1

Elimination of Residual Flux in Transformers

by the Application of an Alternating Polarity
DC Voltage Source
Francisco de León, Senior Member, IEEE, Ashkan Farazmand, Saeed Jazebi, Member, IEEE
Digvijay Deswal, and Raka Levi, Senior member, IEEE

 tions [10]. The low frequency region of the FRA is very much
Abstract— The paper presents the experimental verification of affected by the magnetizing status of the core [11]. The last
a power electronics device for demagnetizing transformers. La- test in factory, performed after FRA, is the dc winding re-
boratory and field tests on single- and three-phase transformers sistance test, which may leave the transformer core magnet-
demonstrate that the residual flux is eliminated. The purpose of
ized [12]. If the core is not properly demagnetized, the foot-
demagnetizing is twofold: (1) to reduce the inrush currents when
transformers are energized; and (2) to make sure that the FRA print of the FRA performed at the installation site is different
(Frequency Response Analysis) tests are consistent to avoid false than the one performed in the factory. This could be mistaken
diagnoses of damage during transportation. A nearly complete as a possible winding movement during transportation. Exam-
demagnetization was observed in the laboratory for a small sin- ples can be found in the CIGRE working group report [10].
gle-phase isolation transformer. The results of demagnetizing an Therefore, demagnetization of large power transformers in the
80 MVA transformer in the field are also presented. The method
field has become routine after performing the dc resistance
proposed in this paper is applied to three-phase transformers
with different core configurations and connections. Topologically tests on the windings.
correct modeling and numerical simulations confirm the full de- A method to reduce residual flux in the core was presented
magnetization of all the branches of three-phase (three- and five- in [13] using an ultra low frequency power source. In [13] the
limb) transformer cores. Inrush current measurements and FRA effectiveness of the method was illustrated with numerical
plots before and after demagnetization confirm the effectiveness simulations using the EMTP. Following a short-circuit, a set
of the process.
of transformers in a feeder were demagnetized before re-
energization. Substantial reduction in the magnitude of the
Index Terms—Demagnetization, FRA, inrush currents, resid-
ual flux, transformers. inrush currents was obtained.
This paper presents the demagnetization results obtained
I. INTRODUCTION experimentally with a laboratory prototype of the power
source proposed in [13]. Laboratory tests are conducted on an
I NRUSH currents often emerge when a transformer is ener-
gized. Depending on the transformer parameters, this transi-
ent phenomenon may last for a few cycles or seconds [1]-[4].
isolation transformer of 1 kVA, 120 V. It is demonstrated that
one can successfully monitor the demagnetization process of
the transformer by plotting the λ-I curve. Results are verified
The magnitude of the inrush currents depends on the phase
by measuring the inrush currents drawn when the transformer
angle of the applied voltage, winding resistance, the air-core
is energized at voltage zero-crossing through a controlled
inductance of the energized winding, and the residual flux in
switch. In addition, it is shown by topological modeling in the
the core. The worst scenario for inrush currents happens when
EMTP that the technique works well for three-phase trans-
the core has residual flux and the switching occurs at voltage
formers with various core constructions and winding connec-
zero-crossing in the direction that builds flux in the core. Un-
tions.
der this condition the transformer core goes into super satura-
The paper also presents experiences accumulated with
tion [5]-[9]. Hence, to avoid large inrush currents transformers
commercially available demagnetizers tested and used on
can be demagnetized before energization.
transformers of up to 1100 MVA rated power and 1000 kV
As a diagnostic tool, FRA (Frequency Response Analysis)
nominal voltage [14].
tests are performed on large power transformers in order to
detect early defects such as core movement or deformation.
II. AVAILABLE DEMAGNETIZING TECHNIQUES AND THEIR
Different ranges of the frequency spectrum correspond to de- PERFORMANCE ON LARGE TRANSFORMERS IN THE FIELD
fects associated with the core, windings, bushings, or connec-
The demagnetization of transformer cores can be per-
F. de León, A. Farazmand, and S. Jazebi are with the NYU Polytechnic formed in several ways as discussed in [13]: (1) Variable
School of Engineering, Six Metrotech Center, Brooklyn, NY, 11201, USA (e- Voltage Constant Frequency (VVCF) source; (2) Constant
mails: [email protected], [email protected], [email protected]). Voltage Variable Frequency (CVVF) source; (3) Decreasing
D. Deswal is with Indian Institute of Technology (IIT) Kharagpur, Kha-
ragpur, West Bengal - 721302, India (e-mail: deswal.digvijay@ gmail.com). the amplitude of an alternating dc current, which is a variation
Raka Levi is with DV Power of Lidingo, Sweden (e-mail: Raka@dv- of the CVVF method. The latter is a modified version of the
power.com).
 2

method suggested in the IEEE Standard C57.152-2013 [15]. method presented in this paper provides identical results with-
This method uses decreasing time as a measure of current. in a few seconds.
Commercially available demagnetizers utilize a constant dc
voltage source of 60 V with fast switching. The process
changes the amplitude of the charging current successively in
steps. A constant voltage is applied to the transformer terminal
and at the moment the current reaches a preset value (for ex-
ample 20 A) the supply is disconnected and a discharging cir-
cuit is connected. Once the current reaches zero, switches
change the polarity of the constant voltage source applied to
the terminal. This time, the source is disconnected when 60%
of the previous current value is reached and consequently the
discharging circuit is connected. This process is repeated until
the limit of current reaches 5 mA. The method is capable to
demagnetize the largest transformers, up to 1100 MVA. The
process may last from about 2 minutes and up to 15 minutes Fig.1. Demagnetization current graph for a three-phase transformer in the
per phase depending on the size and voltage rating of the unit field.
[16].
In this section the experiences gained with the demagnetiza-
tion of large power transformers in the field are presented. The
process is illustrated with a Pauwels 80 MVA transformer,
demagnetized and FRA tested on the LV side (see Figs. 1 and
2). The demagnetization process for this transformer starts by
applying 60 Vdc to the winding, until 20 A current is reached.
The polarity is reversed alternatively until 5 mA is reached.
This is done in three stages. Each of the three events is a sin-
gle phase demagnetization process between two phases: AB,
BC, CA. Fig. 1 shows 3 sets of two phase currents; In the first
stage the current is applied to phase A, and returns through
phase B (that is the reason they are mirror images), in the se-
cond step the current of phase B returns through phase C, and
in the last step the current of phase C returns through phase A (a)
of this delta connected winding.
The demagnetized state of the core is demonstrated by
comparing the two FRA graphs of Fig. 2. The high frequency
portion has been cut since only the low frequency range, up to
around 1 kHz, is dominated by the characteristics of the mag-
netic core. Fig. 2(a) shows the results after the dc winding
resistance test has been performed. One can observe that there
is a difference of peak amplitudes between the responses of
the two outer phases (X1-X2) and (X3-X1), which should be
the same for a healthy transformer. Magnetized traces may
mislead the operator into believing that the core or coils are
deformed or damaged during transport.
After the application of the demagnetizing technique, the
normal traces of the outer phases (which should have two (b)
peaks in this range that coincide), are seen in Fig. 2(b). One Fig. 2. FRA for a 80 MVA transformer. (a) Transformer magnetized after dc
resistance test. Note that the responses of the two external phases (X1-X2) and
can confirm that the plots perfectly match with those obtained (X3-X1) are not identical; (b) Transformer demagnetized. Note that the re-
before the dc resistance test; demonstrating the successful de- sponses of the two external phases (X1-X2) and (X3-X1) are now identical.
magnetization of the core.
The three single-phase demagnetization processes described III. DEMAGNETIZING DEVICE
above has given good results in the field. The process is sim-
ple, but there is no solid proof that the demagnetization is For the physical construction of the demagnetizer in the la-
completed. To check the demagnetization status of the core, it boratory, the more suitable technique is the constant voltage
is necessary to perform the FRA test a second time. The set up variable frequency (CVVF); see Fig. 3. This is so because
is prolonged for a large transformer. constant voltage sources are readily available. The theory be-
In conclusion, the demagnetization process with available hind this algorithm has been described in [13]. The demagnet-
izing device consists of a dc voltage source, a power electron-
commercial devices and techniques necessary to check the
ics switching circuit, and a controller. The functional block
successfulness of the method are time consuming. The new
diagram is shown in Fig. 3(a). The basic components of each
 3

block are: (1) The switching circuit, which consists of four (1) Initially all the switches (S1, S2, S3, and S4) are off. At a
MOSFETs connected in the configuration shown in Fig. 4. A given instant two opposite end switches (S3 and S4) are
freewheeling diode across each MOSFET is needed to avoid tuned on, such that negative voltage (–VDC) is applied
instantaneous breaking of inductive currents; (2) The micro- across the transformer until negative saturation is reached.
controller and gate-drive circuits to switch the MOSFETs The negative saturation is detected by the shunt resistance
on/off and compute parameters such as saturation time. The connected in the path of switches S3 and S4. The sensed
micro-controller also determines saturation and demagnetiza- current is fed to the micro-controller for sampling and de-
tion points of transformers. This is used to determine the time termining the saturation point.
at which voltage reversal is executed; (3) Current sensor, (2) At the negative saturation point, switches S3 and S4 are
which consists of a shunt resistance to measure the current turned off and switches S1 and S2 are turned on. The po-
flowing in the transformer. The current waveform is used to larity of voltage VDC is reversed, so that positive voltage
determine the saturation of the core. A very small shunt re- (+VDC) is applied across the transformer winding. At this
sistance, in the range of milliohms, is used to measure the moment a timer is started. The positive saturation is de-
current as the resistance of the transformer winding itself is tected by the shunt resistance connected in the path of
very small; (4) Signal conditioning circuit, which consists of a switches S1 and S2.
circuit to amplify and filter the noise out from the current sig- (3) When the core reaches positive saturation, switches S1
nal measured across the shunt. It also serves to condition the and S2 are turned off and switches S3 and S4 are again
signal in the working range of the micro-controller; (5) Volt- turned on. So that once again negative voltage (–VDC) is
age sensor, which contains a buffer circuit to prevent loading applied across transformer windings, and time is reset to
effect on the transformer during the measurement of the in- zero.
duced voltage. (4) When the time reaches half of the recorded value in the
timer the voltage source is disconnected from the trans-
Signal Conditioning
Controller
Circuit former.
(5) The transformer is now fully demagnetized.
As discussed above, in this algorithm the current waveform
Current Sensor .
a
Voltage Sensor is observed to determine the saturation point of the transform-
VDC
er. The response is similar to that of an inductive circuit con-
.
b
Transfromer
nected to a dc supply. In Section VI a discussion on the volt-
age source and saturation detection is given.
Switching
Circuit (a)
IV. DEMAGNETIZING OF A SINGLE-PHASE TRANSFORMER
V
+VDC Demagnetization of a 1 kVA, 120 V toroidal transformer is
performed using the demagnetizer via the time for saturation
algorithm described above. The results of simulation using
t EMTP-RV [17], and experiments using the device are present-
-VDC (b) ed in this section.
The π model is selected to present the single phase trans-
Fig. 3. (a) Demagnetization circuit for a single-phase transformer; (b) CVVF
voltage.
former accurately [1]. The diagram of the circuit consists of
the transformer and the demagnetizer as shown in Fig. 5. The
source resistance is measured as 0.144  in the laboratory.
To obtain the parameters of the transformer, the standard
open circuit and short circuit tests are performed [1]. The val-
ues are given in Table I. The air-core inductance of the trans-
former is obtained using a non-ideal low-power rectifier
source in the laboratory as 314 μH [18]. This value is used to
extend the hysteresis loop from the final measured point (ob-
tained from the open circuit test) to infinity.
The applied voltage for the demagnetization process is pre-
sented in Fig. 6. The demagnetization results obtained from
simulation and experiment using the demagnetization device
are presented in Fig. 7. One can see a good agreement between
simulations and experiments. The differences are caused by
the hysteresis fitter of the EMTP that is not capable of accu-
rately reproducing the steepness of the cycle. Five points: A,
B, C, D, and E are marked in Figs. 6 and 7 to describe the de-
magnetization process of the transformer.
Fig. 4. Demagnetizing device circuit for single-phase transformers.
The practical implementation consists of the following
TABLE I
steps: CIRCUIT PARAMETERS FOR THE SINGLE PHASE TRANSFORMER
 4

4 0.6 D

Magnetizing Current (A)

Experiment
D 0.4

Magnetic Flux (Wb)

2 Simulation Simulation
0.2
A C E
0 0 C
A E
-0.2
-2
B
Experiment -0.4
B
-4
-0.6
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
Time (s) Time (s)
(a)
(b)
4 0.6
B Experiment D
Simulation 0.4

Magnetic Flux (Wb)

Induced Voltage (V)

2 C
0.2
A
E E C
0 0
Experiment
-0.2
Simulation
-2
A -0.4
B
D -0.6
-4 -1 -0.5 0 0.5 1
0 0.2 0.4 0.6 0.8 1 1.2
Time (s) Magnetizing Current (A)
(c) (d)
Fig. 7. Demagnetization process of a 1 kVA single-phase transformer. (a) Magnetizing current; (b) Magnetic flux; (c) Induced voltage in the open-circuited
winding; (d) Hysteresis cycle.

R1 [] R’2 [] Ls [mH] Rm1 = Rm2 []

0.277 0.300 0.232 2,832 that time, a negative voltage (–2.3 V) is applied. This voltage
is selected based on parameters derived from open circuit tests
Ls and the characteristics of the voltage source (see Section VI).
Rs+R1 R’2 At point B (t =378 ms) the transformer reaches negative satu-
ration at a value of –0.57 Wb and current of –2.4 A. Conse-
+ quently, the polarity of the voltage is reversed to +2.3 V until
Rm2
VDem Hys1 Hys2 the transformer goes into positive saturation. At point C (t=
- Rm1 622 ms) the flux changes from negative to positive and the
magnetizing current is 0.06 A. The voltage is still positive and
linkage flux continues to build up until it gets to positive satu-
Fig. 5. Model of a single phase transformer connected to the demagnetizer. ration (+0.57 Wb) at point D (t =0.885) with a magnetizing
current of 2.4 A.
3 B C D As it was explained in Section III, in order to demagnetize
2 the transformer core, now the negative voltage must be ap-
plied for half of the time that it takes for the transformer to go
Voltage (V)

1 from the negative saturation point to the positive saturation

A E
0 point. Therefore, by applying the negative voltage from t =885
-1
ms to t=1138 ms the transformer is totally demagnetized.
Therefore, at point D the polarity of the source is changed
-2 again to –2.3 V for half of the time that it took to go from
-3 point B to point D. As it is expected the magnetic flux gets to
0 0.2 0.4 0.6 0.8 1 1.2
zero at point E, and the transformer is demagnetized.
Time (s) When positive voltage is applied across the transformer’s
Fig. 6. Applied voltage for the single-phase transformer through the demag-
netization process by the dc source. winding the current rises. As the flux builds up the core starts
to saturate and the value of inductance L starts to decrease.
Note that in the case presented here, the core is demagnet- This eventually results in an abrupt rise of current at the knee
ized at the beginning of the process. However, results would point, which would be finally limited by the resistance of the
be the same for any initial flux since the core goes into nega- transformer winding, as observed in Fig. 7(a). The slope of the
tive saturation first and then the time that takes between nega- flux changes from positive to negative as the polarity of the
tive and positive saturation is measured. In this example, the applied voltage changes (see Fig. 7(b)). The induced voltage
energization is started at point A (t =115 ms) with zero flux (on the open-circuited winding) is shown in Fig. 7(c), and the
and current (located at the center of the hysteresis cycle). At
 5

hysteresis loop of the transformer during the demagnetization 2 (t =0.5 s). The positive voltage is now applied until the
process is presented in Fig. 7(d). transformer reaches the positive saturation at point 3 (t =0.63
The demagnetization results were verified by performing s). Finally, with the application of the negative dc voltage for
several inrush tests using a zero-crossing switch on a demag- (0.63-0.5)/2= 0.065 s, from t= 0.63 s to 0.695 s and the trans-
netized transformer. The values of inrush current were the former is fully demagnetized (at point 4).
same (254 A) for all tests. Fig. 8 presents the inrush current From Fig. 10(b), it can be seen that all five branches are ful-
waveforms of the transformer obtained from experiments and ly demagnetized after 0.69 s. Fig. 10(c) shows the hysteresis
simulation, which serve as additional validation of the trans- cycle of Hys 1 branch during the demagnetization process
former model. The peak value of the inrush current is 22 times (from t=0.4 s to t=0.7 s). Point 1 is the initial flux as 0.13 Wb
larger than the transformer rated current. This value could be and the magnetic branch is fully demagnetized at point 4. The
much larger if the core is not demagnetized at the time of en- demagnetization process for the other 4 branches results in a
similar hysteresis cycle. Fig. 10(d) shows the transformer pri-
ergization. An example of inrush current for a magnetized
mary current. The energization of the transformer at t=0 with
core is shown in Fig. 8. For this case an initial flux of 0.45 Wb
ac sources produce inrush currents (the peaks have been cut
is present. As shown, the peak of inrush current is 344 A,
from the figure since we are interested in the demagnetization
which is 35% larger than the inrush current for the demagnet- process). The transients are damped and the system reaches
ized core. steady state condition after 0.1 seconds. After the transformer
Magnetized core PI Model-Magnetized core is disconnected from the source at t=0.2 s, some flux remains
300 PI Model-Demagnetized core on different core sections (see Fig. 10(b)). As it is presented,
Experiment-Demagnetized core
the demagnetization process produces two peaks of current of
Current (A)

Demagnetized core
200 32.5 A when the transformer reaches its positive and negative
saturation.
100 The demagnetizing process is also applied to a 5 limb trans-
former. The transformer model is derived based on the princi-
0 ple of duality and consists of seven nonlinear hysteresis
15 20 25 35 30 40 45 50 55 60
branches; see [23]. The model is shown in Fig. 11. The ap-
Time (ms)
Fig. 8. Inrush current waveform of the demagnetized 1 kVA transformer. plied dc voltage is the same as for the three-legged transform-
er; see Fig. 10(a). The changes in the magnetic flux during the
V. DEMAGNETIZING THREE-PHASE TRANSFORMERS demagnetization of the five-limb transformer with Δ-Y con-
nection are presented in Fig. 12. This figure shows that all
In this section the proposed demagnetization technique is hysteresis branches are fully demagnetized at the end of the
applied to three-phase transformers by simulation in the simulation. The hysteresis cycle and the terminal current of
EMTP. Transformer models are obtained from the principle of the transformer are not presented because of their similarity to
duality between electric and magnetic circuits [19], [20]. In the results obtained for the three-legged transformer shown in
this type of models, each magnetic branch has a one-to-one Figs. 10(c) and (d).
relationship with an inductance in the electric circuit. There- The method has been verified on Y-Y connections for three-
fore, duality models are capable to show the magnetizing sta- and five-limb transformers as well. The connection strategy is
tus of each core limb and yoke. shown in Fig. 13. The results are not shown for lack of space,
The connection of the transformer shown in Fig. 9 is Δ-Y. but in all cases all core elements are fully demagnetized.
Note that, the secondary side is open circuited, therefore, the
secondary windings are not shown in this figure. Ra, Rb, and Rc
represent the primary winding resistances. The core is mod- + - + -
eled by five nonlinear branches using the hysteresis fitter in
the EMTP-RV. By having the B-H curve of the core, the λ-I VDem1 VDem2
curve is derived based on the cross sectional area and length of
each section of the transformer core [21], [22]. Ra Rb Rc
For three-phase transformers only two (rater than three) al-
ternating polarity dc sources are required. To verify the de-
magnetization technique with realistic remnant flux, a three-
phase ac source was used to energize the transformer at t =0
and disconnected at various times for each phase (before
t =0.4 s). During the experiments the conditions were varied Hys5 Hys1 Hys2
impressing a wide range of initial conditions (residual flux) on Hys3 Hys4
each component of the transformer core. The applied demag-
netizing voltage waveform is presented in Fig. 10(a). Four
points: 1, 2, 3, and 4 are identified in Fig. 10 to present im- Fig. 9. Demagnetization circuit for a three-phase transformer with Δ-Y con-
portant stages of the transformer demagnetization process. nection.
In this case the selected dc voltage is 5 V. At t =0.4 s (rep-
resented by point 1) the negative dc voltage is applied and the
transformer reaches the negative saturation 0.1 s later at point
 6

3
5 Hys 3, 5
3 Hys 1

Magnetic Flux (Wb)

0.5
Voltage (V)

1
1 4 4
0 0

Hys 2, 4
-0.5
2
-5
2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Time (s) Time (s)
(a) (b)
0.6 40
3 3

Primary Current (A)

0.4
Magnetic Flux (Wb)

20
0.2 1
Initial Flux= 0.13 Wb 1 4
0 4 0
-0.2 End of the demagnetization process
-20
-0.4
2 2
-0.6 -40
-0.1 -0.05 0 0.05 0.1 0.4 0.150.5 0.6 0.7 0 0.1 0.2 0.3
Magnetizing Current (A) Time (s)
(c) (d)
Fig. 10. Demagnetization process of the three-phase, three-leg transformer, (a) Applied voltage by the dc source (VDem1 = VDem2), (b) Flux in all core elements;
(c) Hysteresis cycle (Hys 1) during the demagnetization process (after t=0.4 s); (d) Terminal current.

+ - + -

VDem1 VDem2 +
VDem1
_
Ra Rb Rc
+
VDem2
_

Hys5 Hys1 Hys2 Fig. 13. Connection of demagnetizers to the transformer with Y connections.
Hys6 Hys3 Hys4 Hys7
VI. OPTIMAL VOLTAGE SOURCE
In this section, the minimum voltage required to saturate the
Fig. 11. Duality-derived model for a three-phase five-leg transformer used for
demagnetization. transformer is computed. The transformer is modeled as a se-
ries R-L circuit. The resistance of the circuit includes the in-
ternal resistance of the source, connecting wires, and the re-
Hys 5 Hys 3 sistance of the winding. To facilitate the analysis, transformer
0.6 Hys 1
magnetizing branch is modeled using a piece-wise linear in-
0.4
Magnetic Flux (Wb)

Hys 2 Hys 4 ductor with two slopes; Lm (the magnetizing inductance) for
0.2 the section below the knee and La (the air-core inductance) for
0
the saturated region. Lm and La are derived from standard open
circuit test [15] and air core inductance measurements [18].
-0.2 Hys 6 The current across the inductor includes two components:
Hys 7
-0.4 steady state (iss) and transient (it) as follows:
R
-0.6 V  t

0.4 0 0.5 0.1

0.6 0.7 0.2 0.3 i(t )  iss  it  DC  ke Lm (1)
R
Time (s)
Fig. 12. Demagnetization process of a three-phase, five-limb transformer Constant k is determined from the initial conditions assum-
showing flux in all core elements.
ing that the transformer is disconnected from the source before
demagnetization (I0=0), at t =0:
 7

VDC V 0.37  VDC  2.3 (12)

I0   k  0  k   DC (2)
R R Practical experience demagnetizing large power transform-
Hence, ers using a 60 V source shows that the time required for charg-
R ing and saturating the magnetic core is in the order of 20-30 s.
VDC  t
i(t )  (1  e Lm ) (3) Up to 90 seconds are measured for the charging-to-saturation
R process of units rated at 1000 kV nominal voltage. As the
and voltage across the inductor (VL) is obtained as: charging time depends directly on the applied dc voltage, a 60
di  t
R V source is used to speed up the process. When applying this
VL  Lm  VDC e Lm (4) voltage the saturation current is not limited by the resistance
dt of the winding. The switching-off and discharge is carried out
The magnetic flux is computed by taking the integral of the when the pre-programmed current value is reached.
voltage across the inductor, assuming that residual flux is λ0:
t  t0  t
R R VII. CONCLUSION
LV
 (t )  0   VL (t ) dt  0  m DC (e Lm  e Lm ) (5)
A controlled power electronics device has been designed
t0
R
and built to perform the demagnetization of transformer cores.
The worst case happens when the transformer core is at the Successful demagnetization of single-phase and three-phase
negative saturation point λ0=-λr, and the voltage is positive. transformers has been illustrated by simulation, laboratory
Then the transformer core needs to reach the saturation point
tests, and with an 80 MVA transformer in the field. The per-
at the positive side λs yielding:
formance of the demagnetization has been verified experimen-
R R tally with repetitive inrush current measurements and con-
LV  t0  ts
r  m DC (e Lm  e Lm )  s (6) sistent FRA tests.
R Topological modeling and numerical simulations have
Substituting t0=0, the expression between the voltage of the shown that the demagnetizer properly removes the remanence
source and the time to reach saturation is: of all magnetic branches of three- and five-limb three-phase
R(s  r ) transformers for various connections including Δ-Y and Y-Y.
VDC  (7)
R
 ts
Lm (1  e Lm
) VIII. REFERENCES
[1] F. de León, A. Farazmand, and P. Joseph, ―Comparing the T and
The time to reach saturation ts is obtained by solving (7), pi Equivalent Circuits for the Calculation of Transformer Inrush Cur-
which yields: rents,‖ IEEE Trans. Power Del., vol. 27, no. 4, pp. 2390-2398, Oct.
L  R(s  r )  2012.
ts   m ln 1   (8) [2] S. Jazebi, F. de León, A. Farazmand, and D. Deswal, ―Dual Reversible
R  VDC Lm  Transformer Model for the Calculation of Low-Frequency Transients,‖
IEEE Trans. Power Del., vol. 28, no. 4, pp. 2509-2517, Oct. 2013.
From (7) and (8) one can conclude that there is a direct rela- [3] Y. Wang, S. G. Abdulsalam, and W. Xu, ―Analytical Formula to Esti-
tionship between the amplitude of the voltage source and the mate the Maximum Inrush Current,‖ IEEE Trans. Power Del., vol. 23,
pp. 1266–1268, Apr. 2008.
demagnetization time; the higher the voltage, the faster the [4] M. Steurer and K. Frohlich, "The impact of inrush currents on the me-
demagnetization process. The minimum voltage necessary to chanical stress of high voltage power transformer coils," IEEE Trans.
assure that the core will be saturated (VDC,min) is obtained by Power Del., vol.17, no.1, pp.155-160, Jan. 2002.
the minimization of (7), which results in: [5] J. H. Brunke and K. J. Frohlich, ―Elimination of transformer inrush
currents by controlled switching. I. Theoretical considerations,‖ IEEE
R(s  r ) Trans. Power Del., vol. 16, no. 2, pp. 276-280, Apr. 2001.
VDC ,min  (9) [6] J. H. Brunke and K. J. Frohlich, ―Elimination of transformer inrush
Lm currents by controlled switching. II. Application and performance con-
siderations,‖ IEEE Trans. Power Del., vol. 16, no. 2, pp. 281-285, Apr.
Also to protect the source from over-currents, the terminal 2001.
current in steady state should not exceed the nominal current [7] Y. Cui, S. G. Abdulsalam, S. Chen, and W. Xu, ―A sequential phase
of the source, energization technique for transformer inrush current reduction- Part I:
simulation and experimental results,‖ IEEE Trans. Power Del., vol. 20,
VDC ,max  RI n (10) no. 2, pp. 943-949, Apr. 2005.
[8] W. Xu, S. G. Abdulsalam, Y. Cui, and X. Liu, ―A sequential phase ener-
Therefore, the following condition needs to be satisfied for gization technique for transformer inrush current reduction - Part II: the-
oretical analysis and design guide," IEEE Trans. Power Del., vol. 20, no.
the applied dc voltage: 2, pp. 950-957, Apr. 2005.
[9] A. Farazmand, F. de León, K. Zhang, and S. Jazebi, ―Analysis, Model-
d R(s  r ) RI ing, and Simulation of the Phase-Hop Condition in Transformers: The
 VDC  n (11) Largest Inrush Currents‖, IEEE Trans. Power Del., on Early Access.
Lm d [10] CIGRE Brochure 342, Working Group A2.26 report, Mechanical condi-
tion assessment of transformer winding using Frequency Response
where d =1.1 as a safety factor to assure functionality of the Analysis (FRA), Paris 2008.
device. For the single-phase transformer studied in this paper, [11] N. Abeywickrama, Y. V. Serdyuk, and S. M. Gubanski, ―Effect of Core
Lm=1284 mH, R=R1+Rs=277+144=421 mΩ, λs=0.57 Wb, Magnetization on Frequency Response Analysis (FRA) of Power Trans-
λr=0.45 Wb, and In=6 A, hence, the following condition needs formers,‖ IEEE Trans. Power Del., vol. 23, no. 3, pp. 1432-1438, Jul.
2008.
to be satisfied:
 8

[12] M. F. Lachman, V. Fomichev, V. Rashkovski, and A. M. Shaikh, ―Fre- Ashkan Farazmand was born in Tehran, Iran, in 1983. He received the
quency Response Analysis of Transformers and Influence of Magnetic M.Sc. (Hons.) degree in electrical engineering from the University of Tehran,
Viscosity,‖ 77th Annual International Doble Client Conference, paper Tehran, Iran, in 2009, and the Ph.D. degree (Hons.) in electrical engineering
TX11, Boston MA, Apr. 2010. from the NYU Polytechnic School of Engineering, Brooklyn, NY, in 2013,
[13] B. Kovan, F. de León, D. Czarkowski, Z. Zabar, and L. Birenbaum, where he continued his research as a postdoctoral fellow. Currently he is an
―Mitigation of Inrush Currents in Network Transformers by Reducing R&D electrical engineer at TEG Energy Group, New York, New York. His
the Residual Flux with an Ultra-Low-Frequency Power Source,‖ IEEE research interests are design and analysis of electrical machines and high
Trans. Power Del., vol. 26, no. 3, pp. 1563-1570, Jul. 2011. voltage equipment, power plant control process, power system transients,
[14] DV Power, ―Operating instructions DEM60C demagnetizer‖ Lidingo, power quality, and smart grid technology.
Sweden 2013.
[15] IEEE Guide for Diagnostic Field Testing of Electric Power Apparatus –
Part I: Oil Filled Power Transformers, Regulators, and Reactors, sec-
tion 6.1.3.5, IEEE Standard C57.152-2013.
[16] H. Kristensen and V. Mrdic, ―Comparative analysis of three phase and
single phase dynamic resistance measurement results,‖ CIRED 22nd In- Saeed Jazebi (S’10, M’14) was born in 1983, Kerman, Iran. He received his
ternational Conference on Electricity Distribution, paper 0473, Stock- B.Sc. and M.Sc. degrees in 2006 and 2008 in electrical engineering from
holm, Jun. 2013. Shahid Bahonar University, Kerman, Iran, and Amirkabir University of
[17] J. Mahseredjian, S. Dennetière, L. Dubé, B. Khodabakhchian, and L. Technology, Tehran, Iran, respectively. He received the Ph.D. degree in elec-
Gérin-Lajoie, ―On a new approach for the simulation of transients in trical engineering from NYU Polytechnic School of Engineering, Brooklyn,
power systems,‖ Elect. Power Syst. Res., vol. 77, no. 11, pp. 1514–1520, New York, in 2014, where he continues his research as a postdoctoral fellow.
Sep. 2007. His fields of interest includes: electromagnetic design, modeling and
[18] F. de León, S. Jazebi, and A. Farazmand, ―Accurate measurement of the simulation of electrical machines and power system components, statistical
air-core inductance of iron-Core transformers with a non-ideal low- pattern recognition applications in power engineering, power system
power rectifier,‖ IEEE Trans. Power Del., vol. 29, no. 1, pp. 294-296, protection, and power quality.
Feb. 2014.
[19] E. C. Cherry, ―The duality between interlinked electric and magnetic
circuits and the formation of transformer equivalent circuits,‖ Proc.
Phys. Soc., vol. (B) 62, pp. 101–111, Feb. 1949.
[20] G. R. Slemon, ―Equivalent circuits for transformers and machines in-
cluding nonlinear effects,‖ Proc. Inst. Elect. Eng., IV, vol. 100, pp. 129–
143, 1953. Digvijay Deswal was born in Haryana, India, in November 1992. He is
[21] B. A. Mork, F. Gonzalez, D. Ishchenko, D. L. Stuehm, and J. Mitra, pursuing B. Tech and M. Tech degrees in electrical engineering at the Indian
―Hybrid transformer model for transient simulation—Part I: Develop- Institute of Technology (IIT) Kharagpur, Kharagpur-721302, West Bengal,
ment and parameters,‖ IEEE Trans. Power Del., vol. 22, no. 1, pp. 248– India since 2009. He was a summer intern in the NYU Polytechnic School of
255, Jan. 2007. Engineering, Brooklyn, New York in 2012. His broad research interests
[22] B. A. Mork, F. Gonzalez, D. Ishchenko, D. L. Stuehm, and J. Mitra, include machine drives, switched mode power supply design and analysis of
―Hybrid transformer model for transient simulation-part II: Laboratory electrical machines.
measurements and benchmarking,‖ IEEE Trans. Power Del., vol. 22, no.
1, pp. 256–262, Jan. 2007.
[23] C. M. Arturi, ―Transient simulation and analysis of a three-phase five
limb step-up transformer following and out-of-phase synchronization,‖
IEEE Trans. Power Del., vol. 6, no. 1, pp. 196–207, Jan. 1991.
Raka Levi (M’86, SM’97) received the Dipl.ing in 1978 and the Dr.ing de-
gree in 1994, from University of Belgrade, Yugoslavia, while the ME in 1979
from RPI in Troy, NY, USA, all in the electric power engineering. He has
Francisco de León (S’86–M’92–SM’02) received the B.Sc. and the M.Sc.
over 30 years of asset performance and condition assessment experience,
(Hons.) degrees in electrical engineering from the National Polytechnic Insti-
specializing in apparatus testing, monitoring and diagnostics. As a consultant,
tute, Mexico City, Mexico, in 1983 and 1986, respectively, and the Ph.D.
Dr. Levi has provided service to utility and industry clients in Europe, Africa,
degree in electrical engineering from the University of Toronto, Toronto, ON,
Far East, and USA in the field of diagnostic testing and investigation, as a part
Canada, in 1992. He has held several academic positions in Mexico and has
of comprehensive substation condition assessment programs. He has been
worked for the Canadian electric industry. Currently, he is an Associate Pro-
running committees that assemble asset managers and operation specialists of
fessor at the NYU Polytechnic School of Engineering, Brooklyn, New York.
major European utilities since 1995. Recently he started and has been conven-
His research interests include the analysis of power phenomena under nonsi-
ing a working group on Dynamic Resistance Measurement (DRM) methodol-
nusoidal conditions, the transient and steady state analyses of power systems,
ogy, and is involved in application of dc testing and DRM for power trans-
the thermal rating of cables and transformers, and the calculation of electro-
former OLTC condition assessment.
magnetic fields applied to machine design and modeling.