1636 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 31, NO.

4, AUGUST 2016

Residual Flux Mitigation of Protective Current

Transformers Used in an Autoreclosing Scheme
Ehsan Hajipour, Member, IEEE, Mohammad Salehizadeh, Mehdi Vakilian, Senior Member, IEEE, and
Majid Sanaye-Pasand, Senior Member, IEEE

Abstract—To avoid saturation of a protective current trans- a deep saturation and, thus, the CT output will be severely
former (CT) operating under an autoreclosing scheme, the CT size distorted [4]. The magnitude of residual flux that remained
should be chosen much higher than a similar CT operating under in the CT core under the breaker deadtime interval, directly
a single-step fault clearing scheme. The main reason for this sig-
nificant difference is that the residual flux (caused by the first stage contributes to CT saturation.
of the fault) cannot be noticeably reduced during the deadtime The conventional methods used to control the CT residual
interval of the reclosing process and, therefore, the occurrence of flux include: utilizing a special grade of steel in CT core, using
a subsequent fault can extremely saturate the CT. In this paper, a gapped core CTs, or employing biased core CTs [5]. These
low-cost, low-power electronic device is developed and introduced methods lead to the application of larger and more expensive
to demagnetize the CT under the reclosing deadtime interval. It
will be shown that by using this device, the required size of the cores [6]. Furthermore, when an unanticipated change occurs
CT is reduced by about 40%. To develop this device, an existing in the related power system short-circuit capacity (e.g., due to
flux-based demagnetization technique is improved to operate topological changes [7]), none of these methods are applicable
faster and more efficient. The performance of this technique is for the in-service CTs.
independent of the CT characteristic parameters and provides In this study, the CT secondary terminal voltage at which CT
promising results despite its simplicity. The proposed device can
be employed to de-rate the new CTs to be installed in the system will meet the ratio error limit of 10% at 20 times its rated pri-
and/or to compensate the existing inservice CTs. Comprehensive mary current is named “CT size” [5]. To design a CT with a
computer simulations and laboratory experiments are employed higher secondary terminal voltage, either the core area needs to
to validate the reliability and effectiveness of the proposed com- be increased or the flux density of the core has to be increased
pensation method. by selecting the appropriate material. Both of these would result
Index Terms—Autoreclosing, protective current transformer, in a rise in the CT cost [8]. Therefore, the CT sizing procedure
residual flux reduction, saturation. may be defined as the determination of the minimum secondary
terminal voltage to guarantee saturation-free performance for
the CT [8].
I. INTRODUCTION
In this paper, an effective and low-cost method is proposed

C URRENT-TRANSFORMER (CT) saturation threatens to notably reduce the CT size required for an autoreclosing
the reliable and proper performance of protective relays scheme. In this method, the CT residual flux under autore-
[1]. This phenomenon introduces serious distortions in the closing deadtime is mitigated by the application of an electronic
CT output and, as a result, significantly falsifies information demagnetizing device. Theoretically, if the residual flux is
delivered to the protection devices [2]. CT faces a critical eliminated, the CT transient behavior during each reclosing
regime when working under the application of an autoreclosing shot will be independent of the other shots. Therefore, the re-
scheme [3]. In this reclosing scheme, after initiation of a fault, quired CT size would not depend on the autoreclose open/close
and opening of the related circuit breaker (CB), a large amount duty cycle. Consequently, the required CT size to be employed
of residual flux can remain in the CT core. If the second fault in an autoreclosing scheme would be equivalent to the CT size
occurring after reclosing of the circuit breaker (CB) results in under a single fault case. Therefore, this proposed device will
a flux with the same polarity of the residual flux, CT may enter result in major CT size reduction.
To achieve this goal, a fast and reliable CT demagnetizing
technique is required. According to IEEE C57.13-2008 [9], CT
Manuscript received May 18, 2015; revised September 09, 2015; accepted
September 18, 2015. Date of publication September 22, 2015; date of current demagnetization may be accomplished through the application
version July 21, 2016. Paper no. TPWRD-00605-2015. of a certain level of an external ac voltage to the CT secondary
E. Hajipour, M. Salehizadeh, and M. Vakilian are with the Center of Excel-
terminal to saturate its core; then reducing the voltage magni-
lence in Power System Management and Control, Department of Electrical En-
gineering, Sharif University of Technology, Tehran 11365-11155, Iran (e-mail: tude slowly toward zero. This method needs extensive time for
e_hajipour@ee.sharif.edu; salehizadeh@ee.sharif.edu; vakilian@sharif.edu). demagnetization. However, commercially, a current-controlled
M. Sanaye-Pasand is with the Electrical and Computer Engineering School,
method using a dc voltage source is employed to demagne-
College of Engineering, University of Tehran, Tehran 14395-515, Iran, and also
with the Control and Intelligent Processing Center of Excellence, Electrical and tize the transformer core as stated in IEEE C57.152-2013 [10].
Computer Engineering School, University of Tehran, Tehran 14395-515, Iran Based on [11], at the first step, a fixed voltage is applied to the
(e-mail: msanaye@ut.ac.ir).
transformer terminal; next at the instant the current approaches a
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. preset value, the dc supply is automatically disconnected. Once
Digital Object Identifier 10.1109/TPWRD.2015.2480773 the current value approaches zero, a reverse polarity voltage is

0885-8977 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
HAJIPOUR et al.: RESIDUAL FLUX MITIGATION OF PROTECTIVE CTS USED IN AN AUTORECLOSING SCHEME 1637

reapplied and, finally, when current reaches 60% of its previous

value, the source is again disconnected. This process is repeated
until current approaches a very small amount. The drawbacks of
this method are as follows: first, it is time-consuming [12] and
second, since the magnetization current rises rapidly after core
saturation, this process may be inaccurate [13].
To resolve these drawbacks, a demagnetization process is of-
fered which relies on instantaneous magnetic flux measurement
information of the magnetic core [13]–[15]. The flux-controlled
methods have proven to be the safest and most efficient demag-
netizing approaches [13]. Kovan et al. [14] proposed a flux-con-
trolled technique to demagnetize power transformers through
estimation of the magnetic core saturation flux using a bidi-
rectional core saturation procedure. de Leon et al. [15] devel-
Fig. 1. (a) An unsuccessful single shot autoreclose current and (b) respected
oped an experimental verification procedure to demagnetize the value of the factor for the uncompensated CT versus the proposed method.
core and showed that this technique significantly reduces inrush
current of power transformers. The method presented in [14]
is based on voltage measurement of the transformer secondary breaker (CB) autoreclosing duty cycle. The main contribution of
side under the open-circuit condition. Therefore, in [14], it was this paper is to eliminate the magnetic core residual flux during
not required to model the power transformer equivalent resis- this time interval. In this section, it will be shown that how this
tance in the flux linkage estimation stage. While, in this paper demagnetizing process can help to undersize the CT core. In
(under the application of current transformer), the CT secondary this paper, the transient dimensioning factor that directly
terminal is short-circuited. Therefore, as will be shown later, plays a major role in proper sizing of the CT is employed [8].
since under development of this demagnetization process, the This factor is the ratio of the CT secondary winding flux linkage
equivalent circuit resistance is ignored, this will result in inac- related to through fault current, to the flux linkage developed by
curate demagnetizing performance. the ac component of this fault current [16].
This paper uses a low-voltage dc source, four power-elec- Fig. 1 illustrates the CT primary current during an unsuc-
tronic switches, and an intelligent simple switching controller cessful single shot autoreclosing scheme. As can be seen in
to eliminate the CT residual flux, as it was employed in [14]. Fig. 1(a), at first step, the fault is cleared in a duration of ,
In this proposed demagnetizing strategy, the following addi- followed by a duration in which current is ceased and corre-
tional improvements are implemented: 1) winding resistance is sponds to the deadtime of . Then, the CBs are reclosed and
considered in the CT secondary equivalent circuit and 2) the current passes for the duration of . To avoid CT saturation,
method is modified to be suitable under the autoreclose scheme the following transient dimensioning factor should be con-
application. strained [8]
In the proposed method, the beginning of the deadtime in-
terval is first detected by a transient monitoring index and a zero-
current monitoring filter. Afterwards, regardless of the value of
the CT residual flux, the CT core is derived into saturation re-
gions. At the same time, the CT equivalent circuit resistance (1)
is easily measured through measurement of its instantaneous
voltage and current. Then, CT saturation flux is estimated by where is the power system nominal frequency (i.e., 50 Hz).
saturating the CT in the opposite polarity. Finally, the magnetic and indicate primary system and CT secondary time con-
flux is gradually driven toward zero. stants, respectively.
In addition, this paper introduces a simple and accurate for- Fig. 1(b) shows the plot of the transient dimensioning factor
mula to select the optimum dc voltage source magnitude. It is versus time for 3 s and 60 ms. The fault duration
found that even for large CTs, a low-voltage source can effi- for the first cycle is 120 ms, followed by a deadtime of 200 ms
ciently demagnetize the affected CT during the reclosing dead- and the subsequent fault duration of 400 ms. During the autore-
time. closing deadtime , the CT primary terminal is open-circuited
Through a wide range of simulations carried out on an ac- and the CT secondary circuit time constant is large compared
curate CT model, the efficiency and reliability of the proposed to the deadtime . Therefore, the trapped residual flux in the
method are demonstrated. Furthermore, the performance of the CT core due to the first fault is slowly discharged. Based on
proposed method is verified for three different protective CTs, the definition, it is proportional to the CT magnetic flux
through the experimental results obtained in the laboratory. and, therefore, during the deadtime of , this factor is gradu-
ally moved toward zero with a time constant of , as shown in
II. EFFECT OF THE PROPOSED METHOD ON CT SIZING Fig. 1(b). The second fault can occur while the factor has
According to IEC 60044-6 [16], the deadtime of a reclosing a large nonzero value. Referring to this figure, without applica-
relay is defined as the time interval between interruption and tion of the proposed compensation method, the maximum value
reapplication of the primary short-circuit current during a circuit of the factor reaches 35.
1638 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 31, NO. 4, AUGUST 2016

Fig. 2. CT size reduction factor in percentage when 200 ms and Fig. 3. Proposed demagnetizing device connected to the CT secondary
(a) 3 s and 60 ms. (b) 120 ms and 400 ms. terminals.

While the CT residual flux can be eliminated over the dead-

time interval, it would become similar to the occurrence of a
continuous single fault, and can be defined in terms of the
following relation [8]:

(2)

Fig. 1(b) depicts the corresponding value of factor under

application of the proposed demagnetizing process. During the
deadtime, CT magnetic flux has been mitigated and, therefore,
the factor is settled down to zero. Here, the maximum
value that the factor can reach is 20. Therefore, by em-
ploying the proposed method, the required CT size, which is
defined as the minimum required secondary terminal voltage of
the CT [5], is reduced to about 43% compared with the size of
the uncompensated CT through the following ratio:

(3)

where (in percent) represents the CT size reduction factor.

Fig. 2 shows the contours of CT size reduction factor in
terms of the following parameters: , , , and . In Fig. 2(a), Fig. 4. (Schematic) time variation of the applied dc voltage, CT secondary cur-
the and parameters remained constant, while and rent, and core magnetic flux during the demagnetizing process.
parameters vary over the related reasonable ranges; and vice-
versa, in Fig. 2(b). and remained constant, while and Fig. 3 depicts the proposed demagnetizing device connected
parameters vary. As shown, the proposed method approx- in series with the CT burden and the protective relay. To min-
imately results in a 40% reduction in the CT size and, this is imize the adverse effect of the powerplant noise on the proper
almost independent of the value of the four aforementioned pa- performance of the proposed demagnetizer, it is recommended
rameters. This CT size reduction will be even more prominent to install this device adjacent to the relay in the substation relay
if multishot autoreclosing schemes are investigated. room. As can be seen in Fig. 3, the proposed device consists of
a low-power dc voltage source, four power-electronic switches
III. DEMAGNETIZING ALGORITHM (including freewheeling diodes), a current-limiting resistance
The proposed compensation method requires application , an instant-on switching solid-state relay (IO-SSR), and a
of an electronic device which can demagnetize the CT core microprocessor which manages the switching sequence.
in less than the duration of the reclosing process deadtime During the deadtime, the CT primary winding is open-cir-
. Although various phenomena affect the setting of the cuited and the microprocessor switches off the IO-SSR. Thus,
autoreclosing deadtime, the minimum value implemented the CT core flux can be varied by controlling the applied voltage
in the fast three-poles autoreclosing scheme is 200 ms [1], . Fig. 4 illustrates the variation of applied voltage to-
[17]. Considering this practicing deadtime value, the common gether with the CT secondary current , and magnetic flux
time-consuming demagnetization methods [10], [11] are not versus time when the proposed demagnetization technique is
suitable to be used here. In response to this shortcoming, this employed. The proposed technique consists of three consecu-
section introduces a modified version of the flux-controlled tive steps as follows:
demagnetizing technique previously presented in [14] to elimi- Step 1) Deadtime Start Detection; detects the beginning of
nate the CT residual flux during the reclosing deadtime. the deadtime interval.
HAJIPOUR et al.: RESIDUAL FLUX MITIGATION OF PROTECTIVE CTS USED IN AN AUTORECLOSING SCHEME 1639

The aforementioned observation helps to find the main fea-

ture to detect the beginning of the deadtime. After the tran-
sient inception detection, if the CT secondary current diminishes
to zero over more than 1.5 periods of the power system nom-
inal frequency, it is distinguished that the deadtime has begun
(point in Fig. 4). At this moment, the microprocessor issues a
control command to operate the IO-SSR. Subsequently, the de-
magnetizing device, which has been already been bypassed by
IO-SSR, is inserted into the CT secondary circuit. The typical
response time of an IO-SSR is less than 1 ms [20].
In the proposed demagnetizing technique, under all operating
conditions, the IO-SSR switch will bypass the demagnetizer
before the end of the autoreclosing deadtime. This precaution
should be considered to prevent occurrence of any possible open
circuit in the CT secondary terminal at the beginning of the
second shot of the fault current. Therefore, after detection of
the deadtime start, a countdown timer with a value less than
Fig. 5. (a) Typical scaled primary (dash line) and secondary current (solid line) will be started. Once this timer is triggered, any further action
of a saturated CT due to fault current and (b) CT secondary current during a
typical inrush current transient. of the demagnetizer would be blocked and the close command
will be issued to the IO-SSR switch.

Step 2) Required CT Parameters Determination; cancels the B. Required CT Parameters’ Determination

effect of the residual flux , and measures the cir- In here, first it will be demonstrated how the residual flux
cuit resistance and the CT saturation flux . in the CT core is suppressed, and following that, the methods
Step 3) Demagnetizing by Supply Switching; gradually re- used for detection of CT core saturation and determination of
duces the magnetic flux by controlling the polarity required CT parameters will be presented.
reversal of the applied dc voltage. 1) Residual Flux Suppression: Although the CT core can
The aforementioned steps would be explained in more detail be demagnetized independent of the CT initial flux , as pre-
in the following subsections. sented in [14], prediction of the residual flux sign can help
to reduce the demagnetization process time. To do this, immedi-
A. Deadtime Start Detection
ately after the detection of the short-circuit fault inception (point
Under normal operating condition of the power system, the in Fig. 4), the measured current samples are summed up over
proposed demagnetizing device is bypassed by an IO-SSR. This the duration which starts at the instant of and ends as the
normally closed (NC) contactor will prevent the occurrence of deadtime commences (point in Fig. 4) as shown in (4):
any undesired open-circuit in the CT secondary circuit. The first
fault inception is detected using the Transient Monitor index,
(4)
as presented in [18]. This index shows the quality of a least
error squares (LES) filter to fit the sampled secondary current
of the CT. When the LES filter window contains prefault sam- where refers to the summation of the measured samples.
ples and postfault samples, the Transient Monitor index be- In order to estimate the residual flux sign, this paper pro-
comes larger than a predefined threshold [18] and, therefore, the poses employing the sign of . Assume that the sign of
first fault inception instant is detected (point in Fig. 4). Some is positive (negative); therefore, the residual flux is most likely
other power system transients, such as inrush current, can also positive (negative). To suppress the effect of , a positive (neg-
raise this index to exceed the threshold. However, the following ative) dc voltage will be applied to the CT sec-
zero-monitoring filter helps to prevent the misoperation of the ondary terminals.
deadtime start detection unit. In order to show the effectiveness of the proposed algorithm,
After detection of a transient inception, if the CT secondary a typical 300:1 5P20 8VA CT connected to its nominal burden
current is diminished and became zero over several consecutive has been simulated. Fig. 6 demonstrates the variation of the
samples, one of the following may have occurred: and the CT core flux during the occurrence of a typical fault
1) CT is saturated under fault current conditions. In this case, current. In this figure, the CT pre-fault residual flux is as-
the zero-current interval certainly lasts less than a power sumed to be and the magnitude of the fault current
frequency cycle (typically less than 84% of a cycle [19]), is ; where is the CT nominal current. As can be
as shown in Fig. 5(a). seen, the polarity of the properly reflects the sign of the re-
2) The zero-current portion of the inrush current flows in sulting residual flux. In addition, Table I presents the obtained
the CT primary winding, which certainly lasts less than a results for several trials with different fault current magnitudes
power frequency cycle, as shown in Fig. 5(b). and the CT residual fluxes . In these simulations, a
3) The fault has been interrupted and the deadtime for re- fault with a maximum magnitude of decaying dc component,
closing has begun. having a time constant of 60 ms, has been applied to the
1640 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 31, NO. 4, AUGUST 2016

denotes the nominal secondary current of the CT (1A or 5A).

In Fig. 4, the time delay of the proposed steady-state detector is
represented by the time interval between and .
To implement the method experimentally, MWV can be cal-
culated in a recursive fashion. This will considerably reduce the
computational cost of (5) [22].
In the following text, the CT parameters of overall CT cir-
cuit resistance , and CT saturation flux are estimated or
determined.
3) Measurement of the Resistance: After detection of the
core saturation at point in Fig. 4, overall resistance of the
circuit is estimated by averaging the measured current over
consecutive samples as follows:

(6)

Fig. 6. Typical fault current and the corresponding CT flux and variation. where is the secondary circuit total resistance (including CT
winding resistance , CT burden resistance and demag-
TABLE I netizer limiting-resistance ). determines the average value
PERFORMANCE OF THE RESIDUAL FLUX SIGN DETERMINATION UNIT of the measured current between the points of and in Fig. 4.
In this paper, is equal to one quarter of a power system cycle
(20 samples for a sampling frequency of 4 kHz).
4) Saturation Flux Estimation: After measurement of
the resistance , the applied voltage is reversed. As can
be seen in Fig. 3, a relation could be derived for the voltage
across the magnetic core as follows:

(7)

where takes the value 1 or 1 based on the sign of the applied

voltage .
CT. Here, the dc component of the fault current is assumed to be When the core becomes fully saturated in the reverse direc-
positive, and the sign of the obtained results would be reversed tion (point in Fig. 4), the magnetic core voltage will
if the dc component polarity was negative. Table I shows that diminish to zero. Therefore, saturation can be detected when
the sign of the parameter most likely represents the sign of falls below a small threshold close to zero.
, unless the CT core maintains a large amount of residual flux The value of CT core flux at the points of and in Fig. 4
with opposite polarity right before fault inception. However, in are with a different sign; where indicates CT satu-
practice, the incident of this case is less probable. ration flux. According to the Faraday's law, this paper proposes
As noted earlier, the sign determination of the residual flux the following relation to determine the CT saturation flux:
significantly helps to reduce the time interval between and
in Fig. 4. The applied voltage is maintained until the CT core
saturates (point in Fig. 4).
2) Core Saturation Detection: By applying the dc voltage,
initially the current remains close to zero, then begins to increase (8)
rapidly and, finally, settles to a constant value, when the CT core
becomes fully saturated. In order to identify the core saturation Reference [14] has proposed a similar equation for demag-
inception, and maintain it immune to the measurement noise, a netization of a distribution power transformer core when the
steady-state detector may be employed [21], [22]. In this paper, transformer secondary side is open-circuited. Therefore, it has
the moving window variance (MWV) is used as follows [22]: ignored the effect of resistance in (8) while, in current trans-
former applications, this ignorance could lead to inaccurate re-
sults. Therefore, this paper introduced a resistance measurement
unit to overcome this shortcoming.
(5)
where denotes the last th measured sample and is the C. Demagnetizing by Supply Switching
length of the moving window. At the instant in Fig. 4, the CT magnetic flux is set to the
If the MWV value falls below a predefined threshold (0.05 in estimated saturation flux . This is considered as the lower
this paper) when the current average magnitude is more than band of the integration term in (9). Afterward, the flux can be
, then the CT core is identified to have reached saturation. evaluated by applying a dc voltage of desired polarity (i.e., by
HAJIPOUR et al.: RESIDUAL FLUX MITIGATION OF PROTECTIVE CTS USED IN AN AUTORECLOSING SCHEME 1641

TABLE II
OPTIMAL PARAMETERS OF THE PROPOSED DEMAGNETIZER

Now, assume that depicts the CT-rated secondary ter-

Fig. 7. Effect of different segments of the proposed method in the total demag- minal voltage, which generates the maximum magnetic flux
netizing process time.
in the CT core. According to IEEE C57.13-2008 [9], this mag-
netic flux results in 10% CT ratio error. It is conservatively
assumed that CT saturation flux is 1.2 times of . The re-
setting the parameter to 1 or 1) while measuring the CT
lation between and is given by
secondary current through the following:
(13)
(9)
By combining (12) and (13), the following equation can be
obtained:
where takes the value or based on the sign
of the pre-applied voltage . (14)
In this paper, the magnetic flux is gradually reduced through
a set of consecutive hysteresis cycles as shown in Fig. 4. The The value of the limiting resistance is selected in such a
applied voltage is reversed while the value of the magnetic way that: 1) the maximum current value could be less than the
flux exceeds the defined threshold as given by CT nominal current (1A or 5A); 2) the maximum current
value could be high enough to make sure that the current sensor
(10) can monitor the flowing current accurately; 3) the power deliv-
ered by the dc voltage source and the power dissipated in the
where indicates the threshold for reversing the applied limiting resistor will be limited to the desired value of :
voltage. is the desired total number of voltage reversals. As
can be seen in (10), the magnitude of the threshold is exponen-
tially decayed by a factor of 0.66.
(15)
IV. OPTIMAL DESIGN OF THE DEMAGNETIZING DEVICE
where is a reasonable current value that the current sensor
In this section, the optimal values of and are deter- can precisely measure, while it remains immune to measure-
mined. The value of is selected such that the total time ment noise. is the power delivered by the dc power supply.
required to demagnetize CT becomes less than the au- Table II reports the calculated and preferred values of ,
toreclosing deadtime . As shown in Fig. 7, 1.5 cycles of , and for the standard values of secondary terminal volt-
the power system nominal frequency (in this paper equal to ages stated in [9]. In this table, the deadtime and are
50 Hz) is spent to detect the deadtime beginning. At the worst assumed to be equal to 200 ms (the minimum practical dead-
case, the sign of the residual flux may be predicted incorrectly; time value) and 200 mA, respectively. The preferred values of
in this condition, the integral of the voltage across the CT core are chosen in a form to provide an adequate safety margin
between and would be equal to . Furthermore, for the uncertainties in the demagnetizing process. As can be
as is presented in (8), the integral of between and is seen in Table II, even for 800 V and 200 ms, the
equal to . The saturation detection time delay and the re- required dc voltage source is small; which considerably reduces
quired time for resistance measurement are assumed to be 0.5 the weight and size of the demagnetizing device.
cycles. The integral of during the demagnetizing switching The employment of a low dc voltage source, and the lim-
process is equal to as shown in the following equation: ited level of maximum flowing current and delivered power, as
well as the low computational cost of the proposed technique
(11) ensure that the proposed demagnetizing device is cost-efficient
and easy to produce.

Using Faraday's law, the overall time spent to demagnetize V. SIMULATION RESULTS
CT can be determined using (12); which should be less In this section, simulations are presented to show the effec-
than the autoreclosing deadtime tiveness of the proposed compensation method. An appropriate
CT model including accurate representation of the hysteresis
(12) characteristic is designed using the Jiles-Atherton theory [23].
1642 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 31, NO. 4, AUGUST 2016

TABLE III
CT PARAMETERS

Fig. 9. Applied voltage and measured CT secondary current during the demag-
netizing process.

Fig. 10. Hysteresis cycles during the demagnetizing process.

Fig. 8. (a) Typical unsuccessful autoreclosing under a short-circuit fault.
(b) Corresponding CT magnetic core flux via the proposed method versus the
uncompensated CT.

Table III provides the simulated CT parameters [7]. Considering

(14) and 200 ms, the calculated dc voltage source is equal
to 55.6 V, which is conservatively assumed to be 60 V in this
simulation. The limiting resistance is set to 60 , as pre-
sented in (15). The total number of voltage reversals is as- Fig. 11. Magnetic flux during the demagnetizing process that employs the pro-
sumed to be 10. posed method and [14].
The measured current is preconditioned by an analog second-
order Butterworth low-pass filter with a cutoff frequency of 500
Hz. The sampling frequency is assumed 4 kHz or 80 samples/ followed by the magnetic flux density during the demagne-
cycle (in a 50-Hz system), which is one of the rated values based tization process. The flux density has been gradually reduced
on IEC 60044-8 [24]. through the path of consecutive hysteresis cycles and finally has
A 0.9-kA permanent fault with a maximum magnitude of de- been settled to zero.
caying dc component having a time constant of 60 ms has Fig. 11 shows the magnetic flux variation during the dead-
been applied to CT, as shown in Fig. 8. The fault duration for time, employing the proposed method and [14]. As noted earlier,
the first cycle is 120 ms, followed by a deadtime of 200 ms and [14] assumes that the transformer secondary side is open cir-
again a fault duration of 400 ms. The CT residual flux before cuited during the demagnetization process. Therefore, in [14],
the first fault inception is assumed to be zero. Without the ap- there was no need to consider the effect of CT winding re-
plication of the proposed compensation method, CT would sat- sistance. However, under the application of the proposed CT
urate under the second fault, as can be seen in Fig. 8(a). How- residual flux mitigation of this paper, the CT secondary ter-
ever, the proposed demagnetizing device prevents CT satura- minal is short-circuited. Therefore, the CT winding resistance
tion. Fig. 8(b) illustrates the corresponding CT magnetic fluxes cannot be overlooked. By ignoring the CT resistance effect, the
with and without the application of the demagnetizer. The pro- CT magnetic flux at the end of the deadtime reaches 0.138 Wb
posed method has effectively reduced the residual flux during (14.6% ); while the proposed method provides a satisfac-
the deadtime from 0.884 Wb at the beginning of the deadtime tory result equal to 0.011 Wb (1.2% ).
to 0.011 Wb (1.2% ) at the end of the deadtime. Fig. 9
depicts the applied voltage and measured current during the VI. EXPERIMENTAL RESULTS
deadtime. The demagnetizing process lasts for 162 ms, where a In order to verify the performance of the proposed method,
38-ms safety margin is provided. Fig. 10 demonstrates the path this method has been tested experimentally on three different
HAJIPOUR et al.: RESIDUAL FLUX MITIGATION OF PROTECTIVE CTS USED IN AN AUTORECLOSING SCHEME 1643

TABLE IV
TESTED CTS PARAMETERS

Not specified in the CT nameplate.

Fig. 13. Applied voltage , the measured CT secondary current, the induced
voltage across the CT core , and the CT core flux during the demagnetizing
Fig. 12. Experimental setup: 1) digital oscilloscope card, 2) under test CT, process.
3) resistance burden, 4) demagnetizing device, and 5) monitored signals.

TABLE V
PROPOSED DEMAGNETIZING METHOD PERFORMANCE

In percentage of the corresponding CT saturation flux .

protective CTs. The proposed demagnetizing algorithm is im-

plemented to a low-cost microchip dsPIC33 (digital signal con-
troller) [25]. The CT secondary current is sampled using a high-
precision Hall-effect sensor. Table IV provides the three tested
CTs data ( , , and ) and the corresponding
demagnetizer parameters. In the implemented tests, the required
CT residual flux is initialized using an arbitrary voltage pulse
with desired duration which is applied to the CT secondary just
before the beginning of the demagnetization process. Fig. 12 de-
picts the experimental setup employed in this part of the work.
Fig. 13 displays the applied voltage waveform, the recorded
current waveform, and the waveform of the induced voltage
across the CT core , and the calculated magnetic flux
for . In this figure, the residual flux is 21 mWb (62%
) and the final value of the magnetic flux is equal to 0.11
mWb (0.6% ). Fig. 14(a) depicts the path followed by the
magnetic flux during this demagnetization process. In addition,
Fig. 14(a) illustrates the corresponding result when is set to
zero. The total time required to demagnetize is less than Fig. 14. Measured hysteresis cycles during the demagnetizing process: (a)
108 ms and 123 ms for the equal to 62% and 0% , , (b) , and (c) .
respectively. Fig. 14(b) and (c) illustrates the hysteresis cycles
during the demagnetization of and for two
different arbitrary residual fluxes, respectively. The parameters these tests are summarized in Table V. As can be seen in Fig. 14
demonstrating the performance of the proposed method under and Table V, the proposed method could effectively eliminate
1644 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 31, NO. 4, AUGUST 2016

the residual flux during the deadtime interval independent of [11] H. Kristensen and V. Mrdic, “Comparative analysis of three phase and
the CT residual flux . The total time of the demagnetization single phase dynamic resistance measurement results,” presented at the
CIRED 22nd Int. Conf. Elect. Distrib., Stockholm, Sweden, Jun. 2013.
process is less than 130 ms for all of the studies carried out, and [12] N. Makowski, “Proposal and analysis of demagnetization methods of
the final value of CT flux after implementation of the proposed high voltage power system transformers and design of an instrument
to automate the demagnetization process,” M.Sc. dissertation, Portland
method is less than 4%. State University, Portland, OR, USA, 2011.
[13] M. Putter, M. Radler, and B. Unterer, Reliable demagnetization
VII. SUMMARY AND CONCLUSIONS of transformer cores, Omicron electronics GmbH, 2014. [Online].
Available: https://www.omicron.at
An electronic device has been introduced to reduce the CT [14] B. Kovan, F. de Leon, D. Czarkowski, Z. Zabar, and L. Birenbaum,
sizing requirement where it is employed in a system equipped “Mitigation of inrush currents in network transformers by reducing the
with an autoreclose scheme. An existing demagnetizing tech- residual flux with an ultra-low-frequency power source,” IEEE Trans.
Power Del., vol. 26, no. 3, pp. 1563–1570, Jul. 2011.
nique has been modified to accurately eliminate the residual [15] F. de Leon, A. Farazmand, S. Jazebi, D. Deswal, and R. Revi, “Elimina-
magnetic flux during the deadtime duration of a reclosing tion of residual flux in transformers by the application of an alternating
polarity dc voltage source,” IEEE Trans. Power Del., vol. 30, no. 4, pp.
scheme. Some promising features of the proposed compensa- 1727–1734, Aug. 2015.
tion method could be summarized as follows. [16] Instrument Transformers-Part 6: Requirements for Protective Current
• The ability to compensate both new CTs to be installed and Transformers for Transient Performance, IEC Standard 60 044-6,
1992-3.
in-service CTs. [17] W. Elmore, Protective Relaying: Theory and Applications, 2nd ed.
• Having no dependency on the power system and the CT New York, USA: Marcel Dekker, 2004.
parameters. [18] A. Phadke and J. Thorp, Computer Relaying for Power Systems, 2nd
ed. Chichester, U.K.: Wiley, 2009.
• Fast and accurate demagnetizing unit is developed to im- [19] J. Pan, K. Vu, and Y. Hu, “An efficient compensation algorithm for
prove CT performance under reclosing schemes, realizing current transformer saturation effects,” IEEE Trans. Power Del., vol.
19, no. 4, pp. 1623–1628, Oct. 2004.
its burden resistance effect. [20] Carlo Gavazzi Co., Solid State Relays General Information. [Online].
• The unit preserves a simple low-voltage/low-current hard- Available: www.carlogavazzisales.com
ware structure, having a set of optimal design parameters. [21] T. Jiang, B. Chen, X. He, and P. Stuart, “Application of steady-state
detection method based on wavelet transform,” Comput. Chem. Eng.,
The proposed device is capable of reducing the required size vol. 27, pp. 569–578, 2003.
of the CT about 40% in an economical and reliable way. [22] M. Kim, S. H. Yoon, P. A. Domanski, and W. V. Payne, “Design of a
steady-state detector for fault detection and diagnosis of a residential
air conditioner,” Int. J. Refrig., vol. 31, pp. 790–799, 2008.
REFERENCES [23] U. D. Annakkage et al., “A current transformer model based on the
[1] J. Blackburn and T. Domin, Protective Relaying: Principles and Ap- Jiles-Atherton theory of ferromagnetic hysteresis,” IEEE Trans. Power
plications, 3rd ed. New York, USA: Taylor & Francis, 2007. Del., vol. 15, no. 1, pp. 57–61, Jan. 2000.
[2] W. Rebizant, J. Szafran, and A. Wiszniewski, Digital Signal Pro- [24] Instrument Transformers- Part 8: Electronic Current Transformers,
cessing in Power System Protection and Control. Berlin, Germany: IEC Standard 60 044-8, 2002.
Springer-Verlag, 2011. [25] Microchip Technology Inc., dsPIC Digital Signal Controllers. Chan-
[3] P. Meinhardt and B. Bastigkeit, “Introduction of an easy method to ana- dler, AZ, USA. [Online]. Available: http://ww1.microchip.com
lyze the influence of CT saturation on the protection system,” presented
at the CIRED 21st Int. Conf. Elect. Distrib., Frankfurt, Germany, Jun.
2011.
[4] K. Erenturk, “ANFIS-based compensation algorithm for current-trans- Ehsan Hajipour (M’14), photograph and biography not available at the time
former saturation effects,” IEEE Trans. Power Del., vol. 24, no. 1, pp. of publication.
195–201, Jan. 2009.
[5] IEEE Guide for the Application of Current Transformers Used for Pro-
tective Relaying Purposes, IEEE Standard C37.110, 1996.
[6] W. Knapek, Residual magnetism Omicron electronics GmbH., 2012.
[Online]. Available: https://www.eiseverywhere.com Mohammed Salehizadeh, photograph and biography not available at the time
[7] F. Badrkhani, M. Sanaye-Pasand, M. Davarpanah, A. Rezaei-Zare, and of publication.
R. Iravani, “Compensation of the current-transformer saturation ef-
fects for digital relays,” IEEE Trans. Power Del., vol. 26, no. 4, pp.
2531–2540, Oct. 2011.
[8] P. K. Gangadharan, T. S. Sidhu, and G. J. Finlayson, “Current trans- Mehdi Vakilian (SM’15), photograph and biography not available at the time
former dimensioning for numerical protection relays,” IEEE Trans.
of publication.
Power Del., vol. 22, no. 1, pp. 108–115, Jan. 2007.
[9] IEEE Standard Requirements for Instrument Transformers, IEEE Stan-
dard C57.13, 2008.
[10] IEEE Guide for Diagnostic Field Testing of Electric Power Appa-
ratus—Part I: Oil Filled Power Transformers, Regulators, Reactors, Majid Sanaye-Pasand (SM’12), photograph and biography not available at the
IEEE Standard C57.152, 2013. time of publication.