)
as o0 o0 o0
Fig. 5. The imaginary parts of DL0 for aluminium plates with thickness 1–5 mm (experimental results).
Fig. 6. The peak frequency at different lift-offs for aluminium plates of thickness 3 mm.
ARTICLE IN PRESS
W. Yin, A.J. Peyton / NDT&E International 40 (2007) 43–48 47
3. Results and discussion can be easily calculated. Two experiments were performed,
one using air-cored coil for thin copper foils and one using
To examine the robustness of the feature, much thicker ferrite U-cored coil for thicker aluminium plates.
aluminium plates (1–5 mm) were tested using a ferrite- Thicknesses inferred from a number of measurements
cored U-shaped sensor with the results shown in Fig. 5. As are given in Tables 1 and 2, which are in good agreement
expected, the peak frequency decreases as the thickness of with the real thicknesses with the maximum relative error
the plate increases. less than 3%. The difference between the actual value and
Experiments were also performed with both the ferrite- inferred value is due to the approximate nature of the
cored U-shaped sensor and the air-cored sensor to examine relationship and the error caused by the estimation of peak
the robustness of the peak frequency feature in relation to Im DL0 and op from measurements at a finite number of
lift-off variations. It has been found that with the increase frequency points.
of the lift-off, the magnitude of signal decreases, but the The error caused by conductivity deviations can be
peak frequency remains virtually constant. Fig. 6 gives an estimated using Eq. (10). The first-order approximation of
example for the ferrite-cored U-shaped coil above an the error due to conductivity variations can be expressed as
aluminium plate of 3 mm thickness at lift-off 0.0, 0.5 and ðDs=sÞ ¼ ðDc=cÞ. The error in thickness measurement
1 mm. due to lift-off variations has been found to be within in
The constant in equation c op ¼ const, which is 0.5%. Fig. 7 shows an example for the case of the 3 mm
dependent on sensor geometry, can be obtained through thickness aluminium plate at lift-offs 0, 0.2, 0.4, 0.6, 0.8
calibration. Once the constant is determined, the thickness and 1 mm.
The significance in using such a simplified relationship is
that it only requires simple calculation, which may
Table 1
Actual and measured thickness for copper foils
facilitate the realisation of this method in small portable
instruments using for example micro-controllers with
Plate Actual Measured Relative error limited memory and computation power.
thickness (mm) thickness (%)
magnetic plates. The field performance of the sensor will be eddy current methods. Rev Progr Quant Nondestr Eval
evaluated. 1997;16:1593–600.
[9] Yang HC, Tai CC. Pulsed eddy-current measurement of a conducting
coating on a magnetic metal plate. Meas Sci Technol
Acknowledgements 2002;13:1259–65.
[10] Tian GY, Sophian A, Talyor D, Rudlin J. Multiple sensors on pulsed
eddy-current detection for 3-d subsurface crack assessment. IEEE
The authors would like to thank the UK Engineering
Sensors J 2005;5:90–6.
and Physical Sciences Research Council (EPSRC) for its [11] Tian GY, Sophian A. Defect classification using a new feature for
financial support of this research. And we thank the pulsed eddy current sensors. NDT&E Int 2005;38:77–82.
reviewer for his helpful comments. [12] Yin W, Dickinson SJ, Peyton AJ. Imaging the continuous
conductivity profile within layered metal structures using inductance
spectroscopy. IEEE Sensors J 2005;5:161–6.
References [13] Yin W, Dickinson SJ, Peyton AJ. Imaging the permeability
distribution of a layered conductor using inductance spectroscopy.
[1] Takagi T, Miya K. ECT round-robin test for steam generator tube. J In: Proceedings of the 4th world congress on industrial process
Jpn Soc Appl Electromagn Mech 2000;8:121–8. tomography, Aizu, Japan, 2005. p. 1092–7.
[2] Moulder J, Uzal CE, Rose JH. Thickness and conductivity of metallic [14] Sundararaghavan V. A multi-frequency eddy current inversion
layers from eddy current measurements. Rev Sci Instrum method for characterising conductivity gradients on water jet peened
1992;63:3455–65. components. NDT&E Int 2005;38:541–7.
[3] Nonaka Y. A double coil method for simultaneously measuring the [15] Johnstone S, Binns R, Peyton AJ, Pritchard WDN. Using electro-
resistivity, permeability, and thickness of a moving metal sheet. IEEE magnetic methods to monitor the transformation of steel samples.
Trans Instrum Meas 1996;45:478–82. Trans Inst Meas Control 2001;23:21–9.
[4] Placko D, Clergeot H, Santander E. Physical modelling of an eddy [16] Dickinson SJ, Binns R, Yin W, Davis C, Peyton AJ. The
current sensor designed for real time distance and thickness development of a multi-frequency electromagnetic instrument for
measurement in galvanization industry. IEEE Trans Magn monitoring the phase transformation of hot strip steel. In: Proceed-
1989;25:2861–3. ings of IEEE IMTC, Ottawa, Canada, 2005. p. 1091–6.
[5] Sethuraman A, Rose JH. Rapid inversion of eddy current data for [17] Yin W, Binns R, Dickinson SJ, Davis C, Peyton AJ. Analysis of the
conductivity and thickness of metal coatings. J Nondestructive Eval lift-off effect of phase spectra for eddy current sensors. In:
1995;14:39–46. Proceedings of IEEE IMTC, Ottawa, Canada, 2005. p. 1779–84.
[6] Sophian A, Tian GY, Talyor D, Rudlin J. Design of a pulsed eddy [18] Sethuraman A, Rose JH. Rapid inversion of eddy current data for
current sensor for detection of defects in aircraft lap-joints. Sensors conductivity and thickness of metal coatings. J Nondest Eval
Actuators A: Phys 2002;101:92–8. 1995;14:39–46.
[7] Tai CC, Rose JH, Moulder JC. Thickness and conductivity of [19] Dodd CV, Deeds WE. Analytical solutions to eddy-current probe-
metallic layers from pulsed eddy-current measurements. Rev Sci coil problem. J Appl Phys 1968;39:2829–39.
Instrum 1996;67:3965–72. [20] Yin W, Dickinson SJ, Peyton AJ. A multi-frequency impedance
[8] Tai CC, Rose JH, Moulder JC. Characterization of coatings on analyzing instrument for eddy current testing. Meas Sci Technol
magnetic metals using swept-frequency eddy current and transient 2006;17:393–402.