See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/335193985

Stirring effect in dielectric liquids breakdown voltage determination

Conference Paper · June 2019

DOI: 10.1109/ICDL.2019.8796653

CITATIONS READS

0 53

6 authors, including:

Luigi Calcara S. Sangiovanni

Sapienza University of Rome Sapienza University of Rome
30 PUBLICATIONS   134 CITATIONS    34 PUBLICATIONS   217 CITATIONS   

SEE PROFILE SEE PROFILE

Francesco Muzi
Università degli Studi dell'Aquila
74 PUBLICATIONS   517 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Artificial Intelligence for Smart Energy Management View project

All content following this page was uploaded by Luigi Calcara on 11 November 2019.

The user has requested enhancement of the downloaded file.

 2019 IEEE 20th International Conference on Dielectric Liquids (ICDL), Roma, Italy, June 23-27, 2019

Stirring effect in dielectric liquids breakdown

voltage determination
M. Baur J. Knauel L. Calcara,
BAUR GmbH BAUR GmbH University of Roma “La Sapienza”
Sulz, Austria Sulz, Austria Astronautic, Energetic and Electrical Engineering Dept.
m.baur@baur.at jens.knauel@baur.at luigi.calcara@uniroma1.it

M. Pompili S. Sangiovanni F. Muzi

University of Roma “La Sapienza” University of Roma “La Sapienza” University of L’Aquila
Astronautic, Energetic and Electrical Astronautic, Energetic and Electrical Industrial and Information Engineering
Engineering Dept (Roma, Italy). Engineering Dept. (Roma, Italy). and Economics Dept. (L’Aquila, Italy)
massimo.pompili@uniroma1.it silvia.sangiovanni@uniroma1.it francesco.muzi@univaq.it

Abstract— The latest version of the standard IEC 60156 According to the IEC 60156 the 50-60 Hz breakdown
“Insulating liquids - Determination of the breakdown voltage at voltage mean values are obtained as average of six
power frequency - Test method” was published in August 2018 consecutive breakdowns. At the time of the preparation of the
as Edition 3.0. This standard follows the two previously IEC 60156 in 1995, this simple method was selected for
published versions in 1963 and 1995. The test method described producing quick results with a limited quantity of the
in practically all IEC 60156 versions defines a procedure in insulating liquid. Anyway, the obtained results present
which the insulating liquid is subjected to an increasing AC standard deviations which may be quite high, normally in the
electrical field until breakdown occurs and it is used for both range of 15-20 % or greater [5-9].
acceptance testing on unused liquids at the time of their delivery
and for establishing the condition of samples taken in Studies carried out in the past [10-15] have shown that
monitoring and maintenance of equipment. In this paper the dispersion values up to 30% can be also considered acceptable
experimental results of 50 Hz breakdown voltages of both as also shown in Fig. 1 [1-2]. This picture depicts as the scatter
mineral and natural ester liquids performed following the tests (coefficient of variation as ratio between standard deviation
methods described in the new edition of IEC 60156 are shown. and mean breakdown voltage) of breakdown voltages may be
The tests were performed also with the use of different stirring strongly dependent on the values of the result. The graphical
systems. representation indicates the mean values of breakdown
voltage and the coefficients of variation as they have been
Keywords—insulating liquids, mineral oils, natural esters,
found in a large number of tests in several laboratories using
standardization, breakdown voltage
transformer liquids.

I. INTRODUCTION
Insulating liquids are used in power electrical apparatus
(e.g. power transformers, switchgear, measurement
transformers, etc.) with performance and safety requirements
increasingly stringent. For manufacturers and users of such
electrical equipment, the dielectric strength of the same
insulating liquids is a useful property for checking their
quality and the presence of particles and moisture.
It is well known that the electrical breakdown in insulating
liquids follows probabilistic laws and, consequently, any
possible test for its evaluation provides results characterized
by mean values with related scatters. Moreover, these
statistical dispersions also depend on the test conditions and
measurements setup.
The method for the determination of the breakdown Fig. 1. Graphical representation of coefficient of variation versus mean
breakdown voltage [1-2].
voltages of the insulating liquids at power frequency is Legend: the solid line shows the distribution of the coefficient of variation
describes in IEC 60156. The first edition of this standard has as a function of the mean breakdown value. The dotted lines indicate the
been issued in 1963, revised in 1995 and in 2018 the third expected 2.5 % to 97.5 % range of values of Standard Deviation (SD)/mean
edition has been released [1-2]. ASTM has two similar value as a function of the value of the mean breakdown voltage.
standards D1816 (2014) and D877 (2013) [3-4]. All these
three standards differ in the test procedure and experimental The presence and/or formation of air bubbles in the inter-
set-up and have certain limitations in terms of scatter of mean electrode space is believed to be one of the main reasons for
value of the breakdown voltage. higher dispersion in the determination of the medium value of

978-1-7281-1718-8/19/$31.00 ©2019 IEEE

1
the electrical discharges [11, 12]. Also, other parameters - like
electrode configuration, test procedures, number of the shots,
rising time, oil gassing tendency, etc. - play important roles in
the determination of the breakdown voltages at power
frequency [12, 14-20].
The 2018 revision of IEC 60156 mainly confirmed the
content of the previous edition even if some advances have
been included. The test method has not been changed for
practical reason due to the very large number of
(c)
instrumentations disseminated around the world, even if the
use of stirring is now recommended to allow the air bubbles
dissolution. Moreover, an alternative method is also reported
in one informative annex [2].
In this paper the experimental results of 50-60 Hz
breakdown voltages of both mineral and natural ester liquids
performed following the new edition of IEC 60156 [2] using
or not a stirring system having different features are shown.
(d)

II. EXPERIMENTAL TESTS, RESULTS AND DISCUSSION Fig. 2. Schematic representation of the test cells used.
Legend: (a) test cell without stirrer, (b) test cell with mechanical stirrer on
The breakdown voltages determination was carried out top, (c) test cell with mechanical stirrer on bottom, (d) test cell with magnetic
following the IEC 60156 Ed. 3.0 test method. Also, the tests stirrer on bottom.
were repeated following the innovative method describes in
the Annex A of the same standard believed to be able to reduce
the scatter of the results of breakdown voltage [2]. This The comparison was carried on using two natural esters
method is based on the application of pre-energizing time and (NE1 and NE2) and one mineral oil (MO). Table I reports the
on a larger number of shots (10) for the determination of the main physical and electrical characteristics of these insulating
breakdown voltages. liquids under test.
In particular, the tests have been performed both without
and with different types of stirring systems and following
TABLE I. MAIN CHARACTERISTICS OF THE INSULATING LIQUIDS
different procedures as represented in Fig. 2. The stirring UNDER TEST
systems, both mechanical and magnetic, consisted of plastic
propeller or bar mounted at the top or at the bottom of the cell. Mineral oil Natural ester Natural ester
(MO) (NE1) (NE2)
The adopted stirring systems were activated before the
Viscosity at
beginning of each test. In all the tests, the first application of 40 °C 9,6 37,0 - 43,0 32 - 42
the voltage was started at least 10 minutes after filling the cell (mm2/s)
also checking the absence of air bubbles into the liquid. Density at 20
0,875 0,910 - 0,925 0,92
°C (kg/dm3)
Pour point
-51 -15 - (-27) -18 - (-23)
(°C)
Flash point
152  300  300
(°C)

The oil tester used was able to produce an applied

increasing voltage of 2 kV/s up to 100 kV. Fig. 3 shows the
rising voltages applied during the tests depending on the
(a) method used:
(a) procedure described in IEC 60156, which consists of an
applied increasing voltage of 2 kV/s until the breakdown
occurs applied for six times on the same liquid sample with
a resting time between shots of at least 2 minutes;
(b) procedure described in the Annex A of the IEC 60156,
which consists of an applied increasing voltage of 2 kV/s
until breakdown occurs with 10 seconds pre-energizing
time at 10 kV. The shots were repeated for ten times on the
same sample with a resting time between shots of at least
(b) 2 minutes and disregarding the two highest and the two
lowest results. The mechanical stirring system was
mounted on the top of the cell.

2
 TABLE II. MEAN VALUES OF BREAKDOWN VOLTAGES AND RELATED
STANDARD DEVIATIONS OBTAINED USING BOTH THE METHOD DESCRIBED IN
THE IEC 60156 (2018) WITH DIFFERENT STIRRING SYSTEMS AND THE
METHOD DESCRIBED IN ITS ANNEX A.
LEGEND: MINERAL OIL (MO), NATURAL ESTER 1 (NE1), NATURAL ESTER 2
(NE2) - #1, #2 AND #3 SEQUENCE OF DIFFERENT TRIALS.

# Top Bottom Magnetic

Sample No stirrer Annex A
stirrer stirrer stirrer
Mean #1 64,9 65,4 62,2 61,8 61,7
value #2 63,4 57,9 59,3 56,9 57,1
[kV] #3 59,9 59,8 61,8 59,2 59,6
(a) MO
#1 17,1% 15,6% 10,2% 14,8% 12,2%
Std. dev.
#2 20,0% 19,7% 14,6% 13,6% 7,5%
[%]
#3 13,9% 14,9% 15,2% 15,9% 10,6%

Mean #1 76,7 81,3 78,1 78,7 72,0

value #2 65,2 76,6 72,0 74,0 67,5
[kV] #3 68,3 78,5 72,9 76,6 70,5
NE1
#1 9,8% 11,2% 10,5% 13,2% 3,1%
Std. dev.
#2 22,9% 13,3% 15,9% 9,2% 3,5%
[%]
#3 5,8% 18,4% 10,9% 13,6% 5,7%

Mean #1 85,1 86,1 92,2 84,7 81,5

value #2 68,8 83,3 88,2 84,0 79,2
[kV] #3 77,1 81,1 90,0 82,5 73,6
(b) NE2
#1 13,0% 5,3% 7,2% 9,4% 8,0%
Std. dev.
#2 22,6% 5,4% 6,1% 3,9% 8,5%
[%]
Fig. 3. Applied rising voltages for the determination of the breakdown
#3 16,8% 8,1% 10,9% 5,7% 8,8%
voltage in insulating liquid.
Legend: (a) 2 kV/s until the breakdown occurs (IEC 60156); (b) 2 kV/s until
the breakdown occurs with a 10 seconds pre-energization time at 10 kV (IEC
60156 Annex A). For both (a) and (b), the initial resting time and times 25% 100
between the shots were set at least 10 minutes and 2 minutes respectively. 90
The mean values of the breakdown voltages and related 20% 80
Standard deviation %

standard deviations obtained using these different stirring 70

Mean value kV
systems and procedures are reported in Table II. Fig. 4 reports 15% 60
the comparison of the standard deviations of the breakdown 50
voltages repeating three time each test procedure; in addition, 10% 40
in the same figure also the mean values of the breakdown 30
voltages are also reported. From the analysis of the graphs of 5% 20
Fig. 4 it is possible to note as the use of a stirring system helps 10
to minimize the standard deviations of the breakdown voltages 0% 0
especially when the same scatter in the results is appreciably no stirr. top bottom magnetic annex A
higher. In fact, the stirring system removes decomposition
products and air bubbles formed in the insulating liquid
between the electrodes during the antecedent breakdown. It #1 #2 #3 Mean value kV
was noted that the effectiveness of the stirring systems is
higher when the liquid flow between the electrodes is in
upwards direction; this allows an easier removal of gas (a) Mineral oil
bubbles. The present results show also similar efficiency of
the different stirring mechanisms considered in the present 25% 100
experimental work. 90
Standard deviation %

The results obtained following the test method described 20% 80

Mean value kV

70
in the Annex A of the IEC 60156 (2018), carrying out 10
15% 60
breakdowns on the same liquid and calculating the test results 50
as the average of the remaining six results after disregarding 10% 40
the four dispersed results, have shown significant reductions 30
of the standard deviation, as clearly depicted in Fig. 4. 5% 20
10
The mineral oil (MO) breakdown voltages were found to
0% 0
be slightly lower (with higher standard deviation) than those no stirr. top bottom magnetic annex A
of natural esters (NE1 and NE2), confirming other antecedent
studies [6]. This behavior does not represent an intrinsic
#1 #2 #3 Mean value kV
property of the insulating liquids themselves but it may be
related to the higher moisture content of mineral oil used in
the present experimental work. (b) Natural ester 1

3
 25% 100 sincere appreciation to BAUR GmbH, Austria, for the ac
90
breakdown voltage tester adjustments. The suppliers of the
liquid dielectric samples are also thanked.
20% 80
Standard deviation %

70

Mean value kV
15% 60
REFERENCES
50
[1] IEC 60156, “Insulating Liquids  Determination of the breakdown
10% 40 voltage at power frequency – Test Method”, Ed. 2.0, 1995
30 [2] IEC 60156, Insulating Liquids  Determination of the breakdown
5% 20 voltage at power frequency – Test Method”, Ed. 3.0, 2018
10 [3] ASTM 1816, “Standard test method for dielectric breakdown voltage
of insulating oils of petroleum origin usig VDE electrodes”, 2004
0% 0
no stirr. top bottom magnetic annex A [4] ASTM 877, “Standard test method for dielectric breakdown voltage of
insulating liquids using disk electrodes”, 2013
[5] M. Baur, M. Pompili, R. Bartnikas, "A comment on the test methods
#1 #2 #3 Mean value kV for the breakdown voltage of dielectric liquids", IEEE Transaction on
Dielectrics and Electrical Insulation, vol. 19, p. 1482-1484, 2012
[6] M. Baur, L. Calcara, M. Pompili, “Scatter reduction of the 50-60 Hz
(c) Natural ester 2 breakdown voltage test for insulating liquids”, IEEE Transactions on
Dielectrics and Electrical Insulation, Vol. 22, pp 2401-2407, 2015
Fig. 4. Comparison between the standard deviations of three trials of
mineral oil (MO), Natural Ester 1 (NE1) and Natural Ester 2 (NE2) when [7] M. Baur, J. Knauel, L. Calcara, M. Pompili, "Insulating Liquids
different stirrer types are adopted (top mechanic, bottom mechanic and Breakdown Voltage Determination: Test Method Efficiency",
bottom magnetic) and the mean value of the breakdown voltages of the International Conference on Dielectric Liquid (ICDL), Volume 2017-
January, Pages 1-4, Manchester (UK), 2017
same insulating liquids.
Legend: [8] L. Calcara, M. Pompili, S. Sangiovanni, M. Baur and J. Knauel,
(a) Mineral oil (MO); (b) Natural Ester 1 (NE1); (c) Natural Ester 2 (NE2); “Standard Evolution for the Determination of the Power Frequency
#1, #2 and #3 sequence of different trials. Breakdown Voltages in Insulating Liquids”, IEEE Conference on
Electrical Insulation and Dielectric Phenomena (CEIDP), 2018
[9] L. Calcara, S. Sangiovanni, M. Pompili, “Standardized methods for the
determination of breakdown voltages of liquid dielectrics”, IEEE
III. CONCLUSION Transactions on Dielectrics and Electrical Insulation, Vol. 26, Issue 1,
This paper presented experimental results of power pp. 101-106, 2019
frequency breakdown voltages of both mineral and natural [10] R. Bartnikas, Editor, Engineering Dielectrics, Vol. III, Electrical
Insulating Liquids, Monograph 2, ASTM, Philadelphia/West
ester liquids performed following the tests methods described Conshocken, 1994 (www. astm.org/digitallibrary)
in the new edition (2018) of IEC 60156. The tests were [11] W. Lick, GJ. Pukel, H.M. Muhr, M. Baur, "New test method for
performed also with the use of different stirring systems. dielectric breakdown voltage of insulating oils", Proceeding IEEE
ICDL, France, 2008
The main conclusions can be summarized as in the
[12] G. J. Pukel, “Elektrische Festigkeit von Isolieröl”, PhD work at TU of
following: Graz Austria, OVE (Wien), 2010
- the use of a stirring system allows a slightly reduction of [13] E.O. Forster, “Progress in understanding of the electric breakdown in
the breakdown voltage scatters. This reduction is more condensed matter”, J. Phys. D, Vol. 33, pp. 1506-1514, 1990
appreciable when the breakdown voltages are lower, [14] W. Pfister, J. Dietrich, “Influence of stirring on the breakdown of
insulating liquids”, IEC TC 10 WG4 report, 1990
maybe in consequence of presence in the insulating liquid
of contaminants, particles, sludge and moisture; [15] W. Lick, G. J. Pukel, M. Muhr, “Improvement of the Test Method for
Insulating Oils”, IEEE International Symposium on Electrical
- mechanicals and magnetic stirring systems present similar Insulation, 2006
efficiency in the homogenization of the insulating liquids [16] M. Baur, M. Pompili, “Pre-energizing effect on breakdown voltage
test for insulating liquids”, Proceedings of the 2014 IEEE 18th
under test. The effectiveness of the stirring systems is International Conference on Dielectric Liquids, ICDL 2014, 2014
higher when the liquid flow between the electrodes is in
[17] E. Forster, H. Yamashita, C. Mazzetti, M. Pompili, L. Caroli, S.
upwards direction; this allow an easier removal of air Patrissi, “The effect of the electrode gap on breakdown in liquid
bubbles; dielectrics”, IEEE Transactions on Dielectrics and Electrical
Insulation, Vol. 1, Issue 3, pp. 440-446, 1994
- the use of the innovative test method described in the [18] E.O. Forster, C. Mazzetti, M. Pompili, R. Cecere, “The effect of
Annex A of the IEC 60156 Ed. 3.0 contributes to minimize molecular structure on the properties of dielectric fluids”, IEEE
the scatter of the results of breakdown voltages. transactions on electrical insulation, 26 (4), pp. 749-754, 1991
[19] L. Calcara, M. Pompili, F. Muzi, “Standard Evolution of Partial
Discharge Detection in Dielectric Liquids”, IEEE Transactions on
Dielectrics and Electrical Insulation, Vol. 24, Issue 1, 2017
ACKNOWLEDGMENT
[20] M. Pompili, C. Mazzetti, M. Libotte, “The effect of the definition used
The authors gratefully acknowledge the financial support in measuring partial discharge inception voltages”, IEEE transactions
for this project by the Italian Ministry of University, on electrical insulation, 28 (6), pp. 1002-1006, 1993
Scientific, Technological Research and wish to express their

4
View publication stats