IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 20, NO.

2, APRIL 2005 943

A Sequential Phase Energization Technique

for Transformer Inrush Current Reduction—
Part I: Simulation and Experimental Results
Yu Cui, Sami G. Abdulsalam, Student Member, IEEE, Shiuming Chen, Senior Member, IEEE, and
Wilsun Xu, Senior Member, IEEE

Abstract—This paper presents a new, simple and low cost

method to reduce inrush currents caused by transformer en-
ergization. The method uses a grounding resistor connected
at a transformer neutral point. By energizing each phase of
the transformer in sequence, the neutral resistor behaves as a
series-inserted resistor and thereby significantly reduces the en-
ergization inrush currents. The proposed method has been tested
by computer simulation and laboratory experiments. Both results
show that the method has a performance similar to that of the
resistor pre-insertion scheme. The proposed method is much less
expensive, however, since there is only one resistor involved and
the resistor carries only a small neutral current in steady-state.
Fig. 1. Proposed scheme: sequential phase energization aided with a
Index Terms—Inrush current, power quality, transformer. grounding resistor.

I. INTRODUCTION sistor could provide some damping to the currents. This is the
basis of the proposed idea. The idea is further improved by in-

I NRUSH currents from transformer and reactor energization

have always been a concern in power industry. Over the past
several decades, a few methods have been proposed to limit the
troducing delayed energization of each phase of the transformer.
This improvement has made the proposed scheme almost as ef-
fective as the pre-insertion resistor scheme. The performance
inrush currents. Representative examples are the synchronous and characteristics of the proposed method have been investi-
closing of circuit breakers and the pre-insertion of series resis- gated using simulations and experiments. Very encouraging re-
tors [1]–[4]. sults have been observed. This paper explains the proposed idea
Two recent developments in power industry have re-ignited and presents its performance characteristics. The results are ob-
interests in finding better methods for controlling inrush cur- tained from simulation and experimental studies.
rents. One of them is the deregulation of electricity market.
More and more independent power producers and co-genera- II. PROPOSED SCHEME
tors are taking advantage of the situation. Their proposed gen-
erators often need to have the generator transformers energized The basic mechanism of inrush current generation from trans-
from the system side due to cost considerations. This results former energization has been well documented in [5]. It is well
in the problem of large inrush currents being injected into the known that the inrush currents are highly unbalanced among
supply system. Another development is the increased awareness three phases. If a transformer is Y grounded at the energiza-
on power quality. The power quality consequences of inrush tion side, its neutral current will also contain the inrush current.
currents can be quite detrimental. Examples are motor tripping, One may, therefore, speculate that if a resistor is inserted into
relay misoperation and so on. There is still a need to find simpler the transformer neutral, it may reduce the magnitude of the in-
and low cost schemes to limit the inrush currents. Independent rush current in a way similar to that of the series-inserted re-
power producers are especially interested in such techniques. sistor. This consideration formed the basic idea of the proposed
The idea presented in this paper was conceived when we were scheme, which is shown in Fig. 1. The bypass breaker may not
solving inrush current problems encountered by two co-gener- be needed or can have a low rating as the neutral current is close
ator plants in Canada. In view of the fact that the inrush cur- to zero in normal system operation.
rents are always unbalanced among three phases, a neutral re- Unfortunately, a simple implementation of the above scheme,
simultaneous closing of all three-phase breakers, did not pro-
duce sufficient reductions on the inrush currents. This agrees
Manuscript received May 20, 2003; revised October 18, 2003. This work was
supported by the Alberta Energy Research Institute. Paper no. TPWRD-00240- with the conclusion drawn in [6], [7]. We reasoned, however,
2003. that if one closes each phase of the breaker in sequence with
The authors are with the Department of Electrical and Computer Engi- some delays between them, the neutral resistance could behave
neering, University of Alberta, Edmonton, AB T6G 2V4, Canada (e-mail:
[email protected]). as a series resistor and improve the results. This simple im-
Digital Object Identifier 10.1109/TPWRD.2004.843467 provement has proven to be very effective. In fact, the idea of
0885-8977/$20.00 © 2005 IEEE
944 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 20, NO. 2, APRIL 2005

sequential energization of three-phase equipment can lead to a

new class of techniques to reduce switching transients. In the
subsequent sections, simulation and experimental study results
are presented to demonstrate the characteristics of the proposed
method. Main studies were conducted on a 30 kVA, 120/208 V,
three-phase three-limb laboratory transformer with grounded Y
primary and delta secondary.

III. SIMULATION RESULTS

The proposed scheme was first investigated using computer
simulations. Initially, we used PSCAD/EMTDC to evaluate the
proposed scheme. Unfortunately, the program produced erro-
neous results. Some of the simulation results presented in [8]
are therefore not quite correct. Further investigations revealed
that the program could not simulate transformers with neutral
impedances correctly, though it has that feature. Such trans-
former configuration is known to cause simulation difficulties
for the EMTP programs [9]. As a result, we had to develop a
dedicated simulation program for this project. The transformer
model adopted in this study was published in [10]. It is an equiv-
alent electric and magnetic circuit model. The program was fur-
ther enhanced with the capability of scanning through all pos-
sible voltage phase angles and outputting the maximum inrush
current. Details about the simulation model and model parame-
ters are shown in the Appendix.

A. Sample Simulation Results

Three sets of simulations were conducted. The first set in-
volves the closing of the first phase. The second set simulates
the closing of the 2nd phase after the first phase has reached
steady state. The third set is the energization of the 3rd phase
after the second phase switching has reached steady state. Typ-
ical inrush current waveforms are shown in Fig. 2 for the case
of while different phase is energized. Examination Fig. 2. Typical inrush current waveforms for the case of R = 2 5
when
:
different phase is energized.
of extensive simulation results revealed the following.
• The inrush current occurs in all phases that are electrically
closing instants are scanned through all possible values to ob-
energized. For example, all three phases experience inrush
tain the maximum inrush current. The results for the three se-
currents when the 3rd phase is switched in. The highest
quential switching scenarios are shown in Fig. 3.
inrush current does not necessarily occur in the switched
It can be seen that a large resistance value yields more reduc-
phase.
tion on the inrush current for the first phase switching. This is
• It takes about 5 to 10 cycles for the inrush current to reach
understandable since the resistor serves as a series resistor in
steady state. As a result, there is no need to wait for a long
the energization circuit. For the second phase switching, a large
time to energize the next phase. In other words, the pro-
neutral resistance will cause more second phase inrush current
posed scheme can be implemented by closing each phase
to flow through the first phase. As a result, less inrush current
with 5 to 10 cycle delays.
flows into the neutral resistor, leading to reduced mitigation im-
• The instants of energization have an impact on the mag-
pact of the resistor. This phenomenon is shown in Fig. 4. In the
nitude of the inrush current. However, the inrush currents
extreme case of , the scheme is equivalent to ener-
are small when the neutral resistance is in certain range
gizing two series-connected transformer cores with a line-to-
(see next section). There is no need to control the instant
line voltage. A large inrush current is expected. This explains the
of switching closing.
checkmark-shaped characteristics of the second curve. For the
third phase switching, the inrush current increases with the neu-
B. Impact of Neutral Resistance on Current Magnitude tral resistor. So it is not desirable to have the resistor in the cir-
The most important parameter for the proposed scheme is cuit when the 3rd phase is energized. Further analysis of the phe-
the value of the neutral resistor. For this purpose, the maximum nomenon reveals that once the 3rd phase is energized, the neutral
inrush current as a function of the neutral resistance is inves- current reduces to zero almost instantaneously. Therefore, the
tigated. In this study, the voltage phase angles at the breaker resistor has no direct impact on reducing the inrush current. This
CUI et al.: A SEQUENTIAL PHASE ENERGIZATION TECHNIQUE FOR TRANSFORMER INRUSH CURRENT REDUCTION—PART I 945

Fig. 5. Flux pattern for the case of 3rd phase energization (R = 0).

Fig. 6. Comparison of the neutral resistor scheme with the series resistor.

Fig. 3. Magnitude of inrush current as affected by the neutral resistor. 3rd energization curve with the higher one of the 1st and the
2nd energization curves. For the results shown in Fig. 3, the op-
timal value is found to be about 3 . With this resistance value,
the reduction of inrush current is about 89% for the first phase
energization, 91% for the 2nd phase energization and 93% for
the 3rd phase energization respectively, in comparison with the
cases of zero neutral resistance. This is a significant reduction
on the inrush current. Fig. 6 shows the effectiveness of the pro-
posed scheme in comparison with the series resistor insertion
scheme. For the case of a 3- resistor, the proposed scheme is
Fig. 4. Current flow pattern for the case of second phase energization. essentially as effective as the series resistor scheme and is suf-
ficient for intended application. If one increases the size of the
explains why the scheme of simultaneous breaker closing does resistor, the series resistor scheme can reduce the inrush cur-
not work. However, the size of the neutral resistor can change rent further. The inrush current will increase with the proposed
the steady-state voltage that crosses the third breaker before it is scheme. This is because that the neutral resistor is no longer the
closed. When the neutral resistance is zero, the breaker voltage optimal one.
is zero. This can be understood from Fig. 5. In this figure, phases Note that the maximum inrush currents at are slightly
A and B have been energized, so the flux in the phase C core is different between the two schemes in Fig. 6. This is due to the
fact that a simultaneous closing of three breakers was simulated
for the series resistor scheme, while for the proposed scheme, it
The voltage induced on the phase C winding becomes is a sequential closing.

C. Impact of the Neutral Resistance on Current Duration

In addition to reducing magnitude of inrush current, the neu-
The above equation implies that the induced voltage is equal to tral resistor can also reduce the duration of the inrush current.
the supply voltage. As a result, there are no transients when the This postulation was investigated and the results are shown in
third winding is switched on to the supply. As will be shown in Fig. 7. This figure shows the decaying peak current of each cycle
the companion paper, this is an important requirement for the as a function of time. The curves correspond to the largest in-
proposed scheme to be effective. When R becomes very large, rush current case for the first phase switching.
the case is similar to energizing the second phase. It can be seen that inrush current is damped quickly when
Accordingly, we can conclude that there are conflicting re- there is a resistor in the circuit. This is another benefit of the
quirements for the size of the neutral resistor for each ener- proposed scheme. We also checked the time constant associ-
gization event. The optimal neutral resistance is the one that ated with the second phase switching. A similar phenomenon
can yield compromised reduction of the inrush current for all was observed. But the reduction of time constant is not as sig-
events. Such an optimal value is the interaction point of the nificant as the first phase switching. This is due to the fact that
946 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 20, NO. 2, APRIL 2005

Fig. 7. Decay of peak inrush current with time. Fig. 9. Experimental transformer (30 kVA, 120/208 V).

TABLE I
NO-LOAD TEST DATA FOR A HYUNDAI TRANSFORMER

Fig. 10. Typical current waveforms when the second phase is energized.

which is shown in Fig. 9. The main difficulties encountered for

experiments are 1) the switching instant cannot be controlled at
will, 2) there is an unknown amount of residual flux when the
first phase is energized, 3) the supply system has an impedance
and 4) the supply voltage is distorted with harmonics. To
Fig. 8. Simulation results of a 72/13.8-kV transformer.
overcome the first difficulty, each switching event was repeated
more than 100 times. The results are then grouped according to
there is a second path for the inrush current. In theory, the neutral the recorded phase angles of the supply voltage. The maximum
resistor has no impact on the time constant associated with the inrush current was extracted from the results.
third phase switching. This is because the inrush current does
not flow through the resistor once the third phase is closed. In A. Inrush Current Waveforms
summary, the neutral resistor has some positive effects on re-
ducing the duration of inrush currents. Since the resistor has re- Due to space limitation, only shown in this paper are the
duced the magnitude of inrush current so significantly, the effect sample results for the 2nd and 3rd phase switching when
on duration can be considered as a secondary benefit. . The 1st phase switching is similar to the series resistor
scheme and is omitted here. Fig. 10 shows the phase B and neu-
D. Simulation Results of a Large Transformer tral current when phase B is energized. It can be seen that the
neutral current is larger than phase B current. There is nothing
Simulation studies have also been conducted on a 138 MVA,
special about this phenomenon since, depending on the instant
72/13.8 kV, transformer. The nameplate of this trans-
of switching, phase A current could add to phase B current to
former is shown in Table I. The transformer is energized from
yield a larger neutral current.
the 72 kV side and the simulation results are shown in Fig. 8.
Fig. 11 shows the Phase C and neutral current when phase C
A significant reduction of the inrush current can be seen. The
is energized. The most interesting information one can extract
optimal resistor value is about 2000 , which is approximately
from this figure is that the neutral current reduces to zero al-
10% of the magnetizing impedance of the transformer.
most instantly once the 3rd phase is energized. It justifies why
the neutral resistor has no direct impact on reducing the inrush
IV. EXPERIMENTAL RESULTS current in this case. However, as will be shown in a companion
The proposed scheme has also been investigated through paper, the neutral resistor contributes to the inrush current reduc-
extensive laboratory experiments. The experiments were con- tion through reducing the voltage across the 3rd breaker before
ducted on the same transformer used for simulation studies, it is closed.
CUI et al.: A SEQUENTIAL PHASE ENERGIZATION TECHNIQUE FOR TRANSFORMER INRUSH CURRENT REDUCTION—PART I 947

Fig. 11. Typical inrush current waveforms when 3rd phase is energized. Fig. 13. Magnitude of inrush current as affected by the neutral resistor (Y-Y
connection).

simulation studies cannot take into account the (unknown)

impedance of the supply system. The ‘far end’ of the saturation
curve is not exactly known because of limitations of experi-
mental facility. The harmonic distortion in the supply voltage
is another significant factor. The supply voltage has a harmonic
voltage THD around 4%. This factor is likely the main cause
leading to the fluctuation of the curve representing the second
phase energization. This judgment is supported by an open
delta experiment for the transformer. In this experiment, the
inrush currents versus curves are determined for the test
transformer when its delta-connected secondary is open. In this
case, the transformer cannot trap zero sequence current from
the system. The resultant 2nd phase energization curve has
more noticeable fluctuation (Fig. 13).
In spite of the differences, the effectiveness of the proposed
inrush current suppression scheme has been verified by both
simulation and experimental results. The resultant optimal value
for the neutral resistor has a good agreement.

V. CONCLUSIONS

This paper has presented a new technique to reduce inrush

Fig. 12. Magnitude of inrush current as affected by the neutral resistor. current caused by transformer energization. Extensive simula-
tions and experimental results have confirmed the effectiveness
B. Impact of Neutral Resistance on the Current of the proposed scheme. Main conclusions can be summarized
in the following.
The maximum inrush current as a function of the neutral re-
sistance is also determined from experimental results. As dis- i) There is an optimal neutral resistor value for the proposed
cussed earlier, each switching event was repeated more than 100 scheme. This value is a compromised value between the
times. Approximate maximum inrush current was determined. need to suppress the inrush currents when the first two
The results are shown in Fig. 12. phases are energized and the need to suppress the current
It can be seen that Fig. 12 is quite similar to the one derived when the 3rd phase is energized. It is not essential to use
from simulation studies. The results are good confirmation to an exact optimal value. Resistances around the optimal
the validity of the simulation technique. The optimal resistance value are almost equally effective.
is in the range between 2 and 8 . This value is comparable ii) With the proposed resistance value(s), the neutral resistor
with that derived from simulation. One can also find that there is based scheme can lead to 80% to 90% reduction on the
a relatively wide range to select the value for the neutral resistor. inrush current. The results also show that it is as effective
A resistance value in the range of 2 to 10 has almost the same as the series resistor scheme. As a result, the proposed
effectiveness. Use of exact optimal value is not essential for the scheme has the potential to fully eliminate the power
proposed scheme. quality concerns caused by transformer energization.
The difference between the simulation and experimental It is important to note that the proposed time delay for sequen-
results is due to a number of factors. For example, the exper- tial energization is a simple mechanical delay of 5 to 60 cycles.
imental results cannot catch the highest inrush current. The There are no strict requirements on the delay time. High voltage
948 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 20, NO. 2, APRIL 2005

Fig. A1. Fluxes diagram of a three-phase three-limb transformer.

Fig. A4. Exciting characteristic of three phases.

A. Transformer Model
The transformer model is a set of nonlinear differential
Fig. A2. Equivalent magnetic circuit of a three-phase three-limb transformer. equations as shown below. In these equations all secondary-coil
quantities have been referred to the primary side. N is the
number of turns of the primary side windings.

Fig. A3. Equivalent electric circuit of phase j of a three-phase three-limb

transformer.

circuit breakers are typically set up with independent operating

mechanism for each phase. A delayed closing of phases is
therefore achievable. In the case of a three-phase operated
circuit breaker equipped with a common operating mechanism,
the breaker typically has a built-in mechanical delay among
three phases. This feature can be used to implement the pro-
posed scheme. The voltage or current imbalance introduced by
The above equations have taken into account the delta con-
the delayed operation should not be a concern. This is because
nection in the transformer secondary, which gives
the transformer is unloaded and the duration of imbalance is
. This model consists of 13 equations. It is solved using
very short.
the Matlab as well as MathCAD tools. Two independent pro-
grams were developed for this purpose.
APPENDIX
In this appendix information about the simulation model is B. Model Parameters
presented along with experiments used to determine the model Parameters needed for the above model were determined
parameters. The model was published in [10]. Fig. A1–A3 from experiments. The parameters include the winding resis-
show the fluxes diagram, the equivalent magnetic circuit and tances, leakage inductances, inductances of limb and yoke and
the equivalent electric circuit of a 3-phase 3-limb transformer. the reluctances. The winding resistances and leakage induc-
Magnetic fluxes associated with the transformer can be sub- tances were determined from standard short-circuit tests on
divided into the following categories as indicated in Fig. A1. each winding [11]. The average values among three phases are
• Magnetic fluxes that link different phases and follow the , .
magnetic circuit ( , , , , and ). They are as- A zero sequence open-circuit test was used to deter-
sociated with the nonlinear reluctances , , , , mine the air path reluctance . Measured average value is
and ; , which is used for all three phases.
• Phase leakage fluxes which link all the windings of one To measure the nonlinear reluctances of the limbs and yokes,
phase, and partially go through air ( , , and ). the transformer was reconnected into Y/Y connection and
They are associated with constant reluctances , , was supplied by a 3-phase balanced voltage. A set of open
and . circuit tests is conducted to find the saturation curves, which
In Fig. A3, and are the winding resistor and leakage are then converted to the nonlinear equation of .
inductor of the primary side, while and are those of the Fig. A4 shows the exciting characteristic of different phases.
secondary side, but refereed to the primary side, respectively. Limited by the test facility, we could only get the ‘front’ part
CUI et al.: A SEQUENTIAL PHASE ENERGIZATION TECHNIQUE FOR TRANSFORMER INRUSH CURRENT REDUCTION—PART I 949

[5] Westinghouse, Electrical Transmission and Distribution Reference

Book. Chicago: The Lakeside Press, R. R. Donnelley & Sons Com-
pany, 1944, pp. 411–417.
[6] B. Holmgrem, R. S. Jenkins, and J. Rinbrugent, “Transformer inrush
current,” in CIGRE Proc. 22nd Session, vol. 1, 1968, pp. 1–13.
[7] R. Yacamini and A. Abu-Nasser, “The calculation of inrush current in
three-phase transformers,” IEE Proc. Electr. Power Appl., vol. 133, no.
1, pp. 31–40, Jan. 1986.
[8] L. Cipcigan, W. Xu, and V. Dinavahi, “A new technique to mitigate in-
rush current caused by transformer energization,” in Proc. IEEE 2002
PES Summer Meeting, Chicago, USA, Jul. 2002.
[9] H. M. Dommel, EMTP Theory Book, 2nd ed. Vancouver, British Co-
lumbia: Microtran Power System Analysis Corporation, 1996.
[10] M. Elleuch and M. Poloujadoff, “A contribution to the modeling of three
phase transformers using reluctance,” IEEE Trans. Magn., vol. 32, no.
Fig. A5. Computed inrush current waveforms for a transformer of [10]. 2, pp. 335–343, Mar. 1996.
[11] Members of the staff of the Department of Electrical Engineering,
Massachusetts Institute of Technology, Magnetic Circuits and Trans-
of the saturation curve. The deep-saturation part of the curve is former. New York: Wiley, 1943, pp. 443–453.
extrapolated according to the trend.

C. Model Verification
To verify the model, a test case shown in [10], which is a Yu Cui received the B.Eng. degree from Tsinghua University, Beijing, China
in 1995; the M.Sc. degree from Institute of Electrical Engineering, Chinese
10 kVA, 232 V, 3-phase 3-limb transformer with grounded Y/Y Academy of Sciences, Beijing, China in 2000; and the M.Sc. degree from Uni-
connection, is used. The inrush current waveforms computed versity of Saskatchewan (Canada) in 2003. He is currently working on his Ph.D.
by this project is shown as Fig. A5. The waveforms have a close degree at University of Alberta.
His research areas include power system stability and power quality.
agreement with the results shown in Fig. 9 of [10].
Due to the various factors described in Section IV, we could
not produce inrush current waveforms that can exactly match
the experimental results. However, there is an acceptable agree- Sami G. Abdulsalam (S’03) received the B.Sc. and M.Sc. degrees in electrical
ment for the general shape of the waveforms and their decaying engineering from Elmansoura University, Egypt in 1997 and 2001, respectively.
trends. Furthermore, the validity of the model is confirmed by Since 2001 he has been with Enppi Engineering Company, Cairo, Egypt. He is
currently pursuing the Ph.D. degree in electrical and computer engineering at
the close similarity of inrush current versus curves. the University of Alberta.
His current research interests are in modeling and simulation of power system
transients.
ACKNOWLEDGMENT
The authors wish to thank C. Muskens of ATCO Electric and
T. Martinich of BC Hydro for the suggestions and comments
during the course of this project. The help of A. Terheide, a tech- Shiuming Chen (M’02–SM’02) received the B.Sc. degree in 1990 and M.Sc.
nician at the University of Alberta power lab, on experimental degree in 1993 in electrical engineering from Zhejiang University, and the Ph.D.
degree in 1997 from Tsinghua University, Beijing, P R China.
investigations is fully acknowledged. He is currently an Associate Professor of the Tsinghua University and is
working at the University of Alberta as a post-doctoral fellow. His research in-
terests are over-voltage protection in power systems and electromagnetic com-
REFERENCES patibility in power and electronic systems.
[1] T. Specht, “Transformer inrush and rectifier transient currents,” IEEE
Trans. Power App. Syst., vol. 88, no. 4, pp. 269–276, Apr. 1969.
[2] CIGRE working group task force 13.07, Controlled switching of HVAC
circuit breakers, in 1st Part Elektra, no. 183, pp. 43–73, Apr. 1999.
[3] , “Controlled switching of HVAC circuit breakers,” 2nd Part Wilsun Xu (M’90–SM’95) received the Ph.D. degree from the University of
Elektra, no. 185, pp. 37–57, Aug. 1999. British Columbia, Vancouver, BC, Canada, in 1989.
[4] K. Smith, L. Ran, and B. Leyman, “Analysis of transformer inrush Currently he is a Professor with the University of Alberta, Edmonton, AB,
transients in offshore electrical systems,” Proc. Inst. Elect. Eng., Gen. Canada. He was an Engineer with BC Hydro, Vancouver, BC, Canada, from
Transm. Distrib., vol. 146, no. 1, pp. 89–95, Jun. 1999. 1990 to 1996. His main research interests are power quality and harmonics.