3630 IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 13, NO.

2, JUNE 2003

Numerical Evaluation of AC Losses in HTS Wires

With 2D FEM Formulated by Self Magnetic Field
Kazuhiro Kajikawa, Toshihiro Hayashi, Ryoji Yoshida, Masataka Iwakuma, and Kazuo Funaki

Abstract—The ac losses of high critical-temperature supercon- II. NUMERICAL ANALYSIS

ducting (HTS) wires are numerically calculated by means of a finite
element method (FEM), which is formulated with a self magnetic A. Governing Equation
field due to an induced current as unknown. The numerical model Let us consider superconducting wires with cross-section of
is straight HTS wires carrying an alternating transport current in
an external ac magnetic field perpendicular to the wire axis. In this
arbitrary shape and infinitely long axis, which have a transport
situation, the electromagnetic field around the wires is given by current in a uniform external magnetic field. It is also assumed
two-dimensional (2D) Maxwell’s equations. It is also assumed that that the wires are surrounded by a normal conductor with very
the transport property is represented by either the critical state high resistivity, which is regarded as a vacuum. If the wire axis
model or the power-law model, in which the electric field is propor- is parallel to the -axis and the external field is applied perpen-
tional to the power of the current density. The obtained losses are
dicular to them, the electromagnetic field is given by two-di-
compared with conventional theoretical curves in several simple
geometries. mensional Maxwell’s equations
Index Terms—AC loss, finite element method, high- supercon- (1)
ductor, self magnetic field.
(2)

I. INTRODUCTION where and are the magnetic

flux density and the magnetic field strength, and

I N ORDER to develop electric power devices wound with

high- superconducting (HTS) wires, it is important to re-
duce their ac losses in the actual electromagnetic configuration.
tively, and
the electric field and the current density, respec-
. In (2), the term for a displace-
ment current is ignored as usual. Furthermore, the following re-
This is because the ac losses are directly concerned with the lationships are widely used as constitutive equations with the
efficiency and the volume of cooling system in the devices. permeability and resistivity of material, and .
Since the winding carrying a transport current is simultane-
(3)
ously exposed to an external magnetic field due to the currents
in the others, it is necessary to evaluate the ac losses under (4)
these realistic situations. Furthermore, the HTS wires devel- If the magnetic field due to an induced current is symbol-
oped recently are produced in the form of a flat tape to im- ized by , the total magnetic field is given
prove their critical current densities [1], [2], and therefore they by
have anisotropy with respect to the direction of external mag-
netic field [3], [4]. Thus the electromagnetic properties of HTS (5)
wires are quite complicated, and it has been very useful to es- with the uniform external field . In this
timate the ac losses numerically, especially with finite element case, (2) becomes
methods (FEM’s) [5]–[11].
In the present study, the ac losses of several model wires are (6)
numerically evaluated by means of a two-dimensional (2D) The combination of (1)–(6) leads to the equation
finite element analysis, which is formulated by a self magnetic
field due to an induced current. The obtained results are also (7)
compared with theoretical curves based on the critical state
model [12] or numerical results with the one-dimensional (1D) Equation (7) is a governing equation of finite element analysis
finite difference method (FDM). used in this article, whose unknown is the self magnetic field
.
As known from (6), another self field may be defined by

Manuscript received August 5, 2002.

(8)
K. Kajikawa, M. Iwakuma, and K. Funaki are with the Research Institute
of Superconductivity, Kyushu University, Fukuoka 812-8581, Japan (e-mail:
with an arbitrary scalar function . On the other hand, the di-
kajikawa@ sc.kyushu-u.ac.jp). vergence of both sides of (7) gives
T. Hayashi and R. Yoshida are with the Graduate School of Information Sci-
ence and Electrical Engineering, Kyushu University, Fukuoka 812-8581, Japan. (9)
Digital Object Identifier 10.1109/TASC.2003.812415

1051-8223/03$17.00 © 2003 IEEE

Authorized licensed use limited to: Universidad Nac de Educación a Distancia (UNED). Downloaded on March 22,2023 at 09:52:54 UTC from IEEE Xplore. Restrictions apply.
KAJIKAWA et al.: NUMERICAL EVALUATION OF AC LOSSES 3631

Therefore, if one sets an initial condition

(10)
the subsequent magnetic field is always subject to (10). If the
permeability does not depend on the position, (10) becomes
(11)
When is used as an initial condition satisfying
(10), the self magnetic field is uniquely determined by (7)
under appropriate boundary conditions because both the rota-
tion and divergence of are defined in (6) and (10) [13].
Integrating (6) over the total area for analysis under consid-
eration, , one obtains a boundary condition for as

(12)

where is the closed path on the boundary, the line element,

the surface element, and the total transport current.
Fig. 1. Transport-current losses in round wire.
B. Finite Element Analysis
In order to solve (7) numerically, the following residual is 1 10 m, respectively. Since the wires under considera-
defined. tion consist of only a superconductor in the mixed state, the
permeability of free space, , can be used everywhere.
(13) The square shapes are used as the total areas for analysis in-
cluding the vacuum, and the tangential component of self field
By using a weighting function , the surface integral of weighed is set constant on the closed path for the sake of simplicity.
residual can become zero as follows. It is also assumed that the resistivity of effective vacuum around
the superconducting wires is equal to 1 10 m [8]. The in-
(14) fluence of these assumptions on the numerical results will be
discussed elsewhere [14]. The ac loss per unit volume per cycle
If is assumed on the boundary around the dissipated in the wires, , can be obtained by the equation
effective vacuum, the expression of can be changed into
(18)

(15) where is the total area of the wire region, . The frequency
of ac field is 60 Hz.
In this paper, the total area is discretized with linear trian- Fig. 1 shows the numerical results of ac losses in a round wire
gular elements, and the interpolation function of each node is carrying only alternating transport currents. The wire has the
used for the weighting function . The discretization of time is radius of 0.55 mm, and the size of the area for analysis is 4.0
made with a technique of the backward difference. mm 4.0 mm. The -value in the power-law model is set at
5, 10, or 20. The broken lines in Fig. 1 represent the numer-
III. NUMERICAL RESULTS OF AC LOSSES ical results obtained by means of the finite difference method
in one-dimensional geometry with respect to the radial direc-
The ac losses are numerically calculated for three model tion, whose numerical errors are estimated less than 0.1%. On
wires, whose transport properties are represented by either the the other hand, the solid line is a theoretical curve with the crit-
critical state model [12] ical state model [15]. As seen in Fig. 1, the ac losses calculated
with the finite element method have a good agreement with the
(16) reference data. The linkage currents in the effective vacuum are
less than 0.2%.
or the power-law model The ac losses in a rectangular wire of 3.8 mm 0.25 mm
are plotted in Fig. 2, where the external ac magnetic fields are
(17) applied parallel to the wide face of the wire with no transport
current. The area for analysis has each side of 8.0 mm [8]. The
where is the flux-flow resistivity, the critical current solid line in Fig. 2 represents the theoretical curve [16] for an
density, the electric-field criterion, and the -value. In infinite slab with the width of 0.25 mm, whereas the broken
this paper, it is assumed that the critical current density and lines are the results obtained by the finite difference method. It
-value are independent of the magnetic field, and the values is found in Fig. 2 that the ac losses calculated with the different
of , , and are fixed at 100 V m, 1 10 A m , and methods agree well with each other.

Authorized licensed use limited to: Universidad Nac de Educación a Distancia (UNED). Downloaded on March 22,2023 at 09:52:54 UTC from IEEE Xplore. Restrictions apply.
 3632 IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 13, NO. 2, JUNE 2003

Fig. 2. Parallel-field losses in rectangular wire.

Fig. 4. AC losses of elliptic wire in external ac magnetic field perpendicular
to major axis. The broken lines are a guide to the eye.

Fig. 3. Transport-current losses in elliptic wire.

Fig. 5. Magnetization losses in elliptic wire exposed to external magnetic
fields with angles with respect to major axis.
Fig. 3 shows the numerical results of transport-current losses
in an elliptic wire, whose major and minor axes are 3.6 and The most popular method is the numerical approach with the
0.2 mm, respectively. The size of total area is equal to that for magnetic vector potential , which is defined by ,
the rectangular wire, and the wire has the - characteristics and the electric scalar potential [6], [7]. Since the vector po-
represented by (16). One can see that the numerical losses al- tential has only the -component in the geometry described
most coincide with the theoretical curve based on the critical above, it always satisfies the Coulomb gauge . In this
state model [17]. The ac losses in the elliptic wire are also given case, the scalar potential comes from the presence of elec-
in Fig. 4, where the external ac fields are applied perpendicular tric charge [18], and therefore gives the potential difference
to the major axis of the wire carrying the alternating transport along the wires. Moreover, the interlinkage magnetic flux per
current. In this case, the power-law model is used. Furthermore, unit length into the area parallel to the -axis can be ex-
Fig. 5 shows the ac losses in the same wire exposed to the ex- pressed by
ternal fields tilted against the major axis.

IV. DISCUSSION (19)

There are several different methods for two-dimensional fi- where is the closed path around , and are the mag-
nite element analysis [6]–[10]. Let us discuss the features of the netic vector potential on two lines of parallel to the -axis. In
finite element method formulated by the self magnetic field as this way, the transport current can be given through the gradient
compared with them. of the scalar potential , and the external magnetic field can be

Authorized licensed use limited to: Universidad Nac de Educación a Distancia (UNED). Downloaded on March 22,2023 at 09:52:54 UTC from IEEE Xplore. Restrictions apply.
KAJIKAWA et al.: NUMERICAL EVALUATION OF AC LOSSES 3633

applied by the interlinkage magnetic flux via the difference be- REFERENCES
tween the vector potential on the boundary. Hence, the fine [1] Y. Iijima, K. Kakimoto, M. Kimura, K. Takeda, and T. Saitoh, “Reel
adjustment may be needed in each time step for the application to reel continuous formation of Y-123 coated conductors by IBAD and
PLD method,” IEEE Trans. Appl. Superconduct., vol. 11, no. 1, pp.
of the transport current and/or the external field. In the numer- 2816–2821, Mar. 2001.
ical analysis with the self field, on the other hand, they can be [2] K. Hayashi, T. Hikata, T. Kaneko, M. Ueyama, A. Mikumo, N. Ayai,
applied directly as explained before. S. Kobayashi, H. Takei, and K. Sato, “Development of Ag-sheathed
Bi-2223 superconducting wires and their applications,” IEEE Trans.
The method utilizing the current vector potential and the Appl. Superconduct., vol. 11, no. 1, pp. 3281–3284, Mar. 2001.
magnetic scalar potential is also used [8], where and are [3] H. Teshima, A. Oishi, H. Izumi, K. Ohata, T. Morishita, and S. Tanaka,
defined by and . In this method, one “Dependence of the magnetization in YBa Cu O thin films on the
applied magnetic field direction,” Appl. Phys. Lett., vol. 58, no. 24, pp.
has to use three unknowns per node, and it may take a long time 2833–2835, June 1991.
for computing. If the self-field method is applied, the unknowns [4] M. P. Oomen, J. Rieger, M. Leghissa, and H. H. J. ten Kate, “Field-angle
are two variables on each node. In the - method, Gauss’ law dependence of alternating current loss in multifilamentary high-T su-
perconducting tapes,” Appl. Phys. Lett., vol. 70, no. 22, pp. 3038–3040,
for magnetic field, , can be changed into June 1997.
[5] H. Ishii, S. Honjo, Y. Iwata, T. Yamada, and H. Hashimoto, “Numer-
ical approach for a description of electromagnetic phenomenon in su-
(20) perconducting composites,” in Advances in Superconductivity X, K. Os-
amura and I. Hirabayashi, Eds. Tokyo: Springer-Verlag, 1998, vol. 2,
pp. 1373–1376.
under the assumption of no position dependence of . Equation [6] C. Hofmann and G. Ries, “Modeling the interactions between magnets
(20) gives a gauge condition different from (11). and granular high-T superconductor material with a finite-element
method,” Supercond. Sci. Technol., vol. 14, no. 1, pp. 34–40, Jan. 2001.
There is a method formulated by only the current vector po- [7] N. Nibbio, S. Stavrev, and B. Dutoit, “Finite element method simula-
tential [9]. In this case, is defined only in the conductor re- tion of AC loss in HTS tapes with B -dependent E -J power law,” IEEE
gion, and the gauge condition of is assumed. Although Trans. Appl. Superconduct., vol. 11, no. 1, pp. 2631–2634, Mar. 2001.
[8] N. Amemiya, S. Murasawa, N. Banno, and K. Miyamoto, “Numerical
this method seems to be similar to that presented in the previous modelings of superconducting wires for AC loss calculations,” Physica
section, it has been pointed out [9] that the vector potential C, vol. 310, no. 1–4, pp. 16–29, Dec. 1998.
differs from the magnetic field because they satisfy their dif- [9] K. Miya, Analytical Electromagnetism and Electromagnetic Struc-
ture. Tokyo: Yokendo, 1995, pp. 187–201.
ferent boundary conditions on the conductor surface. However, [10] J. Lehtonen, J. Paasi, and R. Mikkonen, “Computation of losses in a
it can be considered that the discrepancy between them is re- HTS tape carrying AC transport current in external AC magnetic field
duced under the assumption that the wires are surrounded by at temperatures of 20–40 K,” IEEE Trans. Appl. Superconduct., vol. 10,
no. 1, pp. 1200–1203, Mar. 2000.
the effective vacuum with very high resistivity. [11] E. Vinot, G. Meunier, and P. Tixador, “Different formulations to model
Lehtonen et al. [10] have numerically calculated the ac losses superconductors,” IEEE Trans. Magn., vol. 36, no. 4, pp. 1226–1229,
in the superconducting wires by mean of the finite element July 2000.
[12] C. P. Bean, “Magnetization of hard superconductors,” Phys. Rev. Lett.,
method with the magnetic field as unknown variable. Since vol. 8, no. 6, pp. 250–253, Mar. 1962.
they have focused on the one-dimensional electromagnetic [13] K. Ohta, Introduction to Electrodynamics I. Tokyo: Maruzen, 2000,
situation for the magnetic field, this method cannot be just pp. 195–198.
[14] K. Kajikawa, T. Hayashi, R. Yoshida, M. Iwakuma, and K. Funaki, 15th
extended to the object under consideration in the present study. International Symposium on Superconductivity, Nov. 2002.
This point can be also mentioned for the method [9]. [15] R. Hancox, “Calculation of a.c. losses in a type II superconductor,” Proc.
By using the finite element analysis with the self magnetic IEE, vol. 113, no. 7, pp. 1221–1228, July 1966.
[16] H. London, “Alternating current losses in superconductors of the second
field, the ac losses can be estimated for the superconducting kind,” Phys. Lett., vol. 6, no. 2, pp. 162–165, Sep. 1963.
wires carrying the alternating transport currents in the external [17] W. T. Norris, “Calculation of hysteresis losses in hard superconductors
ac magnetic fields. This method also enables us to calculate the carrying ac: Isolated conductors and edges of thin sheets,” J. Phys. D,
vol. 3, pp. 489–507, 1970.
ac losses of tape-shaped wires exposed to the external fields with [18] W. J. Carr Jr., AC Loss and Macroscopic Theory of Superconductors,
arbitrary angles with respect to the wide faces of them. 2nd ed. London: Taylor & Francis, 2001, pp. 45–46.

Authorized licensed use limited to: Universidad Nac de Educación a Distancia (UNED). Downloaded on March 22,2023 at 09:52:54 UTC from IEEE Xplore. Restrictions apply.