2342 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 59, NO.

5, MAY 2012

Analysis of Transformers Working Under Heavily

Saturated Conditions in Grid-Connected
Renewable-Energy Systems
Themistoklis D. Kefalas, Member, IEEE, and Antonios G. Kladas, Senior Member, IEEE

Abstract—In recent years, researchers have proposed trans- v ri Dimensionless relative reluctivity value for Bi2 .
formerless solutions for connecting renewable-energy power µ0 Permeability of vacuum (H · m−1 ).
plants to the grid. Apart from lack of efficiency and increased
µr Dimensionless relative permeability.
cost and weight of the transformer, one of the reasons is the dc
input current that causes transformer saturation. The purpose of
this paper is the development of a finite-element computational I. I NTRODUCTION
tool that is going to aid transformer manufacturers in designing
distribution transformers specifically for the renewable-energy
market. It is based on a generalized macroscopic representation of
electrical steels used in the transformer manufacturing industry
R ECENTLY, many researchers have proposed transformer-
less solutions for connecting renewable-energy power
plants to the grid in order to eliminate the need for con-
that enables the accurate evaluation of electromagnetic field dis- ventional transformers. More specifically, in [1], the authors
tribution of transformer cores under heavily saturated conditions.
Its advantages over conventional formulations include numerical proposed multiple modules of single-switch single-inductor
stability, numerical accuracy, and reduction of iterations of the dc–dc converters for offshore wind farms which can reach “high
Newton–Raphson method. An experimental verification of the voltage gains” without conventional transformers. Transformer-
proposed method is carried out. less offshore wind farm applications are also studied in [2]
Index Terms—Computer-aided analysis, electromagnetic analy- and [3], while in [4], a solution is presented for photovoltaic
sis, finite element methods, magnetic cores, magnetostatics, (PV) plants based on a power electronic transformer which in-
Newton method, nonlinear magnetics, numerical analysis, power cludes dc links, multilevel converters, and a medium-frequency
transformers, soft magnetic materials. transformer. Converters for renewable-energy applications us-
ing high-frequency transformers are developed in [5] and [6].
N OMENCLATURE Finally, multilevel applications for grid-connected converters
a Parameter of the generalized representation of the trans- and PV plants are examined in [7] and [8], respectively. The
former core material (T−2 ). advantages, as well as the disadvantages, of “eliminating the
b Dimensionless parameter of the generalized representa- output transformer from a grid-connected PV system” are de-
tion of the transformer core material. scribed in [9]. The introduction of leakage currents in trans-
B2 Squared flux density (T2 ). formerless PV systems is studied in [10] and [11], whereas
Bi2 ith squared flux density (T2 ). in [12], a modulation technique for three-phase neutral-point-
Bn2 Maximum squared flux density for which vr is experi- clamped multilevel inverters that eliminates leakage currents is
mentally known (T2 ). proposed. In [13], analogous techniques are proposed for mag-
di Second derivative of vr at Bi2 (T−4 ). netic core saturation compensation of grid-connected single-
dn−1 2
Second derivative of vr at Bn−1 (T−4 ). phase power converters. Another disadvantage of eliminating
−1
H Magnetic field intensity (A · m ). the transformer from PV systems is the generation of ground
h Parameter equal to Bn2 − Bn−1
2
(T2 ). currents. It is investigated in [14], where a three-phase PV
−1
M Magnetization (A · m ). inverter with reduced common-mode voltage is proposed.
v Reluctivity (H−1 · m). Due to the transformer robustness, reliability, and efficiency,
v0 Reluctivity of vacuum (H−1 · m). the majority of utilities install conventional distribution trans-
vr Dimensionless relative reluctivity. formers in order to connect large renewable-energy power
plants to the grid [15]. The selection of transformer size can
play a key role in the amount of energy delivered in the grid,
as well as the stability of the network [16]. Furthermore, re-
Manuscript received January 31, 2011; revised May 7, 2011; accepted searchers develop hybrid solutions as the one presented in [17],
June 18, 2011. Date of publication June 30, 2011; date of current version where new step-up dc/dc converter topologies are combined
February 3, 2012.
The authors are with the School of Electrical and Computer Engineer- with a three-phase auxiliary ac link in order to increase the
ing, National Technical University of Athens, 15780 Athens, Greece (e-mail: power density of the converter, whereas in [18], a centralized
[email protected]; [email protected]). nine-switch three-level inverter is developed for a two-string
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. PV and storage system which is interfaced to the grid using a
Digital Object Identifier 10.1109/TIE.2011.2161068 conventional transformer.
0278-0046/$26.00 © 2011 IEEE
KEFALAS AND KLADAS: ANALYSIS OF TRANSFORMERS WORKING UNDER HEAVILY SATURATED CONDITIONS 2343

Nevertheless, the transformerless approach is based on a II. C ONVENTIONAL R EPRESENTATION

number of arguments like reduction of losses, the cost and OF E LECTRICAL S TEELS
weight of power transformers, and the overall efficiency of
If the solution of nonlinear electromagnetic problems by the
the grid-connected system, but most importantly, the trans-
FE method is based on the magnetic vector potential formula-
formerless solution seems promising due to the saturation of
tion, soft magnetic materials are expressed by the relative reluc-
the magnetic cores of transformers that connect renewable-
tivity versus squared flux density curve vr –B 2 rather than the
energy systems to the electricity grid. This is outlined in [19],
magnetization curve B–H, or the relative permeability versus
where a voltage compensation strategy is developed in order
magnetic field intensity curve µr –H. Furthermore, nonlinearity
to eliminate the dc current component that causes transformer
is tackled by the Newton–Raphson iterative method.
core saturation.
The evaluation of vr for B 2 ≤ Bn2 is carried out by cubic
Transformers connecting renewable-energy systems to the
spline interpolation of n + 1 experimentally determined tab-
grid are subjected to the following conditions:
ulated sets of (vri , Bi2 ), where vri is the relative reluctivity
1) DC offset input currents; of the electrical steel under consideration for Bi2 , Bn2 is the
2) distorted supply voltage; maximum squared flux density for which relative reluctivity is
3) load imbalance. experimentally known, and 0 ≤ i ≤ n. In order to compute vr
for B 2 ≥ Bn2 , the conventional approach is the following. The
The aforementioned conditions cause transformer core satu-
first derivative of vr at Bn2 is given by (1), where h is given by
ration that result into increased magnetizing current, increased 2
(2) and dn−1 is the second derivative of vr at Bn−1
reactive losses, overheating, and, in a number of cases, even
 
transformer failure. In particular, the maximum dc current dvr (B 2 )

vrn−1 − vrn dn−1 h
= − + (1)
component injected by power converters and nonlinear loads is dB 2
Bn2 h 6
limited by international regulations since it causes transformer
saturation during a sinusoidal half-period [19]. h = Bn2 − Bn−1
2
. (2)
The solution to the aforementioned problem is the develop-
ment of application-oriented transformer designs specifically For B 2 > Bn2 , relative reluctivity is given by


for the renewable-energy market [20]. Even though transformer   dvr (B 2 )

2
vr (B ) = vrn + B − 2
Bn2 ·
. (3)
manufacturers are well aware of international regulations and dB 2
Bn2
the conditions under which transformers operate, they tend
to underestimate the impact of highly saturated conditions on
transformer operation since they use analytical computational III. P ROPOSED G ENERALIZED R EPRESENTATION
methods or conventional finite-element (FE) commercial codes.
For the Newton–Raphson method to convergence, it is im-
The originality of this paper relies on the development of a
perative to know vr for very high values of B 2 . This is why the
generalized macroscopic representation of electrical steels used
conventional formulation in Section II is widely accepted.
in the transformer manufacturing industry. The advantages of
Nevertheless, as B → ∞, the limit of (3) yields the following
the proposed generalized representation over analytical meth-
equation:
ods and contemporary approaches using a quadratic function 
2



to extrapolate the magnetization curve [21], [22] are the accu-   dv (B )
lim vr (B 2 ) = lim vrn + B 2 − Bn2 ·
r

racy of local field computation of magnetic cores, numerical B→∞ B→∞ dB 2
Bn2
stability, and improvement of convergence characteristics of
the Newton–Raphson iterative method. Also, the approaches in = ∞. (4)
[21] and [22] require the evaluation of four or three parameters,
whereas the proposed generalized representation involves only Equation (4) suggests that for very large excitation levels
two parameters. vr → ∞ or µr → 0. The aforementioned one is, of course, in-
The purpose of this paper is the development of a transformer correct. Considering the constitutive equation for ferromagnetic
core material model [23] which is going to enable the accurate materials (5) and isolating H yield
prediction of the operational characteristics of transformers
B = µ0 (M + H) (5)
working under heavily saturated conditions in the renewable-
energy market. The purpose of this paper is also the develop- H = v0 B − M. (6)
ment of a systematic experimental procedure for the verification
of the specific model. Finally, the authors assembled a FE pack- As B → ∞, the value of magnetization M is negligible
age and integrated the proposed generalized representation of compared to the value of the product v0 B, and the limit of
electrical steels. The specific FE package incorporates magnetic reluctivity v is given by
anisotropy [24]–[27] and loss analysis capabilities [20], [28], H v0 B − M v0 B
[29]. Its main function is the no-load analysis, i.e., the evalu- lim v = lim = lim = lim = v0 .
B→∞ B→∞ B B→∞ B B→∞ B
ation of the magnetizing current, of wound-core, single-phase, (7)
and three-phase shell-type transformers. In contrast, FE codes Equation (7) states the well-known fact that the relative
reported in the literature so far do not take into consideration the reluctivity vr and the relative permeability µr of a soft magnetic
highly saturated region of ferromagnetic materials [30]–[33]. material as B → ∞ tend to unity (vr , µr → 1).
2344 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 59, NO. 5, MAY 2012

From the aforementioned one, it follows that the conven-

tional soft magnetic material representation produces erroneous
results in cases where the excitation level is high. In this paper,
a different formulation is developed. For B 2 ≤ Bn2 , the same
approach is used as shown in Section II. For B 2 ≥ Bn2 , the
authors propose the following function for vr , where a and b are
parameters of the extrapolation function. The dimensions of the
a parameter are T−2 , whereas b is a dimensionless parameter

vr (B 2 ) = 1 − exp −(aB 2 + b) . (8)

The function of (8), as well as its slope, is continuous. Also, it

Fig. 1. Generalized representation of HiB grain-oriented steel M-OH.
can be seen from (9) that, as B → ∞, vr tends to unity since the
exponent term tends to zero. Thus, the proposed extrapolation The a and b parameters were numerically evaluated by a
function yields correct results, and it can be integrated to the preprocessing code which performs the following tasks:
Newton–Raphson iterative technique and to the FE method 1) evaluate di for i = 0, 1, . . . , n, needed for the cubic spline

  interpolation technique;
lim vr (B 2 ) = lim 1 − exp −(aB 2 + b) = 1. (9)
B→∞ B→∞ 2) determine the first derivative of vr at Bn2 using (1);
3) calculate the a and b parameters using (12) and (14),
In order to complete the proposed macroscopic material
respectively.
representation, there are two conditions that must be met:
The specific code was developed by the authors and imple-
1) the continuity of the first derivative of vr at Bn2 ;
ments the conventional cubic spline interpolation technique and
2) the continuity of vr at Bn2 .
the generalized representation developed in this section.
The relative reluctivity at Bn2 and the first derivative of vr at
2
Bn are given by (10) and (11), respectively
     IV. I NTEGRATION TO FE M ETHOD
vr Bn2 = 1 − exp − aBn2 + b (10)

The generalized macroscopic representation in Section III
dvr (B 2 )

   can be integrated to the 2-D or 3-D FE method based on
2
= a exp − aBn2 + b . (11) the magnetic vector potential formulation and the Newton–
dB 2
Bn
Raphson method.
Parameter a is evaluated by substituting (10) into (11) and The exact same procedure used for the integration of the
rewriting, where dvr (B 2 )/dB 2 |Bn2 is given by (1) and vr (Bn2 ) conventional representation of electrical steels to the FE method
is given by (3) by replacing B 2 with Bn2 is used for the generalized representation. The proposed repre-

sentation of electrical steels is achieved by using the composite
dvr (B 2 )

dB 2
B 2 function described in Section III and shown in Fig. 1. The func-
a= n
. (12) tion and its first derivative are continuous for 0 ≤ B 2 ≤ Bn2
1 − vr (Bn2 ) and for B 2 ≥ Bn2 , and the aforementioned two conditions are
Substituting (10) into (12) and rewriting yield essential for the Newton–Raphson iterative method to converge.

The two parameters of the generalized macroscopic represen-
dvr (B 2 )

dB 2
B 2
tation a and b are evaluated only once for each electrical steel,
  
n
= exp − aBn2 + b . (13) and there is no need to reevaluate them at every iteration step of
a the Newton–Raphson method.
Parameter b is evaluated by considering the natural logarithm A large number of FE codes for transformer analysis have
of both sides of (13) and rewriting, where dvr (B 2 )/dB 2 |Bn2 been developed by researchers [30]–[33]. However, none of
and a are given by (1) and (12), respectively them focuses on saturated transformer operation. For that pur-

 pose, a FE package has been assembled by the authors for
dvr (B 2 )

2
2 magnetizing current evaluation of transformers working under
 dB Bn 
b = −aBn2 − ln  . (14) highly saturated conditions. It consists of a total of eight codes.
a Three of the four preprocessing codes are for the FE geometry
construction of the single-phase core-type, single-phase shell-
An application of the proposed generalized core material type, and three-phase shell-type transformers, and the fourth is
representation, for the commonly used one in the transformer used for the FE representation of nonlinear materials and uses
industry, HiB grain-oriented steel M-OH 0.27 mm, is shown in the conventional and the proposed generalized macroscopic
Fig. 1. For B 2 ≤ 4.04 T2 , the vr –B 2 curve is represented by representation in Sections II and III. A first-order triangle
cubic spline interpolation. For B 2 ≥ 4.04 T2 , the vr –B 2 curve mesh generator, a magnetostatic FE solver, and a voltage-
is represented by the proposed function given by (8). The a excited time-stepping FE solver. Both FE solvers are based on
parameter is equal to 3.423 · 10−2 T−2 , and the b parameter is the 2-D magnetic vector potential and incorporate wound-core
equal to −0.132. anisotropy capabilities and the Newton–Raphson method for
KEFALAS AND KLADAS: ANALYSIS OF TRANSFORMERS WORKING UNDER HEAVILY SATURATED CONDITIONS 2345

of 1.19 · 104 , 7.46 · 104 , and 2.15 · 109 A/m2 to the wind-
ing area. Fig. 4 shows the convergence characteristics of the
Newton–Raphson method when the conventional and gener-
alized material representations are used. Figs. 5 and 6 show
the flux density distribution of the single-phase transformer
for the conventional and generalized material representations,
respectively. For a low excitation level, the convergence char-
acteristics of the Newton–Raphson method, as well as the
resulting flux density distribution, are identical for both core
material representations. This is due to the fact that, for low
excitation levels, the vr −B 2 curve is represented by cubic
spline interpolation in both cases. For a medium excitation
level, the convergence error of the Newton–Raphson method is
improved, 2.71 · 10−4 for the conventional representation and
Fig. 2. Single-phase shell-type wound-core transformer. 2.22 · 10−4 for the proposed generalized representation. For a
high excitation level, the Newton–Raphson method needs 29
and 16 iterations in the case of the conventional and generalized
representations, respectively, i.e., the computational effort is
almost doubled when the conventional representation is used.
Fig. 5 shows that the flux density magnitude of the core oppo-
site the winding is lower than that of air between the core and
the winding, i.e., the core material has a relative permeability
lower than unity. The aforementioned problem is solved when
the proposed generalized representation is used. Fig. 6 depicts
the correct flux density distribution and clearly shows that the
flux density in the core material is larger than that of air.

VI. E XPERIMENTAL S ETUPS

A systematic laboratory procedure has been developed in
Fig. 3. FE model of the single-phase shell-type wound-core transformer.
order to determine experimentally the magnetizing currents
under highly saturated magnetic core conditions. Its principle
nonlinearity consideration. The wound-core anisotropy model is quite simple since it uses power supplies of high nomi-
has been developed in [24] and [25] and has been applied nal power for the excitation of wound-core transformers. In
in [20]. Also, the second FE solver is based on coupled this fashion, the impact of highly saturated conditions on the
electromagnetic-circuit analysis and hysteresis models devel- magnetizing transformer current can be determined with low
oped in [28] and has been applied in [20] and [29]. Finally, a cost in a controlled and systematic manner without having
graphics postprocessor is included. to use an experimental apparatus of renewable-energy system
and medium-voltage grid. Two experimental setups were used,
one for wound cores shown in Fig. 7 and one for three-phase
V. C OMPARISON OF C ONVENTIONAL AND
shell-type wound-core transformers shown in Fig. 8. The block
P ROPOSED R EPRESENTATIONS
diagrams of the two experimental setups are shown in Figs. 9
For the comparison of the conventional and generalized and 10.
material representations, the single-phase shell-type wound- In the first experimental setup, a 23-turn excitation coil was
core transformer was used. It is shown in Fig. 2, and it is supplied from a programmable ac and dc power supply (Cal-
constructed of two or four wound cores assembled about a ifornia Instruments MX30) in order to magnetize the wound
preformed winding as described in [20], [34], and [35]. cores. In the second experimental setup, three 20-turn excitation
The preprocessor code of the FE model of the single-phase coils were supplied by a three-phase variable transformer so
shell-type transformer was used. Due to symmetry, only one- as to magnetize the three-phase shell-type transformer. In both
quarter of the transformer was modeled. The model is shown experimental setups, the voltage across the excitation coil ter-
in Fig. 3 and is composed of the wound-core area surround- minals was captured using an active differential voltage probe.
ing a winding area of positive current density. The nonlinear A current probe based on the Hall effect is used for capturing
properties of the conventional grain-oriented electrical steel M4 the magnetizing current.
0.27 mm were assigned to the wound-core area. For the grain- The output of the probes was connected to a noise-rejecting
oriented steel M4 0.27 mm, the a parameter is equal to 1.601 · shielded Bayonet Neill-Concelman (BNC) I/O connector block
10−2 T−2 , and the b parameter is equal to −5.792 · 10−2 . (BNC-2110) via passive probes (TP6060). A noise-rejecting
Magnetostatic analyses were carried out for a low, a medium, shielded cable (SHC68-68-EP) connects the data acquisition
and a high level of excitation by applying a current density (DAQ) device directly to the BNC-2110 connector block.
2346 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 59, NO. 5, MAY 2012

Fig. 4. Comparison of Newton–Raphson convergence for the generalized and conventional core material representations.

Fig. 7. Experimental setup for wound cores.

Fig. 5. Flux density vector plot of the single-phase transformer using the
conventional core material representation, high excitation level.

Fig. 8. Experimental setup for three-phase transformers.

The DAQ device used was National Instruments (NI) 6143.

The specific DAQ device has eight differential channels, an
analog-to-digital converter resolution of 16 bit, a maximum
sampling rate of 250 ksamples/s per channel analog input, and
a ±5 V analog input signal range. Finally, the DAQ device
was placed into a Peripheral Component Interconnect slot of
Fig. 6. Flux density vector plot of the single-phase transformer using the a desktop PC (processor: 1.8 GHz Intel Core 2 Duo; memory:
proposed generalized core material representation, high excitation level. 1 GB DDR2).
KEFALAS AND KLADAS: ANALYSIS OF TRANSFORMERS WORKING UNDER HEAVILY SATURATED CONDITIONS 2347

Fig. 11. (a) Single-phase core-type transformer. (b) FE model.

Fig. 9. Block diagram of the first experimental setup.

Fig. 12. Flux density vector plot of a wound core constructed of M-OH
0.27 mm, 2.0 T.

Fig. 10. Block diagram of the second experimental setup.

Three virtual instruments (VIs) were created with the use of

LabVIEW software. The purpose of the first VI was the real-
time measurement of the excitation coil voltage and no-load
current waveforms and fast Fourier transform (FFT), particu-
larly the peak, rms, and total harmonic distortion (THD) values.
The second VI was used for capturing the voltage and no-load
current waveforms and FFT, for two periods of the fundamental
frequency, and for the maximum sampling rate of the DAQ
Fig. 13. Flux density vector plot of a wound core constructed of M4 0.27 mm,
device. The third VI was used for manipulating the acquired 2.0 T.
voltage and current data in order to compute the no-load loss.
The repeatability error of the experimental setup is ±0.3%, magnetization levels. Two grain-oriented electrical steels were
whereas the absolute error is within ±0.5%. used as core materials, the conventional M4 0.27 mm and the
HiB M-OH 0.27 mm. The anisotropy model in [24] and [25]
VII. A PPLICATION OF THE G ENERALIZED was applied to the wound-core area. Fig. 11(a) shows the single-
M ACROSCOPIC R EPRESENTATION phase core-type transformer which is constructed of a single
wound core assembled about a preformed winding [34]. The
A. Application of the Generalized Macroscopic
preprocessor code of the single-phase core-type transformer
Representation to Wound Cores
was used to set up the FE geometry shown in Fig. 11(b).
The generalized macroscopic representation was applied to Transformer core material selection by the manufacturer
wound cores. Voltage-excited time-stepping analyses using the plays a key role on transformer operation under highly satu-
second FE solver in Section IV were carried out for different rated conditions. This is shown in Figs. 12 and 13 where a
2348 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 59, NO. 5, MAY 2012

Fig. 17. Flux density plot of the three-phase transformer, 2.0 T.

Fig. 14. Three-phase shell-type wound-core transformer.

Fig. 18. Square voltage waveform.

Fig. 15. FE model of the three-phase shell-type transformer.

Fig. 16. Flux density plot of the three-phase transformer, 1.65 T.

considerable difference in local flux density distribution can be

Fig. 19. Experimental and simulated magnetizing currents.
observed due to the different magnetization characteristics of
the M-OH and M4 electrical steels and despite the fact that the C. Experimental Verification
FE analysis was carried out for the same magnetization level
of 2.0 T. In the case of the wound core constructed of the M4 The two experimental setups in Section VI were used in
steel, the local flux density exceeds 2.3 T, and this may lead into order to magnetize a number of wound cores and the three-
increased local losses and temperature rise. phase shell-type transformer in Fig. 14 at highly saturated levels
reaching 2.0 T. In the case of wound cores, a square voltage
supply waveform was used which has a significant impact on
B. Application of the Generalized Macroscopic
the magnetizing current, whereas in the case of the three-phase
Representation to Three-Phase Shell-Type Transformer
transformer, a distorted sinusoidal voltage waveform was used.
Also, the generalized macroscopic representation was ap- Figs. 18 and 19 show the square voltage waveform and the
plied to the three-phase shell-type transformer. Fig. 14 shows resulting experimental and simulated magnetizing currents. The
the three-phase shell-type transformer which is constructed of simulated and experimental results are in good agreement.
four wound cores. The respective preprocessor code of the The computed mean power, i.e., no-load loss, is evaluated by
three-phase transformer was used to set up the FE geometry the input voltage waveform and the simulated magnetizing
shown in Fig. 15. One-quarter of the transformer was modeled current and is underestimated within 5% to 10%.
due to symmetry, and the anisotropy model in [24] and [25] The computed peak magnetizing current, for a magnetization
was applied to the wound-core areas. Figs. 16 and 17 show level of 2.0 T, is overestimated by less than 5% in all cases.
the respective flux density distribution plot of the three-phase In comparison, the computed peak magnetizing currents by
transformer for two magnetization levels. the analytical method used by a transformer manufacturing
KEFALAS AND KLADAS: ANALYSIS OF TRANSFORMERS WORKING UNDER HEAVILY SATURATED CONDITIONS 2349

industry [36] and the conventional electrical steel representa- [13] G. Franceschini, E. Lorenzani, A. Bellini, and A. Fratta, “Compensation
tion in Section II are underestimated by more than 81% and of magnetic core saturation for grid connected single-phase power con-
verters,” in Proc. ICEM, 2010, pp. 1–6.
32%, respectively. [14] G. Vazquez, T. Kerekes, J. Rocabert, P. Rodríguez, R. Teodorescu, and
D. Aguilar, “A photovoltaic three-phase topology to reduce common mode
voltage,” in Proc. ISIE, 2010, pp. 2885–2890.
VIII. C ONCLUSION [15] P. Flores, J. Dixon, M. Ortuzar, R. Carmi, P. Barriuso, and L. Moran,
“Static VAR compensator and active power filter with power injection
This paper has outlined the importance of transformer capability, using 27-level inverters and photovoltaic cells,” IEEE Trans.
operation under highly saturated conditions used in renewable- Ind. Electron., vol. 56, no. 1, pp. 130–138, Jan. 2009.
[16] A. Testa, S. De Caro, R. La Torre, and T. Scimone, “Optimal size selection
energy systems. This paper has proposed an original trans- of step-up transformers in PV plants,” in Proc. ICEM, 2010, pp. 1–6.
former core material representation based on a composite [17] D. Vinnikov and I. Roasto, “Quasi-Z-source-based isolated DC/DC con-
function for the relative reluctivity. The advantages of the pro- verters for distributed power generation,” IEEE Trans. Ind. Electron.,
vol. 58, no. 1, pp. 192–201, Jan. 2011.
posed generalized macroscopic representation over approaches [18] C. S. Edrington, S. Balathandayuthapani, and C. Jianwu, “Analysis of
appearing in the literature are its accuracy, due to the developed integrated storage and grid interfaced photovoltaic system via nine-switch
exponential extrapolation, and the ease of integration in the three-level inverter,” in Proc. IECON, 2010, pp. 3258–3262.
[19] G. Buticchi, E. Lorenzani, and A. Fratta, “A new proposal to eliminate
FE and the Newton–Raphson method. The aforementioned the DC current component at the point of common coupling for grid
advantages were illustrated by a comparison of the proposed connected systems,” in Proc. IECON, 2010, pp. 3244–3249.
transformer core material representation with an analytical [20] T. D. Kefalas and A. G. Kladas, “Harmonic impact on distribution trans-
former no-load loss,” IEEE Trans. Ind. Electron., vol. 57, no. 1, pp. 193–
method used by the transformer industry and the conventional 200, Jan. 2010.
FE nonlinear material representation. [21] L. Janicke and A. Kost, “Numerical modeling for anisotropic magnetic
The proposed generalized macroscopic representation has media including saturation effects,” IEEE Trans. Magn., vol. 33, no. 2,
pp. 1788–1791, Mar. 1997.
been integrated to a FE package for the magnetizing current [22] K. Fujiwara, T. Adachi, and N. Takahashi, “A proposal of finite-element
computation of wound-core transformers. The specific FE analysis considering two-dimensional magnetic properties,” IEEE Trans.
package includes FE solvers that incorporate magnetic aniso- Magn., vol. 38, no. 2, pp. 889–892, Mar. 2002.
[23] T. D. Kefalas, G. Loizos, and A. G. Kladas, “Generalized representation of
tropy and hysteresis models for magnetic loss consideration. transformer electrical steels working under heavily saturated conditions,”
in Proc. ICEM, 2010, pp. 1–6.
[24] T. D. Kefalas, P. S. Georgilakis, A. G. Kladas, A. T. Souflaris, and
R EFERENCES D. G. Paparigas, “Multiple grade lamination wound core: A novel tech-
[1] N. Denniston, A. Massoud, S. Ahmed, and P. Enjeti, “Multiple module nique for transformer iron loss minimization using simulated annealing
high gain high voltage DC–DC transformers for offshore wind energy with restarts and an anisotropy model,” IEEE Trans. Magn., vol. 44, no. 6,
systems,” IEEE Trans. Ind. Electron., vol. 58, no. 5, pp. 1877–1886, pp. 1082–1085, Jun. 2008.
May 2011. [25] T. D. Kefalas and A. G. Kladas, “Robust numerical analysis of wound
[2] M. S. Carmeli, F. Castelli-Dezza, G. Marchegiani, M. Mauri, and core distribution transformers,” in Proc. ICEM, 2008, pp. 1–6.
D. Rosati, “Design and analysis of a medium voltage DC wind farm [26] T. Steinmetz, B. Cranganu-Cretu, and J. Smajic, “Investigations of no-
with a transformer-less wind turbine generator,” in Proc. ICEM, 2010, load and load losses in amorphous core dry-type transformers,” in Proc.
pp. 1–6. ICEM, 2010, pp. 1–6.
[3] M. R. Hanssen, A. G. Ruiz, and M. Molinas, “Operation features of a [27] A. Wilk, J. Nieznanski, and I. Moson, “Nonlinear model of a wound iron
reduced matrix converter for offshore wind power,” in Proc. ISIE, 2010, core traction transformer with the account of magnetic,” in Proc. ICEM,
pp. 2443–2448. 2010, pp. 1–7.
[4] G. Brando, A. Dannier, A. Del Pizzo, and R. Rizzo, “A high performance [28] J. V. Leite, A. Benabou, N. Sadowski, and M. V. F. da Luz, “Finite element
control technique of power electronic transformers in medium voltage three-phase transformer modeling taking into account a vector hysteresis
grid-connected PV plants,” in Proc. ICEM, 2010, pp. 1–6. model,” IEEE Trans. Magn., vol. 45, no. 3, pp. 1716–1719, Mar. 2009.
[5] M. Cacciato, A. Consoli, R. Attanasio, and F. Gennaro, “Soft-switching [29] T. D. Kefalas, G. Loizos, and A. G. Kladas, “Transformer joints FE
converter with HF transformer for grid-connected photovoltaic systems,” analysis using pseudo-source technique,” IEEE Trans. Magn., vol. 47,
IEEE Trans. Ind. Electron., vol. 57, no. 5, pp. 1678–1686, May 2010. no. 5, pp. 1058–1061, May 2011.
[6] S. V. G. Oliveira and I. Barbi, “A three-phase step-up DC–DC converter [30] M. A. Arjona, C. Hernandez, and M. Cisneros-Gonzalez, “Hybrid opti-
with a three-phase high-frequency transformer for DC renewable power mum design of a distribution transformer based on 2-D FE and a manufac-
sources applications,” IEEE Trans. Ind. Electron., vol. 58, no. 8, pp. 3567– turer design methodology,” IEEE Trans. Magn., vol. 46, no. 8, pp. 2864–
3580, Aug. 2011. 2867, Aug. 2010.
[7] O. Lopez, R. Teodorescu, and J. Doval-Gandoy, “Multilevel transformer- [31] X. M. Lopez-Fernandez and C. Alvarez-Marino, “Computation method
less topologies for single-phase grid-connected converters,” in Proc. for transients in power transformers with lossy windings,” IEEE Trans.
IECON, 2006, pp. 5191–5196. Magn., vol. 45, no. 3, pp. 1863–1866, Mar. 2009.
[8] G. Grandi, C. Rossi, D. Ostojic, and D. Casadei, “A new multilevel con- [32] B. Cranganu-Cretu, A. Kertesz, and J. Smajic, “Coupled electromagnetic-
version structure for grid-connected PV applications,” IEEE Trans. Ind. thermal effects of stray flux: Software solution for industrial applications,”
Electron., vol. 56, no. 11, pp. 4416–4426, Nov. 2009. IEEE Trans. Ind. Electron., vol. 57, no. 1, pp. 14–21, Jan. 2010.
[9] R. Gonzalez, E. Gubia, J. Lopez, and L. Marroyo, “Transformerless [33] I. Hernández, J. C. Olivares-Galván, P. S. Georgilakis, and J. M. Cañedo,
single-phase multilevel-based photovoltaic inverter,” IEEE Trans. Ind. “A novel octagonal wound core for distribution transformers validated by
Electron., vol. 55, no. 7, pp. 2694–2702, Jul. 2008. electromagnetic field analysis and comparison with conventional wound
[10] M. C. Cavalcanti, K. C. de Oliveira, A. M. de Farias, F. A. S. Neves, core,” IEEE Trans. Magn., vol. 46, no. 5, pp. 1251–1258, May 2010.
G. M. S. Azevedo, and F. C. Camboim, “Modulation techniques to [34] G. Loizos, T. Kefalas, A. Kladas, T. Souflaris, and D. Paparigas, “Flux
eliminate leakage currents in transformerless three-phase photovoltaic distribution in single phase, Si–Fe, wound transformer cores,” J. Magn.
systems,” IEEE Trans. Ind. Electron., vol. 57, no. 4, pp. 1360–1368, Magn. Mater., vol. 320, no. 20, pp. 874–877, Oct. 2008.
Apr. 2010. [35] G. Loizos, T. D. Kefalas, A. G. Kladas, and A. T. Souflaris, “Flux distribu-
[11] M. Cavalcanti, A. Farias, K. Oliveira, F. Neves, and J. Afonso, “Elim- tion analysis in three-phase Si–Fe wound transformer cores,” IEEE Trans.
inating leakage currents in neutral point clamped inverters for photo- Magn., vol. 46, no. 2, pp. 594–597, Feb. 2010.
voltaic systems,” IEEE Trans. Ind. Electron., vol. 59, no. 1, pp. 435–443, [36] P. S. Georgilakis, N. D. Doulamis, A. D. Doulamis, N. D. Hatziargyriou,
Jan. 2012. and S. D. Kollias, “A novel iron loss reduction technique for distribution
[12] K. C. Oliveira, M. C. Cavalcanti, J. L. Afonso, A. M. Farias, and transformers based on a combined genetic algorithm–neural network ap-
F. A. S. Neves, “Transformerless photovoltaic systems using neutral point proach,” IEEE Trans. Syst., Man, Cybern. C, Appl. Rev., vol. 31, no. 1,
clamped multilevel inverters,” in Proc. ISIE, 2010, pp. 1131–1136. pp. 16–34, Feb. 2001.
2350 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 59, NO. 5, MAY 2012

Themistoklis D. Kefalas (M’09) was born in Greece Antonios G. Kladas (S’80–A’99–M’02–SM’10)

in 1977. He received the Electrical Engineering Ed- was born in Greece in 1959. He received the Diploma
ucator degree from the School of Pedagogical and in electrical engineering from the Aristotle Univer-
Technological Education, Athens, Greece, in 1999 sity of Thessaloniki, Thessaloniki, Greece, in 1982
and the Diploma and the Ph.D. degree in electrical and the D.E.A. and Ph.D. degrees from Pierre and
engineering from the National Technical University Marie Curie University (Paris 6), Paris, France, in
of Athens, Athens, in 2005 and 2008, respectively. 1983 and 1987, respectively.
He is currently with the School of Electrical From 1984 to 1989, he was an Associate Assistant
and Computer Engineering, National Technical Uni- with Pierre and Marie Curie University. From 1991
versity of Athens. His research interests include to 1996, he was with Public Power Corporation S.A.,
transformer and electric machine modeling and Athens, Greece, where he was engaged in the System
optimization. Studies Department. Since 1996, he has been with the School of Electrical
Dr. Kefalas is a member of the Technical Chamber of Greece. and Computer Engineering, National Technical University of Athens, Athens,
where he is currently a Professor. His research interests include transformer
and electric machine modeling and design, as well as the analysis of generating
units by renewable-energy sources and industrial drives.
Dr. Kladas is a member of the Technical Chamber of Greece.