Understanding the Sympathetic Inrush Phenomenon in

the Power Network using Transformer Explorer

Subrat Sahoo, Nilanga Abeywickrama, Tord Bengtsson, Robert Saers
ABB Corporate Research
Västerås, Sweden

Abstract— An in-service transformer in a substation may voltage dip on the primary side voltage. This phenomenon
experience sympathetic inrush phenomenon due to startup could last up to several seconds. Its dependence on switching
inrush current from an incoming transformer, energized into in angle, transformer operating conditions such as residual
the network. The startup inrush causes a half cycle voltage dip flux etc. are discussed for single and three phase transformers
in either cycle, thereby developing an asymmetrical flux
distribution in the nearby transformers in service. The paper
in [2], [3].
discuses about sympathetic inrush current events, originating in The magnitude of the inrush current can be ten times the
a Swedish distribution network due to energization of rated current in the first half cycle and the voltage dip during
transformers in same, or neighboring network. These events this period could be 10-20% of rated peak value. Such events
were further observed to be potentially detrimental to the are distinctly different from the voltage dips observed due to
network operation as well as to the health of the transformers. sudden load current changes. The voltage dips during the
An in-house online monitoring tool- Transformer Explorer was startup can be directly linked to the transient component of
deployed for continuous data acquisition and capture of the the current. Such an event does not trigger voltage controller
anomalous events including sympathetic inrush. The impact of at the generating station such as the automatic voltage
the transformer following such event is also briefly discussed,
based on the combined diagnostic observation made by PSCAD
regulator (AVR) nor is responded by the on-load tap changer
simulation tool and Transformer Explorer. (OLTC). The transformer protection functions are usually
blocked during a start up to avoid false alarm/ tripping.
Index Terms-- Startup inrush, sympathetic inrush, Transformer However, the startup inrush currents in the process of
Explorer, system events, PSCAD. getting into the network disturb the operating profile of the
transformers already in service, in a way that is debatable in
I. INTRODUCTION the research community. In other words, the in-service
transformers sympathize on the transients caused by the
Transformers are not only the single most expensive incoming transformers due to the asymmetrical voltage
devices in a power system, their availability and reliability distribution developed in the network, thereby affecting its
contribute quite significantly to the healthy economy of the magnetization characteristic. This phenomenon is called
energy utilities. Any unplanned outage and subsequent sympathetic inrush event.
downtime costs dearly to various stakeholders, namely-
distribution network operators (DNOs), transmission system II. SYMPATHETIC INRUSH CURRENT OF TRANSFORMERS
operators (TSOs) and the manufacturers [1], especially in the An in-service power transformer in a network can
context of deregulated energy market. The energization of the experience another form of core saturation, when the
transformers in and out of network are quite regular but are aforementioned asymmetrical half-cycle voltage drop is
drawing more relevance in recent times due to the changing created by an incoming transformer [4]-[6]. A schematic
equations of the power network and involvement of large representation of an in-service transformer (#T1) and
scale renewable energy sources such as large solar parks, incoming transformer (#T2) connected to a common bus is
offshore wind etc. This calls for frequent change in network illustrated in Fig. 1.
topology impacting operation and life of both the in-service The start-up inrush current from #T2 creates a half cycle
transformer and the incoming transformer as well as of the voltage dip either in positive or negative slope. The primary
network as a whole. While the network transformers are side voltage of #T2 on the 130 kV side was measured and
seldom switched, the generator transformers are used to such Hilbert transformed during the startup inrush event (around
switching requirements depending on the dispatch conditions. 0.94s) as illustrated in Fig. 2. This sudden asymmetrical
Transformers produce inrush current (also called as voltage dip builds up a magnetic flux offset per cycle in the
startup inrush current) while energized into the network. This core of transformer #T1 that is uneven in the two cycles of
current drags the ferro-magnetic core of the transformer into supply voltage. For a non-zero integral of the supply voltage
deep saturation, during one half cycle of the power frequency. in one cycle, the magnetic flux (being proportional to integral
A half cycle voltage drop (about 90 degree lagging) is created of supply voltage) would drive the core of #T1 into deep
along the upstream system impedance resulting in a half cycle
saturation in one cycle, when #T2 experiences a startup events were simulated similar to the ones observed by the
inrush current in the other cycle. field installations. Fig. 3 depicts the consequence of a startup
event of T8 at station#2. The characteristic difference
between start-up inrush (2nd plot of Fig. 3 from top) and
sympathetic inrush current envelop (3rd plot from top), caused
by the voltage dip (1st plot from top), discussed in the
previous section. The startup inrush can be seen as a single
exponential decay of magnitude over time (few seconds of
decay time), whereas the sympathetic inrush current sees a
double exponential decay (with ~1.5 s rise time and >10 s
decay time from the peak) of magnitude over time.

Figure 1. Generic circuit for studying the sympathetic interaction between an

in-service and an incoming transformer in the common busbar

Figure 2. Measured and Hilbert transformered voltage on the primary side of

130 kV bus of station#2 during a startup event of #T2 at -0.94 s, marked in
blue.

As the sympathetic inrush current picks up amplitude, it

also creates increasing voltage dip per half cycle on the
common bus bar and hence the supply voltage now has two
voltage dips in each half cycle, one of which is created by the
inrush current of #T2 (continuously decreasing) and the other
one created by the sympathetic inrush current of #T1
(increasing in the beginning). At some point, this increasing
and decreasing voltage dips in alternate half cycles lead to a
zero integral per cycle over the voltage which in other words
equals zero flux offset per cycle. That is the time where the
sympathetic current envelop reaches its peak and starts
decaying thereafter (see 3rd plot from top in Fig. 3).
A. Simulation of Symapthetic inrush event using PSCAD
In order to understand the sympathetic inrush current
characteristics, analytical investigation was performed at
different locations within a Swedish utility, following
energization of transformers either in the same or in different
stations. A part of this network is represented in the single
line diagram in Fig. 4. The 130 kV side of the feeding station
is connected to three different stations, designated as Figure 3. PSCAD simulation results showing the impact of startup event of
station#1-3. Each station has two transformers in parallel and T8 at station#2, from top: (1st plot): voltage dip of T8 on 130 kV side, (2nd
the overall network represents a ring structure with 130 kV plot): startup inrush current of T8, sympathetic inrush current in (3rd plot): T7
at the same station#2, (4th plot): T4 at station#1 and (5th plot): T6 at station#3.
on the upstream. Three Transformer Explorer installations
were deployed (with provisions of recording exceptional
If there is an overhead line or cable link between the in-
waveforms- more details in Section III) on those transformers
service transformer (e.g., T6 at station#3 in this case) and the
marked in red in station #1-3 in Fig. 4.
incoming transformer (e.g. T7 at station#2), the line
A part of the above network was representatively
resistance between the transformers should be included as
modelled in PSCAD, with actual overhead line distances and
well. Consequently, the sympathetic inrush current envelop
transformer nameplate parameters. The transformer switching
reaches its peak, quicker (5th plot from top in Fig. 3), as
compared to a case where both interacting transformers exist
on the same bus (3rd plot from top in Fig. 3).

Feeding station
400 kV

T1 T2 Station #1
130 kV

Station #2 ~4.5 km T3
T7

~25 km T4
~12 km
T8
T5

Station #3

T6
Figure 4. Single line diagram of the Swedish utility site for a 130 kV
network, the Transformer Explorer installations are made on station 1, 2 and
3, on the transformers marked as red. Figure 5. Measured sympathetic inrush current in T7 at station#2, (top plot):
due to a startup of T8 at the same station, (middle): top plot zoomed in
Additionally, lower system impedance could also greatly between 0.4-0.5s and (bottom plot): due to energization of T4 in station#1.
reduce the sympathetic inrush current magnitude [4]. The
latter plays an important role, when the ohmic impedance III. TRANSFORMER EXPLORER AS A DIAGNOSTIC
contributes significantly to the total system impedance [7].
This is seen in Fig. 3 (3rd to 5th plot), where the individual TOOL
highest magnitude is almost half of the previous value, for the Transformer Explorer is an online method to continuously
observed sympathetic inrush current. State space approach is monitor the important parameters of a transformer, namely-
used to understand the phenomena in [7], whereas [5] uses a turns ratio, impedance, power loss, etc. It uses a simplified
nonlinear transient field circuit coupled finite element transformer model to deduce the above parameters from the
approach, further discussing various parameters in the voltage and current measurements recorded at either side [9,
network contributing to the sympathetic inrush. 10]. Apart from providing phasors for estimating these
The measured sympathetic inrush current events in T7 parameters, the tool is built on the philosophy that it predicts
from activities in Fig. 4 is reported in Fig. 5. The events are the waveform of the immediate future and continuously
captured by the Transformer Explorer field installations in the compare with what is obtained at present with the frequency,
Swedish utility (as displayed in Fig. 4). The top plot is a amplitude and phase, thereby sensitively detecting any
result of startup of T8 in the same station, whereas the bottom transients or fast changes that would result in a deviation
plot is the contribution from startup of T4 in station#1. The from the predicted waveform. Due to frequency variations in
middle plot demonstrates a zoomed portion of the top plot, the power system, the longest period that can be predicted is
where the excursion of phase-3 in negative half cycle is seen about half a second. Every half second, the acquired
to be followed by the other two phases in the positive cycle. waveforms are analyzed with high precision alternating
The effective dominance of different phases is seen by the current (HPAC) analyzer, as explained below.
envelope for their duration of existence in the network. HPAC is an algorithm for extracting phasors from an AC
This corroborates the fact that with increase in distance wave form. It is similar to DFT but with substantially higher
between the incoming and in-service transformers, the sprecision and can theoretically obtain any frequency (not
sympathetic inrush current experiences a steep rise and fall just harmonics) and number of harmonic components, as it is
and lower envelop peak (bottom plot) due to higher ohmic based on interpolated spectral analysis. HPAC analysis is
impedance offered by the transmission line or cable. In this used in ABB’s injection-based ground fault protection
particular case, station#2 and #1 are located about 30 km functions [8] and sub-synchronous resonance protection [11].
apart. On the contrary, T8 in the same station, contributes to a The HPAC analysis returns high-precision estimates of
lower impedance thus increase in rise and decay time as well frequency, amplitude and phase of all requested approximate
as higher sympathetic inrush current peak as seen in the top frequencies. These values are statistically analyzed for any
plot of Fig. 5. The inrush current magnitude in the top plot is deviations beyond prescribed standard deviation limits, in
one order higher, compared to the bottom plot. Consequences which case these waveforms are stored.
of such phenomena are discussed further in Section III. A log file stores every minute averages of half second
estimates. This log file serves as data input to the
Transformer Explorer platform that is used for monitoring of B. Sympathetic inrush event leading to tripping of the
anomalies in the fundamental quantity estimates, while transformer
comparing with the nameplate data. This concept and its During this special event, station #2 in Fig. 4, underwent a
deployment in several field installations are explained [9-10], trip on the 55 kV side, in the process of energizing T8 and
[12-13]. However, its usability for potential detection of bringing it in parallel with T7 to share the load. Transformer
system events and permanent changes in the transformer Explorer captured the sequence of events which is split into
winding, due to the sympathetic inrush current is reported for four time zone in Fig. 7 for further explanation.
the first time in this paper. The installations include standard  Time t1: T8 switched into the 130 kV bus, causing a
data acquisition cards of 10 kS/s and 16-bit resolution that voltage dip as illustrated for 130 kV side in Fig. 7 (a). This
digitize twelve voltage and current signals from a three phase in turn causes a sympathetic inrush in T7, which starts
two winding transformer. decaying thereafter, until time t2, in Fig. 7 (b). The power
In the following sections, two special events from one out of T7 remains unaffected, since the load is not yet
such installations are discussed, following a sympathetic shared by T8 (see Fig. 7 (c)).
inrush event associated with it, thereby leading to operational
behavior change and even tripping of load side breaker.  Time t2: 55 kV side of T8 is paralleled with T7 to share
the load at t2 (see load drop on T7 in Fig. 7(c), to be
A. Long-term change in transformer operation after a accommodated by T8). As per the magnetizing current
sympathetic inrush event logs in Fig. 7(b), the sympathetic inrush in T7 jumped
The logged event by Transformer Explorer described even higher. The magnetizing current profile at paralleling
below belongs to one transformer under investigation, where operation is not symmetrical among phases compared to
an increased power loss (of the order of a few kW) was the symmetric nature of the normal sympathetic inrush
observed due to a sympathetic inrush current event on the current recorded at t1.
transformer. As illustrated in Fig. 6, where the difference in
power loss to the expected is shown, there is a clear step
around the event. This was due to a confirmed transformer
startup event within the same distribution network as given in
Fig. 4 (involving transformers T5 and T6 in station#3).
Interestingly, it was by far the largest sympathetic inrush
current observed in various Transformer Explorer
installations, with a peak exceeding 100A.
More detailed investigations indicate an occurrence of
permanent increase in power loss of the order of 5 kW (see
the two horizontal lines in Fig. 6 before and after 18th May).
Such increased loss could be attributed to excessive magnetic
flux out of the core during a prolonged period of half cycle
saturation due to sympathetic inrush. This could cause eddy
current in metallic parts and might even result in permanent
eddy current path, developed after the event.

Figure 7. Sequence of events observed through Transformer Explorer for

startup inrush in T8 and sympathetic inrush in T7 at station#2: (a) voltage dip
observed on the high voltage side, (b) sympathetic inrush current of T7 and
Figure 6. Difference in power loss before and after the sympathetic inrush (c) Power distribution of T7 while sharing the load.
current event, marked by a blue bar. The power loss difference is relative to a
reference loss dependent on load, determined before the event.  Time t3: 5s after the parallel operation was initiated, both
T7 and T8 tripped on the 55 kV side by the protection
However, this is only a hypothesis, difficult to confirm system, allegedly due to an asymmetrical operation,
without a detailed disassembly of the transformer. While the exhibiting abnormal magnetizing current profile (where
absolute value of the loss increase is significant, it is only only phase-2 was heavily saturated). After the tripping
about 20% of the no-load losses and therefore does not event at t3, T8 is disconnected from T7 and T7 is also
warrant a deep investigation. Notwithstanding the above pulled back from being loaded (see Fig. 7 (c)). The
corollary, it is clear that the sympathetic inrush event brought sympathetic inrush current follows normal decay as before
in some long-term changes in the transformer operation. (Fig. 7 (b)).
View publication stats

 Time t4: T7 was put back on load (see Fig. 7 (c)) this time of the transformer. This paper exemplifies how Transformer
without T8, about 63 s after the tripping incident. The Explorer monitoring can find subtle transformer property
magnetization current of T7 (Fig. 7 (b)) and the voltage dip changes and the root cause. This kind of monitoring is
of 130 kV side (Fig. 7 (a)) are affected due to the loading therefore an important complement to methods presently
of T7. available commercially. Moreover, Transformer explorer
From the above explanation, it is clear that the sympathetic promotes installation of no additional sensors and can
inrush current lasted long enough to activate the tripping actually be deployed without the need of an outage.
function, 5s after T8 was brought in parallel with T7. During Two system events were demonstrated as a consequence
the paralleling event (time: t2), the voltage produced in of sympathetic inrush phenomenon. In one occasion, long-
secondary of T7 was slightly different than T8, due to term change in operational performance of the transformer
persistent sympathetic inrush current, causing a circulating was noticed, whereas in the other, the event caused tripping
current in the 55 kV side. This current exceeded the prescribed of the secondary side relays, since the adequate event
limit of overcurrent tripping function. The recommendation to dynamics were not taken into account.
overcome such a situation is to allow enough time for the ACKNOWLEDGEMENTS
sympathetic inrush phenomenon to settle down so that it does
not create a voltage difference with respect to the incoming The authors gracefully acknowledge VB Energi Elnät,
transformer. This understanding helped the utility with no Ludvika for making this investigation possible by allowing
further tripping, while sharing the load by the incoming units. Transformer Explorer installations at their substations.
Such an awareness was served only due to the detailed REFERENCES
investigation through Transformer Explorer.
[1] CIGRE UK technical insight 568, “Transformer energisation in power
systems.” CIGRE web contents.
IV. CONCLUSION
[2] P.C.Y. Ling, and A. Basak, “Investigation of magnetizing inrush
The impact of startup inrush currents on the neighboring current in a single-phase transformer,” IEEE Trans. on Magnetics, vol.
transformers in the form of sympathetic inrush current was 24, no. 6, pp. 3217–3222, November 1988.
demonstrated in the paper. The half cycle voltage dip caused [3] A. A. Adly, H. H. Hanafy, and S. E. Abu-Shady, “Utilizing preisach
models of hysteresis in the computation of three-phase transformer
by the incoming transformer and its asymmetric influence in inrush currents,” Electric Power Systems Research, vol. 65, 2003.
the flux distribution for the in-service transformer was [4] H. S. Bronzeado and R. Yacamini, “Phenomenon of sympathetic
discussed. The magnitude and duration of sympathetic inrush interaction between transformers caused by inrush transients,” IEE
current was investigated in a ring network representation, Procedings - Science, Meas and Tech, vol. 142, no. 4, July 1995.
comprising several stations with more than one transformer [5] J. Peng, Assessment of transformer energisation transients and their
impacts on power systems, PhD Thesis, School of Electrical and
each. The impact was found to be more pronounced, if the Electronics Engineering, The University of Manchester, 2013.
influencing and the affected transformer belonged to the same [6] G. B. Kumbhar, S. V. Kulkarni, ” Analysis of Sympathetic Inrush
station, compared to their existence, farther away in the Phenomena in Transformers Using Coupled Field-Circuit Approach,”
feeder, because of the reduced line impedance. IEEE PES General meeting, 24-28 Jun 20007, Tampa, FL, USA.
Transformer Explorer was demonstrated as a useful tool [7] U. Rudez, R. Mihalic, “Sympathetic inrush current phenomenon with
for not only being able to capture the system events including loaded transformers,” Elec. Power systems Research, vol. 138, Sep.
2010, PP 3-10.
sympathetic inrush current, it can also assess its impact on the
[8] S. Roxenborg, H. Johansson, P. Cost, T. Bengtsson and S. Thorburn,
transformer integrity and operational performance after an “Commissioning and verification tool for injection based 100%
event. Such frequent mild overcurrent transients generated stator/rotor ground fault Protection”, PAC World Africa Conference,
from the sympathetic inrush current is known to have a Cape Town, South Africa, 2013.
forbearance on the long-term performance of the transformer [9] T. Bengtsson, N. Abeywickrama, “On-line monitoring of power
and core related issues. transformer by fundamental frequency signals,” A2-110, CIGRE 2012.
Sympathetic inrush currents are inevitable to develop, [10] N. Abeywickrama, T. Bengtsson, R. Saers, “Transformer Explorer-
monitoring transformer status by fundamental frequency signals,”
albeit controllable with appropriate measures in the CMD conference, 25-28th Sept, China.
transmission network. Some counter measures to limit this [11] Z. Gajic, S. Roxenborg, T. Bengtsson, S. Lindahl, P-O Lindström, H.
current are by controlled switching (bringing the switching Eriksson, M. Lindström, ”Design Challenges for Numerical SSR
angle close to 90°) of transformers, improving the short- Protection”, Cigré Study Committee B5 Colloquium, September 20.-
26. 2015, Nanjing, China.
circuit strength of the network or even adding a starting
[12] T. Bengtsson, S. Sahoo, R. Saers, N. Abeywickrama, J. Hedberg,
resistance to the circuit, and working around the flux polarity “Transformer Explorer: Diagnostisk övevakning av tranformatorer i
and residual magnetization of the core. drift, Energigorsk Rapport, 2017:352, Feb. 2017, PP 1-42.
There are three main contributors to aging and failure of [13] S. Sahoo, T. Bengtsson, N. Abeywickrama, R. Saers, J. Hedberg, J.
power transformers: temperature, humidity and system Rossenlind, “ Monitoring Power Transformer Performance, Usage and
events. Present monitoring and diagnostic techniques focus System Event Impacts – A Case Study, CATCON Nov. 2017, Ropar,
India.
on the first two and possibly on secondary effects of the third.
PSCAD simulation tool is able to detect a sympathetic inrush
event but cannot evaluate its potential impact on the integrity