Proceedings of My Transfo 2016 – 15th and 16th November 2016

TERNA management of power transformers:

on-line monitors to investigate the cumulative
effect of through-fault currents
M. Rebolini, C. Serafino
TERNA Rete Italia SpA
Italy

M. Tozzi, A. Salsi, E. Savorelli

Camlin Power Ltd
Northern Ireland

ABSTRACT
Italian Transmission System Operator (TSO) TERNA manages about 700 transformers
and auto-transformers with a failure rate that has been kept <1% for several years.
Due to increasing aging of the fleet and the criticality of the asset, significant efforts
have been made to prevent failures and base the asset maintenance basing on the
real condition of the transformer insulation. One of the main causes of failure has
been addressed to the occurrence of unpredictable external events such as over-
voltages, lightning strikes and through-fault currents. The latter ones, in particular,
are speculated to be the indirect cause of winding displacement on autotransformers
and electrical failures on those transformers interconnected to MV grid. The paper
presents the development of a through fault current (TFC) monitoring system aimed
at detecting short circuit currents and demonstrating that the most of events are un-
known to the utility. By recording such activities, it will be possible to rank the trans-
formers, improve maintenance schedule, correlate multiple on-line data and extract
new parameters to be added into the TERNA Health Index.

Keywords — Transformer, through-fault current, winding deformation

tion and testing. In particular, a home-made Health

1. INTRODUCTION
The rate of failure recorded for all the TERNA trans-
formers in operation has been for several years con-
siderably lower than 1%. However, in recent years
the fault rate has started to increase while the aver-
age age of the failing unit is decreasing [1], leading
TERNA to set up a strategy to ensure the required
level of reliability and safety of medium and large
power transformers. Part of the strategy includes
deployment of on-line monitoring systems, control
and reduction of external electrical stress, specifica-
Index has been adopted [2] based on the combina-
tion of external parameters (keraunic level of the
site, substation lay-out, recurrence of catastrophic
events on that site, etc.) and transformer parame-
ters obtained through on-site testing and laboratory
results. The keraunic level together with the occur-
rence of through-fault currents (TFC) are sensible
parameters considering that lightning strikes and
external short circuits contribute to about 30% of
transformer failures [3]. While the first parameter is
already considered in the TERNA Health Index calcu-

1
 Proceedings of My Transfo 2016 – 15th and 16th November 2016
lation, the second one is now under investigation.
Speculation were made in literature [4-7] about the
aggregate effects of through faults on transformers,
potentially able to produce mechanical stresses
which result in the progressive weakening of the
winding clamping pressure and, thus, movement of
the winding turns. Indeed, these kind of currents
(typically > 50% of maximum short circuit current)
can produce electromagnetic forces, proportional to
the square of the peak of the fault current, which
can weaken the mechanical structure to a point
where the transformer is no longer capable of with-
standing a short-circuit at its terminals. This can
lead to severe winding displacement. Another pos-
sible consequence of the continuous movement of
the winding disks is the occurrence of inter-turn
short circuits.
As a result of several investigations made on failed
transformers, TERNA has recognized that a certain
number of failures could be linked to the electrody-
namic forces caused by the repetitive occurrence of
through fault currents. Evidence was found in both
large auto-transformers, where deformation was
noted in about 10% of units [8], and in smaller
HV/MV transformers suffering turn-to-turn failures
in the MV winding.
These cases led TERNA to explore the possibility to
monitor the occurrence of such events in order to
plan SFRA, Inductance and Winding Resistance test
basing on condition, correlate the data with Dis-
solved Gas in Oil analysis and plan proper corrective
actions on time. Additionally, in case of positive re-
sults, the TFC parameters could be integrated into
the Health Index calculation.
2. BACKGROUND
2.1. Limitations of Disturbance Recorders
Disturbance recorders are widely installed in the
Italian transmission network. Such devices are acti-
vated by the protection relays and, together with
various parameters, record the current waveforms
when the protections are armed. However, such
devices:
- are driven by the protection relay, not by the
current itself; reliability on the data capturing
depends on relay reliability and settings
- are generally used to support a post-failure in-
vestigation; not for predictive maintenance or
condition monitoring
2
- are not recording inrush currents
- do not send any notification
- do not send any data
- do not extract any parameter automatically;
they just store waveform in COMTRADE for-
mat.
Therefore, the current situation is that the Opera-
tion and Maintenance (O&M) people are, most of
the time, unaware of the number and magnitude of
the short circuit events affecting the transformers.
The notification of a short circuit event is, nowa-
days, sent from the control center to the O&M only
if the primary breaker is opened due to a protection
trip. It is then up to the O&M people to go on site
and manually extract the data from the disturbance
recorded. A typical scenario is when a lightning
strike causes a short circuit in the HV line, which is
cleared within <100 ms and the breaker is reclosed
immediately. In such a case the notification of the
event is not sent anywhere, but the transformer has
actually faced a high current, potentially very high.
2.2. Case Study #1 - mechanical stress re-
sulting in winding deformation
A critical inductance variation was measured in a
160 MVA autotransformer, exceeding 5% in one
phase.
According to internal rules within TERNA, a trans-
former with 5% variation must be taken out of ser-
vice. Note that the previous measurement, carried
out 4 years before, was perfect. In order to investi-
gate the possible reasons for such a significant vari-
ation, disturbance recorders were manually interro-
gated by TERNA experts and it was found that the
same phase had experienced, few months before
the measurement, a through-fault current reaching
86% of the maximum short circuit current, Icc. This
event was recorded by the disturbance recorder but
not notified to TERNA because the fault was cleared
before the transformer relay could trip. It was spec-
ulated that the transformer had experienced a sig-
nificant amount of through-fault currents over its
life that could have weakened the structure in a way
that it was not capable to withstand an event of
such a magnitude.
2.3. Case Study #2 – mechanical stress re-
sulting in electrical failure
TERNA owns about 133 transformers which inter-
connect the HV network to the MV grid. The protec-
 Proceedings of My Transfo 2016 – 15th and 16th November 2016

tion at the MV side are managed by the Distribution COMTRADE format and notifications are sent to the
System Operator (DSO). Whenever a short circuit O&M when events are triggered. Three clamp hall-
occurs at the MV side, the DSO clears it before the effect type CTs were installed (Fig 3) in the trans-
transformer HV breaker opens. These faults cause former control room connected to the secondary (5
high currents within the HV/MV transformer, but no A) winding of transformer CTs.
notification is sent to the TSO for two reasons: TABLE I
Short circuit parameters
1) the transformer HV protection relay does
not open PARAMETER DESCRIPTION VALUE
2) sometimes, in old sites, disturbance re-
IN RATED CURRENT 165 A
corders are not even installed on TSO side
VCC% SHORT CIRCUIT IMPEDANCE 13.1 %
Thus, the TSO is unaware of the high currents circu-
lating in the winding, until the transformer trips. A ICC MAXIMUM SHORT CIRCUIT 1260 A
32 MVA transformer has recently experienced a CURRENT (SYMMETRIC)

failure within the MV winding insulation (summer ICCPEAK FIRST PEAK OF THE MAXIMUM 3401 A
2016). Electrical tests were carried out on the failed SHORT CIRCUIT CURRENT
transformers, showing turn-to-turn short circuit in (ASYMMETRIC)
one phase. At least other two similar cases had oc-
TMAX MAXIMUM DURATION TIME 2S
curred in the past 10 years. Also in this case, the DURING WHICH THE
repetitive occurrence of through-fault currents was TRANSFORMER CAN WITHSTAND
speculated to be the main cause of the failure, hav- THE MAXIMUM CURRENT WITH NO
ing caused the inter-turn insulation to degrade due IRREVERSIBLE DAMAGES [8]
to continuous winding movement.
EMAX MAXIMUM SHORT CIRCUIT 3172892
2*
ENERGY A S

Fig. 1 Typical HV/MV configuration. Failures at the MV side cause

high currents at the HV side but are notified to TSO only if HVCB
opens

3. TRIAL PROJECT
3.1. Description of the system Fig. 2 TFC Monitor installed in the TERNA control room
A trial project has run in 2015-2016 installing a TFC
monitoring device on a 63 MVA transformer, 220/22
kV (category IV, [8]). Basing on the transformer
nameplate it was possible to calculate the short
circuit parameters of Table 1. The monitoring sys-
tem (Fig. 2) is able to detect over-currents simulta-
neously in three phases, store the events exceeding
a configurable trigger and extract parameters such
as duration time, maximum peak, energy of the
2
event (I t), RMS and the aggregated effect of the
2
short circuit currents (cumulative I t) from each Fig. 3 CT clamp sensors at the secondary winding of the current
event. Waveforms are stored and available in transformer

3
 Proceedings of My Transfo 2016 – 15th and 16th November 2016

Figure 4 shows the records from the disturbance tection relays and, thus, not recorded by any dis-
recorders (top) compared to the one from the moni- turbance recorders.
toring device (bottom) after a short circuit occurred
at the MV side. The two readings match perfectly,
thus validating the TFC monitor sensitivity and accu-
racy.

Fig. 4 Comparison between the disturbance recorder (top) and

TFC monitor (bottom) readings showing perfect matching
Fig. 5 Three-phase short circuit at MV side recorded by the TFC
Monitor
3.2. Results
During the trial period (11 months) the transformer
circuit breaker never tripped, thus no notification
was sent to TERNA. However, 27 through-fault cur-
rents were recorded by the TFC monitor system,
where 7 of them were exceeding 60% Icc. Further-
more, 11 of them were recorded in just 3 days. The
occurrence of such activities would have been un-
known to TERNA.
Figure 5 shows two records detected during the
monitoring period. The first one (top) is a three- Fig. 6 In-rush current
phase symmetrical short circuit. Initially it presents
Figure 7 shows the historical occurrence of the
as a two-phase short circuit and then becomes
through-fault current events, plotting the RMS in
three-phase. The amplitude of the event is about
p.u. of the maximum Icc, while Fig. 8 shows the
900 A RMS which means about 70% of Icc max and
maximum peak in p.u. of the Iccpeak.
the peak is about 30% of the Iccpeak. The second
event (bottom) is a short circuit between two phas- It can be noted that several events have exceeded
es that evolves into a three phase failure. As well as 50% Icc and, in some case, 78% Icc has been
the previous one, the amplitude is about 70% of Icc reached.
and the duration does not exceed 100 ms. Indeed 2
Figure 9 shows the cumulative I t in p.u of the max-
both events are related to short circuits occurring in
imum energy EMAX, considering a maximum with-
the medium voltage network and cleared by the MV
standing duration time of 2 seconds [9-12]. Note
protections before HV protections are activated.
that the cumulative energy had a step increase be-
Fig. 6 shows an inrush current detected by the TFC tween 7 and 9 July due to the highly repetitive oc-
monitor. This is an interesting event in terms of currence. This was likely due to repetitive circuit
electrodynamic forces. It must be noted that such breaker reclosing attempt at the MV side.
kind of events are, of course, not triggered by pro-

4
 Proceedings of My Transfo 2016 – 15th and 16th November 2016
11 events in 3 days
Fig. 7 Ipeak in p.u. of IccMAX
Fig. 8 RMS in p.u. of Icc
Fig. 9 Cumulative I2t in p.u. of the EMAX
3.3. Practical implications and future de-
velopments
TFC monitors can help TERNA to better schedule
electrical test basing on transformer conditions.
SFRA, DC winding resistance, winding reactance and
DGA analysis can be planned when TFC rises an
alarm.
Alarms can be set:
 basing on the amplitude events: as an internal
criterion in TERNA, electrical tests could be
planned whenever an event exceeding 80% Icc is
5
recorded. The trends shown in Fig. 7 and 8 can
be used to set up triggers. An example could be
to give a first alarm when the RMS reaches 0.8
p.u. of Icc and when the IPeak reaches 0.6 p.u. of
IccMAX
 basing on the theoretical maximum withstanding
capability is reached, which means when the
plot of Fig. 9 reaches 1 p.u.
In order to further reduce the man-work and opti-
mize the maintenance, the TFC monitor is being
now integrated into a more complete monitoring
system including bushing monitoring, Partial Dis-
charges and Dissolved Gas Analysis in Oil. In such a
way, all monitored parameters can be sent to
TERNA on event base, which means every time a
short circuit (exceeding a certain trigger) is detect-
ed. Indeed, correlation of multiple on-line data
aimed at providing meaningful information on the
transformer condition is a key feature for the fleet
management.
Furthermore, another interesting application for the
TFC monitor is the inrush current monitoring in HV
shunt reactors, which are increasingly used to con-
trol reactive power in the network.
Indeed, circuit breakers must be perfectly synchro-
nized in order to prevent the occurrence of critical
inrush currents, considering that reactors are
switched on and off every day. As an example, Fig.
10 shows the inrush currents with a faulty synchro-
nizer. It can be observed that current I4 (blue) is not
symmetrical and its peak is higher than the other
two currents.
The TFC monitor can help in detecting such anoma-
lies and track the misalignment over the time, i.e. an
alarm may be generated if current peak is over than
a set value.
Fig. 10 Inrush current with faulty synchronizer on shunt reactors
 Proceedings of My Transfo 2016 – 15th and 16th November 2016
4. CONCLUSION
The experience has demonstrated that transformers
are repetitively affected by a significant amount of
through-fault currents which can induce mechanical
stresses and, on the long term, can critically jeop-
ardize the winding and blocking system integrity
and reliability. Furthermore, it was demonstrated
that the asset owner is, in most cases, unaware of
the electrical and mechanical stresses affecting the
transformer and, thus, enable to schedule correc-
tive maintenance or perform further investigation.
Through-Fault-Current monitoring represents a val-
uable feature for TERNA in order to rank the trans-
formers based on external event occurrences, corre-
late with on-line data (especially DGA) and plan
condition-based maintenance.
References
[1] L. Colla, A. Di Giulio, V. Iuliani, F. Palone, M. Rebolini, “Auto-
transformers evolution, reliability, safety and modelling in
the Italian transmission grid”, CIGRE A2-103, 2012.
[2] F. Scatiggio, M. Pompili, “Health Index: the TERNA’s Practi-
cal Approach for Transformer Fleet Management”, 2013
IEEE EIC, Ottawa, Canada, pp.178-182
[3] W. H. Bartly “ an analysis of InTERNAtional Transformer
Failures”, 79th Annual InTERNAtional Doble Client Confer-
ence, 2012
[4] Anita Oommen, A Case Study Evaluation of the Causes for
the Premature Failure of Transformer on the Escom Trans-
mission Network, Study Committee B5 Colloquium CIGRE,
Canada, 14-16 September 2005.
[5] R. Moxley, A. Guzman, Transformer Maintenance Interval
Management, Schweitzer Engineering Laboratories, Inc.,
2005.
[6] R.P.P Smeets, L.H.te Paske, T. Fogelberg, Short Circuit With-
stand Capability of Large Power Transformers.
[7] K. Yule, D. Brock, J. Purdy, Accountability and Evaluation of
Aggregate Effects of Through Faults On Power Transform-
ers. Unclassified Open Source
[8] M. Tozzi, C. McIlroy, C. A. Serafino, M. Rebolini, “Condition
Based Maintenance on 400 MVA Autotransformers: a case
study”, CIGRE VII Workspot, A2 PS1, Rio de Janeiro, 23-26
Nov, 2014
[9] IEEE Std C57.12.00 -- IEEE Standard General Requirements
for Liquid-Immersed Distribution, Power, and Regulating
Transformers
[10] IEC, Power Transformer – Ability to withstand short circuit,
IEC 60076-5, 2000.
[11] IEEE Guide for Transformer Through Fault Current Duration,
IEEE C57.109-1995.
6
[12] H. Maryono, H.I. Septiyani, M. Muhlis, M.N. Nugraha
“Through Fault Current Monitoring System to Predict the
Degradation of Transformer Withstand Capability”, InTER-
NAtional Conf. on Electr. Eng. and Informatics, 17-19 July
2011, Bandung, Indonesia
Massimo Rebolini joined ITALTRAFO in 1978
and then served ENEL from 1980 to 2000 ini-
tially as expert in Electrical System following
the design and construction of electrical system
of NPP mainly in the qualification class 1E of
electrical components. From 1988 to 1996 was
appointed as Project Manager following the
HVDC Interconnection between Italy and Greece. Afterwards was
appointed in 1997 as responsible of HV Sistems ( Substation and
OHL ) in Enelpower company until 2000 when enter in GRTN
(Italian Independent System Operator) . He was appointed as
Project Manager for Preliminary Project and Permitting Proce-
dures for interconnection 380 kVca between Italy and Switzer-
land ( S. Fiorano – Robbia ) and for SAPEI Project ( new HVDC link
between Sardinia and Mainland Italy). In 2006 was appointed as
Head of Preliminary Projects in Terna Planning and Grid Devel-
opment Direction following all the main strategic works until
2009 when assumed the actual responsibility of Development
Technologies. Author of several national and international tech-
nical papers, he is coordinator member of Electrical Energy
Transmission WG, advisor of AEE ( Associazione Energia Elettrica )
and Chairman of Italian Cigrè.
Claudio Angelo Serafino was born in Rivoli (TO)
on 1961 May the 29th. High School certificate
obtained in Turin on 1980; electrician expert.
Employed in Enel from 1982 to 1999; tests on
High Voltage devices, Transformers, Substation
Automation and Control systems. From 1999 to
2013 employed in Terna SpA maintenance department: Routine
and special tests on High Voltage devices, Transformers, Substa-
tion Automation and Control systems. Since 2013 employed in
Terna SpA engineering department: Routine and special tests on
Transformers; Transformers standardization; fault investigation;
member of CT14.
Marco Tozzi received the M. S. degree in Elec-
trical Engineering from the University of Trieste
in 2005 and the Ph. D. degree in electrical
engineering in 2010 from the University of
Bologna. From 2007 to 2011 was Project Man-
ager and Technical Advisor in Techimp, Italy, being involved in
research activity on diagnostic of insulating systems by partial
discharges analysis. In 2012 he joined Camlin Power Ltd., part of
the Camlin Power group, in Northern Ireland, where he is Product
Manager for transformer monitoring systems. Involved in IEEE
transformer committee, he is author or co-author of more than
30 technical and scientific papers.
Alessandro Salsi received the M. S. degree in
Electrical Engineering from the University of
Bologna in 2007. From 2004 to 2011 was Ser-
vice Engineer and later Service Manager in
Techimp, Italy, being involved in partial dis-
charge measurements and relevant analysis;
after many years on field R&D for sensors and
 Proceedings of My Transfo 2016 – 15th and 16th November 2016

instruments was also a main task. In 2012 he joined Camlin Pow-

er Ltd., part of the Camlin Power group, in Northern Ireland,
where he is Product Designer for transformer and generator
monitoring systems.

Enrico Savorelli received the M. S. degree in

Electrical Engineering from the University of
Bologna in 2007. From 2003 he has been mainly
involved in research activity on diagnostic of
insulating systems by Partial Discharges (PD)
analysis. In 2012 he joined Camlin Power Ltd,
where he is involved in developing PD monitor-
ing systems for Medium and High Voltage asset.