1040 IEEE Transactions on Dielectrics and Electrical Insulation Vol. 8 No.

6, Dec(�mber 2001

Fundamental Investigations on Some

Transformer Liquids under Various
Outdoor Conditions
I. Fofana
Research Group on Atmospheric Ici ng
Universite du Quebec a Chicoutimi
Quebec, Canada

H. Borsi and E. Gockenbach

Institute of Electric Power Systems
Division of HV Engineering
Schering·Institute
University of Hanover, Hanover, Germany

ABSTRACT
This paper deals with the humidity uptake of some transformer liquids and its influence on
the electrical breakdown. The investigations were performed under some selective outdoor
conditions on different insulating liquids, e.g. mineral oil, ester and silicone used in HV trans­
formers. The temperature was at 23 or 60"( while the relative air humidity was varied between
20 and 90% to simulate climatic variation. The subject is important in a kind of HV transformer
w ith an air breathing system and, even if in such transformer, where normally the air passes
through a dehydrating apparatus with silica gel, a bad run of this apparatus can induce an
accident. The scenario considered corresponds to such a defect in the breathing filter system,
and thus when the insulating liquid in the transformer becomes in contact with humidity of
atmospheric air. The work presented is in two parts. The first part relates the evolution of
the breakdown voltage in terms of the insulating liquid humidity content. The second one
describes the evolution of insulating liquid humidity content in terms of surrounding relative
air humidity. This evolution is observed during 30 days and the results are discussed regard­
ing the ac electric strength according to International Electrotechnical Commission (lEe) or
Verband der Elektrotechnik Elektronik lnformationstechnik e. V. (VDE) standards. A correla­
tion between the surrounding relative air humidity, temperature and humidity uptake rate is
derived.

1 INTRODUCTION down. Water content increases electric conductivity, di:lsipation factor

RANSFORMER life/ aging is related mainly to the degradation of the
T insulation, caused dominantly by thermal degrading of the insu­
lating paper, together with the decomposition of the paper. The by­
pro ducts are water and other substances of par tly polarizable and ion­
izable character. For example, the life of insulating cellulose paper at
11 0"( is calculated as ten years [1]. Aging depends not just on loading,
but also is influenced significantly by the type of paper, pulp compo­
sition, humidity and oxygen content as well as the acidity level within
the insulating liquid [2].
Water is par ticularly detrimental to dielectric properties of both liq­
uid and paper insulation systems and thus to their resistance to ag­
ing. Moisture in insulating liquid may under fast decreasing temper­
ature transients, result in free water that can lead to electrical break-.
and worsens the electric strength. A mass transfer process of water
results from the equil ibrium imbalance, in which at higher tempera­
tures moisture leaves the paper to enter the liquid. Water in the liquid
originates from air moisture in the case of 'open-breather' or 'liquid­
conservator' types, and the thermal decomposition of cellulose-based
solid insulating materials. It has a strong influence on the life ex­
pectancy and load capacity of a transformer.
The insulation liquid in the transformer changes volumetrically ac­
cording to the variation in its temperature arising from changes in at­
mospheric temperature and/ or transformer load. From the expansion
coefficient of the liquid and the range of temperature changes (- 20
to 90"C), its volumetric change is � 9% maximum. Therefore, par­
ticular attention has to be given to prevent the introduction of mois-

1070-9878/1/ $3.00 © 2001 IEEE

IEEE Transactions on Dielectrics and Electrical Insulation
ture into the transformer during operation. Electrical transformers are
principally of two types, hermetic and air breathing transformers. The
hermetic transformer will always have a great advantage over the air­
breathing type, for as long as it remains air proof. The vast majority of
air breathing transformers are equipped with an exterior expansion ves­
sel (necessary for this volumetric change) called a conservator, which
is installed on the transformer main body. When contacting the atmos­
phere directly; insulating liquid absorbs humidity, which deteriorates
its dielectric strength as well as that of the insulating paper when mois­
ture is absorbed into the insulation material in the liquid. Therefore, to
prevent inhalation of moisture from the atmosphere into the conserva­
tor, a dehydrating (silica gel) breather is fitted on the end of the air inlet
pipe of the conservator. The atmospheric air goes in and out automati­
cally via the dehydrating agent. The dehydrating breather contributes
to safe and reliable operation of the transformer. In case of defects in the
breathing system filter, the insulating liquid will absorb moisture from
the atmosphere. The rate of moisture uptake depends on the surround­
ing atmospheric conditions i.e. relative air humidity and temperature.
It is therefore useful to predict the dielectric behavior of the insulation
liquid under standard outdoor conditions as well as the moisture up­
take of the insulated liquids under some selective conditions.
We present here some fundamental investigations concerning hu­
midity uptake rate of different insulating liquids used in HV transform­
ers under selective conditions. To the best of our knowledge, such in­
vestigations have not yet been performed.
The temperature was chosen at 23 and 60"C while the relative air
humidity was varied between 20 and 90% to simulate climatic varia­
tion, since the relative humidity during the winter can decrease below
30% and increase above 75% in summer days. Water content of liquid
specimens were controlled during 30 days. The results are discussed
regarding the ac electric strength according to lEe (International Elec­
trotechnical Commission) or Verband der Elektrotechnik Elektronik In­
formationstechnik e. V. (VDE) standards.
2 DESCRIPTION OF THE
LIQUIDS
The investigations were carried out using three different insulating
liquids, commonly used in HV transformers, and a mixture of two of
them recently proposed as insulating liquid for transformers [3]. A
short description of them is presented below.
2.1 MINERAL OIL
Mineral oil, so called 'transformer oil', is made by refining a frac­
tion of the hydrocarbons collected during the distillation of petroleum
crude stock. The crude oil stock and the refining process used in pro­
ducing these oils are typical of those used in producing many common
petroleum lubricating oils. Chemically, the mineral oil consists of a
complex mixture of basic hydrocarbon liquids such as paraffin (40 to
60%), naphthene (30 to 50%),aromatic (5 to 20%) and olefin (roughly
1%). The chemical structures are described in the literature [4-7].
The early transformer oils were paraffin based but, in about 1925,
they were replaced with naphthene oils because of the high pour point
of paraffin oils. It was soon recognized that paraffin oils are prone
Vol. 8 No. 6, December 2001 1041
to sludging, and this resulted in the development of oil reclamation
techniques [2].
The relatively good aging behavior and low viscosity make mineral
oil a good insulating and cooling liquid. Its electrical and dielectric
properties are strongly temperature and moisture content dependent.
The main advantages of this petroleum-based oil are its wide availabil­
ity and low cost. However, mineral oils posses a relativel� low permit­
tivity, a low flash point and are slightly toxic. Mineral oil also has the
disadvantage of endangering the environment in case of a transformer
leakage [9].
When searching substitutes for poly chlorinated biphenyls (PCB),
ecological considerations raised the problem of searching for incom­
bustible and non-toxic insulating liquids, and ester liquids and silicone
fluids were proposed. They belong to the high fire point (HFP) 'liq­
uids also known as less inflammable' liquids. It is to noted that, by
definition, a HFP liquid must have a minimum fire point of 300"C [4].
Their relatively high cost and availability has limited their use to special
transformer applications.
2.2 ESTER LIQUID
The ester liquid used for these investigations consists of pentaery­
thritol tetraester and different fatty acids. The chemical structure of it
can be found in the literature [4-7]. Ester liquids are not toxic, are well
digested by micro-organisms and posses low vapor pressure at oper­
ating temperature of IIV transformer. In fire they generate no dioxins
or toxic products and possess a good biodegradability [8]. Previous
work [9] has shown that ester liquids can be used for the retro-filling
of mineral oil filled transformers. Ester liquids posses good properties,
and this feature, coupled with the ability of drying the solid insulation
(impregnated paper), is considered as a positive one. However, they
are also prone to possible hydrolytic detachment through their high
moisture saturation limit.
2.3 SILICONE FLUID
Silicone transformer fluid is one of a family of fluids known chem­
ically as poly-dimethyl siloxanes [10]. The chemical structure can be
found in the literature [6,7]. Silicone filled transformers meet most en­
vironmental and fire protection requirements. Indeed, silicone fluid is
environmentally neutral and flame retardant. It posseses good aging
and oxidation properties. Moreover, it does not endanger living or­
ganisms in case of emission [11]. As far as its cooling and insulating
properties are concerned, silicone can be compared with mineral oils.
Its stability is related to the energy of the Si-O bond (374 instead of
245 kJ/mol for C-C bond). Silicone fluid is colorless, and characterized
by a very low pour point compared to that of mineral oils, even if their
viscosity at 20"C is higher.
2.4 MIXTURE OF MINERAL OIL AND
ESTER LIQUID
The dielectric and aging properties of mineral oil are well known to
depend strongly on moisture content [11, 12]. High relative humidity
enhances a high water content in insulating paper that increases the ag­
ing rate of the associated solid insulation. Experience shows that dou­
bling the water content in paper accelerates the aging rate by a factor
10412 Fofana et at.: Fundamental Investigations on Some Transformer Liquids under Outdoor Conditions

ten [12). In a recent work [3), we have shown that through the mix­ Table 2. Mat erial constants of the investigated liquids.
ture of mineral oil with various amounts of an hygroscopic insulating Insulating liquid K(pp m) H(K)
liquid such as ester liquid, the dielectric and aging properties of the Mineral oil 19.2x106 3805
solid/liquid insulation of the HV apparatus were improved. Moreover, Silicone Fluid 6.58x105 2372

the water solubility of the mixture were improved compared to that of
the mineral oil itself. When coupled with solid insulation, it increases
its drying compared to pure mineral oil. For the present investigations
we will use the mixture of mineral oil with 20% of ester liquid, which
seems to be more suitable for use in HV transformers when taking into
account economic aspects as well as electrical properties [3).
Some of the properties of the investigated liquids are summarized
in Table 1.
3 THE RELATIVE HUMIDITY
By definition, the relative humidity of a liquid is the dissolved water
content of the liquid relative to the maximum capacity of moisture that
it can hold at that specific temperature.
The relative humidity Wr (%) at a given absolute temperature T
(K), is defined in terms of the absolute water content in liquid Wabs
Mineral oil, 20% ester 6.25x105 2229
Ester liquid 2.61x105 1340
The material variables Wo and H of the investigated insulating
liquids can be taken from Table 2. Figure 1 shows the water saturation
limit of the investigated liquids for an absolute temperahire range of 0
to 100"C. As illustrated,the water solubility of the ester is much higher
than that of mineral oil, followed by the mixture of mineral oil with
20% ester liquid, and silicone fluid. The difference in the behavior of
silicone and ester liquids compared to mineral oil can be explained by
the difference in water absorption in these liquids. Ester or silicone
liquid can take up water in chemically bounded and in dissolved form,
while in mineral oil, water is only dissolved [7,11). Indeed, it is well
known that water contamination in the liquid may be present in three
different states,namely dissolved, emulsified, and dispersed [6,7,11).
Moisture can accumulate chemically in the oxygen-containing liqUid.
Especially, ester molecules contain chemically bond oxygen of 20% �

(ppm) VS. the saturation limit WL(T) such as by weight [13) and water linkages of the polar side valences lead to a
Wabs high water saturation limit [6,14). However,investigations in the field
W
r - WdT) (1) of chemical species diffusion would be very helpful in understanding
the process.
The saturation limit of the insulating liquids depend on the insulat­
ing liquid type, its chemical composition and mean molar weight [7). 4 EXPERIMENTAL
The temperature dependence of maximum water solubility of the in­ PROCEDURES
sulating liquids is described by the following well-known exponential
equation [7, 12) The liquids were dried and degassed in a two-stage drying unit [4,
11] to ensure a very low initial water content. The investigations were
(2) performed at relative air humidity of 20,50 and 90% for various temper­
atures (23 and 60"C) to simul a te operating conditions as well a s critical
where Wo (ppm) and H (K) are material paramaters which have to ones.
be determined experimentally for each insulating liquid [4,7,11). In
i 100
I I
a logarithmic diagram,Equation (2) depicts a linear curve. Therefore,
0
only two points are necessary to describe the insulating liquid behav­ I 9
so
---.f"--.
...,
ior. For this reason,we only performed investigations at three different � 70
� r--
temperatures (20,60,88"C) and extrapolated the results from 0 to 100"C. 5 60
�
� . 50 '"',
'0000 � 0 '"
4
1 l000������----��������
�
a 30

20

" '--- 1 -�
0 !
l00�------�==��--==-=--�
I
10 20 30 40 50 60 70 80 90 100
Air relative humidity RH (%) -_

Figure 2. Relative air humidity in a closed environment using a

water-glycerin mixture according t6 [10]. The relative humidity is ob­
an uncertainty of 1% for air temperatures between +15
2.9E;.-3
tained with
and 60'C.
37E-3 3.5E-3 3.1E-3 2.7E-3

1rr[1/K] +------

20
Relative humidity for air is the water vapor content of air relative
__
30 60 100
'
T[ C] to its content at saturation at the specific temperature Ill]. To obtain
Figure 1. Water saturation limit WL as a function of the absolute an expected relative air humidity, water has been mixed with a certain
temperature T for differe nt i nsulating liquids. amount of glycerin (C3Hs(OHh), a water white colorless liquid. The
mixture ratio can be taken from Figure 2 [15).
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 8 No.6, December 2001 1043

Table 1. Some important properties of the investigated insulating liquids compared to the limits according to lEe and VDE . • The data are mean values
obtained from many manufacturers.
Property I Unit I Oil I Oil, 20%ester I Silicone I Ester
General physical properties
Density, 23'C kg/m3 856 890 960 960
Density, 90C( kg/m3 810 851 -
915
Pour point c( -40 - -33 -50
Toxicity slightly slightly non·toxic non·toxic
Biodegradability high high high very high
Water solubility, 20C( ppm 45 310 200 2700
Water solubility, IOOC( ppm 650 1600 1100 7200

Heat transfer
Cinematic viscosity, 20C( mm'/s 16 19.43 50 63
Cinematic viscosity, lOOC( mm'/s 2.3 3.45 16 7.7
Heat capacity, 20C( W ((mK)
.
0.135 -
0.151 0.165
Heat capacity, 90C( W ((m K) 0.125 -
- 0.155
Specific heat, 20C( kJ(kg.K 1.85 -
1.55 1.81

Fire properties
Flash point 'C 150�175 - >335 310
Flame point c( 130�135 -
>300 257
Combustion heat WkJlkg 40 -
32.2 36. 8
Self ignition c( 330 - - 405

Electrical properties (at 23'C)

Breakdown strength (ac) Uv JjJ<: kV >60 >55 >50 >55
Permittivity (25C(, 50 Hz)
Or 2.2 2.35 2.9 3.3
Dissipation factor t"" " (90q <lOXlO-
4 <20xlO-4 1.6xlO-4 1xlO-4
Volume resistivity Dcm lOOxlO-
12
8xlO-
14 20x10-'2

5 RESULTS AND DISCUSSIONS

5.1 SIMULATION OF A DEFECT IN
Scaled vessel containing THE BREATHING SYSTEM
insuluting liquid
Io--_-I!-_specimens and a
hygrometer The methods for evaluating the properties of insulating liquids that
guarantee their quality and life duration are standardized. All the prop­
• ""mm��""",,,_��;; ,, ;;;Jt- water-glycerine mixture
-
erties listed in the literature [2, 16] are important, but some have a spe­
ors.ilicagC!1
cial merit for the characterization of an insulating liquid. The important
Figure 3. Experimental procedure used to asses the moisture uptake ones are those that are likely to vary significantly with liquid purity
rate of different insulating liquids under selective conditions, and composition as well as with external parameters, such as temper­
ature and electric field. T he dielectric properties of insulating liquids
are well known to depend on water content [2, 5, 8, 9, 12]. Water con­
For the relative air humidity of 20%, the amount of glycerin was tent increases electric conductivity and dissipation factor and worsens
extrapolated from Figure 2 or a certain amount of silica gel was used dielectric strength. Since the two parameters of our investigations are
in place of the water-glycerin mixture. The uncertainty for a linear temperature and moisture, we investigated their ac electric strength
interpolation between experimental points is <1% [15]. This mixture (or the dielectric breakdown voltage at power frequency), which is the
was placed in a sealed vessel containing samples of the different insu­ most often controlled parameter describing the liquid's function as an
lating liquid, where a hygrometer controlled continuously the relative insulant [2]. The obtained results are then used to asses insulating liq­
humidity, Indeed, the water-glycerin mixture changes in case of water uid insulant behavior under the selective conditions,
absorption by the insulating liquids, It is therefore of prime impor­ To determine the electric strength at various temperatures and for
tance to check regularly the relative air humidity. Figure 3 depicts the various water contents, the respective specimens were placed in the vir­
experimental procedure,
tually homogeneous field of the Verband der Elektrotechnik Elektronik
The uncertainty of the relative humidity was 1%. The tempera­ Informationstechnik e, V. (VDE) hemispherical setup [17] and subjected
ture was controlled by a proportional integral differential (PID) system, to a 50 Hz voltage, increasing at a rate of 2 kV/s to breakdown, The
with an uncertainty of lOC. The water content of the liquid specimens time interval between individual measurements was set at 5 min. Be­ .
were measured with a microprocessor controlled Karl-Fischer titration' tween individual measurements, a glass stirrer was used to remove any
method. The relative uncertainty of the measuring device is <0.5%, In solid decomposition products appearing between the electrodes, and to
contrast to the iodometric method, this procedure can be used to deter­ take away any gaseous decomposition products back in the liquid. Be­
mine the physically dissolved water as well as bonded water [11]. fore and after each measurement series, the electrodes were purged of
the decomposition products of preceding breakdowns. The basic dia-
1044 Fofana et al.: Fundamental Investigations on Some Transformer Liquids under Outdoor Conditions

�-------,
o Minera!oU I

I
• Esoorliquid
- L�mltrorunagedliQuid
,-,� lIlflllror agt!o liqUId

Figure 4. Measuring circuit for breakdown tests. 1: Motor actuating

transformer, 2: manual transformer, 3: HV transformer, 4: protective
resistor, 5: capacitive voltage divider (100 pF/100 nFl, 6: breakdown
recognition (optional), 7: test vessel with VDE electrodes, 8: heating 120 140 160
Relallli& water content [%]
40 60 80 100

device.
Figure 6. Breakdown voltage vs. insulating liquid relative water
gram for dielectric testing of the insulating liquids is shown in Figure 4. content for mineral oil and ester liquid.
In the VDE guidelines [17], six individual measurements in series are
recommended. When investigating silicone fluid, a problem arose at
electric breakdown with the formation of gelatinous crosslinked poly­ �
4J�m' �
J
o Mn!'Jraloil

meric siloxanes which impair the breakdown strength of the fluid con­ - limit for u1aged 011
.- MineraloU.j.20%Esterli':juid

�,- limitforaged oit

siderably, leading to individual breakdown. A modification in the test 60

apparatus has already been proposed by Borsi [11]. �u �______���____________________�

F
�

r[ �
20

10

o . -------.- "
20 R, eo R2
��-1
Rel2uI.ewAierconmnt 1%]
80 100 120
-_
140 160

Figure 7. Breakdown voltage us. insulating liquid relative water

content for mineral oil and the mixture of mineral oil with 20% ester
liquid.

Figure 5. Schematic circuit diagram for measuring breakdown volt­

age of silicone fluid. 1: Series resistor in primary circuit, 2: current Table 3. Values of relative water content related to the limits sug­
limiting resistor, 3: coupling resistors , 4: electronic switch off. gested by lEe or VDE standards [18-20J for different insulating liquids
depicted in Figures 6 to 8.
The arcing energy was reduced by limiting the magnitude and Liquid Unused specimens Aged s� ecimens
duration of the short-circuit current, so that subsequent readings are VI (kV) RI(%) V2 (kV) R2(%)
no longer influenced by decomposition products. Figure 5 shows a Ester liquid 45 20-+30 30 22�40
Mineral oil 50 43 30 6(1
schematic diagram of the circuit design developed for this purpose.
Silicone fluid 40 >15 30 >25
The investigations were performed at different water content and Mineral oil, 20% ester 50 40-�,45 30 55-->80
temperatures. The temperatures was varied between 20 and 120"( to
simulate normal operating conditions as well as critical ones, while the investigations, showing clearly the limits (Ul and U2) suggested by
water content of each insulating liquid has been varied with respect lEe (International Electrotechnical Commission) or VDE [18-20]. On
to their saturation limit at room temperature. More particularly, water these Figures, U1 and U2 correspond respectively to the limits related
content has been varied from 5 to 80 ppm for mineral oil, from 30 to to unused and aged specimens, while Rl and R2 are the relative water
5000 ppm for ester liquid, from 20 to 460 ppm for silicone fluid and content accordingly related to these limits (Table 3). The investigations
from 20 to 120 ppm for the mixture of mineral oil with 20% ester liquid. were only performed for new liquid specimens. However they can be
These conditions allowed reaching relative water contents of > 100%. useful for serviceable insulating liquids since experimental evidence
indicates that liquids which are in serviceable condition or liquids sub­
Because the saturation mixing ratio is also a function of pressure,
mitted to accelerated oxidation tests show little change in their water
and especially of temperature, the relative humidity is a combined in­
solubility characteristics [6,14]. Only when the liquid is severely aged
dex of the environment and reflects more than water content [12]. The
or contaminated does the saturation level increase significantly.
breakdown voltage is thus described as a function of relative water con­
tent. Figures 6 to 8 summarize the results of the ac electric strength Polar compounds present in severely aged liquids are thought to
IEEE Transactions on Dielectrics and Electrical Insulation Vol. 8 No. 6, December 2001 1045

1= 70

�.6(]

\!
.. so-

jU1
•
m U,

20

Figure 9. Moisture uptak e of mineral oil vs. exposition time. The

10-
temperature and ambient relative air humidity acted as parameters,

Rl R1 50 100 150 200 250

0,

1%J
0

RH=20%-1=23"C
RelatNe water content

• RH=50%- T=23"C
Figure 8. Breakdown voltage 115_ insulating liquid relative water
•

RH"'20%- T.ooO·C
• RH=1I0'!o-T=23"C

content for mineral oil and silicone fluid. RH=50%- r=so·c

o
o
D. R�=90%-T"'60'C

�-.-.-..-.--. t- ------
i !
- - •..

influence the water solubility characteristics since water molecules can

be captured by hydrogen bonds with carboxyl groupings [6].
-!
For mineral-based oil, the breakdown voltage UB may be described
as a function of relative water content (very small scatter) and thus all
ExposlhOrldurahool(diI�)
breakdown data at the different temperatures and different absolute
humidity contents can be shown in mutual dependence, This does not Figure 10. Moisture uptake of silicone fluid us, exposition time, The
apply to either silicone fluid or ester liqUid, Contrary to mineral oil, temperature and ambient relative air humidity acted as parameters.
there is a relatively wide scatter range for ester liquid and a wide scat­
ter range for silicone fluid, Especially, investigations were performed
for high relative water content in case of silicone fluid in order to check
if its dielectric behavior de cays with inc reasing relative water content
(>100%) since its be ha v i or is tot ally different from those of other inves­
tigated insulating liquids.
For ester and silicone liquids, the twin dependency of breakdown
vol t age
on temperature and water content cannot be explained by the
dependence on relative water content. Thus the b rea kd own of these
&positilmdurSMI1t(c!ays)
10 15 20

two liquids is evid en tly not du e to physical dissolving of the water.

From Table 1, it can be seen that both liquids posses a high viscos­
ity compared to mineral oil, but results reported by Beroual et al. [21]
indicate that viscosi ty has little effect on the breakdown process_
The difference in their behavior compa red to min eral oil can be ex­
plained by the difference in water absorption and/or water solution
states in these liquids,and consequent change in their molecular struc­
ture, depending on the temperature, Ester or silicone liquid can take up
water in chemically bonded form and in dissolved form, while in min­
eral oil, water is only dissolved [7, 11]. Because the electric breakdown
of insulating liquids is determined by electronic conduction processes
[6,7,11], the molecular structure has a major effect on the level of the
breakdown voltage.
However, the behavior of the mixture of mineral oil with 20% ester
shows an intermediate behavior between mineral oil and ester liquid
(Figures 6 and 7).
Figures 9 to 12 depict the results of the investigations concerning the
moisture uptake rate of the different insulating liquids vs. time, The
temperature and ambient relative air humidity acted as parameters.
It can be seen that when testing the liquids as new ones, the limits
suggested by VDE or IEC standards are respected to a 30 day exposition
Figure 11. Moisture uptake of mineral oil mixed with 20% ester liq­
uidvs, exposition time, The temperature and ambient relative air
humidity acted as parameters.
only for a relative air humidity of 20%. For higher relative humidity
(? 50%) the limit is respected only for a few days. But when testing
the liquids as an aged one,the limits remained respected up to a 30 day
exposition at an relative air humidity of 50% for mineral oil and the
mixture of mineral oil with 20% ester liquid. Table 4 shows a compari­
son of the investigated liquids according to the limits suggested by IEC
or VDE. It can be seen that the coupling moisture uptake rate and the ac
electric strength limit suggested by lEe or VDE standards [18-20], the
mixture of mineral oil and 20% ester liquid seems to be more suitable,
since it will mostly retard possible failure under operating condition.
5.2 FUNDAMENTAL
INVESTIGATIONS
As can be seen in Figures 9 to 12, the higher relative air humidity,
the higher moisture uptake rate of the insulating liquids. Also, the
temperature has a very significant influence on the moisture uptake of
each insulating liquid. For a given relative air humidity, the higher the
1046 Fofana et al.: Fundamental Investigations on Some Transformer Liquids under Outdoor Conditions

RH"'2C%· T=23·C Table 6. Parameters a and b of the investigated insulating liquids,

RH=50%· T="J�G

Este�
•

depicted in Equation 4.
•
... RH::90%.T=23"C
o RI-l�20%. T..6C�C
C RH�50%. T;:60°C
t. RH�90%. T=fl[]"C

0.77±0.18
-l.l±3

in the relative liquid humidity. The reason of the large scatter in the
values obtained for mineral oil could be related to the difference in the
initial relative water content used for the investigations performed at
Figure 12. Moisture uptake of ester liquid vs. exposition time. The 20 and 60'C. But additional investigations are needed before drawing
temperature and ambient relative air humidity acted as parameters. more general conclusions.
A mathematical approach of data suggests the existence of some
Table 4. Exposition duration time (in days) of the investigated insu­ kind of relationship between relative moisture content of liquid and ex­
lating liquids at which the limit suggested by the standards (lEe or position time. Indeed, relative liquid humidity Wr can be described as
VDE [18-20]) are no more respected. Note that these limits were calcu­
a mathematical function of the exposition time t (in days) representing
lated taking into account the lower value of the relative water content
the solid·line curves, which fits the experimental data plotted in Figures
as depicted in Table 3.
������
9 to 12
Wr A(l exp[-t/k])
= - (3)
The constants A (in%) and k (in days) depending on the liquid itself,
50 >2 are deduced from Figures 9 to 12.
90 >1
On Table 5, it can be seen that k depends only on the insulating liq­
20 >30 uid while A depends on the liquid and on ambient relative air humidity
50 >2 RH(in%).
90 >1

90
Table 5. Parameters A and k of the investigated insulating liquids, 80 - o Mineral 011
�.
depicted in Equation 3.
70 - • Silicone fluid /' �
RH(%)I Oil I Oil, 20%ester I I Ester 6 Mineral oil I /V:/
-60

,I 1
Silicone -
+ 2D% ester

A(%) e: • ester liquid

-:: 50 /�V

I I
20 18±5.21 18.6±2.5 16±2.4
� V
��
28±13
50 50±7.6 1l±2.3 47±10.3 35±3.7 E 40
83±17.5 79±3 . 9 73±13.5 70±15.3
� -r- �
----....-7
90 0. 30
k (day)
20
'�
3 I:/"

I I I I
20 3 3 5
10
3 3 5
I
I
50 3
90 3 3 3 5
10 20 30 40 50 60 70 ao 90
Ambiant air relative humidity RH (%)
temperature, the higher the moisture absorption rate. The reason is the
saturation limit of the insulating liquid which is highly temperature­ Figure 13. Parameter A VS. ambient relative air humidity RH for
dependent a nd a lrea dy shown in Fig u re 1. different insulating liquids
On Figures 9 to 12, it can be observed generally that, during an ex­
position under some selective conditions (ambient relative air humidity The dependency of the const ant A VS. RH can be taken from Fig­
and temperature), insulating liquids uptake moisture and saturate after ure 13. A close linear relationship exists between parameter A and the
a few days depending on the kind of liquid (moisture uptake veloc­ ambient relative humidity RH. The curves can be linearly fitted such
ity, saturation limit, temperature and initial water content), and on the as
outdoor conditions (relative humidity and temperature). A= aRH+b (4)
The investigations show that mineral oil takes moisture to a rela­
where a and b a re dimensionless parameters that depend on the liquid
tive water content of 9 3% (Figure 9 ) while silicone fluid and ester
itself as seen in Table 6.
�

liqUid reach 70% as relative water content (Figures 10 and 12). Out
�

of Figure 11, it can be seen that the mixture of mineral oil and ester liq­
uid reaches 80% of relative water content, i.e. an intermediate value
�
between mineral oil and ester liquid.
Apart from mineral oil, a close relationship seems to exist between
the relative moisture uptake rate of the different liquids and the exposi.
tion duration, information concerning the temperature being included
The behavior of insulating liquids under these selective conditions
(different exposition temperatures and ambient relative air humidity)
can thus be modeled combining Equations (3) and (4) such as
Wr (aRH +b)(l - exp[-t/k])
=
(5)
The absolute moisture content Wabs can be deduced knowing the satu·
ration level of the insulating liquid at a given temperature T (K).
IEEE Transactions on Dielectrics and Electrical Insulation
This relationship can be useful to predict the moisture uptake rate
of an insulating liquid submitted to some selective conditions (tem­
perature, ambient relative air humidity) provided the initial relative
humidity of the insulating liquid (to be calculated according to Equa­
tion 1) is known. Indeed at a given ambient relative air humidity RH
and temperature T, moisture uptake rate of the insulating liquid VS.
the time t can be deduced such as
Wabs
Wr 100
WdT)
=
(6)
(aRH+ (
=
b)[_ t: to])
1 _ exp
with
to. - [ - WdTW)'(aRH
= k In
·
1
+
] 100
b)
(7)
where a, b, k and WL (1') depend on the kind of liquid. The standard
deviation of Equations (6) and (7) can be deduced from the parameters
a and b from Table 6. The parameter to contains information related to
the initial absolute moisture content Wi of the liquid itself before the
exposition.
These relationships are of prime importance since they allow to de­
scribe the moisture uptake rate of an insulating liquid under selective
conditions (temperature, relative humidity), provided the liqUid pa­
rameters (a, b, k, WdT)) are known. However, the relatively large
standard deviation associated to the parameter a and b (Table 6) shows
that additional investigations will be very useful for refining these val­
ues.
6 CONCLUSION
HE work presented dealt with some fundamental investigations
Tconcerning insulating liquids actually used in HV transformers and
a mixture of two of them, recently proposed as insulating liquid.: It is
generally found that the less hygroscopic the insulating liquid is, the
higher the moisture uptake rate is. The mixture of mineral oil with 20%
of ester liquid seems to be promising. lndeed, when adding ester liquid
to mineral oil, an intermediate behavior of the two insulating liquid
is obtained: the ac electric strength remain almost unchanged while
the moisture uptake velocity is reduced compared to that of the pure
mineral oil.
When considering the scenario of a defect in a breathing system
(for air-breathing type transformers) it was found that at a very low
ambient relative air humidity (20%), the ac breakdown voltage limit
suggested by IEC or VDE remains respected to a 30 day exposition. For
a relative humidity >50% these limits are respected only for a few days
depending on the kind of insulating liquid (moisture uptake velocity,
saturation limit, temperature and initial water content) as well as on the
outdoor conditions (relative humidity and temperature). These inves­
tigations can provide practical information for monitoring conservator
type transformer in case of a bad run in the breathing system.
Also, the mathematical approach depicted empirical relationships
between moisture uptake velocity, outdoor conditions (temperature
and ambient relative air humidity). These relationships can be use­
ful also for serviceable insulating liquids since experimental evidence
indicates that liquids which are in serviceable condition or liquids sub­
mitted to accelerated oxidation tests show little change in their water
Vol. 8 No.6, December 2001 1047
solubility characteristics [16]. Only when the liquid is severely aged
or contaminated, does the saturation level increase significantly. Po­
lar compounds present in severely aged liquids are thought to influ­
ence their water solubility characteristics. However, the relatively large
standard deviation associated to the parameters a and b shows that
additional investigations are needed to refine these values.
REFERENCES
[IJ Hisaa Kan and Terua Miyamoto, "Proposals fa r an improvement in transformer
diagno sis using Dissolved Gas Analysis", IEEE Eleetreical Insulation Magazine Vol.
11,)\a. 6, pp. 15 -21, 1995.
[2J A.Sierota and J.Rungis, "Electrical Insulating Oils- Part 1: Characterization and
Pre-treatment of New Transformer Oils", IEEE Electrical Insulation Magazine Vol. ,
l!, No.8,pp. 8-20, 1995.
[31 H. Borsi, E. Gockenbach, 1. F of ana and j. Bastian, "Verfahren zur Verbesserung der
clektrischen und Alterungseigenschaften von mine ral Olgefiill ten Hachspatmungs­
geriiten durch Beimischung einer hygroskopischen IsolierfiUssigkeit", Deutsches
Patent und Markenamt DE 199 46 641 AI,Ap ril 2000.
[41 K.Dumke, Un rs te uchungen an einer Esterjliissigkeit als lsolierstojffor 'Iransformatoren,
PhD Thesis University of Hanover,1998.
[5J I. Fofana, V Wasserberg, H. Borsi and E. Gock enbach "Retrofilling Condition of
,
HV transf ormers", IEEE Electrical Insulation M ag azine, Vol. 17, No.2, pp.17-30,
MarchiApril 2001
[6J r J.Vuarchex, "Huiles et liquides isolants", Techniques de l'ingenieur,pp. D-230 to
D-233,12-1986.
[7J M. Beyer, W. Boeck, K. Moller and W.Zaengel, Hochspannungstechnik Theoretische
und praktische Grundlagen, Springer-Verlag, 1986.
[8J E. D. Senkevitch, V G. Arakelian, T V Glasunova, V A. lipshtein, T I. Morozova
and N. M. Panova, "New synt het ic liquids for transformers", OGRE S ym po sium ,
Vietma (Austria),pp. 500-504, 1987.
[9J H. Borsi, "EsterflUssigkeit Midel 7131 als Ersatz liir Mineral61 in Transformatoren",
Elektrizitatswirtschaft,Jg. 93, Vol. 24, pp 1523-1528,1994.
[101 IEEE Guide for Acceptance of Silicone Insulating Fluid and its Maintenance in
Transformers, C57 111-1989.
[l1J H.Borsi, "Dielectric Behavior of Silicone and Ester Fluids for Use in Distribution
Transformers", IEEE Trans.on Elec. Insul.Vol.26,No.4, pp.755-762, 1991.
[12J Y. Du, M. Zahn, B. C. Lesieut re , A. V Mamishev and S. R. Lindgren, "Moisture
Equilibrium in Transformer Paper-oil Systems", lEEE Electrical Insulation Maga­
zine, Vol.15, No.1,pp. 11-20, 1999.
[13J W. Breuer and G. Hegemann, "Behavior of on-load tap changers in Midel 7131 - A
PCB-alternative for transformers", ClGRE Symposium, Vietma (Austria), pp. 500-
509,1987.
[14 1 P. A.von Guggenbcrg and j. R.Melcher, "An immersible relative saturation mois­
ture sensor with application to transformer oil",Proc.of the 3rd Int.ConI. on Prop­
erties and Applications of Dielectric Materials, Tokyo, Japan, Ju ly 8-12,1991.
[15J lEC 260,"Test enclosures of non-injection type for constant relative humidity",Ed.
1, 1968,withdrawn in 2000.
[16J T.O. Rouse, "Mine ral oil in transformers", IEEE Electrical Insulation Magazine, Vol.
14, No.3 pp. 6-16, 1998.
[17J VDE 0370 Part 5,IEC 156,Insulating oils: D etermination of the breakdown voltage
at line frequency, (lEC 156 1995),1996-03.
[18J IEC 60296, Specifications for new insulating oils for transformers and switchgear,
(similar to VDE 0370 Part 1 and 2), 1992.
[19J IEe 61099,Specifications for unused synthetic organic esters for electrical purposes,
(similar to VDE 0375 Part 1),1992.
[20J IEC 836 (1988), Specifications for silicone liquids for electrical purposes, (similar to
VDE 0374 part 1: 1991-09).
[21 J A Beroual,M.Zahn,A. Baden!, K. Kist, A. J. Schwabe, H. Yamashita,K.Yamazawa,
M. Danikas, W. G. Chadband and Y. Torshin, "Propagation and structure of strea m ­
ers in liquid dielectrics", IEEE E lectrical Insulation Ma gazine Vol.14, No .2, pp.
,
6-17,1998.
Manuscript was received on 27 November 2000, in final form 16 Tillie 2001.