ISA Transactions 47 (2008) 25–31

www.elsevier.com/locate/isatrans

Automation of the temperature elevation test in transformers with

insulating oil
José Manuel Esteves Vicente, Angelo José Junqueira Rezek ∗ , Antonio Tadeu Lyrio de Almeida,
Carlos Alberto Mohallem Guimarães
UNIFEI/ISEE – Federal University of Itajubá, P.O Box: 50, Zip Code 37500-903, Itajubá, MG, Brazil

Received 24 June 2005; accepted 1 June 2007

Available online 24 August 2007

Abstract

The automation of the temperature elevation test is outlined here for both the oil temperature elevation and the determination of the winding
temperature elevation. While automating this test it is necessary to use four thermometers, one three-phase wattmeter, a motorized voltage variator
and a Kelvin bridge to measure the resistance. All the equipments must communicate with a microcomputer, which will have the test program
implemented. The system to be outlined here was initially implemented in the laboratory and, due to the good results achieved, is already in use
in some transformer manufacturing plants.
c 2007, ISA. Published by Elsevier Ltd. All rights reserved.

Keywords: Transformers; Automated tests; Temperature elevation

1. Introduction
In the past, the temperature elevation test was done only
in the first prototypes from the manufacturers because there
were only a few electrical power utilities, on the purchase.
This way, despite being a long and cumbersome test, as it was
seldom done, it did not payoff the investment to be automated.
A better option was to address it to an accredited laboratory on
request with follow up from a manufacturer representative and
a purchaser inspector.
Another relevant factor is that the older transformers, mainly
the distribution ones, that had to comply to standards in force
with respect to their losses, had their design completed with
older techniques and tooling, resulting in a bigger and heavier
core than the ones obtained with state-of-the-art techniques.
This way, for a certain power the older transformers are
bigger than the current ones; as the allowed losses values almost
do not change, the current transformers do not dissipate the
∗ Corresponding author. Tel.: +55 3536291377x1180;
3536291380x1187.
E-mail addresses: manuel@unifei.edu.br (J.M.E. Vicente),
rezek@unifei.edu.br (A.J.J. Rezek).
heat generated internally so easily. Thus the design of the tank
and the radiators became more critical. So, the manufacturer
must perform more temperature elevation tests, that justifies the
implementation of this test automation at the laboratory.
2. The classic test
The temperature elevation test is made up of two distinct
parts; the temperature elevation top of oil and the temperature
elevation of the windings [1].
Through the routine tests previously performed, the value
of the total losses is calculated; this value must be maintained
constant throughout the test of temperature elevation top of oil.
By beginning the test, the temperature on top of the oil
and at least three ambient temperatures must be measured,
with thermometers placed around the transformer so that, at
regular time intervals, the operator notes down the temperature
values, checks if the total losses value is correct and calculates
the temperature gradient of the oil concerning the ambient
fax: +55 temperatures measure.
The test is considered stable when the temperature elevation
top of oil has not varied more than 1 ◦ C for the last three hours.

c 2007, ISA. Published by Elsevier Ltd. All rights reserved.

0019-0578/$ - see front matter

doi:10.1016/j.isatra.2007.06.002
26 J.M.E. Vicente et al. / ISA Transactions 47 (2008) 25–31
The value of the temperature elevation top of oil must be less
than or equal to the values tabulated according to the standard
NBR3356 [2].
The second part of the test consists in circulating the
rated current in the windings for one hour, switching off,
disconnecting the cables, connecting the Kelvin bridge in a
winding and making a set of readings of resistance and their
respective times within a time frame of four minutes at most.
With the pairs resistance x times it is possible to make a plot
and by making a logarithm extrapolation for t = 0, to obtain
the winding resistance value on switching off.
This way, on switching off, the average temperature of a
winding, through the method of the resistance variation, is
determined by the equation:
Rhot (t = 0)
Thot = (K + Tcold ) − K (1)
Rcold
where:
K = 234.5 for copper and 225 for aluminum.
This test introduces two main setbacks: the first is the
constant need for an operator always paying attention to the
value of the total losses applied to the test subject and correcting
it if necessary, besides reading the temperatures and noting
them down. The second setback and maybe the most critical one
is usually related to the disconnection, where experience and
skill is required for handling and reading of the measurement
equipment.
3. Test automation
So as to have this test automated it is necessary to
use a personal computer and equipment to measure active
power, temperature and ohmic resistance with features of
communication with the personal computer.
3.1. Test procedure
The purpose of the temperature elevation test is to determine
the temperature elevation gradients both for the oil top and
the windings. The gradient obtained at the end of the test
must be smaller than the maximum allowed in the standard—
NBR5380/93 and IEEE C57.12.90.
The test procedure consists in short-circuiting a winding,
generally the low voltage one and applying voltage to the other
winding, in this case the high voltage one until it reaches the
total losses value (no-load losses + load losses) previously
calculated for the smaller tap.
This condition of total losses applied to the transformer
must be kept until thermal stability occurs between this one
and the environment temperature. After this balance occurs,
the temperature elevation test of the oil top is concluded. The
gradient obtained from this test, for mineral oil, by standard,
must be small and equal to 50 ◦ C and of 55 ◦ C for the windings,
these values being the most commonly used.
Some advantages of the automation methodology used are:
(a) Since the power control is precise, the test tends to stabilize
more quickly;
(b) The test can be shortened by using the control option
allowed by the standard NBR5380/ANSI by using an initial
overload of 30% (approximately for 5 straight hours);
(c) The test does not need an operator following it at all times.
The test is tiring for the operator, because he has to control
the power and note the temperatures every 15 min. The test
usually lasts for 15 h;
(d) As the test workbench is exclusive for this test procedure,
it is interesting to be performed at night and the automatic
performance frees the operator from tiring test procedures;
(e) Every 15 min the operator has to complete a spreadsheet
with measured temperatures, power, voltage, and current;
on the other hand, in the automated mode, these procedures
are all done by equipment and descriptive control methods.
3.2. Temperature elevation of top of oil
The control program initially commands the closure of the
motorized varivolt relay, making the voltage to elevate until
the active power read from the wattmeter is the same as the
power of the total losses previously calculated and inputted
in the program. If the active power is higher than the desired
one, a closure command will be issued to another relay which
will make the voltage to decrease. This process assures that the
value of the total losses remains constant throughout the entire
test. At the same time, temperatures are also measured and the
temperature elevation of top of oil is calculated.
This way, the control program cycles through the process
outlined, storing all the measured and calculated data until the
thermal balance occurs, that is, until the temperature elevation
of top of oil is smaller than 1 ◦ C in the last three hours.
3.3. Temperature elevation of the windings
After the thermal balance takes place, the temperature
elevation of the windings must be determined. The method
used is the resistance variation, which is the temperature
determination by comparing the electrical resistance of the
winding in temperature to be determined with its resistance at a
known temperature [1].
The known resistance and temperature is obtained during
routine tests and is technically called cold resistance, that is,
measured at ambient temperature before any test is begun.
For this second part of the test, the rated current of the
windings must be circulated for at least one hour. This way,
now the control program adjusts the voltage variator voltage
until the current is equal to the desired one. This process is
repeated for one hour, after which it is necessary to disconnect
the transformer source supply and to connect the Kelvin bridge
in a winding.
The control program, in this point of the test, while
communicating with the Kelvin bridge, makes a set of readings
of ohmic resistance of the winding selected, storing them with
their respective times, always within the time frame of four
minutes.
For the other windings, the test is repeated always applying
the rated current for one hour.
 J.M.E. Vicente et al. / ISA Transactions 47 (2008) 25–31
Fig. 1. Equipments used in the test.
For the determination of the ohmic resistance value, on
switching off, an exponential regression is done by the least
square method [1].
4. Experimental results
4.1. Equipments used
The following equipments were used for the test automa-
tion:
(a) Three-phase digital wattmeter with the following character-
istics (Yokogawa WT 130):
– Three voltage channels from 0 to 350 V each;
– Three current branches from 0 to 5 A each;
– Four inputs for the thermometers from 0 to 100 ◦ C
(thermometer PT-100 output 4–20 mA);
– Two independent relays with contacts usually open;
– Communication with microcomputer through serial RS-
232.
(b) Three-phase digital “Kelvin Bridge” with the following
characteristics:
– Six input channels, “four wires each”, three for the high
voltage bushings and three for the low voltage bushings;
– Measurement range from 0.001 m to 1000 ;
– Serial RS-232, for communication with the microcom-
puter.
This equipment reads the individual resistance of each
channel, according to the control program request. Its
internal source, on turning on and off varies the voltage
ramp-like to minimize the inductive effects in the windings.
It also has protection at the inputs against voltage surges.
(c) Measurement table, containing basically three potential
transformers and three current transformers, with several
taps.
(d) Advantech PCL 711 data acquisition card (4 analog inputs
4–20 mA, 2 digital outputs to relay up/down voltage).
(e) Motorized voltage variator.
Fig. 1 shows the equipments used in the test, where: T1, T2,
T3, T4—Thermometers, T1 (top of oil), T2, T3, T4 (localized
around the transformer, 1–1.5 m, half height).
27
4.2. The “software”
The control software, called eleva.exe, was developed by
using the Turbo Pascal programming language and it has
approximately twenty thousand programming lines.
In the loop of the program the measurement is made every
250 ms. However, if necessary, the setting of the active power
is done every minute, if it is out of ±2% of the adjusted value.
The average of the measurement values (voltage, current, active
power, and temperatures—one for the top of oil and three others
for the environment), are kept in file, every 5 min.
Initially the program was developed in turbo pascal, due
to the fact that the first author has already done some works
using this language, and his experience in this language is good.
Nowadays there is a version in test, running in Delphi, with the
same functions. However, in the case of interest, the program
can be written in C++.
It was developed in the 7.0 version of turbo pascal. The
serial communication routines have been written in assembler
language, due to the experience of the author in serial
communication protocol and also in this referred language. The
first author has himself developed the communication routines
in the assembler.
In terms of speed, the measured variables are updated
approximately every 250 ms. As the test demands a long time,
in the range of 12–15 h, the automation of that is of great
interest, due to the fact that the possibility of liberation of the
operator from the work of test verification, and data acquisition
(power (P), voltage (V ), current (I ), and temperatures) at
maximum every 15 min.
The main functions and features implemented are the
following ones.
Fig. 2 shows the flowchart of the developed program.
4.2.1. Input for test
This option allows to input a transformer in the software
database to perform the test. It can be done either automatically
or manually. In the last case, the operator is requested to type
all the data necessary to the test such as kind of connection,
power, HV (high voltage), LV (low voltage), total losses, cold
resistance values with the respective temperatures and also time
in seconds that the Kelvin bridge took to stabilize the cold
resistance readings.
In the automatic mode the program searches all the
information necessary to the input of the test in the
TRANS4.EXE program database, which is used for the routine
tests management.
4.2.2. Test of elevation of the oil temperature
As the name implies, this option is used to perform the first
part of the test. However, it still allows the operator to do the
test in three distinct ways, namely:
– Standard test, with total losses from the onset;
– Test with initial overload of 30% for five hours;
– Test with reduced total losses of 20%.
In the second case, the test is abridged to some hours, due to
initial overload.
28 J.M.E. Vicente et al. / ISA Transactions 47 (2008) 25–31

Fig. 2. Flowchart of the developed program.

The third case is allowed by standard and it is used when the $o = Total losses in W
power source applied to the transformer is not enough for it to $or = Reduced losses in W.
reach the value of the total losses calculated.
In this case, the following correction is applied by the 4.2.3. Test of elevation of the winding temperature
program [1]: This option of the program allows to apply the rated current
0,8 of the windings for one hour, at the end of which we can
$o

∆θo = ∆θr (2) proceed to the hot resistance measurement.
$or
where: 4.2.4. Measurement of the hot windings resistance
∆θo = Elevation of the oil temperature over the coolant In this option the program, using the Kelvin Bridge, makes
temperature, with total losses in ◦ C; ten readings, with their respective times in the windings
∆θr = Elevation of the oil temperature over the coolant selected. One or two windings can be read in the same
temperature, with reduced losses in ◦ C; disconnection respecting the limit of four minutes.
 J.M.E. Vicente et al. / ISA Transactions 47 (2008) 25–31
Fig. 3. Test arrangement.
Fig. 4. Temperature elevation without initial overload.
For the other windings it is necessary to repeat the test
outlined in Section 4.2.3.
For ease and to take the least possible time in the
disconnection of the transformer it is interesting to prepare
a fast commutation system between the transformer and the
Kelvin bridge. Fig. 3 illustrates a fast commutation option.
During the temperature elevation test, switch SH1 is closed,
maintaining the secondary in short-circuit and the switch SH2
is turned to the position A, feeding the transformer. The
commutation is done in the following sequence:
(a) The voltage variator is disconnected and a chronometer is
manually triggered;
(b) Switch SH2 is commutated from A to B;
(c) Switch SH1 is opened;
(d) The program is allowed to begin reading in the windings
selected and the chronometer is stood still. The program
later will request the time recorded in seconds. If only the
reading of a winding was selected, the ten readings will
be made along the four minutes time frame allowed. If the
29
Fig. 5. Temperature elevation with initial overload.
option selected was reading in two windings, ten points are
initially read in the LV’s winding, in the interval of one
minute and ten more readings in the HV’s winding in the
remaining interval of three minutes.
It is important to comment that the measurements of cold
resistance were made with the same Kelvin bridge, by using a
specific program for that.
This program besides making the readings of the ohmic
resistances of the windings, also recorded the stabilization time
of the inductive effect.
In the test of measurement of the hot resistance, this time is
used to guide the beginning of readings.
4.2.5. Calculations and reports
This option allows the test results to be seen on the screen of
the computer and to be printed. These test results are:
– Plot of the temperature elevation along the time without
initial overload (Fig. 4) and with overload (Fig. 5). In Figs. 4
30 J.M.E. Vicente et al. / ISA Transactions 47 (2008) 25–31
Table 1
Winding temperatures with respect to the oil and the winding with respect to the ambient temperature
Fig. 6. RH1H2, RH2H3 and RH1H3 versus time.
and 5 one has temperature elevation (top curves) and total
losses (bottom curve).
Plot containing the pairs of points, resistance and time,
and the extrapolation curve for t = 0 (Figs. 6 and 7).
– The values calculated for the winding temperatures with
respect to the oil and for the winding with respect to the
ambient temperature are shown in Table 1.
5. Conclusion
The automation process for the temperature elevation test
has shown to be very useful and practical because it contributes
to ease and standardize two critical points of this test, which
are: long stabilization follow-up for the temperature top of oil
and for disconnection and hot resistances measurement.
With this system some manufacturers start the test right after
office hours and it is completed the next morning, ready to make
the hot resistance measurement.
With the use of the initial overload option, the test can be
started in the morning and finished the same day.
Besides the program eleva.exe, the authors have developed
all the measurement equipment: three-phase digital wattmeter,
Fig. 7. RX1X2, RX2X3 and RX1X3 versus time.
three-phase Kelvin bridge, and controlling circuit for the
voltage variator and temperature reading. Thus a maximum
interaction of the set has been achieved.
The program, however, can be adapted to control other
similar equipments from the manufacturer, provided they have
communication serial interface. So, for other equipments it is
possible to adapt the described work. For that it is necessary to
know the commands of the serial interface or GPIB.
Several Brazilian manufactures are using the described au-
tomated equipment for transformer test, such as: Romagnole—
PR; ITD, Birigui—S.P.; Trafo, Ortolândia—S.P.; Cemec,
Fortaleza—CE; Itaipu, Itápolis—S.P.; Vijai, Aracaju—SE.
Electricity Companies are also using the automated equip-
ment, such as: CEMIG—M.G.; CELPE—P.E.; Enersul—R.S.;
Energipe—SE; Celg—GO [3–11].
Acknowledgement
The authors would like to thank very much our dear
colleague and friend Ronaldo Ivo Firmo of Alcoa (Poços de
Caldas – MG, Brazil), for the important help done by him to
us, in providing the elaboration of the version of the paper
 J.M.E. Vicente et al. / ISA Transactions 47 (2008) 25–31 31

in english, that has been approved to be published in ISA [6] ANSI C57.12.00. Distribution, power and regulating transformers.
transactions. [7] ANSI C57.12.20. Requirements for overhead type distribution transform-
ers 67.000 volts and below 500 KVA and smaller.
References [8] ANSI C57.12.90. Test code for distribution, power and regulating
transformers.
[1] Technical note ABNT NBR5380. May 1993. [9] de Oliveira JC, Cogo JR, de Abreu JPG. Transformers, theory and tests.
[2] Technical note ABNT NBR3356. August 1993. Edgard Blücher Ltda.; 1998 [in Portuguese].
[3] Technical note ABNT NBR5440. July 1999. [10] Ruggiero MAG, Lopes VLR. Numerical calculation: Computational and
[4] Technical note ABNT NBR10295. May 1988. theoretical aspects. 2nd ed. Makron Books; 1997 [in Portuguese].
[5] Technical note ABNT NBR 5440.-PB(99) Distribution Transformers – [11] Vicente JME. A contribution to the automation of tests in medium voltage
Standardization. power transformers. Ph.D. thesis. UNIFEI, Brazil; 2006.