IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 31, NO.

5, AUGUST 2021 5700305

Theoretical and Experimental Studies of SMES

Configurations for Design Optimization
Jaime Gómez, Belén Pérez, Pilar Suárez , Alfredo Álvarez , and Belén Rivera

Abstract—In the current energy scenario, due to the increment

in power generation from renewable sources, the importance of
electrical storage systems has increased significantly, and as a
consequence, the study of the improvement of its efficiency and
the design of new storage systems also increased. Superconducting
material permits the design of Superconducting Magnetic Energy
Storage (SMES). The main problem of SMEs is the low energy
density they have, what make the optimization of design to be one
of the keys for inclusion of this elements in the power grid and
other specific applications as, for instance, flux pumps. As the only
basic forms for SMES, and with the objective of its mathematical
optimization, this work (i) evaluates the mathematical equations for
a real solenoidal winding, and (ii) develops the equivalent equations Fig. 1. Energy storage systems classification.
for a toroidal winding, from the electromagnetic laws. Then, (iii) a
practical structure formed by short solenoids connected in series
along a circular axis (quasi-toroidal structure) is studied. Due to the
large number of equations involved in this case, the finite-element
method in used here. Finally (iv), in order to validate the results
without building the complete solenoid (impossible at the time),
one of the magnetic coupling between two solenoids in the quasi-
toroidal winding was developed according with the theoretical
method, and experimentally tested. The study was carried out by
programming different dimensions in order to make conclusions
for a further development of an optimization algorithm. These
conclusions are presented. This work is the first stage for the
optimized design of a SMES, and presents the complete equations Fig. 2. Comparison energy storage systems.
of the real toroidal winding as the base of the outline dimensions of
a practical quasi-toroidal SMES.
efficiency [3] or response velocity [4], but, on the other hand,
Index Terms—Efficiency, electrical storage systems, energy the weakness of this technology is the low energy density that it
density, finite-element models, optimization of design, SMES, exhibits [5]. For this reason, when a SMES is calculated, a good
solenoids. design is especially necessary, and many authors have reported
interesting solutions [6]–[9]. The optimization of dimensions,
I. INTRODUCTION having into account an adequate constraint of parameters, helps
HE energy scene in the world is in a critical spot, fossil to improve such a design. There are some interesting computer
T fuels, the main source of energy in many countries, are run-
ning out. Renewable energy sources are supposed to fill that role
tools for optimization (MS Excel or Matlab, for instance, have
specific tools for this task) but, in all of cases, the mathematic
formulation of the problem is the key of the successful. In the
in the upcoming future, but they need energy storage systems in
order to succeed. There are quite a lot of different technologies follow section we develop the equations for the two main real
available when it comes to storing energy [1], [2], as illustrated configurations used for SMES design (solenoidal and toroidal
in Figs. 1 and 2. SMES play a key role in that scene, present- configuration), which have been compared before [10] under
ing some elements that other technologies cannot provide, as different points of view. Starting from the outline dimensions
obtained from the theoretical equations, a fem (finite elements
method) model for a “quasi-toroidal” configuration, made from
Manuscript received November 26, 2020; revised January 14, 2021; accepted short solenoidal elements, is developed highlighting the param-
February 2, 2021. Date of publication February 16, 2021; date of current eters which must be used in a finer optimization of SMES.
version April 2, 2021. This work was supported in part by the Government
of Extremadura through the funding GR18092, IB18076 and in part by FEDER.
(Fondo Europeo de Desarrollo Regional). (Corresponding author: Belén Pérez.) II. THEORETICAL FRAME
The authors are with the Department of Electrical Engineering, and Elec-
trical Applications of Superconductors (EAS) Group of the University of Ex- The magnetic energy stored in a coil, with self-inductance L,
tremadura, 06006 Badajoz, Spain (e-mail: [email protected]). supplied by a current I, is given by the following expression:
Color versions of one or more figures in this article are available at https:
//doi.org/10.1109/TASC.2021.3059609. 1
Digital Object Identifier 10.1109/TASC.2021.3059609 W = LI 2 (1)
2
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
5700305 IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 31, NO. 5, AUGUST 2021

Fig. 3. Ideal solenoid parameters for magnetic flux density integration.

Fig. 4. Real solenoid. Reference triangle for the calculation of κ.

The inductance of a coil is defined as the variation of its own

magnetic flux, Φ, with its current. Taking N as the number of where the correction factor KL (0 < KL < 1), is known as
turns in the coil, the result for the inductance is: Nagaoka coefficient, and has the following form:
 
dΦ d    8r 2κ2 − 1 1 − κ2
L = N = N B·A , (2) KL = E (κ) + K (κ) − 1 . (7)
dI dI 3lπ κ3 κ3
 is the surface vector of the normal section of the coil.
where A
 is parallel to B,
 (2) can be written as following: In (7) E(κ) and K(κ), are elliptic integrals of first and
When A second kind respectively, being κ the module of the integrals, a
d parameter that depends on the dimensions of the solenoid, and
L = N (BA) (3) is given by:
dI
D
Thus, in order to calculate the magnetic energy stored in a κ = sin θ = √ . (8)
coil, it is necessary to determine the expression of the magnetic D 2 + l2
field and then the inductance. where D = 2r is the diameter of the solenoid cross section, as
We present here the more common configuration to build a Fig. 4 shows.
real SMES, i.e., the solenoidal and toroidal configurations Substituting (7) in (6) and the result in (1) we get the energy
stored in the real solenoid:
 
A. Real Solenoid μ0 N 2 I 2 D3 2κ2 − 1 1 − κ2
W = E (κ) + K (κ) − 1 (9)
The magnetic field created by a current I in a solenoid with 3l2 κ3 κ3
radius r and length l is well known when l >> r, and its N turns
This is the equation, with the adequate constrictions (e.g.,
are really close to each other (ideal solenoid). Fig. 3 shows a
minimum volume to improve the energy density) that have to
scheme of this solenoid. In order to integrate the Ampere’s law,
be used in a mathematical optimization of the design as above
a close path C like in Fig. 3 is chosen.
 inside the solenoid is mentioned.
The magnetic flux density vector B Nagaoka approach is often used in the study of solenoidal
supposed to be constant in all the points: SMES (see e.g., [6] or [9]), but no an equivalent equation is
found for a toroidal geometry. In the next section we present the
 = μ0 IN k
B (4) study of such an equation.
l
(k is the axial unit vector). B. Real Toroid
Using (4) in (3), the inductance can be written: As in the case of the solenoid, the magnetic flux density in an
N turns ideal toroid with main radius R, cross section radius r <<
dBA μ0 N 2 πr2
L = N = . (5) R, and current I, is easy to evaluate by application of Ampere’s
dI l law in cylindrical coordinates:
In real cases, the obtained result (5) is valuable only in the
central region of long solenoids, in which the magnetic field is B = μ0 IN ϕ  (10)
2πR
parallel, and the border effects are negligible. Near the solenoid
ends (5) is no longer valid, and its calculus becomes excessively ϕ is the angular unit vector).
(
complex. For this reason, in order to fit the ideal result to the real In the ideal toroid, (10) is valid for all the points of the cross
value, correction parameters are generally used in engineering. section. If this does not happen, the magnetic field B  on each
The most widely correction used was developed by the Japanese point depends on the length of the field line used to integrate
Physicist Hantaro Nagaoka [11]. According to Nagaoka, the the Ampere’s law. Fig. 5 helps to understand this. The magnetic
inductance of a short solenoid like that in Fig. 4, can be expressed field in a point separated by R’ from the center of the toroid, is:
as following:
 (R ) = μ0 IN ϕ
B  (R − r < R < R + r) . (11)
2
μ0 N πr 2 2πR
L = KL , (6)
l
GÓMEZ et al.: THEORETICAL AND EXPERIMENTAL STUDIES OF SMES CONFIGURATIONS FOR DESIGN OPTIMIZATION 5700305

Fig. 5. Toroidal parameters for calculus. Fig. 7. Scheme of a modular toroid made up of short solenoids.

Finally, using (16) in (2) and the result in (1), the equation of
the energy in a real toroidal coil is found out:
 R+r  1
2 2
1 2RR
W = μ0 N 2 I 2 arctan − 1 dR
2 R−r R 2 + R2 − r 2
(17)
As in the solenoidal coil, this is the equation to be used in a
mathematical optimization of a toroidal SMES.

III. QUASI-TOROIDAL STRUCTURE

Fig. 6. Non-ideal torus cross section. From a practical point of view, the toroidal configuration is
the most convenient form for a SMES. However, manufacturing
issues and structural requirements recommend using a modular
quasi-toroidal structure forms by short straight solenoids [12].
The magnetic flux, Φ, can be calculated with the aid of Fig. 6.
Fig. 7 approximates the shape of this structure.
There, λ(R ) represents an arc with radius R inside the cross
As can be noted, the angle between elements determines the
 is constant in λ(R ) and parallel to
section A of the toroid. B number of them, and the aspect ratio of the toroid determines
 so we can integrate the magnetic flux as following:
vector A, the aspect ratio of the element.
  For optimization of this configuration, a first study of the
Φ = B (R ) dA = B (R ) λ(R )dR . (12) complete real solenoid as in Section II must be done. After taking
A A a decision of the toroid main radius and cross-section radius, the
effect of the modular construction must be studied. Nevertheless,
In order to determinate λ(R ), the angle α that sustains it (see the theoretical formulation of an N-coils modular toroid involves
Fig. 6) must be determined. To do that, the coordinates (xP , yP ) N! equations corresponding to all N! possible couplings between
of the point P have to be obtained. This point is the intersection, pairs of coils.
for x > 0, of the circumferences in Fig. 6, the equations of which The computer resources recommend addressing this study by
appear next to them. Equaling these equations and taking just modeling the system and experimentally validating the model.
positive values of x, the following result is obtained: Here we present an approach to this study.

R 2 + R2 − r 2 A. Modelling the Modular Toroid

yP (R ) = ,
2R
 Using Comsol Multiphysics software, the model of a mod-
xP (R ) = R 2 − yp2 (13) ular toroid was programed to be studied by the finite element
method (fem). Because no specific superconducting properties
From these results, the angle α can be calculated as: are involved in this stage, we have used the Electromagnetic
Fields module (Physic). This Physic uses the Maxwell equations
 1 to solve the vector potential. The possibility of defining variable
2RR 2
2
 xP and calculable parameters in the model permits easily to compare
α (R ) = 2 arctan = 2 arctan −1
yP R 2 + R2 − r 2 results.
(14) To illustrate the proceeding, two modular toroids, with the
and, therefore, same coils but different angle between them, were simulated.
The parameters are listed in Table I.
 1
Fig. 8 shows the graphical results for the magnetic field and
2RR 2
2
   
λ (R ) = R α (R ) = 2R arctan −1 Table II the results for the energy with different currents.
R 2 + R2 − r 2
(15) B. Testing the Model for a Single Coupling
The flux can be found replacing (11) and (15) in (12):
Due to the impossibility of building a real superconducting
  1
modular toroidal coil to compare with the simulation results,
R+r 2 2
2RR
Φ = μ0 IN actan −1 dR (16) a third system was studied. The objective is to validate the
R−r R 2 + R2 − r 2 model (initial and boundary conditions, mainly) using it with
5700305 IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 31, NO. 5, AUGUST 2021

TABLE I
PARAMETERS OF TWO MODULAR TOROIDAL SMES TO BE COMPARED

(∗) Angle is the only programed difference between models. The toroids radii
Fig. 10. Magnetic flux density (T) for the 30° coupling with 2.8 A.
below are calculated as a function of this by the model.

TABLE III
ENERGIES STORED IN TWO COILS COUPLING AT 15° AND 30°

Fig. 8. Magnetic flux density (T) of the modular toroid for a current of 7 A.
(a) 15° angle between coils and (b) 30° angle between coils.

Fig. 11. Electrical circuit and sinusoidal steady state vector diagram corre-
sponding to the experimental setup.
TABLE II
MAGNETIC ENERGY STORED IN THE MODULAR TOROID AT 15° AND 30°

The model previously used was adapted for this case and the
computational results were obtained. As an example, Fig. 10
shows the magnetic field in the 30° coupling with 2.8 A.
The energies stored in both configurations, for different cur-
rents, are reported in Table III.
As an experimental test to verify the validity of this results, the
setup in Fig. 11 was used. This is the conventional no-load test for
couplings or transformers whose vector diagram in sinusoidal
steady state (sss) is shown there too.
As can be reviewed in literature (see, e.g., [13]), by measuring
in sss the real power P, and the effective (rms) values of the
current I and the source voltages V, the value of R (in primary)
can be calculated and then, L and M –note that the voltage in
the secondary voltmeter is VL2 , as this side is under no-load
conditions during the test. The energy in the couple of identical
coils carrying the same current is calculated as:
W = LI 2 + M I 2 . (18)
All the measurements and calculus programmed in the spread-
Fig. 9. Detail of the coupling tested for an angle of 15° between coils. sheet are summarized in Table IV.
Finally, Table V compares the energies obtained in the sim-
ulation with those experimentally calculated. The relative error
of the measurements has also been included in the table. As
a simplified system that presents a geometry feasible to build. can be seen, the experimental results fit satisfactorily the model
This system (Fig. 9) consist of two individual coils separated by results, which gives validity to it and, therefore, to the model of
the same two angles used before. the complete superconducting quasi-toroidal SMES.
GÓMEZ et al.: THEORETICAL AND EXPERIMENTAL STUDIES OF SMES CONFIGURATIONS FOR DESIGN OPTIMIZATION 5700305

TABLE IV REFERENCES
MEASUREMENTS AND RESULTS OBTAINED FROM THE EXPERIMENTAL TESTS
[1] A. Evans, V. Strezov, and T. J. Evans, “Assessment of utility energy storage
options for increased renewable energy penetration,” Renewable Sustain.
Energy Rev., vol. 16, 2012. [Online]. Available: https://doi.org/10.1016/j.
rser.2012.03.048
[2] S. Koohi-Fayegh and M. A. Rosen, “A review of energy storage types,
applications and recent developments,” J. Energy Storage, vol. 27, 2020.
[Online]. Available: https://doi.org/10.1016/j.est.2019.101047
[3] R. Holla, “Energy storage methods - Superconducting magnetic energy
storage - A Review,” J. Undergraduate Res., vol. 5-1, 2015. [Online].
Available: https://doi.org/10.5210/jur.v8i1.7540
[4] A. K. Rohit et al., “An overview of energy storage and its importance in
indian renewable energy sector: Part I—Technologies and comparison,”
J. Energy Storage, vol. 13, 2017. [Online]. Available: https://doi.org/10.
1016/j.est.2017.06.005
[5] D. O. Akinyele and R. K. Rayudu. “Review of energy storage technologies
for sustainable power networks,” Sustain. Energy Technol. Assessments,
vol. 8, 2014. [Online]. Available: https://doi.org/10.1016/j.seta.2014.07.
TABLE V 004
COMPARISON BETWEEN THE SIMULATION AND EXPERIMENTAL RESULTS [6] P. R. Raut, H. J. Bahirat, and M. D. Atrey, “Analytical approach for
optimal HTS solenoid design,” IEEE Trans. Appl. Supercond., vol. 31,
no. 2, Mar. 2021, Art. no. 4900209.
[7] A. W. Zimmermann and S. M. Sharkh, “Design of a 1 MJ/100 kW high
temperature superconducting magnet for energy storage,” Energy Rep.,
vol. 6, 2020. [Online]. Available: https://doi.org/10.1016/j.egyr.2020.03.
023
[8] S. Noguchi, Y. Inaba, and H. Igarashi, “An optimal configuration design
method for HTS-SMES coils taking account of thermal and electro-
magnetic characteristics,” IEEE Trans. Appl. Supercond., vol. 18, no. 2,
pp. 762–765, Jun. 2008.
[9] N. Sugimoto et al., “Compact SMES with a superconducting film in a
IV. CONCLUSION spiral groove on a Si wafer formed by MEMS technology with possible
high-energy storage volume density comparable to that of rechargeable
A mathematical model of the energy in a real toroidal coil have batteries,” Supercond. Sci. Technol., vol. 30, 2017. [Online]. Available:
been developed following the methodology used by Nagaoka https://doi.org/10.1088/0953-2048/30/1/015014
[10] A. Morandi, M. Fabbri, B. Gholizad, F. Grilli, F. Sirois, and V. M. R.
[11] for the real solenoidal coil. Zermeño, “Design and comparison of a 1-MW/5-s HTS SMES with
The formulation contains all the elements necessary for the toroidal and solenoidal geometry,” IEEE Trans. Appl. Supercond., vol. 26,
SMES optimization by means of the available tools, and with no. 4, Jun. 2016, Art. no. 5700606.
the adequate constrictions. [11] H. Nagaoka, “The inductance coefficients of solenoids,” J. College Sci.,
vol. 27, pp. 18–33, 1909.
For the practical quasi-toroidal coil made from short [12] A. Hekmati and R. Hekmati, “Double pancake superconducting coil design
solenoidal coils, a model built using Comsol Multiphysics soft- for maximum magnetic energy storage in small scale SMES systems,”
ware is a simple but effective tool to substitute the high number Cryogenics, vol. 80-1, 2016. [Online]. Available: https://doi.org/10.1016/
of equations involved in the theoretical analysis. j.cryogenics.2016.09.009
The model has been experimentally tested with satisfactory [13] S. Nasar, “Electric machines and electromechanics,” in Schaum’s Outline
Series. Mcgraw-HILL. 2nd ed. 1997.
results.