INTERNATIONAL TRANSACTIONS ON ELECTRICAL ENERGY SYSTEMS

Int. Trans. Electr. Energ. Syst. (2014)

Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/etep.1964

Frequency domain model for zero-sequence electromagnetic

harmonic filter performance analysis

José C. Oliveira1, Luís C. O. Oliveira2, Fernando N. Belchior3,

Rodrigo N. Oliveira2 and João Areis F. Barbosa Jr.1*,†
1
Faculty of Electrical Engineering, Federal University of Uberlândia, 38408-100, Uberlândia, MG, Brazil
2
Power Quality Study Group, Federal University of Itajubá, Itajubá, MG, Brazil
3
Faculty of Electrical Engineering, State University of São Paulo, Ilha Solteira, SP, Brazil

SUMMARY
Using frequency domain approach, this paper focuses on a zero-sequence harmonic filter model. The model
is based on electromagnetic arrangement created through a three-legged core construction and zigzag
windings that provide a low-priced, robust and efficient equipment to reduce zero-sequence fundamental
and harmonic distortion. The methodology adopted here provides a useful technique for evaluating the
steady-state operation of the filter with emphasis on its efficiency at reducing harmonics and at identifying
the main constructive and electrical parameters of influence on the product performance. The representation
is based on sequence harmonic transmittances that were found helpful in estimating the filter operational
characteristics under both ideal and non-ideal conditions. Furthermore, the effectiveness of the filter and
its representation are fully validated using an equipment prototype and laboratory results. Copyright ©
2014 John Wiley & Sons, Ltd.

key words: frequency domain model; electromagnetic device; harmonics filters; zero-sequence current;
power quality

1. INTRODUCTION

The increasing number of nonlinear and unbalance loads in distribution systems has been
reported as the cause of many occurring problems for the utilities [1,2]. This is the case of the
classical zero-sequence components being the triple order and its integer multiples the more
relevant ones [3]. These components are recognized as typical harmonics produced by single-
phase non-linear loads such as switched and linear power supplies, compact fluorescent lighting
and specific variable speed drives [4,5]. The proliferation of these loads in distribution networks
has directly contributed to the increase in the level of neutral current, even under conditions of
balanced three-phase load.
In general, the neutral fundamental and harmonic current may be responsible by effects comprising
undesirable operation conditions, failures and damages to the installation [6,7]. This is the case of
overheating of neutral conductors, malfunctioning of equipment due to neutral to earth voltages, low
efficiency and overheating of Uninterrupted Power Supply, tripping of protection units for no obvious
reason, rising interference in communication networks and power consumption.
From this background arises the search for approaches aimed at controlling the level of neutral
current circulation along the electric feeders, especially the zero-sequence components. This subject
has been investigated in [8–11], in which, in addition to traditional techniques based on passive filters
[3,12], the effectiveness of electromagnetic filters and other approaches are highlighted.

*Correspondence to: João Areis Ferreira Barbosa Júnior, Faculty of Electrical Engineering, Federal University of
Uberlândia (UFU), 38408-100, Uberlândia, MG, Brazil.
†
E-mail: [email protected]

Copyright © 2014 John Wiley & Sons, Ltd.

 J. C. OLIVEIRA ET AL.

Reference [11] describes and compares several methods for neutral current compensation using
shunt electromagnetic technologies. In [13] a zigzag transformer configuration for the reduction of
harmonics and neutral current is given. It is shown that the use of such devices provides a cost-effective
method, as compared to other procedures of neutral current compensation. Another approach, as given in
[14] proposes a novel electromagnetic strategy based on a series device to block the zero-sequence
components. The results reported in these papers are quite encouraging to demonstrate the effectiveness
of utilizing the shunt and series electromagnetic strategies.
A few references consider, besides the electromagnetic approach, the zero-sequence harmonic
current mitigation throughout electronic techniques or a combination of both electromagnetic and elec-
tronic devices. In this regard, [15] presents a power filter comprising a three-phase three-wire active
filter with a zigzag transformer. In [16], a hybrid filter is claimed to offer advantages over the conven-
tional schemes to suppress neutral-current harmonics in three-phase four-wire electrical distribution
systems. Considering the possibility of harmonic compensation by either a passive filter or a hybrid
four-branch star filter, based on the FBS topology, [17] showed that the proposed scheme could offer
an effective and economical solution for current conditioning in three-phase four-wire networks. As
for more advanced techniques based on power electronics, [18] deals with a neutral network controlled
Distribution Static Compensator using dSPACE processor. This technique was implemented for power
quality improvement in a three-phase four-wire distribution system. A three-leg voltage source
converter with a zigzag transformer was used in [19] and [20] as a distribution static compensator
for compensation of the reactive power, harmonic currents, unbalanced loads and the neutral current
in three-phase four-wire distribution system. Although the electronic devices, while dealing with a
zigzag transformer under ideal and non-ideal power conditions, can effectively provide zero-sequence
harmonic suppression in accordance with [21], it has been recognized that the electromagnetic unit can
effectively attenuate the neutral current and zero-sequence harmonic currents throughout a lower cost
equipment than the three-phase four-wire active power filter.
Therefore, considering the already obtained performance success of computational and experimental
electromagnetic proposal [9], such principle is still an attractive approach for the suppression of zero-
sequence harmonic and fundamental currents. In spite of the experimental success demonstrated in the
reference cited, development of a comprehensive mathematical model enabling full knowledge of
operation and of the parameters that influence product performance is of great value for both physical
analysis and design considerations. In this way, the use of frequency domain techniques is considered
to be of great value as the information measured is often related to this domain.
Focusing on the above, this paper outlines the proposed mathematical model to represent steady-
state electromagnetic zero-sequence harmonic filter performance, studies towards the validation of
the proposal efficacy and the approach validation using an experimental arrangement consisting of a
filter prototype in a typical installation of dominant zero harmonic generation. In this regard, this study
makes improvements over [9] through the use of a simpler and more efficient strategy for the represen-
tation of the filter by using electrical equivalents on the frequency domain, based on sequence
components and transmittance matrices. This offers simplicity in representation, greater ease of perfor-
mance interpretation and of the device correlation with its electrical parameters, as well as providing
guidelines for targeting designs for the equipment in question.

2. FILTER FREQUENCY DOMAIN MODEL

The effectiveness of the harmonic filtering system can be evaluated using computational models for
calculation of transmittances and the respective responses in frequency [22–24]. This methodology
is based on the techniques that use frequency domain and superposition theory. It is particularly
attractive because it allows prompt identification of the most influential factors that affect the overall
system, load and filtering unit behavior. Using three-phase representation, any non-ideal situation
regarding phase imbalances may also be taken into account.
In accordance with [9], the electromagnetic device under consideration is constituted by two
windings, namely I and II. Both coils have the same number of turns. The physical winding terminal con-
nections can be identified as the classical zigzag arrangement. Following this structure, each magnetic

Copyright © 2014 John Wiley & Sons, Ltd. Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep
 ZERO-SEQUENCE ELECTROMAGNETIC HARMONIC FILTER

core column will produce a magnetomotive force (MMF) from two sources corresponding to coils I and
II. This interaction, combined with constructional and operational electrical system considerations, is
expected to produce a low impedance path, which, in turn, attracts the zero-sequence components from
being injected into the mains. Clearly, the lower the filter zero-sequence impedance is in comparison to
the supply impedance, the higher the level of zero-sequence current will have to be deviated to the shunt
connected device [9,10]. Figure 1(a) illustrates the physical arrangement of the device, while Figure 1(b)
highlights the magnetic coupling combinations associated with the first column main winding and the
remaining ones. The same applies to any other coil.
Figure 2 represents an electrical circuit equivalent to the previous arrangement. Using this circuit,
and by ignoring the resistances, the voltage across the windings I and II, for a general harmonic order
h, will be described by Equations (1) and (2).

LIR þ M I;II M I;II M I;I

  I;II
M I;I
3h 2     3 2 3h
V IR LR
2
RS RT RS RT IR
6 I7 6  I;II I;I   I I;II   I;II I;I  76 7
4 V S 5 ¼ jhω:64 LS M SR LS þ M ST M SR M ST 7 5: 4 I S 5 (1)
I  I;II I;I   I;II I;I I;II
VT IT
  I 
M TS M TR LT M TS LT þ M TR
M II;I M II;II M II;I M II;II LIIR þ M II;I
2 II 3h 2       3 2 3h
VR TR RS RS RT RT IR
II;I  II;I II;II   II;I II;II  7 6
6 II 7 6  II  7
4 V S 5 ¼ jhω:4 LS þ M SR M RS M ST M ST M SR 5: 4 I S 5 (2)
6 7

V IIT
 II;I
M TR M II;II LT þ M II;I M II;I M II;II IT
  II   
TS TS ST TR

where Lxp is the self-inductance related to winding x, connected to phase p; M x;y

pq is the mutual inductance
x
between x and y windings connected to phases p and q, respectively;V p is the complex voltage drop across
winding x, connected to phase p; Ip is the complex current in phase p; and h is the harmonic order.
By having the individual voltages at windings I and II, it is possible to obtain the busbar supply
voltages here defined as R, S and T phase voltages, for a specific h harmonic order. The result is given
by Equation (3).
2 I 3h 2 II 3h
VR h
2 3
VR VS
6 I7 6 II 7
4 VS 5 ¼ 4 VS 5 þ 4 VT 5 (3)
6 7

VT V IT V IIR

From the foregoing, using vector representation, the theoretical basis to support the filter operational
principles is to be related to symmetrical components, instead of R, S and T phases. Therefore, the
previous expression needs to be referred to the classical positive, negative and zero components. This
can be provided throughout the linear transformation of phase components to the symmetrical ones.
Using this approach, Equation (4) leads to Equation (5).
VhRST ¼ jhω:LRST :IhRST (4)

Figure 1. Zero-sequence electromagnetic harmonic filter.

Copyright © 2014 John Wiley & Sons, Ltd. Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep
 J. C. OLIVEIRA ET AL.

Figure 2. Electrical zero-sequence filter connections.

Vh012 ¼ jhω:T 1 :LRST :T:Ih012 (5)

where V hRST are the busbar supply voltages related to phases R, S and T; LRST are the inductances
related to phases R, S and T; IRST are the line currents associated with phases R, S and T; h is the
harmonic order; V h012 are the zero, positive and negative busbar voltage sequence components; and
I h012 are the zero, positive and negative line current sequence components.
Still talking about Equations (4) and (5), T is the transformation matrix obtained from the theorem
fortescue for symmetrical components; α is the alpha operator and also belongs to the fortescue
theorem for the symmetrical components.
From Equations (1)–(5), it can be noticed that the inductances matrix, which describes the relation-
ship between the voltages and currents in terms of sequential components for the electromagnetic
device, for a specific harmonic order h, can be represented by Equation (6):
Vh012 ¼ jhω:L012 :Ih012 (6)
In a general way, the L012 matrix elements will be non-zero and will be asymmetrical. This leads to
mutual inter-dependence between the voltage and current components of each frequency. However, based
on geometric aspects of the magnetic core, as also the constructive similarities of the windings, some sim-
plified hypothesis can be used. Considering, initially, that windings I and II are identical and that the supply
voltages do not lead the magnetic core into saturation region, then Equations (7) and (8) can be written thus:
Lxp ¼ L (7)
M x;
pq
y
¼ M x;qpy (8)
In addition to the above, the geometric and constructive core suggests that other simplifications can
be applied to mutual inductances. They consist of a few numerical identities illustrated in Figure 3. The
arrows mean equal values for the indicated mutual inductances.
The previous assumptions lead to Equations (9)–(11):
• Mutual inductances between windings I and II built on the same column:

M I;II I;II I;II

RT ¼ M SR ¼ M TS ¼ Lm (9)

• Mutual inductances between windings I and II of distinct and adjacent columns:

M I;II I;II I;II I;II I;I I;I I;I II;II II;II

RR ¼ M SS ¼ M ST ¼ M TR ¼ M RS ¼ M ST ¼ M RS ¼ M TR ¼ M RS ¼ M adj (10)

Copyright © 2014 John Wiley & Sons, Ltd. Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep
 ZERO-SEQUENCE ELECTROMAGNETIC HARMONIC FILTER

Figure 3. Mutual inductances at zero-sequence harmonic filter.

• Mutual inductances between windings I and II of distinct and external columns:

M I;II I;II I;I II;II

RS ¼ M TT ¼ M RT ¼ M TS ¼ M ext (11)

From the foregoing, it is possible to obtain the final expressions for the zero, positive and negative
filter inductance matrix elements, as given by Equations (12) and (13).
L00 ¼ 2ðL Lm Þ (12)
L11 ¼ L22 ¼ 2L þ Lm þ 2M adj þ M ext (13)
where M x;y
pq are the mutual inductance between x and y windings connected to phases p and q, respec-
tively; L are the windings self-inductance; Lm are the mutual inductance between x and y windings for
a given phase; L00,11,22 are the zero, positive and negative sequence self-inductances; Madj are the
mutual inductance related to adjacent magnetic columns; and Mext are the mutual inductance related
to external magnetic columns.
Considering the final elements forming the inductance matrix, it can be noticed that the mutual
inductances between the zero sequence and the positive and negative ones L10 and L20 are null. This
is a guarantee that zero-sequence voltage and current behavior can be investigated separately without
any interference from positive and negative values. This, however, does not apply to other sequences,
because elements L12 and L21 are not zeroes, Equations (14) and (15).
L12 ¼ 2α2 M ext M adj
 
(14)
 
L21 ¼ 2α M adj M ext (15)
The interdependence of the positive and negative sequence circuits is attributed to the intrinsic
unbalanced construction caused by the planar geometry of three columns. It causes small difference
between the mutual inductances related to the adjacent and the external columns, Madj and Mext,
respectively. The differences between the values of Madj and Mext are still substantially less than the
inductances themselves (L) and equal in the same magnetic column (Lm), which is utterly predominant
in the currents calculation of the current sequences, positive and negative (13). Aiming at analytical
simplification afforded by the decoupling sequence, the authors considered the existence of a single
mutual inductance, that, M = Madj = Mext.
With this in mind, it is possible to admit the equality of these positive and negative inductances
without introducing major inaccuracies to the analyses. Despite this consideration, the analysis of filter
performance for the purpose of this study depends exclusively on the zero-sequence equivalent
inductance.
Then, Equation (16) can be used to relate the harmonic voltages and currents with the filter internal
impedances thus:
2 3h 2 32 3h
V0 2ðL Lm Þ 0 0 I0
4 V 1 5 ¼ jhω4 0 2L þ Lm þ 3M 0 5:4 I 1 5 (16)
6 7 6 76 7

V2 0 0 2L þ Lm þ 3M I2

Copyright © 2014 John Wiley & Sons, Ltd. Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep
 J. C. OLIVEIRA ET AL.

Once Equation (16) represents independent zero, positive and negative voltage and current relation-
ship, they can be represented by the detached sequential circuits shown in Figure 4.
Considering the inductance matrix elements shown in Equation (16), it can be seen that the
effectiveness of zero-sequence electromagnetic device is directly dependent on the coupling induc-
tance between the windings built up on the same magnetic column. This interaction produces the most
important parameter to define the value of the zero-sequence filter impedance. Thus, by having a value
lower than that of the corresponding supply, most of the zero-sequence current will be deviated to the
shunt connected device. The ideal condition of filtering occurs when the coupling factor will be equal
to the unit, that is, Lm = L.

3. MODEL PERFORMANCE

The circuit that was set up to evaluate the zero-sequence filter performance can be represented by an
equivalent frequency domain circuit shown in Figure 5. The single line diagram of Figure 5(a)
illustrates the supply, the non-linear load and the electromagnetic filter. In a complementary way,

Figure 4. Sequential circuits for electromagnetic filter.

Figure 5. Simplified electrical arrangement used.

Copyright © 2014 John Wiley & Sons, Ltd. Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep
 ZERO-SEQUENCE ELECTROMAGNETIC HARMONIC FILTER

Figure 5(b) relates to the equivalent circuit itself. It must be emphasized that the circuit is to be applied
to the individual frequencies involved in the process.
The symbols Zs and Zfe are the complex impedances of the supply and the electromagnetic filter.
The variables E and U are the supply and complex load voltages, while I and J are the corresponding
currents. Again, the h and k indexes represent, respectively, the specific frequency focused and the
sequential circuit in the analysis.
Under these conditions, for each sequential circuit k, the dependence between supply current
harmonics, load voltage harmonics, load current harmonics and supply system voltage harmonics is
expressed by the three transmittances given by the equivalent circuits illustrated in Figure 6. The
Figure 6(a) takes into account the supply contribution, whilst Figure 6(b) illustrates the effect attributed
to the load.
The relationship between supply and load currents, as also the one between the load and supply
system voltages, is given by Equation (17), for a specific sequential circuit k and harmonic component
h. This expression provides the so-called transmittance factor.
 h  h  h
h Ιj Ue Zf e
Ak ¼ ¼ ¼ (17)
J k E k Zs þ Zf e k

where, for a specific sequential circuit k and harmonic component h, Ahk is the transmittance factor; Ij
is the load current contribution; J is the load current; Ue is the supply voltage contribution; E is the
supply voltage; Zfe is the electromagnetic filter complex impedance; and Zs is the supply system
complex impedance.
According to Equation (17), if Ah is greater than unity, it implies that harmonic current amplification
was injected by the load into the supply. Concerning voltage analysis, if Ah is greater than unity, it
implies voltage amplification at the load busbar.
The harmonic currents may occur not only due to non-linear loads (Ij) but also due to harmonic
voltages on the supply system (Ie). The magnitude of harmonic currents depends on the equivalent
admittance seen by the busbar and by what is calculated throughout Equation (18).
 h  h
Ιe 1
Yhak ¼ ¼ (18)
E k Zs þ Zf e k

The harmonic voltages may similarly result from a combination of the distortions attributed to the
supply (Ue) and the nonlinear load (Uj). From Figure 6(b), the harmonic contribution related to the
individual sequence component k is given by Equation (19):
 h 
Z s :Z f e h

h Uj
Zbk ¼ ¼ (19)
J k Zs þ Zf e k

where, for a specific sequential circuit k and harmonic component h, Yhak is the equivalent admittance;
Ιe is the current contribution from the supply; Zhbk is the equivalent harmonic impedance; and Uj is the
voltage contribution from load current.

Figure 6. Equivalent circuits to evaluate the supply and load contributions using superposition principle.

Copyright © 2014 John Wiley & Sons, Ltd. Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep
 J. C. OLIVEIRA ET AL.

As the final voltage and current distortions are given by a combination of the load and supply har-
monic sources, the following Equations (20) and (21) can be obtained by applying the superposition
principles [10,23,25].
h
Ihk ¼ I j þ I e k

(20)

Table I. Zero-sequence filter prototype main parameters (at 60 Hz).

Rated voltage 220 V

Rated power 1 kVA
Winding turns 237
Core area 18 cm2
DC resistance 1.6 Ω
Self-inductance L 250 mH
Mutual inductance Lm 249.95 mH
Mutual inductance M 110 mH

Figure 7. Electromagnetic zero-sequence filter prototype.

Figure 8. Laboratory physical structure.

Copyright © 2014 John Wiley & Sons, Ltd. Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep
 ZERO-SEQUENCE ELECTROMAGNETIC HARMONIC FILTER

h
Uhk ¼ U j þ U e k

(21)

4. EXPERIMENTAL VALIDATION

To provide a means for verifying the proposed filter frequency domain representation, a zero-sequence
filter prototype of 1 kVA/220 V was constructed. The magnetic core was so dimensioned as to
guarantee the filter operation within the linear region of the B × H curve. Besides, a constructive
procedure was adopted to improve the magnetic coupling between the windings built on the same
magnetic column. The objective was achieved by building windings I and II with interposed coils
during the assembling procedure [24].

Table II. System data.

Voltage at X busbar (V) 220

Frequency (Hz) 60
Reactance-to-resistance ratio of the supply for h = 1 1.72
Short circuit power/load active power 14.16

Figure 9. Experimental current waveforms.

Copyright © 2014 John Wiley & Sons, Ltd. Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep
 J. C. OLIVEIRA ET AL.

Table I presents the prototype’s main parameters, and Figure 7 illustrates the equipment prototype.

4.1. Experimental results

Studies were then carried out by using the simplified arrangement shown in Figure 8. The non-linear
load comprises three identical single-phase rectifiers, star-grounded and connected. They were the
sources required for zero-sequence harmonic current.

Table III. Harmonic load currents.

H Ja(A) ϕa(o) Jb(A) ϕb(o) Jc(A) ϕc(o)
1 1.835 9.5 1.846 248.4 1.863 129.2
3 1.263 203.0 1.239 200.7 1.284 201.1
5 0.490 26.4 0.451 148.8 0.477 269.2
7 0.130 122.1 0.122 35.3 0.1062 235.8
9 0.175 254.7 0.204 262.9 0.239 264.9
RMS (A) 1.63 1.62 1.65
THD (%) 74 72 75
Power factor 0.88 0.85 0.85

Figure 10. Computational results in terms of symmetrical components.

Copyright © 2014 John Wiley & Sons, Ltd. Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep
 ZERO-SEQUENCE ELECTROMAGNETIC HARMONIC FILTER

The filter was shunt connected in accordance with earlier information. A programmable source was
used to guarantee a balanced and linear three-phase voltage supply that was practically ideal with
harmonic distortion below 0.1% and unbalanced below 0.05%. An inductance line series was inserted
to represent the supply impedance.
The experimental structure used was based on the following components: programmable power
supply California Instruments—model 6000L/6kVA; data acquisition system Wavebook 16
IOtech, A/D 1.0 Msample conversion, 16 bits, 16 channels; AC/10 current transducers—class
2%, with 40 Hz to 3 kHz, 1 mA to 10 A, sensibility 100 mV/A and voltage transducers
WBK61/IOTech, class 0.1%, to 500 kHz, of 200/1 scale. The software used for signal processing
was DasyLab 4.0.
The main data relating to the studied case are presented in Table II.
By feeding the arrangement with 220 V (phase to phase rms value) and 60 Hz, the non-linear load,
supply and filter currents, and the corresponding harmonic spectra, are shown in Figure 9.
Table III shows the fundamental, harmonic and other information related to the nonlinear load currents.
The results clearly show that the third harmonic current value, produced by the load, was about 70%
of the fundamental one. On the other hand, from the results given in Figure 9(b), it is possible to
visualize that only 16% of this harmonic component was injected into the supply. This reduction
emphasizes filter efficacy in meeting the expected performance.

Figure 11. Computational results associated to R, S and T line currents.

Copyright © 2014 John Wiley & Sons, Ltd. Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep
 J. C. OLIVEIRA ET AL.

4.2. Computational results

By using the sequential harmonic transmittance, the harmonic components of the currents and voltages
were calculated (Equations (20) and (21)). The waveforms for the different sequential current compo-
nents are obtained from Equation (22).
H
X max

I hk sin hωt þ ϕ hk
 
ik ðtÞ ¼ (22)
h¼1

where, for a specific sequential component k, I hk is the amplitude of h order harmonic current compo-
nent i; ϕ hk is the phase angle of h order harmonic current component i.
Figure 10 illustrates the current sequential components on the load, supply and filter.
The foregoing results highlight the computational current waveforms in terms of their sequential
components. The corresponding harmonic spectra clearly show that the filter is quite effective at reduc-
ing the zero harmonic current injected into the source. In fact, this tool is very useful for investigating
the harmonic compensation achieved for each sequential component.
By applying the inverse procedure, it is possible to obtain the phase components throughout
traditional procedures as given by (23). Figure 11 provides the final results.

Figure 12. Experimental and computational results related to line R current.

Copyright © 2014 John Wiley & Sons, Ltd. Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep
 ZERO-SEQUENCE ELECTROMAGNETIC HARMONIC FILTER

IhRST ¼ Τ:Ih012 (23)

4.3. Experimental versus computational results analysis

For a better comparison of the results for the load, supply and filter currents, the line R experimental
and computational results were superposed as shown in Figure 12.
A comparison of the present results with the previous ones reveals a close agreement between them.
It shows that the model utilized here for estimating the filter performance is accurate enough to predict
the efficacy of the filter at reducing the zero-sequence harmonics and others.

5. CONCLUSIONS

The zero-sequence electromagnetic filter using only electric and magnetic components has proved to
be feasible and effective in reducing the level of zero-sequence harmonics injected into the supply.
It is inexpensive, rugged and does not pose any resonance problem.
By comparing the theoretical and laboratory results, the computational approach proposed here has
been proved adequate in representing the overall system and in evaluating filter performance under
ideal and non-ideal conditions.
The results demonstrate that the effectiveness of filter depends on the level of supply short circuit.
To guarantee the necessary filtering performance under a high short circuit level, a low-leakage imped-
ance filter must be used. In this way, the approach employed here has been proved very useful in
identifying possible optimization actions of the studied system operational behavior.

6. LIST OF SYMBOLS AND ABBREVIATIONS

6.1. Symbols
Lxp Self-inductance related to winding x, connected to phase p
M x;y
pq Mutual inductance between x and y windings connected to phases p and q, respectively
V xp Complex voltage drop across winding x, connected to phase p
Ip Complex current in phase p
h Harmonic order
V hRST Busbar supply voltages related to phases R, S and T
LRST Inductances related to phases R, S and T
IRST Line currents associated with phases R, S and T
V h012 Zero, positive and negative busbar voltage sequence components
I h012 Zero, positive and negative line current sequence components
T Transformation matrix obtained from the Fortescue Theorem for symmetrical components
α Alpha operator and also belongs to the Fortescue Theorem for the symmetrical components
L Windings self-inductance
Lm Mutual inductance between x and y windings for a given phase
L00,11,22 Zero, positive and negative sequence self-inductances
Madj Mutual inductance related to adjacent magnetic columns
Mext Mutual inductance related to external magnetic columns
Ahk Transmittance factor
Ij Load current contribution
J Load current
Ue Supply voltage contribution
E Supply voltage
Zfe Electromagnetic filter complex impedance
ZS Supply system complex impedance
Yhak Equivalent admittance
Ie Current contribution from the supply

Copyright © 2014 John Wiley & Sons, Ltd. Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep
 J. C. OLIVEIRA ET AL.

Zhbk Equivalent harmonic impedance

Uj Voltage contribution from load current
I hk Amplitude of h order harmonic current component i
ϕ hk Phase angle of h order harmonic current component i.

6.2. Abbreviations
FBS Four-branch star
MMF Magnetomotive force
DC Direct current
RMS Root mean square
THD Total harmonic distortion
A/D Analog-to-digital
AC Alternating current

ACKNOWLEDGEMENT

The authors would like to thank CNPq, CAPES and FAPEMIG for the Ph.D. scholarship.

REFERENCES

1. Ibrahim MA. Disturbance Analysis for Power Systems. Wiley-IEEE Press eBook Chapters: Hoboken, New Jersey,
2012; 85–181. DOI: 10.1002/9781118172094.ch3
2. Abdullah MF, Hamid NH, Baharudin Z, Khamis MFI. Triplen harmonics currents propagation through medium
voltage distribution network. 4th International Conference on Modeling, Simulation and Applied Optimization
(ICMSAO) 2011; 1–5. DOI: 10.1109/ICMSAO.2011.5775604.
3. El-Saadany EF, Salama MMA, Chikrani AY. New Passive Filter Design for Neutral Current Cancellation in
Balanced 3-Phase 4-Wire Non-Linear Distribution Systems. European Transactions on Electrical Power 2003;
13:79–89. DOI: 10.1002/etep.4450130203.
4. Singh GK. Power system harmonics research: a survey. European Transactions on Electrical Power 2009; 19:151–172.
DOI: 10.1002/etep.201.
5. APPA. DSTAR project 13–5: Harmonic Load Impact and Mitigation. American Public Power Association, 2012.
6. Schoene J, Walling R, Bo Y, Niemann B, Zheglov V, Guinn D, Peele S, Grappe J, Zavadil B, Freeman L. Analysis
and mitigation of excessive zero-sequence harmonic currents in distribution systems. Transmission and Distribution
Conference and Exposition (T&D) IEEE PES, 2012; 1–6. DOI: 10.1109/TDC.2012.6281508.
7. Demoulias C, Goutzamanis D, Gouramanis K. Analysis of the voltage harmonic distortion at buses feeding office
loads. IET Science, Measurement & Technology, IET 2009; 3:286–301 DOI: 10.1049/iet-smt.2008.0069.
8. Dahono PA, Widjaya RE, Marsi S, Qamaruzzaman. A practical approach to minimize the zero sequence current
harmonics in power distribution systems. Proceedings of the Power Conversion Conference 1997; 2:683–686.
DOI: 10.1109/PCCON.1997.638278.
9. Belchior FN, Ferreira JFV, Apolônio R, Vasconcelos AB, Oliveira JC. Three phase electromagnetic filter for zero
sequence harmonics. IEEE Transactions on Magnetics 2006; 42:2201–2207. DOI: 10.1109/TMAG.2006.880398.
10. Belchior FN, Ferreira JFV, Oliveira JC, Oliveira LCO, Bernardes VA. Zero sequence electromagnetic filter
performance with non ideal conditions. IEEE-PES-ICHPQ - International Conference on Harmonic and Power
of Quality, Cascais, Portugal, 2006.
11. Singh B, Jayaprakash P, Kothari DP. Magnetics for neutral current compensation in three-phase four-wire distribu-
tion system. International Conference on Power Electronics, Drives and Energy Systems (PEDES), New Delhi,
India, 2010; 1–7. DOI: 10.1109/PEDES.2010.5712545.
12. Ortega JMM, Payán MB, Mitchell CI. A new design method of passive tuned filters for harmonic mitigation and
reactive power compensation. European Transactions on Electrical Power 2006; 16:219–233. DOI: 10.1002/
etep.77.
13. Kumar SR, Surendhar S, Negi A, Raja P. Zig Zag Transformer performance analysis on harmonic reduction in distribu-
tion load. International Conference on Electrical, Control and Computer Engineering (INECCE), 2011; 107–112. DOI:
10.1109/INECCE.2011.5953859.
14. Laka A, Barrena JA, Chivite-Zabalza J, Vidal MAR, Calvo G. Novel Zero-Sequence Blocking Transformer (ZSBT)
using three single-phase transformers. IEEE Transactions on Energy Conversion 2013; 28:234–242. DOI: 10.1109/
TEC.2012.2229282.
15. Jou HL, Wu KD, Wu JC, Chiang WJ. A three-phase four-wire power filter comprising a three-phase three-wire ac-
tive power filter and a zig–zag transformer. IEEE Transactions on Power Electronics 2008; 23:252–259. DOI:
10.1109/TPEL.2007.911779.

Copyright © 2014 John Wiley & Sons, Ltd. Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep
 ZERO-SEQUENCE ELECTROMAGNETIC HARMONIC FILTER

16. Choi S, Jang M. Analysis and control of a single-phase-inverter–zigzag-transformer hybrid neutral-current suppres-
sor in three-phase four-wire systems. IEEE Transactions on Industrial Electronics 2007; 54:2201–2208. DOI:
10.1109/TIE.2007.899831.
17. Rodriguez P, Candela JI, Luna A, Asiminoaei L, Teodorescu R, Blaabjerg F. Current harmonics cancellation in
three-phase four-wire systems by using a four-branch star filtering topology. IEEE Transactions on Power Electron-
ics 2009; 24:1939–1950. DOI: 10.1109/TPEL.2009.2017810.
18. Singh B, Jayaprakash P, Kumar S, Kothari DP. Implementation of neutral-network-controlled three-leg VSC and
a transformer as three-phase four-wire DSTATCOM. IEEE Transactions on Industry Applications 2009;
47:1892–1901. DOI: 10.1109/TIA.2011.2153811.
19. Singh B, Kumar S. Modified power balance theory for control of DSTATCOM. International Conference on Power
Electronics, Drives and Systems (PEDES), New Delhi, India, 2010; 1–8. DOI: 10.1109/PEDES.2010.5712547.
20. Singh B, Kumar S. Control of DSTATCOM using IcosΦ algorithm. 35th Annual Conference of IEEE Industrial
Electronics (IECON’09), 2009; 322–327. DOI: 10.1109/IECON.2009.5414942.
21. Jou BHL, Wu JC, Wu KD, Chiang WJ, Chen YH. Analysis of zig–zag transformer applying in the three-phase four-
wire distribution power system. IEEE Transactions on Power Delivery 2005; 20:1168–1173. DOI: 10.1109/
TPWRD.2005.844281.
22. Czarnecki LS. An overview of methods of harmonic suppression in distribution systems. Proceedings of the IEEE
PES Summer Meeting, Seattle, USA 2000; 2:800–805. DOI: 10.1109/PESS.2000.867456.
23. Czarnecki LS, Ginn H. Effectiveness of resonant harmonic filters and its improvement. Proceedings of the IEEE
PES Summer Meeting, Seattle, USA, 2000; 2:742–747. DOI: 10.1109/PESS.2000.867443.
24. Oliveira LCO, Ravagnani L, Milanez DL. Modelling for passive hybrid filter performance analysis. Proceedings of
the EPQU’2007-9th International Conference on Power Quality and Utilization, Barcelona, Spain, 2007.
25. Oliveira LCO, Ravagnani L, Oliveira RN, Oliveira JC, Belchior FN. Experimental analysis of the zero sequence
filter impedances (in portuguese), VII CBQEE - Power Quality Brazilian Conference. Santos-SP, Brazil, 2007.

Copyright © 2014 John Wiley & Sons, Ltd. Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep