-
tir
e &
atio
al
nlin
oposed method can unambiguously detect interturn faults even
operation, and improve power supply and service to custo-high
fault currents within the short circuited part of the
winding, accompanied by a relatively low value of current
in the remainder of the winding. On the other hand, if for
for fault detection and diagnosis of the electric machines
is
EUROPEAN TRANSACTIONS ON ELECTRICAL POWER
Euro. Trans. Electr. Power 2011; 21:196211
Published online 15 April 2010 in Wiley Online Library
(wileyonlinelibrary.com). DOI: 10.1002/etep.431more severe damage
to the power transformer.
The problem of the traditional differential protection
In literature, using the negative sequence as an
indicatoraccount for approximately 70% of all the failures in
power
transformers. Furthermore, from these reviews, it can also
be understood that protection of the power transformers
against winding interturn faults is still a challenge. One
major problem in protecting large power transformers from
interturn winding faults is that the current equipment being
used to protect the transformers from interturn faults often
fails until minor faults developed into high level faults
with
pickup current of the restraint differential protection be
set
relatively high, then minor interturn faults cannot be
detected until it evolves into a more severe fault with
higher differential currents. This work focuses mainly on
the low level interturn faults on the windings of power
transformer. Fast and early detection of this type of faults
is
critical in preventing a major damage to the power system
and the transformer itself.winding faults takes a great
participation. From a number
of surveys [1,2], it can be deduced that interturn faultssome
reason, for example, because of uncompensated
movements of an on-load tap changer, the minimummers. Among
several of transformer faults, the transformerusing data generated
by simulations on a finite element model. Copyright # 2010 John
Wiley & Sons, Ltd.
KEYWORDS
power transformer; interturn fault; negative sequence current;
finite element model
* Correspondence
Vahid Behjat, Department of Electrical Engineering, Iran
University of Science & Technology, Narmak 16846, Tehran,
Iran.
E-mail: [email protected]
1. INTRODUCTION
Nowadays, there is an increasing interest in online
monitoring
of power transformers because of its potential to provide
early warning of electrical failures, enhance the reliability
of
transformers has been that just these low level interturn
faults could not be detected with the overall sensitivity
represented by the percentage restraint differential
protection. A short circuit of a few turns, in spite of
veryheavily distorted conditions. The performance of the proposdown
to two shorted turns along the winding. The method is not
influenced by the supply and load harmonics even under
ed technique was studied for a variety of operating
conditionsinto high level faults with more severe damage to the
pow
suggested to overcome this problem to a great extent. The
prtransformers using the ratio of negative sequence components of
the primary and secondary line currents. The ratio is equal
to the turn ratio during external faults as well as in the
supply or load imbalance conditions, while it differs from the
turn
ratio when interturn winding faults occur. The main feature of
the proposed method is its capability to detect low level
interturn faults which typically cannot be detected by the
traditional transformer protection devices before they
developed
er transformer. In this work, a major improvement has
beenRESEARCH ARTICLE
Online monitoring of powerof internal winding short csequence
analysis
Abolfazl Vahedi1,2 and Vahid Behjat1,2*
1Department of Electrical Engineering, Iran University of
Scienc
2Center of Excellence for Power System Automation and Oper
ABSTRACT
As a recent trend, online monitoring techniques for electric
considered very important. This paper presents a new owhich
recognized as the basis protection of the power
196ransformers for detectioncuit faults using negative
Technology, Tehran, Iran
n, Tehran, Iran
machines, mainly including power transformers, has been
e method for detection of interturn faults in the powerprimarily
focused on induction motors. The occurrence of a
Copyright 2010 John Wiley & Sons, Ltd.
-
to test fault diagnostic techniques. Amajor breakthrough in
winding three-phase power transformer has been con-
sidered for the present study. The transformer was
employed in simulations with all parameters and con-
figuration provided by the manufacturer. Ratings of the
transformer are presented in Table I. A transversal section
representation of the power transformer structure is shown
in Figure 1.
The magnetic field inside the transformer is governed by
Table I. Ratings of the transformer.
Rated power 8 MVA
Rated frequency 50 Hz
Primary rated voltage 20 kV
Secondary rated voltage 11 kV
Turns ratio 510/162
Connection D/Y
Figure 1. Geometry of the transformer model (all units are in
mm).
A. Vahedi and V. Behjat Online monitoring of power
transformersthe detection of stator interturn faults has been
achieved by
utilizing the effective negative sequence impedance as an
indicator of these faults [8,9]. In Ref. [10], a comparative
analysis is provided for online detection of stator winding
interturn short fault in induction motors. The authors
compared several detection methods in terms of diagnostic
efficiency and requirements for practical implementation.
Among all the analyzed techniques, those based on the
sequence components are considered the most promising.
It is claimed in Ref. [10] that the negative sequence
components present a high diagnostic efficiency based on
the compensation of non-idealities, good experimental
results, and medium simplicity for practical implementa-
tion. While the negative sequence current is able to detect
and diagnose a stator winding short, the method is unable
to discriminate between an interturn short fault and the
imbalance in the power supply, which is common in the
operation of power systems.
In the case of power transformers, one of the earliest
works was introduced by Sidhu et al. [11,12]. The techniqueuses
the arguments of the positive and negative sequence
impedances of the power system in a fault detection
algorithm. Since then, some other authors have taken
advantage of symmetrical component capabilities to develop
protection schemes for power transformers [13,14].
An attempt has been made in this paper to obtain a new
andmodified fault detectionmethod using negative sequence
quantities. Such a modification could bring significant
improvement in the detection of transformer interturn
faults and especially overcome the drawbacks of the
existing fault detection schemes to detect minor interturn
faults. To this end, a numerical simulation of power
transformer is completed firstly by the use of finite
element
model. Using transformer FEMmodel, it would be possible
to obtain a comprehensive view of the overall transformer
magnetic and electrical behavior under normal and
disturbance conditions. In addition, it provides a
significant
degree of accuracy, which is very useful to explore the
terminal behavior of the transformer. Based on that, this
paper points out that negative sequence components of the
transformer line currents can be viewed as the feature
criterion and utilized to detect interturn faults.
2. NUMERICAL SIMULATION
The fault simulations were carried out by means of solving
a finite element model of a power transformer. Such model
was defined through data sheet obtained from the
transformer manufacturer. An 8MVA, core type, twonegative
sequence component in the motor supply currents
is identified when shorted stator windings are present [3].
In Refs. [4,5], online techniques were presented for
detection of stator winding faults in three-phase induction
motors from observation and measurement of negative
sequence supply current. Refs. [6,7] presented an induction
machine model for simulation of stator interturn faults andEuro.
Trans. Electr. Power 2011; 21:196211 2010 John Wiley & Sons,
LtdDOI: 10.1002/etepthe well-knownMaxwell equations. FEM is applied
here in
order to solve electromagnetic field problems described by
the Maxwell equations. When a fault occurs, the magnetic
flux distribution is fundamentally altered as well as the
current in the circuit domain. But, the transformer terminal
behavior still satisfies the governing equations. Thus,
obtaining the faulty transformer behavior is achieved by
solving these equations. We used direct coupling between
the field and circuit equations in order to define and
simulate the transformer behavior under normal and
interturn fault conditions.
Figure 2 represents finite element model of the studied
power transformer. It is composed of 17 598 surface and
7406 line elements and includes 38 827 nodes. We decided
to use first-order elements, since there was no significant
difference in results compared with second-order elements.
The regions in the finite element model are coupled to the
circuit model. Both the primary and secondary windings
are modeled as stranded coil conductors in the circuit
model. In order to verify the models reliability and
precision, the model was validated in the steady-state
magnetic formulation by comparison with values of. 197
-
Online monitoring of power transformers A. Vahedi and V.
Behjatterminal current and voltages, short circuit impedance,
etc.,
provided by the manufacturer for the steady state of the
considered transformer. For the transformer model, the rms
values of the terminal voltages and currents and also the
other parameters obtained from the simulation and the
manufacturer data were almost equal. Table II presents the
results of this comparison and demonstrates the excep-
tional ability of the model to reproduce the real behavior
of
the machine. Figure 3a and b shows the flux plot (equiflux
lines) of the transformer and color shaded plots of the flux
density on the transformer regions generated by the FEM
model under normal operating conditions and at the rated
load. A detailed description about the finite element model
is given in Appendix.
Figure 4a and b shows the corresponding circuit domain
and FE domain representation of the transformer coil for an
interturn fault on the transformer phase B secondary
winding. When an interturn fault is on the primary or
secondary winding, the fault winding is divided in three
parts a, b, and c in the FE domain as well as in the
circuit domain as shown in Figure 4. A time controlled
Figure 2. 2D Finite element model of the transformer.switch plus
a limiting fault resistance was utilized to
initiate an internal fault. Therefore, by means of the
coupled electrical circuit, faults can be introduced at
different locations along the windings. The severity of the
fault can be controlled by different values of fault
resistance between turns. Indeed, the fault resistance
are denoted by the indices pn and sn, respectively (p stands
Table II. Comparison of the values of the transformer.
Magnitude Specified manufacturer Simulation
Primary voltage 20 000 V(D) 20 000
Primary current 230.9 A(Y) 228.97
Secondary voltage 11 000 V(Y) 10925.939
Secondary current 419.9 A(Y) 415.989
Short circuit impedance 2.33% 2.2%
Total power loss 54.92 kW 52.9 kW
198 Euro.for primary, s for secondary, and n for negative
sequence).
The two negative sequence currents are expressed asrepresents
the resistive component of the dielectric
material in the dielectric equivalent parallel circuit
model.
3. DETECTION METHOD
The theoretical foundation of the new detection method is
based on the theory of symmetrical components, or more
exact, on the negative sequence currents. Existence of
relatively high negative sequence currents is in itself an
indication of a disturbance, and quantitatively represents
asymmetries coupled to a specific condition of an electrical
system, in our case, the transformer. The proposed method
for interturn fault detection is based on the principle that
for a healthy (unfaulted) transformer supplied by
symmetrical multiphase voltage sources without turn
faults in its windings, no negative sequence component
of the line and phase current occurs. An interturn fault
will
break that symmetry and give rise to a negative sequence
current which may then be used as a measure of fault
severity or to initiate a warning alarm for the monitoring
or
protection devices to make proper decision. However,
asymmetries in the three-phase quantities of the transfor-
mer may arise due to some other reasons such as
instrumentation asymmetries and operation with imbal-
anced load or supply voltages. Such asymmetries can be
reflected in negative sequence components and yield a
significant increase in the negative sequence components
of line and phase currents. The proposed method in this
paper can not only detect and diagnose a winding interturn
short circuit fault, but also is able to discriminate
between
an interturn fault and other disturbances such as imbalance
in the power supply and load and external fault conditions.
In general, the negative sequence current is obtained
from the measurement of the three-line currents of the
transformer using:
Ian 13Iaf a2Ibf aIcf (1)
In the above equation, the negative sequence component
is denoted by the index an (a stands for the line current
and
n for the negative sequence), Iaf, Ibf, and Icf are the
phasorsof fundamental components of three-line current signals,
and a is a phase rotation operator equivalent to ej2p=3.The
underlying principle for interturn fault detection is
based on the measurement of negative sequence com-
ponent of the line currents from both sides of the
transformer. The ratio of the negative sequence com-
ponents (RNSC)is compared with the turn ratio for the
purpose of fault detection. Defining the RNSC gives:
RNSC Ipn3
pIsn
(2)
In the above equation, the negative sequence com-
ponents of line currents from primary and secondary sidesTrans.
Electr. Power 2011; 21:196211 2010 John Wiley & Sons, Ltd.DOI:
10.1002/etep
-
A. Vahedi and V. Behjat Online monitoring of power
transformersphasors. The H3 coefficient on the right hand side of
theequation allows the compensation of turn ratio due to delta
winding in the primary.
Theoretically, in case of external fault and imbalanced
load or supply voltage, the RNSC is equal to the turn ratio
(N1/N2) while for an interturn fault it is not. In order
toexplain this phenomenon in greater detail, reconsideration
of the symmetrical components theory will be useful.
Figure 3. (a) Equiflux lines under normal operating conditions.
(b) C
Euro. Trans. Electr. Power 2011; 21:196211 2010 John Wiley &
Sons, LtdDOI: 10.1002/etepAccording to symmetrical components
theory [15,16],
the negative sequence currents are transferred through at a
power transformer. Further, for any external disturbance,
such as external fault and imbalanced load or supply, the
negative sequence source will be located outside the
transformer at the fault point, load or supply side,
respectively. Thus the negative sequence currents will
enter the healthy power transformer on the fault side, and
olor shaded plots of the flux density on the transformer
regions.
. 199
-
been presented in the next section.
Online monitoring of power transformers A. Vahedi and V.
Behjatleave it on the other side, properly transformed. Hence,
the
RNSC is the same as turn ratio. On the contrary, for an
interturn fault with the negative sequence source within the
transformer, the negative sequence currents will flow out of
the faulty power transformer on both sides and con-
sequently, the RNSC is different from N1/N2.From this point, the
RNSC of (2) can be used for
detecting interturn faults in a simple and accurate manner.
However, in reality, for an external disturbance, the RNSC
can be a little different from N1/N2 because there might besome
small difference between two negative sequence
currents due to possible different negative sequence
impedance values on the respective sides. The Detector
described by (3) is used to detect a fault. The Detector
measures the percentage difference between the two
estimated negative sequence currents:
Detector 3
pIsn RNSC N2N1
Ipn 100% (3)
Figure 4. Circuit coupled FEM model of the power transformer
for modeling interturn faults: (a) circuit domain and (b)
FEM
domain.If (3) is less than a threshold, the transformer is
not
faulted; if greater, it is. Further, even for normal
operating
conditions, small values of the negative sequence currents
can be measured in line currents. Thus, for a trustworthy
decision regarding the interturn fault occurrence, the two
negative sequence currents must be above a certain
minimum value otherwise no comparison is allowed. If
both negative sequence currents exceed the threshold,
which in itself is as a sign that a disturbance must have
happened, as the negative sequence currents are super-
imposed, pure-fault quantities, the directional comparison
is carried out.
For the system studied in this paper, the recommended
value for the detector threshold is 1% and minimum
allowable negative sequence current is 0.1% of the
transformer rated current.
It should be pointed out that the main feature of the
proposed algorithm is its capability to detect low level
response of the detector is above the threshold, con-
sequently a fault alarm is activated.
200 Euro.4.1.2. Fault size 1%, Rf equal to 0.1V, ratedresistive
load.Figure 6ad represents the results for the Case B. The
difference between cases A and B lies in the fault size and
severity. In this case, an interturn fault involving 1% of
the
turns on the top end of HV3 (primary) winding is studied
which representing a relatively decreased fault size and
severity. Significant decreasing in the values of terminal4.1.
Case 1: Interturn fault on HV winding
4.1.1. Fault size 5%, Rf equal to 0.5V, ratedresistive
load.Figure 5a and b shows three primary and secondary line
currents of the transformer working at rated load and for an
interturn fault involving 5% of the turns on the top end of
HV3 (primary) winding, respectively. The studied case has
a fault resistance, Rf, equal to 0.5V. Such a fault will
giverise to a large increase of primary current and severe
distortion of it; while the secondary current does not
change very much. Figure 5c and d shows the value of
negative sequence currents obtained from this study and
the operating response of the fault detector, respectively.
As expected, the RNSC from primary and secondary
sides is not the same as the turn ratio and thus the operating4.
RESULTS AND CASE STUDIES
The performance of the proposed detection method was
verified on various disturbance conditions such as winding
interturn faults, external fault, as well as load or source
asymmetries. To demonstrate the algorithms ability on
detecting even minor interturn fault, a series of case
studies
was designated aimed at covering a wide variety of
interturn fault conditions of different fault severities and
locations. Also, the performance of the faulted transformer
was examined under varying conditions of load level and
power factor. In all of the case studies, the disturbance
occurs at t 0.1 second. Although a much larger number
ofsimulations were carried out, only some selected cases are
included here.interturn faults. While a short circuit of a few
turns of the
winding will give rise to very small changes in the terminal
currents, it results in an increased negative sequence
components in the line currents. Thus, the proposed
algorithm is a good complement to the traditional power
transformer protections for detecting minor faults, with a
high sensitivity and speed.
Some test results verifying the performance of the
technique for different faults and operating conditions
haveTrans. Electr. Power 2011; 21:196211 2010 John Wiley &
Sons, Ltd.DOI: 10.1002/etep
-
a very complicated problem for traditional power
the condition is similar to that which occurs when the
transformer is damaged in the HVwinding. Figure 9c and d
respective negative sequence components from the HVand
LV power transformer sides will be nearly equal in
from the results, the primary and secondary line currents
are heavily distorted. However, the detector remains below
A. Vahedi and V. Behjat Online monitoring of power
transformerstransformer protection devices. It can be distinctly
seen
from Figure 6a that short circuiting of a few turns (1% of
turns on the winding) will give rise to a very small
variation
in the terminal currents; so absolutely the differential and
over current protection will be stable in this case. While,
using the negative sequence current as a fault indicator, a
proper alarm can be generated in the fault detection system.
Thus, this study also demonstrates both the proposed
method dependability and sensitivity for detecting minor
interturn faults. The results of much more simulations
demonstrate that the proposed method is capable of
detecting an interturn fault involving even two shorted
turns in an unambiguous manner.
4.1.3. Fault size 5%, Rf equal to 0.5V, ratedload, 0.66 lagging
power factor.Figure 7ad shows the results obtained for the Case C;
this
is identical to Case A except for the load index and power
factor. The case study is carried out with a load index
equal
to 50% of the rated load and 0.8 lagging power factor. The
comparison of thee two cases demonstrates that the
modulus and phase angle of the negative sequence current
are independent of the angle and level of the impedance
load. Similar to Case A, the response of the detector has a
value greater than the operating threshold.
4.1.4. Fault size 5%, Rf equal to 0.5V, ratedresistive load,
harmonic condition.The proposed detection method was simulated for
an
interturn fault involving 5% of the turns on the top end of
HV3 (primary) winding and harmonic conditions.
Figure 8ad shows the results for Case D where the
primary and secondary currents are heavily distorted and
contain harmonic components. In practice, such harmonics
may be caused by CT saturation or nonlinear BH charac-teristic
of the transformer core. The results clearly indicate
that the two calculated negative sequence components of
the fundamental frequency phasor are not the same in
interturn fault conditions even if the primary and secondary
currents contain harmonic components. Thus the method is
completely independent of time harmonics and fault alarm
is issued as shown in Figure 8d.
4.2. Case 2: Interturn fault on LV winding
Figure 9a and b shows three primary and secondary line
currents of the transformer working at rated load and for an
interturn fault between 1% of the turns on the top end of
LV3 (secondary) winding, respectively. The studied casecurrents
and negative sequence component is particularly
visible when the size of the interturn fault decreases, as
can
be seen by comparing Figure 6a with Figure 5a. Figure 6d
shows the operating response of the detector which meets
the 1% operating threshold.
It is worth pointing out that the proposed detection
method can easily detect a low level interturn fault which
isEuro. Trans. Electr. Power 2011; 21:196211 2010 John Wiley &
Sons, LtdDOI: 10.1002/etep1% threshold, consequently the fault
detection system
would be stable in this condition and no interturn fault
alarm will be issued.
4.5. Case 5: Supply imbalance
Another case is studied when the transformer is loaded
with the rated load, but supplied with asymmetrical three-
phase voltage source. Supply imbalance conditions produce
almost fully asymmetrical current signals as shown in
Figure 12a and b. The negative sequence currents caused
by these asymmetric currents have been shown in
Figure 12c. The stability of the proposed detection method
against supply imbalance can be observed in Figure
12d.magnitude, after the compensation of the transformer turns
ratio, as shown in Figure 10c. Figure 10d shows the
operating response of the detector; remains its value below
1% threshold. This led to the decision that the fault is
outside the protection zone of the transformer and
consequently no interturn fault alarm is activated.
4.4. Case 4: Load imbalance
The proposed algorithm was simulated in the case of an
imbalanced load. Figure 11ad shows the results obtained
for this case when the transformer has rated load at phases
a and b and no load at phase c. As can be seenshows the value of
negative sequence currents obtained
from this study and the operating response of the fault
detector, respectively.
4.3. Case 3: External fault
This study is used to prove the proposed fault detection
method stability under external faults. A symmetrical three-
phase fault was simulated on the HV side of the transformer
outside the transformer protection zone. As expected, thehas a
fault resistance equal to 0.01V. It can be easily foundout from the
simulations (Figures 5 and 9) that for an
interturn fault on the transformer windings, either on the
primary or secondary winding, the primary current will
rise. However, the secondary current does not change;
remain fairly constant, for the fault on the primary winding
and decreases a little when the fault is on the secondary
winding. Anyhow, asymmetries in the primary side caused
by the interturn faults, regardless of the fault location,
will
produce significant values of negative sequence com-
ponents. So when the short circuit arises in the LV winding.
201
-
-500
-400
-300
-200
-100
0
100
200
300
400
500
0.20.180.160.140.120.10.080.060.040.020
Time(s)
a) P
rimar
y Li
ne C
urre
nts(
A)
iapibpicp
-800
-600
-400
-200
0
200
400
600
800
0.20.180.160.140.120.10.080.060.040.020
Time(s)
b) S
econ
dary
Lin
e C
urre
nts(
A)
iasibsics
-60
-40
-20
0
20
40
60
0.20.180.160.140.120.10.080.060.040.020
Time(s)
c) N
egat
ive
Sequ
ence
Com
pone
nts(
A)
Primary Negative Sequence CurrentSecondary Negative Sequence
Current
0
10
20
30
40
50
60
70
80
90
100
0.20.180.160.140.120.10.080.060.040.020
Time(s)
d) O
pera
ting
Res
pons
e of
the
Det
ecto
r
(a)
(b)
(c)
(d)
Figure 5. (a) Primary line currents, (b) secondary line
currents, (c) negative sequence components, and (d) operating
response of the
detector for Case 1A.
202 Euro. Trans. Electr. Power 2011; 21:196211 2010 John Wiley
& Sons, Ltd.DOI: 10.1002/etep
Online monitoring of power transformers A. Vahedi and V.
Behjat
-
-400
-300
-200
-100
0
100
200
300
400
0.20.180.160.140.120.10.080.060.040.020
Time(s)
a) P
rimar
y Li
ne C
urre
nts(
A)
iapibpicp
-800
-600
-400
-200
0
200
400
600
800
0.20.180.160.140.120.10.080.060.040.020
Time(s)
b) S
econ
dary
Lin
e C
urre
nts(
A)
iasibsics
-15
-10
-5
0
5
10
15
0.20.180.160.140.120.10.080.060.040.020
Time(s)
c) N
egat
ive
Sequ
ence
Com
pone
nts(
A)
Primary Negative Sequence CurrentSecondary Negative Sequence
Current
0
10
20
30
40
50
60
70
80
90
100
0.20.180.160.140.120.10.080.060.040.020
Time(s)
d) O
pera
ting
Res
pons
e of
the
Det
ecto
r
(a)
(b)
(c)
(d)
Figure 6. (a) Primary line currents, (b) secondary line
currents, (c) negative sequence components, and (d) operating
response of the
detector for Case 1B.
Euro. Trans. Electr. Power 2011; 21:196211 2010 John Wiley &
Sons, Ltd.DOI: 10.1002/etep
203
A. Vahedi and V. Behjat Online monitoring of power
transformers
-
-800
-600
-400
-200
0
200
400
600
800
0.20.180.160.140.120.10.080.060.040.020
Time(s)
a) P
rimar
y Li
ne C
urre
nts(
A)
iapibpicp
-1500
-1000
-500
0
500
1000
1500
0.20.180.160.140.120.10.080.060.040.020
Time(s)
b) S
econ
dary
Lin
e C
urre
nts(
A)
iasibsics
-60
-40
-20
0
20
40
60
0.20.180.160.140.120.10.080.060.040.020
Time(s)
c) N
egat
ive
Sequ
ence
Com
pone
nts(
A)
Primary Negative Sequence CurrentSecondary Negative Sequence
Current
0
10
20
30
40
50
60
70
80
90
100
0.20.180.160.140.120.10.080.060.040.020
Time(s)
d) O
pera
ting
Res
pons
e of
the
Det
ecto
r
(a)
(b)
(c)
(d)
Figure 7. (a) Primary line currents, (b) secondary line
currents, (c) negative sequence components, and (d) operating
response of the
detector for Case 1C.
204 Euro. Trans. Electr. Power 2011; 21:196211 2010 John Wiley
& Sons, Ltd.DOI: 10.1002/etep
Online monitoring of power transformers A. Vahedi and V.
Behjat
-
-600
-400
-200
0
200
400
600
0.20.180.160.140.120.10.080.060.040.020
Time(s)
a) P
rimar
y Li
ne C
urre
nts(
A)
iapibpicp
-800
-600
-400
-200
0
200
400
600
800
0.20.180.160.140.120.10.080.060.040.020
Time(s)
b) S
econ
dary
Lin
e C
urre
nts(
A)
iasibsics
-60
-40
-20
0
20
40
60
0.20.180.160.140.120.10.080.060.040.020
Time(s)
c) N
egat
ive
Sequ
ence
Com
pone
nts(
A)
Primary Negative Sequence CurrentSecondary Negative Sequence
Current
0
10
20
30
40
50
60
70
80
90
100
0.20.180.160.140.120.10.080.060.040.020
Time(s)
d) O
pera
ting
Res
pons
e of
the
Det
ecto
r
(a)
(b)
(c)
(d)
Figure 8. (a) Primary line currents, (b) secondary line
currents, (c) negative sequence components, and (d) operating
response of the
detector for Case 1D.
Euro. Trans. Electr. Power 2011; 21:196211 2010 John Wiley &
Sons, Ltd.DOI: 10.1002/etep
205
A. Vahedi and V. Behjat Online monitoring of power
transformers
-
-400
-300
-200
-100
0
100
200
300
400
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
a) P
rimar
y Li
ne C
urre
nts(
A)
iapibpicp
-800
-600
-400
-200
0
200
400
600
800
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
b) S
econ
dary
Lin
e Cu
rren
ts(A
)
iasibsics
-15
-10
-5
0
5
10
15
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
c) N
egat
ive
Sequ
ence
Com
pone
nts(
A)
Primary Negative Sequence CurrentSecondary Negative Sequence
Current
0
10
20
30
40
50
60
70
80
90
100
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
d) O
pera
ting
Resp
onse
of t
he D
etec
tor
(a)
(b)
(c)
(d)
Figure 9. (a) Primary line currents, (b) secondary line
currents, (c) negative sequence components, and (d) operating
response of the
detector for Case 2.
206 Euro. Trans. Electr. Power 2011; 21:196211 2010 John Wiley
& Sons, Ltd.DOI: 10.1002/etep
Online monitoring of power transformers A. Vahedi and V.
Behjat
-
-20000
-15000
-10000
-5000
0
5000
10000
15000
20000
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
a) P
rimar
y Li
ne C
urre
nts(
A)
iapibpicp
-40000
-30000
-20000
-10000
0
10000
20000
30000
40000
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
b) S
econ
dary
Lin
e Cu
rren
ts(A
)
iasibsics
-200
-150
-100
-50
0
50
100
150
200
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
c) N
egat
ive
Sequ
ence
Com
pone
nts(
A)
Primary Negative Sequence CurrentSecondary Negative Sequence
Current
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
d) O
pera
ting
Resp
onse
of t
he D
etec
tor
(a)
(b)
(c)
(d)
Figure 10. (a) Primary line currents, (b) secondary line
currents, (c) negative sequence components, and (d) operating
response of the
detector for Case 3.
Euro. Trans. Electr. Power 2011; 21:196211 2010 John Wiley &
Sons, Ltd.DOI: 10.1002/etep
207
A. Vahedi and V. Behjat Online monitoring of power
transformers
-
-400
-300
-200
-100
0
100
200
300
400
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2Time(s)
a) Pri
mar
y Lin
e Cu
rren
ts(A)
iapibpicp
-800
-600
-400
-200
0
200
400
600
800
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2Time(s)
b) Se
con
dary
Li
ne
Curr
ents(
A)
iasibsics
-250
-200
-150
-100
-50
0
50
100
150
200
250
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2Time(s)
c) Neg
ative
Se
quen
ce Co
mpo
nen
ts(A)
Primary Negative Sequence CurrentSecondary Negative Sequence
Current
0
0.002
0.004
0.006
0.008
0.01
0.012
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2Time(s)
d) Op
erat
ing
Resp
onse
of t
he D
etec
tor
(a)
(b)
(c)
(d)
Figure 11. (a) Primary line currents, (b) secondary line
currents, (c) negative sequence components, and (d) operating
response of the
detector for Case 4.
208 Euro. Trans. Electr. Power 2011; 21:196211 2010 John Wiley
& Sons, Ltd.DOI: 10.1002/etep
Online monitoring of power transformers A. Vahedi and V.
Behjat
-
-400
-300
-200
-100
0
100
200
300
400
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
a) P
rimar
y Li
ne C
urre
nts(
A)
iapibpicp
-800
-600
-400
-200
0
200
400
600
800
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
b) S
econ
dary
Lin
e Cu
rren
ts(A
)
iasibsics
-200
-150
-100
-50
0
50
100
150
200
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
c) N
egat
ive
Sequ
ence
Com
pone
nts(
A)
Primary Negative Sequence CurrentSecondary Negative Sequence
Current
0
0.002
0.004
0.006
0.008
0.01
0.012
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
d) O
pera
ting
Resp
onse
of t
he D
etec
tor
(a)
(b)
(c)
(d)
Figure 12. (a) Primary line currents, (b) secondary line
currents, (c) negative sequence components, and (d) operating
response of the
detector for Case 5.
Euro. Trans. Electr. Power 2011; 21:196211 2010 John Wiley &
Sons, Ltd.DOI: 10.1002/etep
209
A. Vahedi and V. Behjat Online monitoring of power
transformers
-
5. CONCLUSION
sequen
6. LIST OF SYMBOLS AND
REFERENCES
Online monitoring of power transformers A. Vahedi and V.
BehjatABBREVIATIONS
6.1. Symbols
HV high voltage side of the transformerIaf phasor of the
fundamental component of the phase
a line current signal
Ibf phasor of the fundamental component of the phaseb line
current signal
Icf phasor of the fundamental component of the phasec line
current signal
Ian negative sequence component of the line currentsIpn negative
sequence component of the line currents
from primary side
Isn negative sequence component of the line currentsfrom
secondary side
LV low voltage side of the transformerN1 turns number of the
primary windingN2 turns number of the secondary windingRf fault
resistance between shorted turnsa phase rotation operator
equivalent to ej2p=3.
6.2. Abbreviation
RNSC Ratio of negative sequence componentscurrent data. Also, no
information concerning the
transformer and power system parameters is needed for
the application of the technique.method a
impleme210nd no additional measurements are required to
nt the technique since it only needs the terminalbeen preue for
different faults and operating conditions have
sented. The proposed method is a non-invasiveSome ca
techniqsting power transformer fault detection methods.
se studies verifying the performance of thefaults. H
the exiotection schemes in detecting low level interturn
ence, it is found to be a very good complement toconquers
mer prthe limitations of the traditional power transfor-heavily
dtely independent of time harmonics even under
istorted conditions. Further, the proposed methodvariation
complen fault and an imbalance due to load and supply
s and external fault conditions. Thus the method isof depen
interturdability, between an asymmetry caused by aninterturn
faults, but also differentiate, with a high degreefault detection
method can not only detect windingcaused serious imbalance in the
line currents. The proposedfound toce components of the transformer
line currents are
be sensitive to interturn fault occurrence whichmethod for
detecting interturn faults of power transformers
has been presented in this paper. The effective negativeA
negative sequence current-based sensitive detectionEuro.1. Bartley
W. Analysis of transformer failures. Inter-national Association of
Engineering Insurers 36thAnnual Conference, Stockholm, Sweden,
2003.
2. Stigant SA, Franlin AC. The J&P Transformer Book:
A Practical Technology of the Power Transformer;
10th edn, Wiley: New York, 1973.
3. Cruz SMA, Cardoso AJM. Multiple reference frame
theory: a new method for the diagnosis of stator faults
in three-phase induction motors. IEEE Transactionson Energy
Conversion 2005; 20: 611619. 10.1109/TEC.2005.847975.
4. Arkan M, Unsworth PJ. Stator fault diagnosis in
induction motors using power decomposition. Pro-ceedings of the
IEEE Industry Applications Confer-ence 34th Annual Meeting, Vol. 3,
Phoenix, USA,1999; 19081912.
5. Arkan M, Kostic-Perovic D, Unsworth PJ. Online
stator fault diagnosis in induction motors. IEE Pro-ceedings of
the Electric Power Applications 2001;148(6): 537547.
6. Shuo C, Rastko Z. Modelling and simulation of stator
and rotor fault conditions in induction machines for
testing fault diagnostic techniques. European Trans-actions on
Electrical Power 2009; Published online inWiley InterScience.
10.1002/etep.342.
7. Arkan M, Kostic-Perovic D, Unsworth PJ. Modeling
and simulation of induction motors with inter-turn
faults for diagnostics. Electric Power System Research2005; 75:
5766. 10.1016/j.epsr.2004.08.015.
8. Kohler JL, Sottile J, Trutt FC. Alternatives for asses-
sing the electrical integrity of induction motors.
IEEETransactions on Industry Applications 1992; 28(5):11091117.
10.1109/28.158836.
9. Kohler JL, Sottile J, Trutt FC. Condition monitoring of
stator windings in induction motors. I. Experimental
investigation of the effective negative-sequence impe-
dance detector. IEEE Transactions on Industry Appli-cations
2002; 38(5): 14471453. 10.1109/TIA.2002.802935.
10. Albizu I, Zamora I, Mazon AJ, Tapia A. Techniques
for online diagnosis of stator shorted turns in induction
motors. Electric Power Components and Systems2006; 34(1): 97114.
10.1080/15325000691001359.
11. Sidhu TS, Gill HS, Sachdev MS. A transformer pro-
tection technique with immunity to CT saturation and
ratio-mismatch conditions. IEEE Canadian Confer-ence on
Electrical and Computer Engineering, Vol. 1,Waterloo, Ontario,
Canada, 1998; 2428.
12. Sidhu TS, Gill HS, Sachdev MS. A numerical tech-
nique based on symmetrical components for protect-
ing three-winding transformers. Electric PowerSystem Research
2000; 54: 1928. 10.1016/S0378-7796(99)00069-3.
13. Guzman D, Ignacio D, Pablo A, Javier G-A. Zero-
sequence-based relaying technique for protectingTrans. Electr.
Power 2011; 21:196211 2010 John Wiley & Sons, Ltd.DOI:
10.1002/etep
-
power transformers and its performance assessment
using unsupervised learning ANN. European Trans-actions on
Electric Power 2006; 16: 147160.10.1002/etep.72.
14. Guzman D, Javier G-A, Pablo A. Diagnosis of a turn-
to-turn short circuit in power transformers by means of
zero sequence current analysis. Electric Power Sys-tems Research
2004; 69: 321329. 10.1016/j.epsr.2003.07.014.
15. Wagner CF, Evans RD. Symmetrical Components,
McGraw-Hill: New York & London, 1933.
16. Blackburn JL. Symmetrical Components for Power
System Engineering, Marcel Dekker: New York,
Basel, Hong Kong, 1993.
Appendix
conductors. From the physical point of view, stranded
conductors are characterized by a value of the skin depth
much greater than the dimensions of the conductor cross-
section and, as a consequence, by an almost uniform
distribution of the current density over all the conductor
cross-section. Solid conductors are characterized by a
value of the skin depth comparable to or smaller than the
dimensions of the conductor cross-section. The density of
supplied or induced currents is non-uniform in the cross-
section of such conductors. In this study, both of the
primary and secondary windings, which have a small
cross-section and high number of turns and consequently
negligible eddy current losses in the windings, are modeled
as stranded coil conductors.
To analyze the transient dynamic behavior of the
transformer, the equations of electromagnetic and electric
circuit fields are directly coupled and solved simul-
taneously at each time step. The method of weighted
residuals has been applied to the coupled field-circuit
A. Vahedi and V. Behjat Online monitoring of power
transformersThe 2D transient magnetic solver in the FLUX
software
package was used to implement the transformer finite
element model. It is possible to carry out a 2D plane study
if the magnetic flux of the device is supposed to
concentrate
on the cross-section plane and there may be no extremity
effect or magnetic flux leakage in the third direction. A 2D
study is recommended in modeling of the transformers,
because the magnetic fluxes, created by the primary and
secondary conductors, is strongly confined in the magnetic
circuit and therefore in the cross-section plane.
In general, the magnetic field model of the transformer
can be distinguished into three parts: the core, the
windings, and the oil surrounding the core and the
windings. Neglecting the stray losses in the transformer
tank walls, the transformer tank has excluded in the
computation domain. The core and surrounding oil has
been entirely included in the model. Representation of the
winding is related to the modeling of skin effect.
Ordinarily, the conductors concerned by the field-circuit
coupling are of two types: stranded conductors and solidEuro.
Trans. Electr. Power 2011; 21:196211 2010 John Wiley & Sons,
LtdDOI: 10.1002/etepequations to yield the finite element matrix
equations.
Also, the finite element time stepping scheme has been
used to discretize the equations in the time domain. To
obtain a unique solution for the governing equation based
on AVA formulation, the zero Dirichlet boundary
condition is applied on the external border of the
computation domain.
The mesh of the regions of the transformer should be
created depending on the physics of the problem, since the
quality of the results depends on the quality of the mesh.
The mesh is much more refined in zones of strong variation
and high intensity of the magnetic field than in the zone
close to the computation domain boundary. A mesh
generator that creates first-order rectangular elements has
been used for meshing of the windings. The core,
surrounding oil, and the boundary domain were fully
discretized by triangular first-order elements. The finite
element model of the studied power transformer is
composed of 17 598 surface and 7406 line elements and
includes 38 827 nodes as illustrated in Figure 2.. 211