Professional Documents
Culture Documents
Impedance Calculation of A Bobbin Coil I
Impedance Calculation of A Bobbin Coil I
The complex impedance of an air-cored coil in a conductive tube with eccentric inner and outer cylindrical surfaces is calculated. The
analytic expressions for the induced fields and the impedance variation due to the eddy-current flow inside the tube wall are derived using
a second-order potential approach. The addition theorem of Bessel functions is employed to perform the transition between the local
coordinate systems that conform to the boundaries of the structure. Although the model can be used for any coil shape and orientation,
we focus our study on the configuration of a bobbin coil with axis parallel to the axes of the tube surfaces, but not necessarily coinciding
with either of them. The results of the presented analysis are verified by a finite-element-method (FEM) solution.
Index Terms—Eddy currents, impedance calculation, nondestructive evaluation, tube testing.
I. INTRODUCTION the scattering effect, the eddy-current problem not being explic-
itly discussed. Furthermore, the field excitation takes the form
of impinging plane waves and has hence different characteristics
than those of the usual excitations utilized in NDE applications.
XISTING semianalytical models for the simulation of
E eddy-current nondestructive evaluation (NDE) of con-
ductive tubes are restricted to cylindrical tubes with coaxial
The analytical formulation of the eddy-current problem in
concentric tubes excited by a coaxial air-cored bobbin coil was
studied by Dodd et al. [8]. The symmetry of this problem al-
internal and external interfaces. Deviations from this geometry lows its treatment by means of a scalar potential (or a single
in the form of interface eccentricity have not been studied component of magnetic vector potential). Arbitrary 3-D excita-
so far, although they can be of practical interest, e.g., when tions in problems possessing cylindrical symmetry, however, re-
construction imperfections have to be taken into account. Tube sult in coupling between transverse electric and transverse
configurations with eccentric walls can also be considered magnetic solutions, which inhibits the use of a scalar
to model a special case of nonuniform wall thinning in tube potential formulation. The problem can be tackled by treating
inspection. The straightforward extension of such a model to the axial components of the electric and the magnetic field si-
eccentric multilayered structures can be of considerable interest multaneously by means of a dyadic formalism [9], [10]. Fol-
for applications involving tubes with coatings or casings in lowing this approach Lovell and Chew analyzed the effect of
wells when the external layers present a relative drift with tool eccentricity in inductive well-logging [11]. An alternative
respect to the embedded tube. Even if such geometries can be approach was followed by Theodoulidis [12], where the wobble
treated by means of fully numerical tools, like for example the effect of bobbin coils in eddy-current tube testing is analyzed
finite-element method (FEM), they have the disadvantage of a by introducing a second-order potential which is similar to the
raised computational effort. Debye potentials used in [4]. Using the same approach, Burke
Eccentric multilayered cylindrical structures have been ex- and Theodoulidis [13] treated the more elaborate problem of
haustively studied in high-frequency applications. The plane an horizontal coil in a conductive borehole. Their solution for
wave scattering from a dielectric cylinder with an eccentric di- the eddy-current field in the undamaged tube has been used in
electric enclosure was treated by Uzunoglu and Fikioris [1]. The conjuction with the boundary-element method for the efficient
more general problem of scattering from an eccentric multilay- treatment of cracks in cylindrical tubes [14]. Appropriate ex-
ered circular dielectric cylinder was tackled by Kishk et al. [2]. tensions of the second-order potential approach have also made
Stratigaki et al. [3] considered the different senario of a dielec- feasible the analytical treatment of cylindrical coils with arbi-
tric host cylinder embedding multiple separate cylindrical di- trary orientation above conducting plates [15], [16], problems
electric inclusions, while in [4] an extension of this problem of increased practical interest. This method can also be applied
by considering inclusions that are also eccentrically stratified to tackle finite length cylindrical structures like that of a coil
is studied. Roumeliotis et al. [5] studied the scattering from an near the opening of a borehole [17] and the similar problem of
eccentrically coated metallic cylinder. Similar efforts were also a coil above a right-angled conductive wedge [18].
reported in the literature [6], [7]. Even if conductivity effects can This work deals with the semianalytic solution of the eddy-
also be taken into account in most of the above formulations (by current problem in an eccentric, conductive, nonmagnetic tube,
considering lossy dielectrics) the solution is merely focused on following the second-order potential approach. By “eccentric
tube” we hereafter refer to a tube with eccentric inner and outer
cylindrical surfaces. Eddy currents in the tube are excited by an
Manuscript received November 27, 2009; revised March 18, 2010; accepted air-cored bobbin coil whose axis is parallel to the tube axis, and
July 28, 2010. Date of publication August 09, 2010; date of current version Oc-
tober 20, 2010. Corresponding author: A. Skarlatos (e-mail: anastasios.skarlatos which is allowed to be also eccentric with respect to the tube
@cea.fr). surfaces. The presented analysis can be considered as a gener-
Digital Object Identifier 10.1109/TMAG.2010.2064331 alization of the simpler case of the concentric tube eddy-current
0018-9464/$26.00 © 2010 IEEE
3886 IEEE TRANSACTIONS ON MAGNETICS, VOL. 46, NO. 11, NOVEMBER 2010
(1)
(2)
The potential expressions inside the tube wall ( and The boundary conditions on the outer interface ,
) read read
(19)
for ,
(10)
(20)
for , and
(11)
(13)
for , (24)
(14)
for , and
(15)
(26)
(16)
B. Calculation of the Excitation Term: Coil in Free Space
The expression for the excitation term of an air-cored bobbin
coil is derived by the integral representation of the azimuthal
component of the magnetic vector potential produced in free-
space [12]
(17)
(27)
(18)
3888 IEEE TRANSACTIONS ON MAGNETICS, VOL. 46, NO. 11, NOVEMBER 2010
(32)
and
(29)
Applying the addition theorem to the triangle OAB (cf. Fig. 2),
the Bessel function can be written in analogy with
(23)
(30)
where and have been exchanged with and , respectively,
and is the angle between and . Substituting (30) into
(29) and comparing it with the system can be written in the form of (33) at the bottom of
the page.
Notice that for , i.e., for tube without eccentricity, only
the terms of the , survive, and the matrices
reduce to the unit matrix. The system (33) reduces then to the
(31) system of equations with noncoupled modes of a normal tube
with coaxial surfaces. Since more than half of the submatrices
(33)
SKARLATOS AND THEODOULIDIS: IMPEDANCE CALCULATION OF A BOBBIN COIL IN A CONDUCTIVE TUBE 3889
in (33) are diagonal or zero, the system can be solved very ef- offsets and tube eccentricities. For a typical bobbin coil, the
ficiently by applying a variation of the Gauss elimination. The coil offset cannot be large and, moreover, eddy-current testing
inversion of (33) is examined in more detail in the Appendix. is typically applied to thin-wall tubes, i.e., tubes with a wall
thickness much smaller that the average tube radius. Thus,
III. CALCULATION OF THE COIL IMPEDANCE appropriate values for and were found to be small
Having determined the expansion coefficients, the coil and in all cases they were both set to 5. Note that
impedance can be easily obtained following the procedure of for a well-defined linear system. It is interesting to note that the
[12]. The total impedance can be written as the sum of the coil number defines the size of the matrix that needs to be
impedance in free space plus its variation due to the presence inverted. This number is , so in our case we
of the tube, namely solve 30 30 systems for each in the integral calculation.
The configuration studied is shown in Fig. 1. In all cases ex-
(34) amined, the coil has inner radius , outer radius
, height , and is wound with
where wire turns. The coil inductance can be calculated from (35)
and is found to be 943.6 . The tube is made of 304-type
austenitic stainless steel (nonmagnetic) having a nominal con-
ductivity of and a constant inner surface radius
of . The tube eccentricity is created by changing
(35) the outer surface offset with respect to the inner surface. In ad-
The impedance variation can be obtained applying the reci- dition, the coil is also given an offset.
procity theorem Theoretical and numerical results are compared in Table I for
two outer tube surface radii ( and 13 mm), for two outer
(36) tube surface eccentricities ( and 2 mm) and for two coil
position offsets ( and 1 mm). For , the coil is
located next to the thin part of the tube wall, i.e., . For
where the integration extends over the inner surface of the tube each combination, we compute the normalized coil resistance
. , denote the electric field and the magnetic flux den- and coil inductive reactance at four frequencies. The normal-
sity in the absence of the tube, whereas , denote the corre- izing factor is the inductance of the bobbin coil in free-space and
sponding quantities when the tube is present. This equation has the four frequencies are related to the so-called frequency
been shown to take the following form [12]: which is a standard frequency for performing eddy-current tube
testing and is calculated by , where is the
tube conductivity in % IACS and is the tube wall thickness
in mm. The four frequencies, i.e., 13, 26, 52, and 104 kHz cor-
respond to the frequency, which
(37) is about 52 kHz for the particular combination of specific con-
which in turn reduces to the following form upon substituting (8) ductivity (2.5% IACS) and tube wall thickness of the concentric
and (9) and applying Parseval’s theorem for Fourier transform tube with .
and Fourier series Regarding the FEM results, for each geometrical config-
uration a different grid is used with the number of degrees
of freedom (with quadratic elements) ranging from 70 000 to
110 000. The same grid, however, is used for all four frequen-
cies, which means that as frequency is increased the results may
not be as accurate for the real part of the impedance change
(38) due to the smaller skin depth and the abrupt change of the field
across the tube radius. Actually, this is what we observe in the
IV. RESULTS results of Table I. Other than that, the agreement is excellent.
In this section, first we verify the theoretical results by Computation times for each parameter combination are of the
comparing them to numerical ones derived from the 3-D FEM order of 5 min for FEM and 10 s for the proposed model with
package Comsol 3.3. For the evaluation of (38), both the in- a typical personal computer. Note that we do not simulate the
tegration range and the summation limit need to be truncated combination and because in this case
to suitable numbers and . In addition, the there will be a region in the circumference where the tube wall
maximum number of terms in the summation of the addition thickness will be zero and close to zero and it is extremely diffi-
theorem needs to be established. In [13], automatic cult to get results with FEM due to grid construction problems.
integration routines were used for the integration and hence Some results (in the form of impedance loci) are shown next,
did not need to be established. A thorough study on in the normalized complex impedance plane, which is the ac-
and was done in [20]. We followed the latter cepted way of presenting eddy-current testing data as they sim-
and used a Nyquist-type criterion for Fourier transforms to ulate the display of eddy-current instruments. Fig. 3 shows the
decide upon . Integration was done by trapezoidal rule. impedance loci produced by increasing the eccentricity of the
Regarding and , these are greater for larger coil outer surface at the four frequencies. In all cases, the coil has no
3890 IEEE TRANSACTIONS ON MAGNETICS, VOL. 46, NO. 11, NOVEMBER 2010
TABLE I
COMPARISON OF THEORETICAL RESULTS TO RESULTS FROM FEM FOR THE NORMALIZED COIL IMPEDANCE. THE PERCENTAGE DIFFERENCE IS GIVEN SEPARATELY
FOR THE REAL AND IMAGINARY PARTS
V. CONCLUSION
An exact model is presented for the interaction of a coil in an
eccentric tube. In the core of the calculations lie the use of the
addition theorem for Bessel functions and the efficient numer-
ical inversion of the resulting linear system for the calculations
of the unknown expansion coefficients. The system of equations
(33) is solved numerically, that is the reason for identifying the
method as semianalytical. The model is very fast and is verified and
by FEM. It can be used in studies of eddy-current tube inspec-
tions and moreover offers the possibility of several extensions.
For example, when adopting integral methods, a Green’s func-
tion can be constructed that will be helpful for modeling defect
inspections in such eccentric tubes. Moreover, the required in-
cident electric field, which is the field produced by the coil in
the tube without defect can be calculated with the model of this
The submatrices , , are diagonal, so their inversion
paper.
is trivial. The subsystem
APPENDIX
The system of equations (33) can be written in the following
way:
is full, hence there is no benefit in trying to reduce it further.
After direct inversion of this subsystem (using Gauss elim-
ination or LU decomposition) and the computation of ,
, the remaining coefficients can be easily obtained by
backsubstitution.
Let us now consider the axisymmetrical mode more closely.
Let , be the diagonal matrices and the excitation vector if
we hold only the zero-mode elements. From (33) we obtain
(39)
The present form of the matrix is inconvenient since it pos- [6] R. P. Parrikar, A. A. Kishk, and A. Z. Elsherbeni, “Scattering from an
sesses zero elements around the main diagonal (second line) impedance cylinder embedded in a nonconcentric dielectric cylinder,”
IEE Proc—Microw Antennas Propag., vol. 138, no. 2, pp. 169–175,
and the previous procedure will collapse. Rearranging the rows Apr. 1991.
of the zero mode elements we can arrive at an equivalent form [7] J. A. Roumeliotis and N. B. Kakogiannos, “Scattering from an infinite
of the matrix which will be invertible using the manipulations cylinder of small radius embedded into a dielectric one,” IEEE Trans.
presented above for the nonzero modes. Swapping the second, Microw. Theory Tech., vol. 42, no. 3, pp. 463–470, Mar. 1994.
[8] C. V. Dodd, C. C. Cheng, and W. E. Deeds, “Induction coils coaxial
third, and fourth lines of the above system and eliminating the with an arbitrary number of cylindrical conductors,” J. Appl. Phys., vol.
element, we arrive at (42) at the bottom of the page. This 45, no. 2, pp. 638–647, Feb. 1974.
form of the system is compatible with the process introduced [9] W. C. Chew, Waves and Fields in Inhomogeneous Media. New York:
above for the nonzero modes. It suffices to replace the lines of IEEE Press, 1995.
[10] J. R. Lovell and W. C. Chew, “Response of a point source in a multi-
the initial system which correspond to the zero mode with those cylindrically layered medium,” IEEE Trans. Geosci. Remote Sens., vol.
of (42) and to apply the algorithm described above. Attention 25, no. 6, pp. 850–858, Nov. 1987.
must be given to the inversion of the submatrix which is no [11] J. R. Lovell and W. C. Chew, “Effect of tool eccentricity on some elec-
trical well-logging tools,” IEEE Trans. Geosci. Remote Sens., vol. 28,
more diagonal, but it has an additional line of nonzero elements no. 1, pp. 127–136, Jan. 1990.
apart from its main diagonal. Its inverse however can be easily [12] T. P. Theodoulidis, “Analytical modeling of wobble in eddy current
obtained and is given by (43) at the bottom of the page. tube testing with bobbin coils,” Res. Nondestr. Eval., vol. 14, no. 2, pp.
111–126, Jun. 2002.
[13] S. K. Burke and T. P. Theodoulidis, “Impedance of a horizontal coil
in a borehole: A model for eddy-current bolthole probes,” J. Phys. D:
Appl. Phys., vol. 37, no. 3, pp. 485–494, Feb. 2004.
[14] J. R. Bowler and T. Theodoulidis, “Boundary element calculation of
REFERENCES eddy currents in cylindrical structures containing cracks,” IEEE Trans.
Magn., vol. 45, no. 3, pp. 1012–1015, Mar. 2009.
[15] T. Theodoulidis, “Analytical model for tilted coils in eddy-current
[1] N. K. Uzunoglu and J. G. Fikioris, “Scattering from an infinite dielec- nondestructive inspection,” IEEE Trans. Magn., vol. 41, no. 9, pp.
tric cylinder embedded into another,” J. Phys. A: Gen. Phys., vol. 12, 2447–2454, Sep. 2005.
no. 6, pp. 825–834, 1979. [16] T. Theodoulidis and R. J. Ditchburn, “Mutual impedance of cylindrical
[2] A. A. Kishk, R. P. Parrikar, and A. Z. Elsherbeni, “Electromagnetic coils at an arbitrary position and orientation above a planar conductor,”
scattering from an eccentric multilayered circular cylinder,” IEEE IEEE Trans. Magn., vol. 43, no. 8, pp. 3368–3370, Aug. 2007.
Trans. Antennas Propag., vol. 40, no. 3, pp. 295–303, Mar. 1992. [17] T. P. Theodoulidis, “Impedance of an induction coil at the opening
[3] L. G. Stratigaki, M. P. Ioannidou, and D. P. Chrissoulidis, “Scattering of a borehole in a conductor,” J. Appl. Phys., vol. 103, no. 2, pp.
from a dielectric cylinder with multiple eccentric cylindrical dielectric 024905–024905-9, Jan. 2008.
inclusions,” IEE Proc—Microw Antennas Propag., vol. 143, no. 6, pp. [18] T. Theodoulidis and J. R. Bowler, “Interaction of an eddy current coil
505–511, Dec. 1996. with a right-angled conductive wedge,” IEEE Trans. Magn., vol. 46,
[4] M. P. Ioannidou, K. D. Kapsalas, and D. P. Chrissoulidis, “Electromag- no. 4, pp. 1034–1042, Apr. 2010.
netic-wave scattering by an eccentrically stratified, dielectric cylinder [19] G. N. Watson, Theory of Bessel Functions. Cambridge, U.K.: Cam-
with multiple eccentrically stratified, cylindrical, dielectric inclusions,” bridge University Press, 1952.
J. Electromagn. Waves Appl., vol. 18, no. 4, pp. 495–516, 2004. [20] G. Micolau, G. Pichenot, D. Premel, D. Lesselier, and M. Lambert,
[5] J. A. Roumeliotis, J. G. Fikioris, and G. P. Gounaris, “Electromagnetic “Dyad-based model of the electric field in a conductive cylinder at
scattering from an eccentrically coated infinite metallic cylinder,” J. eddy-current frequencies,” IEEE Trans. Magn., vol. 40, no. 2, pp.
Appl. Phys., vol. 51, no. 8, pp. 4488–4493, Aug. 1980. 400–409, Mar. 2004.
(42)
.. .. .. .. .. .. .. ..
. . . . . . . .
(43)
.. .. .. .. .. .. .. ..
. . . . . . . .