PIERS ONLINE, VOL. 7, NO. 8, 2011 766 A New Analytical Method for Calculation of Eddy Current Distribution and Its Application to a System of Conductor-slab and Rectangular Coil T. Itaya1 , K. Ishida2 , A. Tanaka3 , N. Takehira2 , and T. Miki4 1 Suzuka National College of Technology, Japan 2 Tokuyama College of Technology, Japan 3 Ube National College of Technology, Japan 4 Yamaguchi University, Japan Abstract— This study proposes an analytical method for the eddy current distribution analyses, and provides the eddy current distribution in conductor slab with rectangular coils arranged perpendicular to the slab. Our analytical method utilizes double Fourier transform to derive a set of equations for determining the eddy current distribution. The eddy current density is derived from the analytical solution called stream function. The spatial distribution of eddy current, which is dependent upon the frequency of coil current, the thickness and velocity of a conductor slab, is successfully obtained. Our analytical method is usable for calculations of varieties of eddy current problems, and we demonstrate the eddy current distributions in a conductor slab facing to a rectangular coil. 1. INTRODUCTION The eddy current analysis is widely used in solving the problems on magnetic interaction between an electrically conductive material and an excitation coil that carries an AC current, e.g., the eddy current problems of magnetic resonance diagnosis in medical field, nondestructive testing (NDT) and magnetic sensing in industrial measurement field, induction heating in industrial power application field, etc. Numerical methods such as the finite element method (FEM) and/or boundary element method (BEM) are often employed, and recent increase of computer power have enabled three-dimensional eddy current analysis [1–3], Recently, eddy current distribution (ECD) imaging has also been developed by several research groups [4–6]. ECD imaging of a moving conductor facing to an excitation coil is useful for developing more precise and sophisticated magnetic sensing as well as other varieties of applications. Although the FEM is a potential tool for solving eddy current problems, it is hardly applied to the eddy current problems with moving conductors because of long computational time. Instead, the authors have proposed a new analytical method that enables to obtain the exact solution of magnetic field in a moving conductor facing to a rectangular excitation coil [7, 8]. Using this analytical method, we have developed optional NDT and magnetic sensing techniques, for instance, the thickness and velocity measurements of moving conductors. This article presents an analytical method of ECD in a moving conductor slab facing to a rectangular coil. The ECDs have been obtained using a set of analytical formulae derived by double Fourier transform of Maxwells equations. A shape function is introduced for analyzing ECD excited by an arbitrarily-shaped coil [9, 10], and numerical ECD calculations were carried out with a conventional personal computer. In this article, we demonstrate the ECDs dependent upon the excitation frequency of the coil as well as the thickness and velocity of the moving conductor slab. 2. THEORETICAL ANALYSIS Figure 1 shows a schematic drawing of the coil-conductor arrangement studied in this work. The plane of a rectangular coil is perpendicularly arranged to the conductor slab. In our ECD analyses, we have simply assumed: (1) The moving conductor is isotropic and infinitely wide. (2) The coil is one-turn, and it carries an AC current with a given effective RMS value and angular frequency ω. The coil wire is assumed to be infinitely thin. (3) Conductivity σ, permeability µ and conductor velocity v are all constant. PIERS ONLINE, VOL. 7, NO. 8, 2011 767 z b z0 a I vy 0 y x0 v vx x d (σ , µ 0 . µr ) Figure 1: Arrangement of analytical model. 2.1. Magnetic Flux Density In our previous papers [9], we have introduced shape function to analyze the magnetic field produced by an arbitrarily-shaped coil. However, because of high symmetry of rectangular coil, one can simply express the magnetic flux density for this coil-conductor system in Fig. 1. According to Refs. [7, 8], the x-, y- and z-components of the magnetic flux density B 1 in the conductor slab are given by the following equations: B1x B1y B1z Z∞ Z∞ o √ sin (bη) hn ξ 2 ejx0 ξ µ0 µr I −( ξ 2 +η 2 −γ )d 2γd − (1 + λ ) e + ν e erz = − 0 0 4π 2 η (ξ 2 + η 2 ) 1 − e2γd −∞ −∞ n o i √ 2 2 √ 2 2³ √ 2 2 √ 2 2´ + 1+λ0 −ν0 e−( ξ +η −γ )d e−rz e−z0 ξ +η ea ξ +η −e−a ξ +η e−j(xξ+yη) dξdη, (1) Z∞ Z∞ o √ µ0 µr I ξejx0 ξ sin (bη) hn −( ξ 2 +η 2 −γ )d 2γd = − erz − (1 + λ ) e + ν e 0 0 4π 2 ξ 2 + η 2 1 − e2γd −∞ −∞ n o i √ 2 2 √ 2 2³ √ 2 2 √ 2 2´ + 1+λ0 −ν0 e−( ξ +η −γ )d e−rz e−z0 ξ +η ea ξ +η −e−a ξ +η e−j(xξ+yη) dξdη, (2) Z∞ Z∞ jx0 ξ o √ µ0 µr I sin (bη) hn ξe −( ξ 2 +η 2 −γ )d 2γd erz = −j − (1 + λ ) e + ν e 0 0 4π 2 ηγ 1 − e2γd −∞ −∞ n o i √ 2 2 √ 2 2³ √ 2 2 √ 2 2´ − 1+λ0 −ν0 e−( ξ +η −γ )d e−rz e−z0 ξ +η ea ξ +η −e−a ξ +η e−j(xξ+yη) dξdη.(3) Here, q ξ 2 + η 2 − jσµ0 µr (vx ξ + vy η) + jωσµ0 µr , © 2 ¡ ¢ª ¡ ¢ γ − µ2r ξ 2 + η 2 1 − e−2γd λ0 = ³ , ´2 ³ ´2 p p γ + µr ξ 2 + η 2 − γ − µr ξ 2 + η 2 e−2γd √ 2 2 p 4µr ξ 2 + η 2 γe( ξ +η −γ )d v0 = ³ . ´2 ³ ´2 p p γ + µr ξ 2 + η 2 − γ − µr ξ 2 + η 2 e−2γd γ = (4) (5) (6) Here, σ is the conductivity of the conductor; d is its thickness; vx and vy are the x- and ycomponents of the conductor velocity, respectively; µr is the relative permeability of the conductor; µ0 is permeability of vacuum. The quantities ξ and η are the integration variables of the Fourier transform. The values and depend on the angular frequency of the excitation current, conductor thickness, velocity, and material properties as well as the coil geometry. PIERS ONLINE, VOL. 7, NO. 8, 2011 768 2.2. Eddy Current Density The x- and y-components of ECD in the conductor slab is expressed by the following equations: µ ¶ ∂B1y 1 ∂B1z (7) J1x = − , µ0 µr ∂y ∂z µ ¶ 1 ∂B1x ∂B1z J1y = − . (8) µ0 µr ∂z ∂x 2.3. Stream function We introduce stream function to derive ECDs. The following equation expresses streamline of eddy current that satisfies Jx = J1x (x, y, z) and Jy = J1y (x, y, z) in the x-y plane is Re (Jy ) dy = . dx Re (Jx ) (9) Re (Jy ) dx − Re (Jx ) dy = 0, (10) Therefore, where, Re(. . .) returns the real part of the arguments and gives the instantaneous value of the eddy current density. The stream function U (x, y) in the x-y plane is given by µZ ¶ U (x, y) = Re Jy dx = k = constant, (11) or µZ U (x, y) = Re ¶ −Jx dy = k = constant. (12) 2.4. Stream Function Including Time t Since the eddy current in conductor slab is a function of time t, the stream function U (x, y, z, t) is given by ¶ · µZ ¶ µZ ¶ ¸ µZ √ √ jωt = 2 Re Jy dx cos ωt − Im Jy dx sin ωt = k (13) U (x, y, z, t) = Re Jy dx 2e where, Im(. . .) returns the imaginary part of its arguments and gives the instantaneous value of the eddy current density. The stream function on the z 0 -plane at time t0 is given by ¡ ¢ U x, y, z 0 , t0 = k = constant. (14) From Eq. (14), the constant k is obtained by changing (x, y) and connecting point (x, y) with the equivalent value k. In this way, ECDs on various z-planes are obtained. 3. RESULTS AND DISCUSSION From Eq. (13), one obtained ECDs dependent upon the excitation frequency f , conductor velocity v and conductor thickness d. The ECDs were calculated by adjusting the integration range of the integral using the Gauss-Legendre integration method, and the contour plots were obtained by using MATLAB. In this research, we have calculated the ECDs in metallic aluminum slabs. The specifications of the conductor slab are given in Table 1, and the dimensions of rectangular coil are shown in Table 2. The contour plots shown hereafter are the ECDs at z = 0 and t = 0. For Table 1: Specifications for conductor slab. Table 2: Specifications for coils. Aluminum d = 10 mm σ = 3 × 107 S/m µ = µ0 = 4π × 10−7 H/m Rectangular coil a = 25 mm b = 25 mm z0 = 55 mm x0 = 0 mm PIERS ONLINE, VOL. 7, NO. 8, 2011 769 Table 3: Comparison between our technique and FEM. Proposed method exact solution within two seconds Computational accuracy Computational time 0.04 -4 20 10 20 -20 15 20 6 y [m] -2 4 6 -15 -20 -15 0 5 -6 -6 0 -20 15 -4 0 2 0 -5 -10 4 -4 15 4 -15 0.02 0.02 -0.02 -0.02 4 -4 -0.04 5 0 2 0 x [m] 0.05 0 -1 -0.06 -0.1 0.1 -15 -10 -5 10 -0.05 -2 -0.06 -0.1 15 -4 4 -0.04 0 y [m] 0 10 0.04 -1 0.06 10 0.06 FEM dependent on mesh size 10 times longer 10 -0.05 0 x [m] (a) 0.05 0.1 (b) Figure 2: Effect of excitation frequency on ECD of standstill conductor. (a) ECD at f = 100 Hz and (b) ECD at f = 1 kHz. 0.06 0.06 -4 4 4 -4 0.04 -4 0.04 6 -2 -4 -6 -8 6 -10 0 -4 6 -8 -6 y [m] 6 y [m] -4 0 -6 0 2 2 0 -2 6 4 -4 0.02 4 0.02 -0.02 -4 -0.02 6 4 2 4 2 0.05 -0.06 -0.1 0.1 -0.05 -4 0 -6 -2 -0.05 -2 0 -0.06 -0.1 -4 0 4 -0.04 4 -0.04 0 x [m] 0.05 0.1 x [m] (a) (b) Figure 3: Effect of sliding motion of conductor slab on ECD in the slab with a thickness of 10 mm. The excitation frequency is f = 100 Hz. (a) ECD for vx = 10 m/s and (b) ECD for vx = 20 m/s. 0.06 0.06 6 -6 6 -6 -10 8 -8 10 -60 6 40 -20 -60 y [m] -6 0.05 0.1 -0.06 -0.1 -6 0 4 2 -6 6 -0.05 -4 40 20 -8 -2 0 y [m] -8 -10 8 (a) 8 6 0 x [m] -0.04 0 -4 -20 40 -0.05 0 -0.02 -40 -0.06 -0.1 10 60 -0.02 0 2 4 -40 0 20 60 0 -2 -4 0.02 0.02 -0.04 -8 8 -40 -40 0.04 40 40 0.04 0 x [m] 0.05 0.1 (b) Figure 4: The ECDs of conductor-slab with thicknesses of (a) d = 1 mm and (b) d = 5 mm. The excitation frequency is f = 1 kHz. separate computation of the real and imaginary part of the ECDs, in double precision, the CPU time for one ECD calculation on a 3.2 GHz Core i5 PC is about 1.2 s. The comparison between our PIERS ONLINE, VOL. 7, NO. 8, 2011 770 technique and the FEM are shown in Table 3. The accurate solution of ECDs can also be obtained using the FEM. However, it requires a long computational time than our technique. First, the velocity of conductor slab was set to be vx = vy = 0. Figs. 2(a) and (b) show the ECDs at f = 1 kHz and 10 kHz, respectively. The direction of eddy currents on either side of the x = 0 line is opposite each other, which is concordant with our common sense. Upon increasing frequency f , the magnitude of eddy current increases, and the ECD becomes more concentrated around the portions where two intense eddy currents arise. Next, we investigated the effect of the sliding motion of conductor slab on ECDs. Figs. 3(a) and (b) show the ECDs at vx = 10 and 20 m/s, respectively, and vy = 0. The excitation frequency was chosen to be f = 100 Hz. The contour plots of Fig. 3 indicate that the sliding motion causes asymmetric ECDs. Two types of eddy currents, one is shown by counter-clockwise vectors and the other by clockwise vectors in Fig. 3, simultaneously shift towards the −x direction. The ECD shift becomes larger, as the slab velocity increases. The effect of excitation frequency on the ECD shift was also examined. It is found that the sliding motion of conductor slab gives rise to no significant change in ECD at f = 1 kHz, when vx is less than 20 m/s. Figs. 4(a) and (b) show the ECDs of conductor slabs with thicknesses of d = 1 mm and d = 5 mm. The ECDs are calculated at vx = vy = 0 and at f = 1 kHz. The magnitude of the surface eddy current of thin conductor is larger than that of thicker slab. The eddy current is strongly influenced when the conductor slab thickness is smaller than the flux penetration depth. Since the penetration depth is about 2.68 mm in aluminum at f = 1 kHz, high eddy current for the aluminum slab with d = 1 mm reflects the penetration effect. 4. CONCLUSION In this paper, a theoretical solution was derived that exactly displays the eddy current distribution occurring when a rectangular coil is arranged perpendicular to the conductor slab, which was until now unsolved. The flow of the eddy current, which arises in the conductor as a function of excitation frequency, thickness of the conductor slab, and speed of the slab, has been clarified by this theoretical solution. This analysis method enables the immediate determination of the eddy current irrespective of whether the conductor slab is moving. Moreover, it is considered useful not only in sophisticated analyses of eddy currents but also in other types of eddy current analyses. ACKNOWLEDGMENT The authors gratefully acknowledge Mr. K. Nawata for his cooperation. REFERENCES 1. Cacciola, M., S. Calcagno, G. Megali, D. Pellicano, M. Versaci, and F. C. Morabito, “Eddy current modeling in composite materials,” PIERS Online, Vol. 5, No. 6, 591–595, 2009. 2. Tsuboi, H., N. Seshima, I. Sebestyn, J. Pv, S. Gyimthy, and A. Gasparics, “Transient eddy current analysis of pulsed eddy current testing by finite element method,” IEEE Trans. on Magnetics, Vol. 40, No. 2, 1330–1333, 2004. 3. Preis, K., O. Biro, and I. Ticar, “FEM analysis of eddy current losses in nonlinear laminated iron cores,” IEEE Trans. on Magnetics, Vol. 41, No. 5, 1412–1415, 2005. 4. Theodoulidis, T. P. and E. E. Kriezis, “Impedance evaluation of rectangular coils for eddy current testing of planar media,” NDT and E International, Vol. 35, 407–414, 2002. 5. Yin, W. and A. Peyton, “Sensitivity formulation including velocity effects for electromagnetic induction systems,” IEEE Trans. on Magnetics, Vol. 46, No. 5, 1172–1176, 2010. 6. He, Y., F. Luo, and M. Pan, “Defect characterisation based on pulsed eddy current imaging technique,” Sensors and Actuators, Vol. A, No. 164, 1–7, 2010. 7. Tanaka, A., N. Takehira, and K. Toda, “Analysis of a rectangular coil facing moving sheet conductor,” IEEJ. Trans. A, Vol. 101, No. 8, 405–412, 1981. 8. Itaya, T., K. Ishida, A. Tanaka, and N. Takehira, “Analysis of a fork-shaped rectangular coil facing moving sheet conductors,” IET Science, Measurement and Technology, Vol. 3, No. 5, 279–285, 2009. 9. Ishida, K., T. Itaya, A. Tanaka, and N. Takehira, “Magnetic field analysis of an arbitrary shaped coil using shape functions,” IEEE Trans. on Magnetics, Vol. 45, No. 1, 104–112, 2009. 10. Ishida, K., T. Itaya, A. Tanaka, N. Takehira, and T. Miki, “Arbitrary-shaped single-layer coil self-inductance using shape functions,” IET Science, Measurement and Technology, Vol. 45, No. 1, 21–27, 2010.