Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Electr Eng (2013) 95:135–145

DOI 10.1007/s00202-012-0248-4

ORIGINAL PAPER

An experimental approach for investigating low-level interturn


winding faults in power transformers
Vahid Behjat · Abolfazl Vahedi

Received: 2 October 2011 / Accepted: 28 April 2012 / Published online: 13 May 2012
© Springer-Verlag 2012

Abstract Interturn faults are one of the most prevalent and 1 Introduction
potentially destructive electrical faults in power transform-
ers. In spite of an extensive interturn fault detection literature, Power transformers are critical components of the energy
the issue still constitutes an open problem. The difficulty is transmission and distribution process for electric utilities.
mainly caused by lacking of enough information on the char- In view of increasing demand for reliable and high-quality
acteristic signatures associated with interturn faults which energy supply, electrical utilities are more interested in avoid-
were not clearly brought out by the previous studies. This ing transformer failures. Whatever the consequences of fail-
contribution is aimed at obtaining a better understanding of ures, such events are undesirable since they constitute a
the physical behavior of the power transformers in the pres- decrease in overall system reliability and additional demands
ence of interturn faults as well as extracting several features on manpower, finance, and time, in order to rectify the prob-
which would be useful to specify the transformers inter- lem. However, degradation as a result of aging under ser-
turn faults. The approach keeps at disposal a 100 kVA distri- vice conditions is inevitable. Arising primarily from turn
bution transformer on which interturn faults were imposed insulation degradation, interturn winding faults are one of
and an experimental setup consisting of an instrumentation the leading root causes of failures in power transformers.
system and suitable transducers which enables monitoring Deterioration of the turn insulation, in turn, may be caused
selected characteristics of the transformer. The paper exam- by a combination of electromechanical forces induced by a
ines the faulted transformer performance under various fault variety of factors, including frequent transformer overload-
and transformer operating conditions. Unique characteristic ing, mechanical vibrations, high-transient voltage stresses,
fault-signatures derived from the experiments will provide high-current stresses particularly in the presence of exter-
an important basis for developing more reliable and sensitive nal short circuits, thermal overloading, and contamination
methods to detect interturn faults on the transformer wind- [1–3].
ings well before such faults lead to a catastrophic failure with Early stages of winding interturn faults may often have
serious damage to the windings. negligible effects on the transformer performance; however
such faults may rapidly lead to more serious permanent forms
Keywords Power transformer · Interturn winding fault · such as phase to phase or phase to ground faults. Once the
Experimental approach · Fault detection interturn faults was occurred, a large circulating fault current
is induced in the shorted turns, leading to localized thermal
overloading in the defective region of the winding. Over some
period of time, the generated heat in the defective region will
V. Behjat · A. Vahedi
Center of Excellence for Power System Automation and Operation, cause the fault to increase in size, i.e., fault energy level,
Iran University of Science and Technology, Tehran, Iran until a catastrophic fault involving another phase or ground
occurs. Accordingly, it would be advantageous to detect
V. Behjat (B)
interturn fault in its earliest stage to prevent further damage
Department of Electrical Engineering, Engineering Faculty,
Azarbaijan University of Tarbiat Moallem, Tabriz, Iran to the transformer, thereby reducing repair costs and trans-
e-mail: behjat@azaruniv.edu former outage time. In addition to the benefits gained from

123
136 Electr Eng (2013) 95:135–145

eliminating irreversible damage to the transformer windings 2 Experimental setup


through timely prediction of the impending failure, signifi-
cant advantages would accrue by orderly shutting down the Experimental tests were conducted in a high voltage labo-
transformer and doing some advanced planning for repair ratory on a three phase, two winding, 35 kV/400 V, Yzn5,
or replacement of the equipment and hence enhancing the 50 Hz, 100 kVA, oil immersed, ONAN, core type distribu-
reliability of the continuous power supply operation. Obvi- tion transformer. Interturn faults were imposed on the turns
ously, in a power system, where a transformer is playing a of the outermost layer of the HV discs, which was the only
key role as an energy transmission device, a sudden trans- accessible part of the transformer’s windings. To develop in-
former failure will likely result in an overall network fail- terturn short circuit faults, in steps, the transformer oil was
ure. pumped out and the front wall of the transformer tank was
Due to its popularity, there have been many investigations removed to expose the windings. After the windings were
on interturn faults in power transformers throughout the liter- allowed to dry, two conductors on the farthest layer of one of
ature, especially in the recent years [4–20]. Owing to many the HV discs, located at two ends of the layer, were chosen
technical and economic limitations in generating interturn and the insulation over them at a point on each was carefully
faults in an actual transformer, most of the investigations removed to make tapping points. The next step was extract-
are focused upon techniques developed to model a power ing leads from the tapping points on the chosen conductors.
transformer under interturn fault conditions for the purposes Low impedance insulated wires were attached to the conduc-
of obtaining diagnosis criteria. However, it is well known tors by means of specific clamps embracing the conductors
that the utility of models for fault diagnosis is restricted at the tapping points. The leads were then brought out of
because it is even theoretically impossible to include all non- the transformer to allow easy access to the internal turns and
idealities that exist in a real machine. On the other hand, also providing possibility for externally producing interturn
a literature review indicates that the few published contri- faults. The leads were long enough to be closed from outside
butions pertinent to the present study [6,7], find it enough the transformer through a switch.
to show the overall changes of the transformer behavior Since it was difficult to quantify exactly what number
due to fault occurrence. The modification of the transformer of turns were involved in the fault, after the connections
behavior as a result of interturn fault occurrence has nei- were completed, the winding was energized by a low-voltage
ther been systematically analyzed nor reasons for it ascer- power supply and the open circuit voltage between the tap
tained in the existing literature. In an attempt to overcome conductors was recorded. This measured voltage between
the shortcomings of the existing literature, in this paper, a the taps, divided by the line to neutral value of the measured
detailed experimental investigation of the interturn faults in voltage applied to the winding, was an exact measure of the
power transformers is carried out by using an experimen- fraction of the winding that was involved. The fault level
tal setup, and the influence of a whole variety of different which could be realized by shunting the tap conductors was
system and fault conditions on the transformer characteris- equal to 0.2 % of the turns on the winding which involves a
tics is clarified. The present study is of utmost importance very small percentage of the winding. Before reassembling
since as an essential pre-requisite to the successful devel- the transformer, an insulation resistance test was performed
opment of fault diagnosis techniques, it is vitally important to verify that the resistance of the tap conductors to ground
to be able to obtain the characteristic fault signatures, and was greater than 1 M. The front wall of the transformer
experimental data associated with the interturn fault phenom- tank was replaced with a glass wall fixed to the tank by screw
enon. bolts and then the transformer was refilled with oil. Thereaf-
The paper first outlines the experimental setup in a ter, interturn faults could be staged by connecting two taps
high voltage laboratory used for acquiring the experi- to each other through a low-impedance knife switch to be
mental data. This includes a brief description address- able to handle the extremely high circulating fault currents
ing the electrical characteristics of the transformer under flowing through the shorted turns. To adjust the fault sever-
test on which the interturn faults were imposed and the ity in the shorted turns, a variable resistor was used in series
instrumentation system used for monitoring the electrical with the switch in the conductive path between the terminals
characteristics of the transformer. Section 3 is on the meth- of the fault region. Various levels of fault severity could then
odology used for doing the experiments. Section 4 presents be attained by changing the value of the fault resistance in
the results obtained from exhaustive tests on the studied this leakage path.
transformer and detail discussions to extract some features The sketch in Fig. 1, illustrates an exaggerated presenta-
which would be useful to specify the transformer winding tion of the corresponding geometrical and circuit domain of
interturn faults. Finally, conclusions will be given in the the transformer coils, assuming that an interturn fault exists
Sect. 5. on phase “U” of the transformer HV winding. In the figure,

123
Electr Eng (2013) 95:135–145 137

HV Winding LV Winding
To provide a platform for testing the transformer under
varying conditions of load and fault level, a test setup was
Phase U
Rf
Phase u
constructed which consisted of the transformer under test,
S supplying transformers, two sets of resistive load banks, a
Shorted Turns 400 V distribution lateral and protective devices, i.e., fuses
HV Winding
LV Winding

and over-current protection relays. Three single-phase high-


Core

Phase V Phase v voltage transformers and a LV voltage-regulated autotrans-


former were used to supply power to the tested transformer
from HV and LV sides, respectively. Since, it was neces-
Shorted Turns
sary to conduct the tests at different levels of supply volt-
Phase W Phase w
age, both sets of the supplying transformers were modified
to allow the voltage to be controlled to the required level up
(a) Geometrical Domain (b) Circuit Domain to the nameplate value for the tested transformer by convert-
ing the three-phase voltage in the 400 V distribution lateral
Fig. 1 Geometrical and circuit domain of the transformer coils, assum-
ing that an interturn fault exists on phase “U” of the HV winding of the laboratory.
Two sets of three-phase resistive load banks consisted
the external limiting fault resistance and the time-controlled of 100 kV and 400 V paralleled single-phase resistors in
switch for initiating the fault are denoted by “s” and “Rf ” each phase that could be individually controlled to provide
symbols, respectively. A close view of the considered trans- a wide range of loads for the transformer were used for
former’s HV winding and tap positions of the introduced loading the HV and LV sides of the transformer, respec-
fault, before and after refilling the oil to the transformer tank, tively. An instrumentation and measurement system was
is given in Fig. 2. developed to sense the terminal signals as well as the cir-
The above artificially imposed interturn fault on HV wind- culating current in the shorted turns and also to store and
ing of the transformer belongs to a class of turn-to-turn process the data. A set of seven current transformers in
faults which would be created by direct contact between conjunction with resistors were used for capturing each of
the winding conductors, normally mentioned as galvanic the three-phase current signals’ waveforms, six for the line
short-circuits. Faults of this kind develop as a very poor ini- currents in both primary and secondary side of the trans-
tial point contact, immediately heated to the melting tempera- former and one, embracing the shorted turns. The resis-
ture, open at one point and finally permit contact between the tors were used to convert each CT output to an appropriate
conductors of different potentials without arc ignition [19]. It voltage signal. Terminal voltage signals were acquired by
should be pointed out that, as stated in [21], probably just only using voltage dividers connected between transformer ter-
in transformers above 10 MVA which are characterized by minals and ground. A set of resistive voltage dividers were
voltages per turn exceeding 50 and small values of turn con- used to obtain the voltage signals on the LV side of the
ductor’s resistance, occurrence of the interturn fault is accom- transformer. In the case of HV side, due to the relatively
panied by maintaining an electric arc across the shorted turns high level of the voltage signals even after passing through
in normal frequency voltage, which can burn in either a stable the voltage dividers, first, shunt capacitors of large values
or intermittent way. Obviously, for the 100 kVA transformer (100 kV) were added across the HV terminals to reduce
considered here, the circuit conduction and turn voltage are the terminal voltages to measurable levels of the record-
adequate small to avoid arc ignition in the shorted turns. ing apparatus. After that a capacitor divider was utilized

Fig. 2 HV winding of the


transformer and tap positions on
the winding, before and after
refilling the oil

123
138 Electr Eng (2013) 95:135–145

Fig. 3 Experimental setup used


for performing the tests. a Three
single phase transformers for
supplying HV side of the
transformer and HV shunt
capacitors, b resistive load for
loading the HV side of the
transformer

disassembly to verify that the modifications had not changed


the transformers’ normal operating characteristics. Once this
test was completed, selected fault scenarios were staged in
the laboratory using the setup discussed in the previous sec-
tion.
The first series of tests was designed to characterize the
electrical behavior of the transformer during interturn faults
on the primary and secondary sides while the transformer was
supplied with the rated nameplate values of the supplying
voltage. In the next step of the experimental tests, additional
runs were conducted with the transformer operating under
various levels of supplying voltage and load as well as fault
severity to determine the effects of theses parameters on the
electrical behavior of the faulty transformer. This stage was
accomplished by repeating the first series of tests but with dif-
Fig. 4 Data acquisition setup for conducting the experiments ferent values of the mentioned parameters; however a same
procedure was followed for both series of the tests, that is,
to further lower the signals to levels that could be directly after the desired values for the load and fault resistance were
recorded. adjusted, the transformer was then energized and the volt-
The results from tests were recorded through a data acqui- age on the supplying transformer was varied slowly till the
sition system, with a sampling frequency of 10 kHz. The requested value for the tested transformer was reached. The
obtained signals were stored on data files in the recorder and transformer allowed stabilizing over about a minute before
then imported into standard packages for performing addi- the fault was initiated to take the measurements correspond-
tional analyses. Other variables such as load and voltage level ing with the failure-free operation, so called normal state
as well as fault severity were also saved in each experiment of the transformer. Then the switch was closed to stage the
case and appended to the data files for further processing. short circuit fault and make the faulty state measurements.
Figures 3 and 4 show the experimental setup and data acqui- The transformer under test was allowed to experience not
sition system used for conducting the tests. more than a few cycles of short circuit time in order to avoid
evolving the fault into a probable transformer failure.
Since, depending on the introduced interturn fault located
3 Experiment methodology on the primary or secondary side of the transformer, dif-
ferent characteristic pattern appears in the physical behav-
Over the course of the experimental tests, a series of exper- ior of the faulty transformer, it was necessary to cover the
iments were conducted, in some cases with several trials of fault locations on both sides of the transformer. Keeping
each to verify the correctness of the measurement records. this in view that the considered interturn fault was imposed
The first step of the experimental procedure consisted of run- on the outermost turns of the HV winding discs, which
ning a full-load test on the transformer with the taps open and was the only accessible part of the tested transformer’s wind-
then comparing the recorded three-phase voltage and current ings, the transformer must be supplied from HV and LV
waveforms in this condition with those ones obtained prior to windings, respectively, for locating the staged fault on the

123
Electr Eng (2013) 95:135–145 139

Tested Transformer 60

Primary Line Currents (A)


Iu
LV AutoTransformer U V W 40 Iv

HV Resistive Load
Iw
u v w 20

Shorted -20
Turns
-40
LV Winding HV Winding
-60
(a) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
Tested Transformer (a)

Secondary Phase Voltages (kV)


U VW 40

LV Resistive Load
volU volV volW
u v w 30
HV Supplying
Transformers

20
10
Shorted
Turns 0
-10
HV Winding LV Winding -20

(b) -30
-40
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20
Fig. 5 Test setup for locating the imposed interturn fault on a second- Time(s)
ary side, b primary side of the considered transformer (b)
Secondary Line Currents (A)
0.45
IU IV IW
0.3
primary and secondary sides. Figure 5 describes a schematic
0.15
representation of the test setup including the tested trans-
former, supplying transformers and resistive loads for load- 0

ing the transformer, in two conditions where the fault is -0.15

located on the primary and secondary sides of the trans- -0.3


former. -0.45
The tests were repeated until the entire desired fault sce- 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
narios had been carried out. After each experiment case, the
(c)
condition of the transformer was assessed by visual inspec- 150
If
Circulating Current (A)

tion and performing a full load test to ensure that the staged 100
fault had not damaged the transformer or in any other way
50
changed its characteristics. In all of the experiments con-
0
ducted for this research, the primary and secondary voltages
and currents as well as the circulating current were recorded. -50

The remainder of this paper will describe the results of this -100
experimental study and their analysis. -150
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
(d)
4 Findings and results
Fig. 6 Terminal waveforms of the transformer for an interturn fault
As mentioned earlier, the experimental tests were carried out located on secondary side of the transformer. a Three phase primary
line currents, b three phase secondary phase voltages, c three phase
in two stages. The first stage of the experiments was related
secondary line currents, d circulating current in shorted turns
to investigation the characteristics of the transformer in rated
voltage conditions and in presence of interturn fault located turns is denoted by the index f. Three phase resistive load
on the primary and secondary sides. Figures 6 and 7 present equal to 83 and 17  was connected to the HV and LV sides,
the results, including recorded waveform of the terminal val- respectively, in each experiment case. As well, the value of
ues as well as circulating current in the shorted turns, when the external fault resistance was chosen to zero to account
the transformer was supplied from the LV and HV sides and for a metal-to-metal contact and dispose of an extreme value
hence the fault was located on secondary and primary sides to help evaluate trends. However, it should be remarked that
of the transformer respectively. In the figures, the three phase the short circuit loop itself had a resistance equal to 690 m,
HV and LV side currents are denoted by the index U, V, W including the resistance of the connection point, connecting
and u, v, w, respectively. The circulating current in the shorted clamps and leads brought out of the transformer.

123
140 Electr Eng (2013) 95:135–145

Primary Line Currents (A) 0.8


IU law states that the electromotive force (emf), induced in a
0.6 IV
0.4 IW turn equals the rate of variation of the electromagnetic flux
0.2 inside it. As a well-known conclusion from this law, at a fixed
0 frequency, the amount of flux entering the turn is fixed by the
-0.2 emf at its terminals. From this assumption, when a short cir-
-0.4
cuit occurs at a given turn, the voltage is forced to drop in
-0.6
-0.8
it and consequently a circulating current flows in the short
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
circuited part, so that it produces opposite mmf to the phase
Time(s)
(a) currents’ mmf to limit the entering flux into the turns. It is
Secondary Phase Voltages (V)

450 obvious that when an interturn fault involves a low fraction of


volu volv volw
300 the winding, the circulating current must be high enough to
150
be able to oppose the mmf of the phase winding in the short
circuited region in order to force the flux to surround the
0
damaged turns. The higher the turn ratio between the whole
-150
winding and short circuited turns, the larger is the circulat-
-300
ing current to be flowed in the shorted turns. The amount of
-450 circulating current in the shorted turns is also related to the
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s) fault severity which will be discussed in more detail in the
(b) following section.
Secondary Line Currents (A)

30
Iu Iv Iw The results of the experiments demonstrate that for an
20
interturn fault on the transformer windings, either on the pri-
10 mary or secondary side, the primary current will rise, Figs. 6a
0 and 7a. The increased primary current is due to the reduced
-10 value of the mmf needed for establishing the linkage flux in
-20 the transformer as a result of fault occurrence. It is worth
-30
pointing out that, according to Faraday’s law, with the pri-
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 mary voltage held constant, the linkage flux passing through
Time(s)
the primary and secondary winding is constant. The effect of
(c)
150
the interturn fault is to remove a turn from the transformer
If
winding. Therefore, when a short circuit occurs, the phase
Circulating Current (A)

100
winding has less turns and, therefore, less MMF. In addition,
50
as stated earlier, the circulating current due to the shorted
0
turn is the source of an opposite mmf. Thus, interturn short
-50 circuits have a cumulative effect in decreasing the mmf near
-100 the short-circuited turns. Larger current must, therefore, be
-150 drawn from the primary to bring the mmf and so linkage flux
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s) to the value proportional to the constant supply voltage. Com-
(d) parison the given results in Figs. 7a with 6a indicates when
the star-connected HV winding was used as the primary side
Fig. 7 Terminal waveforms of the transformer for an interturn fault of the transformer, just the faulty phase current increases after
located on primary side of the transformer. a Three phase primary line
currents, b three phase secondary phase voltages, c three phase second- the fault occurrence; while for the case of zigzag-connected
ary line currents, d circulating current in shorted turns LV winding, tow line currents increase as a result of fault.
Obviously, different connection type of the windings con-
tributes to the observed difference in the primary currents
As can be clearly understood from observation of the behavior.
Figs. 6 and 7d, the interturn fault on the transformer wind- Observation of the Fig. 7b indicates that in the case of
ings, either on the primary or on the secondary side, causes a interturn fault on the primary winding, the secondary volt-
large circulating current, in this case of the order of 45 of the age does not change remains fairly constant. This is due to
transformer rated current, to flow in the shorted turns. Such constant supply voltage and hence constant linkage flux pass-
a great circulating current is a result of high ratio of transfor- ing through the secondary winding. Since the current flow-
mation between the whole winding and the short-circuited ing through the load is in direct proportion to the secondary
turns. The key to understand this behavior can be found in voltage in all cases, so it has the characteristics of the sec-
explanation of Faraday’s law in the shorted turns. Faraday’s ondary voltage, Fig. 7c. For the fault on the secondary side,

123
Electr Eng (2013) 95:135–145 141

Primary Line Current (A)


Table 1 Relative change of the terminal values after fault occurrence 60 IuNL
on secondary side of the transformer IuL1
40 IuL2

Terminal parameter Relative 20


change (%) 0
-20
Primary line currents
-40
Iu 78.27
-60
Iv 00.07 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time(s)
Iw 77.05
(a)

Secondary Phase Voltage (kV)


Secondary phase voltages 45
volU −00.48 30
volUNL volUL1 volUL2

volV −00.18 15
volW 00.06 0
Secondary line currents -15
IU −00.48 -30
IV −00.18 -45
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
IW 00.07
Time(s)
(b)

Circulating Current (A)


150
Table 2 Relative change of the terminal values after fault occurrence IfNL
100 IfL1
on primary side of the transformer IfL2
50
Terminal parameter Relative 0
change (%)
-50

Primary line currents -100

IU 218.06 -150
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
IV 005.88 Time(s)
IW 009.88 (c)
Secondary phase voltages
Fig. 8 Terminal waveforms of the transformer for three different load
volu −0.002
levels. a Primary line currents, b secondary phase voltages, c circulating
volv 0.009 current in shorted turns
volw −0.006
Secondary line currents transformer and the supplying voltage was kept constant at its
Iu −0.002 nominal value. Figure 8 presents the recorded waveforms of
Iv 0.009 the transformer in no-load and two-load conditions when the
Iw −0.006 transformer was supplying 83 and 125 k resistive loads at its
HV side. In the figures, NL stands for the no-load condition
and L2, L1 for the 125 and 83 k loads. As one can see
Fig. 6b and c, the secondary voltage and current will decrease. from the comparison of waveforms in Fig. 8c, the circulat-
The shorted turns on the secondary winding will cause the ing current in the shorted turns remains same for all levels
effective number of the turns to be reduced which results of the transformer loading. This would be expected since
in decreasing the secondary voltage and current. Since the the circulating current in the shorted turns is mainly deter-
staged fault involves just very low fraction of the HV winding mined by the linkage flux established in the magnetic core,
turns, so the fault effect on the secondary voltage and current so with a constant supply voltage and hence constant link-
is too small to identify clearly. To better observe the modi- age flux in the core, the circulating current remains constant
fications of the terminal values as a result of fault, Tables 1 regardless of the load current. Although, owing to the con-
and 2 summarizes relative change of the terminal values after stant circulating current in the shorted turns, the effect of
the fault occurrence on the secondary and primary windings, the fault on the primary current is the same at different load
respectively. conditions, but with the transformer unloaded, the relative
The second stage of the testing was related to the deter- change remains significant, Fig. 8a. As the transformer load-
mination of the effects of the load and supply level as well as ing was increased, the relative change of the primary current
fault severity on the electrical characteristics of the faulty for the shorted versus normal condition was reduced because
transformer. In all of the experiment cases performed in the fault effect is too small when compared to the transformer
this section, the fault was located on secondary side of the higher load currents. This would seem to present a problem

123
142 Electr Eng (2013) 95:135–145

60

Primary Line Current (A)


Table 3 Relative change of the terminal values for three different load Iu100
Iu150
levels 40 Iu200
Iu230

Terminal parameter Relative 20


change (%) 0

Primary line currents -20

IuNL 787.33 -40

IuL2 119.03 -60


0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
IuL1 078.27 Time(s)
Secondary phase voltages (a)

Secondary Phase Voltages (kV)


volUNL −00.48 45
volU100 volU150 volU200 volU230
volUL2 −00.48 30

volUL1 −00.50 15

-15

for fault detection in full-load conditions where it would be -30

very difficult to detect a change in the terminal current caused -45


0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
by the interturn fault in spite of the drastic circulating current Time(s)
in the shorted turns. Owing to the constant linkage flux pass- (b)
ing through the secondary winding, before and after the fault 150
Circulating Current (A)
If100
occurrence, the relative change of the secondary voltage is 100 If150
If200
the same for all the load levels, Fig. 8b. As it was mentioned 50 If230

earlier, decreasing of the secondary voltage is just due to


0
decreasing the effective number of the turns on the winding
-50
as a result of fault; so while the number of shorted turns and
-100
their resistance is constant, the relative change of the sec-
ondary voltage and current remains unaltered. A summary -150
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
of the results illustrated in Fig. 8 are listed in Table 3. It Time(s)
should be remarked that the small differences between the (c)
relative changes of secondary voltage for different load lev-
Fig. 9 Terminal waveforms of the transformer for four different sup-
els, observed in Table 3, were caused by natural variation in plying voltage levels. a Primary line currents, b secondary phase volt-
the measuring process rather than by the fault effect. ages, c circulating current in shorted turns
Figure 9 shows the recorded waveforms of the transformer
for different levels of supplying voltage, applied on the pri-
mary while the secondary connected to the 83 k resistive of the secondary voltage is the same for all the supplying
load. Four levels of voltage used for supplying the LV side voltage levels, Fig. 9b. In fact, such a behavior is not sur-
of the transformer, rated value (230 V rms) and three levels prising, as the number of shorted turns is constant in all the
lower than the rating (200, 150, and 100 V rms), are denoted cases. The modifications of the terminal parameter values as
by 230, 200, 150, and 100 indexes in the figures. From Fig. 9c, a result of fault for the various supplying voltage levels are
it can be clearly observed how for small values of the sup- gathered in Table 4.
plying voltage, the circulating current in the shorted turns The tests were repeated retaining the fixed rated voltage in
tends to decrease. As can be shown, at the lowest level of the primary and the 83 k resistive load in the secondary, but
the supplying voltage, the circulating current was limited to with different fault resistance values to create the desired fault
32 A in rms value which is far less than attained in the rated severity level in the shorted turns. Figure 10 shows the cor-
conditions. This was predicted since the low level of the sup- responding terminal values and circulating current recorded
plying voltage means the low level of the linkage flux in the during interturn faults with four different fault-resistance val-
magnetic core and hence low level of the circulating current ues, denoted by Rf1–Rf4 in the figures. Observing Fig. 10c,
flowed in the shorted turns to oppose the mmf of the phase it becomes evident that during the exercise of increasing the
winding. Since, in each experiment case with different levels value of the fault resistance, i.e. decreasing the fault severity
of supplying voltage, the primary current must increase in a level, a decreasing is visible in the value of the circulating
manner to bring the linkage flux to a value proportional to the current flowed in the shorted turns. The magnitude of the
supplied voltage; thus, the relative change of the primary cur- primary current increases as much as 78 % in the worst con-
rent is the same for all the cases, Fig. 9a. The relative change dition where the fault-resistance is equal to 690 m to 12 %

123
Electr Eng (2013) 95:135–145 143

Table 4 Relative change of the terminal values for four different Table 5 Relative change of the terminal values for four different fault
supplying voltage levels severity levels (Rf1 = 0.69 , Rf2 = 1 , Rf3 = 2.1 , Rf4 = 7.4 )
Terminal parameter Relative Terminal parameter Relative
change (%) change (%)

Primary line currents Primary line currents


Iu100 78.22 IuRf1 78.27
Iu150 78.24 IuRf2 62.84
Iu200 78.26 IuRf3 36.78
Iu230 78.27 IuRf4 16.84
Secondary phase voltages IuRf5 12.15
volU100 −0.48 Secondary phase voltages
volU150 −0.48 volURf1 −0.48
volU200 −0.48 volURf2 −0.38
volU230 −0.48 volURf3 −0.22
volURf4 −0.10
60 volURf5 −0.07
Primary Line Current (A)

IuRf1 IuRf2
40 IuRf3 IuRf4

20

0 tion of largest value of the fault resistance representing the


-20 lowest degree of the fault severity. Indeed, the fault resis-
-40
tance represents the resistive component of the turn dielec-
-60
tric material in the dielectric equivalent parallel circuit model
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 [22]. When the transformer is in good condition (large val-
Time(s)
ues of the fault resistance), the dielectric is almost perfect,
(a)
the terminal current is very close to the rated values and the
Secondary Phase Voltage (kV)

45
volURf1 volURf2 volURf3 volURf4 fault circulating current is nearly zero. As the severity pro-
30
gresses and the fault becomes more severe (low values of fault
15
resistance), the terminal and fault currents increase more sig-
0 nificantly. Obviously, maximum values for the terminal and
-15 circulating currents are obtained when the dielectric material
-30 completely breakdowns and a metal-to-metal contact occurs
-45 between the turns. A summary of the results obtained in the
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
performed tests with different fault severity levels are tabu-
Time(s)
(b) lated in Table 5.
150 While on this point it is worth investigating how con-
Circulating Current (A)

IfRf1
100 IfRf2 ventional power transformer protection devices often fail in
IfRf3
IfRf4
detecting low-level interturn winding faults. The problem of
50
the traditional differential protection which recognized as the
0
basic protection of the power transformers against internal
-50
faults is that the low-level interturn faults cannot be detected
-100 with the overall sensitivity represented by the restrained dif-
-150
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
ferential protection until they developed into high-level faults
Time(s)
with more severe damage to the power transformer. In order
(c) to explain this issue in greater detail, reconsideration of the
results obtained in this section and a review of operating
Fig. 10 Terminal waveforms of the transformer for four different fault characteristics of the differential relay system will be useful.
severity levels (Rf1 = 0.69 , Rf2 = 1 , Rf3 = 2.1 , Rf4 = 7.4 ).
a Primary line currents, b secondary phase voltages, c circulating cur-
The current approach for differential protection of regulating
rent in shorted turns transformers is as follows: by selecting the minimum bias to
be greater than sum of the possible CT measurement errors
in the condition where the fault resistance has a value of and other factors introducing noise into the measurement
7.4 . The relative change of the secondary voltage decreases signals, maloperation due to this cause is avoided. Commonly,
from 0.48 % in the worst condition to 0.07 % in the condi- a general set to operate for a current imbalance of 10–50 %

123
144 Electr Eng (2013) 95:135–145

of rated current is recommended for the minimum pickup detection system relies on this understanding of the physical
differential current. The differential relay normally has a phenomena linked to the fault. Based on the analysis of the
dual slope bias characteristic. The initial percentage slope faulty transformer behavior in this paper, the future work will
from zero to rated current should be high enough to accom- be development of an online, reliable, and sensitive interturn
modate the full range of voltage change as the on-load tap fault detection method.
changer moves over its tapping range. At currents above the
rated current, extra errors may be gradually introduced as a
result of CT saturation. The second bias slope is, therefore, 5 Conclusion
increased to high levels to compensate for this [23]. It was
found from the experiments that an interturn fault involv- Experimental investigation of interturn faults in power trans-
ing 0.2 % of the turn on the HV winding of the considered formers was studied in this work. The results obtained with
transformer only causes a differential current of 12 % rated a 100 kVA distribution transformer are provided to show that
current for the lowest level of the fault severity in the shorted terminal values of the transformer convey information which
turns. Now, if the minimum pickup current of the restraint would be useful in detecting interturn faults. The transformer
differential protection, be set to 30 %, this low-level interturn has been tested in several fault and operating conditions
fault, cannot be detected until it evolves into a more severe aiming to account for the different device characteristics.
fault with higher differential currents. Exhaustive experiments on different fault scenarios revealed
The present work described the first experience of the that although the interturn short circuit fault was being of
authors in developing an online monitoring system for detec- little consequence on the terminal value of the transformer
tion of interturn winding faults in power transformers. It is as a whole, but there was a clear pattern in the trend of termi-
believed that, based on the present findings, the terminal nal behavior under all of the operating and fault conditions.
values of the transformer including the current and voltage The primary current increased during an interturn fault on
signals of the primary and secondary windings and also cir- the transformer winding either on the primary or secondary
culating current flowed in the shorted turns convey useful side. The secondary voltage and current did not change for
information regarding the fault occurrence on the transformer the fault on the primary winding and decreased when the fault
windings. The standpoint extracted from the results of the was on the secondary winding. These variations were corre-
entire tests, Tables 1–5, is the fact that there is a clear trend lated with the load and supplying voltage level as well as the
in the terminal values of the faulted transformer. Increasing fault severity in the shorted turns. In all of the fault cases,
the primary current of the transformer for the faults on the a large circulating current flowed in the shorted turn which
windings, either the primary or secondary winding, decreas- was of the magnitude of 145 times larger than the transformer
ing the secondary voltage and current for the faults on the rated current under the worst condition created. The circu-
secondary side and unchanged secondary voltage and current lating current was found to be dependent critically to some
for the faults on the primary side are discernible trends in all variables including the supplying voltage of the transformer,
of the experiment cases performed in this study. The com- or clearly the linkage flux established in the core and the fault
parison between the results of the second series (Tables 3–5) severity. The larger supplying voltage applied on the faulty
and first series of the tests (Tables 1, 2), demonstrates that the transformer, the larger the linkage flux and hence circulating
pattern in the trend of terminal voltages and currents is the current flowed in the shorted turns. As the fault became more
same at different fault scenarios and operating conditions. severe, the circulating current in the shorted turns tended to
The only remarkable difference is the reduced or increased increase. The transformer loaded with different load levels
effect of the fault in different operating and fault conditions. exhibited a constant circulating current in the shorted turns,
Using the information in the Tables 3–5, it can be concluded regardless of the load level. Although the primary current
that the worst case in view of the fault-detection ability is changes following the fault inception, it does not change
when a low-level interturn fault at its incipient stage is applied in a predictable manner under certain operating conditions.
on the transformers’ winding, while the primary side of the Under full-load conditions, reduced supplying voltage and
transformer is supplied by a reduced voltage and the second- incipient stages of low-level interturn faults, there was no
ary side connected to the rated load. Under this condition, remarkable change in the primary current of the transformer
the fault had no perceptible affect on the transformer oper- because the effect of the fault is too small when compared to
ation and so there was no detectable change in the perfor- the transformer larger load currents. In fact, this situation cor-
mance of the transformer. Of course, in practice, the decision responds with the worst case scenario to be detected. Char-
on whether or not a fault is underway depends on whether acteristic fault signatures associated with the interturn faults
the difference between the two conditions, before and after derived from the experiments will provide an important basis
fault, is statistically significant. The information provided for developing reliable and sensitive fault detection methods
is of great importance as a complete monitoring and fault- which constitutes one of the future directions of this research.

123
Electr Eng (2013) 95:135–145 145

Acknowledgments The authors are highly grateful to the Schering 11. Wang H, Butler KL (2002) Modeling transformers with internal
Institute of Leibniz University of Hanover for the provision of test incipient faults. IEEE Trans Power Deliv 17(2):500–509
facilities and great support for experimental tests. 12. Butler-Purry KL, Bagriyanik M, Mousavi MJ, Palmer-Buckle P
(2004) Experimental investigation of internal short circuit faults
leading to advanced incipient behavior and failure of a distribution
transformer. Proc Power Syst Conf Expo 3:1407–1416
References 13. Kezunovic M, Guo Y (2000) Modeling and simulation of the power
transformer faults and related protective relay behavior. IEEE Trans
1. Bartley W (2003) Analysis of transformer failures. In: Interna- Power Deliv 15(1):44–50
tional Association of Engineering Insurers 36th annual conference, 14. Abed NY, Mohammed OA (2007) Modeling and characterization
Stockholm, Sweden of transformers internal faults using finite element and discrete
2. Wang M, Vandermaar AJ, Srivastava KD (2002) Review of con- wavelet transforms. IEEE Trans Magn 43(4):1425–1428
dition assessment of power transformers in service. IEEE Electr 15. Liu Z, Mohammed OA, Liu S (2007) A practical method for build-
Insul Mag 18(6):12–25 ing the FE-based phase variable model of single phase transformers
3. Stigant SA, Franlin AC (1973) The J&P transformer book: a prac- for dynamic simulations. IEEE Trans Magn 43(4):1769–1772
tical technology of the power transformer, 10th edn. Wiley, New 16. Liu S, Liu Z, Mohammed OA (2007) FE-based modeling of single-
York phase distribution transformers with winding short circuit faults.
4. Bastard P, Bertrand P, Meunier M (1994) A transformer model for IEEE Trans Magn 43(4):1841–1844
winding fault studies. IEEE Trans Power Deliv 9(2):690–699 17. Díaz G, Arboleya P, Gómez-Aleixandre J (2006) Analytical
5. Aggarwal RK, Mao P (1998) Digital simulation of the transient approach to internal fault simulation in power transformers based
phenomena in high voltage power transformers with particular ref- on fault- related incremental currents. IEEE Trans Power Deliv
erence to accurate fault detection. In: Proceedings of the Institute 21(1):142–149
of Electrical Engineering international conference on simulation, 18. Díaz G, Gómez-Aleixandre J, Arboleya P (2004) Electromagnetic
pp 390–397 model of turn-to-turn short circuits in transformers. COMPEL Int
6. Palmer-Buckle P, Butler KL, Sarma NDR (1999) Characteristics of J Comput Math Electr Electron Eng 23(2):562–575
a transformer parameters during internal winding faults based on 19. Díaz G, Gómez-Aleixandre J, Arboleya P (2004) Diagnosis of a
experimental measurements. In: Proceedings of IEEE transmission turn-to-turn short circuit in power transformers by means of zero
distribution conference, New Orleans, LA, pp 882–887 sequence current analysis. Electr Power Syst Res 69(2–3):321–329
7. Butler KL, Kuforiji A (2001) Experimental results from short 20. Perez-Rojas C, Garcia-Martinez S (2005) Internal winding faults in
circuit faults on distribution transformers. In: Proceedings of three phase five limb transformer. In: 2nd International conference
2001 IEEE transmission and distribution conference, Atlanta, GA, on electrical and electronic engineering (ICEEE) and XI confer-
pp 299–306 ence on electrical engineering (CIE2005), Mexico City, Mexico,
8. Wang H, Butler KL (1999) Modeling transformer with internal pp 314–317
winding faults by calculating leakage factors. In: Proceeding of 21. Bean RL, Chackan N, Moore HR, Wentz EC (1959) Transformers
the 31th North American power symposium, October, San Luis for the electric power industry. McGraw-Hill, New York, pp 217
Obispo, CA, USA, pp 176–182 22. Malik NH, Al-Arainy AA, Qureshi MI (1997) Electrical insulation
9. Wang H, Palmer-Buckle P, Butler KL (1998) Transformer models in power system. Marcel Dekker, New York
for detection of incipient internal winding faults. In: Proceedings of 23. IEEE Guide for Protective Relay Applications to Power
the 30th North American power symposium, October, Cleveland, Transformers, Std. C37.91-2000, Inst Elect Electron Eng
OH, pp 90–98
10. Wang H, Butler KL (2001) Finite element analysis of internal
winding faults in distribution transformers. IEEE Trans Power
Deliv 16(3):422–427

123

You might also like