IEEE Transactions on Dielectrics and Electrical Insulation

Vol. 20, No. 3; June 2013

887

Performance of Coconut Oil as an Alternative

Transformer Liquid Insulation
B. S. H. M. S. Y. Matharage, M. A. R. M. Fernando
Department of Electrical and Electronic Engineering,
University of Peradeniya,
Peradeniya, 20400, Sri Lanka.

M. A. A. P. Bandara, G. A. Jayantha
Ceylon Electricity Board,
Sri Lanka.
and C. S. Kalpage
Department of Chemical and Process Engineering,
University of Peradeniya,
Peradeniya, 20400, Sri Lanka.

ABSTRACT
Transformer mineral oil has been replaced by alternative oils such as synthetic oil and
natural esters due to their biodegradability and environmentally friendly nature. This
paper presents performance of coconut oil as such an alternative. Generally, as the
other alternative oils, coconut oil has high conductivity due to the presence of free fatty
acids. In this study, three different types of coconut oil samples consisting of virgin,
copra and RBD (refined, bleached and deodorized) were initially tested by frequency
dielectric spectroscopy (FDS) measurements to see how the conductivity was improved
by dehydration and neutralization. The FDS results were fitted by inverse power
dependence and Cole-Cole models to estimate the conductivity and response functions.
Afterwards, a set of new coconut oil samples extracted from copra were thermally aged
at 120 C under sealed conditions and compared with that of mineral oil. The
performances of oil samples were evaluated using the test results of breakdown
voltage, acidity, interfacial tension and FDS measurements under different aging
periods such as 2, 5 and 7 weeks. Another 4 sets of new coconut and mineral oil samples
were subjected to simulated thermal faults and electrical faults which include aging for
12 hours at 160 0C, one hour at 200 0C, exposing to partial discharges for four hours
and subjecting to 20 low energy breakdowns. The performance comparisons were done
by FDS measurements and dissolved gas analysis. In parallel, field-aged coconut oil
samples collected from a sealed distribution transformer with 11 years of service were
also tested by FDS measurements. In general, coconut oil shows its suitability as an
alternative to the mineral oils for transformers, despite limitations found in some of
their physical properties. It was found that the FDS results were in good agreement
with chemical test results and with the estimated conductivity values.
Index Terms - Power transformers, insulation, mineral oil, coconut oil, aging,
frequency dielectric spectroscopy, dissolved gas analysis, conductivity.

1 INTRODUCTION
POWER transformers consisting of oil/paper/pressboard
insulation are considered as one of the main components in a
power system. However during service, their insulation gets
deteriorated due to applied stresses such as thermal, electrical,
mechanical, etc., so that majority of failures are caused by the
Manuscript received on 25 October 2012, in final form 7 March 2013.

failures of their insulations [1, 2]. In addition to the insulation, the

oil is not only used as a liquid insulation but also as a coolant of
the transformer.
The mineral oil extracted from crude oil, because of its better
performance, has been used as the liquid insulation for more than
70 years [2, 3]. However, due to its poor biodegradability and
future scarcity, attention has been paid to alternative natural esters
including oil from sunflower, soya beans, vegetable, etc [4-12]. The
studies found that those natural esters need to be purified to

1070-9878/13/$25.00 2013 IEEE

888

B. S. H. M. S. Y. Matharage et al.: Performance of Coconut Oil as an Alternative Transformer Liquid Insulation

improve their insulation properties [4]. For example, commercially

available refined, bleached and deodorized type of natural esters
had been purified to produce BIOTEMP and Envirotemp FR3
which are commercially available alternative oils [4, 7]. In general,
the breakdown voltages of most of the vegetable oils are higher
than mineral oils. Despite that, the properties such as viscosity,
conductivity, acidity and water solubility are still to be improved
[4]. As far as the aging is concerned it has been reported that the
natural esters produced high content of dissolved gases such as CO
and CO2. [6-8, 11]. To improve those properties methods such as
partial hydrogenation, adding special antioxidants to reduce
oxidation, minimizing the unstable tri-saturates, adding special clay
to reduce conductivity levels, etc. have been proposed [4]. As far as
overall performances are concerned, the investigations of
alternative oils under laboratory conditions are at a satisfactory
level [8, 9]. Transformers filled with new alternative oil have also
been tested at different voltage levels in both transmission and
distribution [10] and have verified better performance under sealed
conditions [4-6].
The coconut oil is also a vegetable oil (natural ester) used for
cooking purposes and is widely available in tropical countries
such as India and Sri Lanka. However, most of its physical and
chemical properties characterizing liquid insulation have been
proved at satisfactory levels [13-15]. The coconut oil differs
from some of the other types of vegetable oils as they are rich
in saturated fats (about 90%) and out of that about 65%
includes medium and short chains. Usually high amounts of
saturated fat cause increases in the melting point, but for the
positive side, due to increased oil-air stability the oxidation
resistance is weaker [9] and the same applies to coconut oil as
well. As for most natural esters, the coconut oil also has higher
content of free fatty acids which increases the conductivity
levels [4, 14, 16]. This paper presents the suitability of
coconut oil as a transformer liquid insulation and its
performance in comparison with mineral oil. Purification
process such as dehydration and neutralization has been used
to improve the oil conductivity.
In service, the transformer insulation deteriorates with time
when subjected to single or combined stresses such as electrical,
thermal, mechanical, etc. due to aging of the main insulation
and mineral oil. Although field investigations provide realistic
condition of the insulation aging, practically it takes a long time
to see the real aging effects. In such situations, laboratory
investigations are conducted with accelerated aging and
simulated faults [6-8]. Electrical and chemical (non-electrical)
tests are usually conducted for performance investigation for
such situations. In non-electrical analysis, moisture content,
acidity, interfacial tension and viscosity are some of the
common test methods to see the insulation deterioration [6, 8,
17] whereas the content of dissolved gases are measured to see
the effects under faulty conditions [7, 8, 18]. Out of electrical
tests, frequency dielectric spectroscopy (FDS) measurements
are widely used as a non-destructive test to assess the condition
of the insulation [19-21]. In this work, the oil samples were
aged as well as subjected to thermal and electrical faults. Their
performances were investigated by chemical and electrical tests.
FDS measurements were conducted in all cases and the results
were modeled to estimate the conductivity and response
function.

2 CHEMICAL AND PHYSICAL

PROPERTIES
2.1 CHEMICAL PROPERTIES
Coconut oil as with most of the other natural esters consists
of triglycerides. Figure 1 shows a typical reaction occurring
between glycerol and acid groups to form triglycerides. These
triglycerides contain different fatty acids (R1, R2 and R3)
which might be saturated or unsaturated depending on the oil
type. Table 1 shows the comparison of acid composition in
coconut oil with other commonly used vegetable oils in
transformers.

Figure 1. Chemical reaction between glycerol and fatty acids when forming
triglyceride in coconut oil [9].
Table 1. Acid composition in coconut oil and other alternative oils [15].
Fatty acid name R1, R2,
R3
Sunflower
Coconut
Rapeseed
Soya Oil
(high
Scientific
Comm
Oil
Oil
oleic) Oil
on
name
Caproic
C6:0
ND-0.7
ND
ND
ND
Caprylic
C8:0
4.6-10.0
ND
ND
ND
Capric
C10:0
5.0-8.0
ND
ND
ND
Lauric
C12:0
45.1-53.2 ND
ND-0.1
ND
Myristic
C14:0
16.8-21.0 ND-0.2
ND-0.2
ND-1
Palmitic
C16:0
7.5-10.2
1.5-6.0
8.0-13.5
2.0-6.0
Palmitoleic
C16:1
ND
ND-3.0
ND-0.2
ND-0.05
Margaric
C17:0
ND
ND-0.1
ND-0.1
ND-0.05
Heptadecenoi
c (cis-10)
C17:1
ND
0.5-3.1
ND-0.1
ND-0.06
Stearic
C18:0
2.0-4.0
8.0-60.0
2.0-5.4
2.9-6.2
Oleic
C18:1
5.0-10.0
11.0-23.0 17-30
75 -90
Linoleic
C18:2
1.0-2.5
11.0-23.0 48.0-59.0 2-17
Linolenic
C18:3
ND-0.2
5.0-13.0
4.5-11.0
ND-0.3
Arachidic
C20:0
ND-0.2
ND-3.0
0.1-0.6
0.2-0.4
ND-non detectable, defined as 0.05%

According to Table 1, unlike in most of the other natural

esters, it is clear that coconut oil contains a high percentage of
saturated fats (C12:0, C14:0). This makes the coconut oil more
stable when contacting with air [8, 13]. Therefore the
oxidative rancidity is less significant in coconut oil than other
vegetable oils [22]. However, high pour point is the main
disadvantage caused by the high percentage of saturated fat
[8].
2.2 PHYSICAL PROPERTIES
Table 2 shows some important physical properties of
coconut oil, commercially available vegetable oil for
transformers [4] and recommended levels for uninhibited
mineral oil of class I [23]. Most of the physical properties such
as dielectric strength, flash point, moisture content, etc. are
within the recommended limits for mineral oil stipulated by

IEEE Transactions on Dielectrics and Electrical Insulation

Vol. 20, No. 3; June 2013

IEC 60296 [23]. Even though the viscosity of coconut oil is

higher than the recommended value, it is less than some of the
vegetable oils currently used as alternative oil [8]. The pour
point of coconut oil has the highest diversity from the
recommended level.
Table 2. Comparison of physical properties of coconut oil and typical
vegetable oil.
Coconut
Vegetable
Uninhibited class
Property
oil [13]
oil [4]
I mineral oil [23]
Dielectric strength [kV]
60
74
50
29
33-45
13
Viscosity [cSt at 40C]
Pour point [C]
23
-15 to -25
-40
Flash point [C]
170-225
310-325
154
0.917
0.91-0.95
0.895
Specific gravity at 20C
Moisture content [mg/kg]
1
50-10
1.5

As proven by some other laboratory experiments partial

solidification of coconut oil at temperatures below the melting
point does not significantly affect its breakdown voltage [13].
It has been proved that the pour point could be reduced to 12
C by adding 15% of styrenated phenol to the coconut oil (by
weight) [24]. Further it could be reduced to -3 C and even up
-42 C, by chemical modification (acid catalyzed condensation
reaction with coconut oil/castor oil mixture) and by adding
lauric and oleic acids as the synthesis respectively [24].

3 EFFECT OF MANUFACTURING
PROCESS
Three main types are categorized according to the
manufacturing process of the virgin coconut oil, the copra oil
and the RBD type oil. The virgin coconut oil, usually
produced in small scale, is extracted from fresh coconut meat,
milk or residue. This process is considered as a wet method.
Chemicals and high heating are not used for in the refining
process. The copra oil, known as dry method, is produced by
heating and drying the meat extracted from the coconut shell.
The oil extracted from copra is then refined, bleached and
deodorized to make RBD coconut oil. In general, the
manufacturing process of the coconut oil has limited effect on
the chemical composition but may affect free fatty acid
content significantly [22]. When considering the free fatty acid
percentage as lauric acid, copra oil shows the highest acid
percentage i.e. 1.41% whereas the RBD coconut oil shows the
lowest i.e. 0.02%. Virgin coconut oil has an acid percentage of
0.13% [22].
3.1 SAMPLES
Preliminary investigations were conducted on different
types of coconut oil samples to observe the effect of
manufacturing process on their properties. Three different
types of commercially available coconut oil, namely Baraka,
Marina, and N-joy were selected for the study. Baraka oil is a
virgin coconut oil which is made from the white scraped
coconut. First it is dried below 60 0C to remove the moisture
and then cooled back to room temperature. Then the oil is
extracted with a cold press method. Finally, the oil is filtered
and heated in a water bath between 55 0C 60 0C. N-joy brand
belongs to copra oil type. In the extraction process coconut
meat separated from the shelf is dried in a kiln to reduce the

889

moisture content to less than 6%. The copra is then crushed,

cooked using indirect steam heating and mechanically
expelled to produce crude coconut oil. The crude oil is then
filtered using a micron-sized filter. Marina is a R.B.D. type
coconut oil. It is produced from coconuts, converted into
Copra (dried meat or kernel of the coconut), by a cold process.
Then it is refined physically by steam and vacuum.
3.2 PROCEDURE
Oil samples of 100ml volume from all three types were
taken for the investigation. The FDS measurements were taken
at room temperature of about 25 0C for all the samples using
an Insulation Dielectric Analyzer IDA 200. A test cell
consisting of three electrode systems with a geometric
capacitance of 70 pF was used for the measurements. The
frequency was varied from 1 kHz to 1 mHz at 50 V ac and the
corresponding complex capacitance was monitored with
respect of the frequency. Then the results were modeled to
find out the quality of the oil samples in terms of the
conductivity, the permittivity and the response function. In the
modeling, frequency variation of complex susceptibility was
modeled with Cole-Cole model with two relaxation processes
(1, 1, 1, 2, 2, 2) and inverse power dependence (A and
n). The inverse power dependence was used at low
frequencies whereas Cole-Cole models were used for high
frequencies. The measured complex permittivity (capacitance)
values were fitted by adding permittivity at infinity and
conductivity terms on the real and imaginary parts of the
susceptibility respectively (see equations (1) and (2)).
'

"

Re

A cot

1n

Im

(1)

(2)

3.3 RESULTS
Figures 2 and 3 show the variation of measured and
modeled real and imaginary permittivity values with respect of
frequency. Table 3 shows the estimated modeled parameters
for the three different brands. According to Figure 2 all the oil
types had nearly equal real permittivity values at high
frequencies and the permittivity values increased when the
frequency reduced towards the low frequencies. The Baraka
oil brand (type B) had the highest permittivity values within
the measured frequency range. When considering the
dielectric losses, all the oil types followed a linear behavior
approximately about -1 gradient towards the low frequencies
confirming the effect of the dc conductivity as the dominant
part.
According to Figure 3 N-Joy oil brand (type C) showed the
highest dielectric losses whereas the Marina oil brand (type A)
showed the lowest. The estimated conductivity values (dc) in
Table 3 also matched with the FDS results i.e. the type A
brand having the lowest conductivity and the type C brand
having the highest.

890

B. S. H. M. S. Y. Matharage et al.: Performance of Coconut Oil as an Alternative Transformer Liquid Insulation

Barakaoilmeasured

100

Marinaoilmeasured

Oilmodeled

10

1
0.0001

0.001

0.01

0.1
1
Frequency[Hz]

10

100

1000

Figure 2. Variation of measured and modeled real part of permittivity with

respect of frequency for different coconut oil brands.
100000

Barakaoilmeasured

10000

Marinaoilmeasured
NJoyoilmeasured

Permittivitye"

1000

Oilmodeled

100
10
1
0.1
0.01

0.001
0.0001

4.1 SAMPLES AND PROCEDURE

The selected oil type brand A was used to improve the
conductivity by two methods, namely dehydrating and
neutralization. Five coconut oil samples with a volume of
100 ml each from type A brand was selected. In dehydrating,
two samples were heated at 105 0C in an air circulating oven
(Yamato DS-63 with inner dimensions: 450 mm (width) x 480
mm (depth) x 450 mm (height)) for 24 hours in unsealed
conditions to remove absorbed moisture. Afterwards, the
samples were cooled for 24 hours in two ways i.e. naturally or
under low RH inside a desiccator. In the neutralization, a precalculated amount of NaOH was added to another two samples
to reduce the acids content. Afterwards the samples were
subjected to dehydration. One new oil sample was selected as
the reference. Table 4 shows the selected oil samples together
with the purification procedure. More details of the test
procedure are available in [16]. After the samples were
prepared, the FDS measurements were conducted and the
results were modeled as described in the section 3.2.

0.001

0.01

0.1
1
10
100
1000
Frequency[Hz]
Figure 3. Variation of measured and modeled imaginary part of permittivity
with respect of frequency for different coconut oil brands.
Table 3. Estimated response function parameters and conductivities for
different coconut oil brands.
Marina
Baraka new
N-Joy new
Brand name
new Oil
Oil
Oil
Type
A
B
C
A [10-3]
20
9.1
18.1
n
0.5
1.8
1.2
1
4.2
20
18.4
1 [s]
34.6
5.0
8.5
1
0.07
0.13
0.09
2
0.45
0.1
0.6
2 [ms]
133.0
19.5
1.0
0.1
0.2
0.1
2
dc (pS/m)
69.0
383.0
844.9

The dielectric response also showed a similar behavior by

having relaxations in the low frequencies (highest 1 and 2
values) for type A brand compared to others (see Table 3). In
general, the copra coconut oil (type C brand) showed the
highest conductivity (844.9 pS/m) and it might be due to the
presence of high content of free fatty acid levels as no refining
process is used after manufacturing. Out of the tested
samples, type A brand, having the lowest conductivity values,
was selected for further improvement of the conductivity (see
section 4). The type C, having the worst conductivity, was
selected for performance testing under aging and faults
conditions (section 5 and 6).

Table 4. Details of the coconut oil samples of brand A and the purification
processes to improve the conductivity.
Sample
Description
A
New oil
A1
New oil, Dehydrated and cooled in open air
A2
New oil, Dehydrated and cooled under low RH
A3
New oil, Neutralized, Dehydrated and cooled in open air
A4
New oil, Neutralized, Dehydrated and cooled under low RH

4.2 RESULTS
Figures 4 and 5 show the variation of measured and
modeled real and imaginary permittivity values with respect to
frequency. Table 5 shows the estimated modeled parameters
for the different treatments used to improve conductivity.
100

Permittivitye'

Permittivitye'

NJoyoilmeasured

4 IMPROVEMENT OF CONDUCTIVITY

10

Ameasured

A1measured

A2measured

A3measured

A4measured

Oilmodeled

0.1
0.0001

0.01

100

Frequency[Hz]
Figure 4. Variation of measured and modeled real part of permittivity with
respect of frequency by different purification methods.

When comparing dried oil with new oil, it is clearly visible

that the samples, A1 and A2, had limited changes in the real
permittivity values whereas the imaginary permittivity
reduced significantly by about 60% in low frequencies
confirming the removal of the majority of the absorbed
moisture in the oil (see Figures 4 and 5). When the oil samples
were neutralized prior to the drying, the real permittivity
values reduced significantly about 50% at low frequencies.

IEEE Transactions on Dielectrics and Electrical Insulation

10000

Vol. 20, No. 3; June 2013

A1measured

1000

A2measured

A3measured

100

A4measured

Oilmeasured

Permittivitye"

Ameasured

10
1
0.1

0.01
0.001
0.0001

0.001

0.01

0.1
1
Frequency[Hz]

10

100

1000

Figure 5.Variation of measured and modeled imaginary part of permittivity

with respect of frequency by different purification methods.
Table 5. Estimated response function parameters and conductivities under
different purification processes.
Sample name
A
A1
A2
A3
A4
A [10-3]
20
1.1
0.6
0.03
0.5
n
0.5
1.7
1.99
1.99
1.1
1
4.2
1
1.1
0.038
0.025
1 [s]
34.6
30
30
13
0.5
0.07
0.7
0.6
0.41
0.56
1
2
0.45
0.05
0.04
0.001
0.002
2 [ms]
133.0
0.1
0.01
0.14
0.1
0.1
0.1
0.1
0.1
0.1
2
dc (pS/m)
69.0
23.7
35
30
17.6

At the same time the imaginary frequency values also reduced

by 60% at low frequencies, showing the removal of fatty acid
content to some extent. More importantly neutralization also
reduced imaginary permittivity about 50% at high frequencies.
It is also clearly visible that two cooling methods had different
effects on the complex permittivity values especially on the
imaginary component. The cooling under low RH inside the
dessicator indicated a reduction of the imaginary permittivity
values by about 50% more than the cooling in normal
atmosphere. This difference should be due to the fact that the
possibility of moisture absorption during cooling under former
case is limited compared to the latter one. The estimated
conductivity values also showed a reduction with the
purification processes (see Table 4). The neutralized,
dehydrated and cooled under low RH sample (A4) showed the
best conductivity values i.e. about 75% improvement.
According to the estimated response function parameters, the
relaxation processes seems to happen at low frequencies for
sample A4 compared to the new sample A.

5 THERMAL AGING
5.1 SAMPLES AND PROCEDURE
The oil brand C (N-Joy brand) which showed the worst
conductivity levels was selected for the aging test and
compared with mineral oil (naphthenic type) [23]. First the
moisture content (MC) of both oil types was measured and the
values were 47 ppm and 42 ppm for mineral oil and coconut
oil respectively. Despite the MC levels were high compared to
recommended values, no drying process was used to remove
the moisture as the purpose was to age under the worst
conditions as much as possible. Three coconut oil samples and

891

six mineral oil samples of volume of 1200 ml were selected.

For all the samples, different metal parts (i.e. copper 2.5 g/l,
iron 2.5 g/l, aluminum 0.5 g/l and zinc 0.5 g/l) were added to
represent the actual conditions inside a transformer [9]. In
addition, naturally-wet pressboard pieces (100 g/1) were
added to the prepared coconut oil samples and three samples
of mineral oil to investigate the effect of moisture during
aging. The rest of the three mineral oil samples were kept in
the dry condition. All the oil samples were aged by drying
inside the air circulating oven at120 0C under sealed condition.
The sealed condition was used as the worst case compared to
aging under unsealed condition [25, 26]. Different aging
periods such as 2, 5 and 7 weeks were selected for the three
different sets of samples and 7 weeks was taken as the longest
duration with sufficient aging [25]. Details of the samples are
given in Table 6.
After the samples were aged, their performances were
evaluated by visual scrutiny, electrical and non-electrical tests.
First, color of the oil was obtained by visual inspection. Two
non-electrical tests were performed to determine interfacial
tension (IFT) and acidity values. The IFT was evaluated by
Easy Dyne measuring device which use the du Nouy ring
method. Average of five tests had been taken as the final value
of IFT. The acidity level was evaluated by titrating the oil
sample with 0.1% w/v NaOH solution. Then the results were
converted into equivalent milligrams of KOH per gram [16].
Two electrical tests; breakdown voltage and FDS
measurements were done. The breakdown voltage was
obtained by MEGGER OTS 80AF/2 using an oil container
with sphere gaps of 2.5 mm distance. The FDS measurements
were performed by using the oil container and the Insulation
Dielectric Analyzer (IDA 200) at 50 V from 1 kHz to 1 mHz
(same procedure described in section 3.2). The FDS results
were finally modeled using the Cole-Cole and inverse power
dependences.
Table 6. Details and aged durations of coconut oil and mineral oil samples.
Dry Transformer
Wet Coconut
Wet Transformer
Duration
mineral oil
oil
mineral oil
[weeks]
TD
CA
TW
0-new
TD2
CA2
TW2
2
TD5
CA5
TW5
5
TD7
CA7
TW7
7
T Mineral oil, CA Coconut oil, D Dry and W - Wet

5.2 RESULTS
Table 7 shows the color variations of the samples during
aging. It is clear that all the samples have gone through color
changes during aging mainly attributed to the oxidation
process. However, in coconut oil as most of other vegetable
oils, the oxidation process is weaker due to the higher
percentage of saturated fatty acids [9, 22].
Table 7. Color variation of coconut oil and mineral oil samples with respect
of the aging period.
Sampl
Color
Sample
Color
Sample
Color
e No.
No.
No.
TD
Pale yellow
CA
Pale yellow
TW
Pale yellow
TD2
Pale yellow
CA2
Pale yellow
TW2
Pale yellow
TD5
Pale yellow
CA5
Yellow
TW5
Yellow
TD7
Amber
CA7
Brown
TW7
Amber
T Mineral oil, CA Coconut oil, D Dry and W - Wet

892

B. S. H. M. S. Y. Matharage et al.: Performance of Coconut Oil as an Alternative Transformer Liquid Insulation

Figure 6 shows the IFT test results obtained for both coconut
oil and mineral oil samples. Some results at higher aging
periods could not be obtained. From the results, it is clearly
visible that at the beginning both mineral and coconut oils had
similar and higher IFT values than required by the standard [23]
which is marked as RL. However, the IFT values reduced in
both oil types during aging and coconut oil showed a drastic
change in that respect. Usually sludge and wax formation could
be a cause for the reduction of IFT. When coconut oil heats for
a longer duration, weaker substances with low density might
have formed and rise to the oil-air interface so that IFT values
reduce with the aging. Water solubility is also an important
factor. Further studies should be conducted to understand these
observations.
Figure 7 shows the variation of acidity levels during aging.
The acid levels increased in both mineral and coconut oil during
aging, but the increase for the latter case was significant to
about 10 times as the mineral oil. However, when comparing
acidity levels of commercially available alternative oils such as
BIOTEMP [4], the acidity levels are reported to be in a similar
range. According to literature, acidity levels of other types of
alternative oils too have increased with aging [8].
Recent reports have shown that new treatment methods such
as Fullers Earth method can be used to reduce acidity level of
natural esters significantly [27]. Based on that information, the
high acidity level of coconut oil seems to be less significant and
further studies are recommended to reduce the acidity level.
1000

CA

TD

RL

20.8

11.8

23.8
11.3

16.1

29.9
11.7

42.1

IFT(mN/m)

100

42.1

174.7

TW

10

Figures 8 and 9 shows the real and imaginary permittivity

values obtained from the FDS measurements. Table 8 shows
the corresponding modeled parameters. When considering the
Figures 8, 9 and Table 8, coconut oil showed aging behavior
by increasing its permittivity, dielectric losses and the
estimated conductivities with respect of the aging time. The
real permittivity showed a significant increase at the lowest
frequency which was about 350% after 7 weeks with respect
of the new samples. However, at high frequencies, the
permittivity values were in the same order. On the other hand,
both dielectric losses and estimated conductivity values after 7
weeks aging had similar values as the beginning. In coconut
oil, as for other natural esters, hydrolysis (reverse of the
esterification) occurs during aging [9]. This process creates
fatty acids while consuming water as shown in Figure 1.
Therefore, the moisture which helps to increase oil
conductivity for new oil decreases with aging. However, the
acid level increases with aging so that the conductivity
increases. As far as both dry and wet mineral oil samples were
concerned, the permittivity, dielectric losses and estimated
conductivities initially decreased and then increased with the
aging time. It is reasonable to state that in mineral oil, aging
process is followed by drying process.
In general, the estimated conductivity values and dielectric
losses for coconut oil were comparatively higher than mineral
oil. This was true for both new and aged samples. Increased
conductivity is mainly due to presence of moisture and/or acid
substances. When the oil samples were new, the absorbed
moisture in the new wet coconut oil sample increased the
conductivity. When the sample was aged, the formation of free
fatty acids caused the increase of conductivity. After
considering Figures 6 to 9, the FDS results were in good
agreement with the acidity and IFT results.
Figure 10 shows the variation of the breakdown voltages
during aging for both coconut oil and mineral oil samples. It is
clearly visible that, same as most of the other vegetable oils
coconut oil has the highest breakdown voltage than mineral oil
with aging [8, 9].

1
0

Time(weeks)

1000

Cmeasured
CA2measured
CA5measured
CA7measured
Cmodeled
Tmeasured
TW2measured
TW5measured
TW7measured
TWmodeled
TD2measured
TD5measured
TD7measured
TDmeasured

Figure 6.Variation of the IFT values for both coconut oil and mineral oil
samples with respect of the aging period.

TD
2.28

2.19

1.00

0.04

0.05

0.04

0.05

0.04

0.02

0.05

0.08

0.10

100

10

0.02

Acidity(KOHmg/oilg

10.00

RL

Permittivitye'

CA

2.36

TW

1
0.0001

0.01
0

2 Time(weeks) 5

Figure 7. Variation of the acidity values for both coconut oil and mineral oil
samples with respect of the aging period.

0.001

0.01

0.1

10

100

1000

Frequency[Hz]

Figure 8.Variation of measured and modeled real part of permittivity with

respect of frequency for mineral oil and coconut oil at different aging periods.

IEEE Transactions on Dielectrics and Electrical Insulation

C
CA2measured
CA5measured
CA7measured
Cmodeled
T
TW2measured
TW5measured
TW7measured
TWmodeled
TD2measured
TD5measured
TD7measured

100000
10000
1000
100
permittivitye"

Vol. 20, No. 3; June 2013

10

893

6 THERMAL AND ELECTRICAL FAULTS

Usually transformers are subjected to thermal and electrical
faults during their operations. The thermal faults are
categorized T1, T2 and T3 according to temperature band as
T1 (<300 0C), T2 (>300 0C) and T3 (>700 0C). The electrical
faults are categorized as corona type (PD), low energy
discharges (D1) and high energy discharge (D2) [7, 28, 29]. In
this study thermal faults of T1 type with different intensities
and electrical faults of PD and D1 were simulated [7, 29].

1
0.1
0.01
0.001
0.0001
0.0001

0.001

0.01

0.1
1
Frequency[Hz]

10

100

1000

Figure 9. Variation of measured and modeled imaginary part of permittivity

with respect of frequency for mineral oil and coconut oil at different aging
periods
Table 8. Estimated response function parameters and conductivities for
different aging periods.

TW5

A
[10]
18.1
1.7
4.4
6
64.2
3E04
0.9

TW7
TD2
TD5

0.2
60
7

1.8
1.1
1.1

0.2
9
0.1

20
2
5.1

1
5
1

0.001
0.01
0.008

TD7

0.1

1.1

0.2

0.01

Sample
No.
C
CA2
CA5
CA7
T
TW2

1 [s]

1.2
1.8
1.8
1.9
1.1
4

18.4
9
8
6
3
45

8.5
311
144.6
51.53
1.33
5.1

0.9
1.5
2
2
2
1

0.6
0.01
0.052
0.01
2.08
1.56

1.4

0.11

1.3

0.01

1
[10-1]

2
[m
s]
1
0.1
6.1
0.1
2
5.1

2
[10-1]
1
5
1
1
2.1
1.4

dc
(pS/
m)
845
368
512
877
54
77.6

0.0
1
0.1
1
0.0
1
0.0
1

3.3

2.2

0.8
5
1

4.1
64.9
10

0.1

6.1 SAMPLES AND PROCEDURE

Copra type coconut oil (brand C) and mineral oil (naphthenic
type according to [23]) in volumes of 500 ml were used for the
investigation. All samples were prepared inside stainless steel
containers of volume 1500 ml under sealed conditions. Metal
substances (Copper, Aluminum, Iron and Zinc) and pressboard
pieces were included with all samples in the same amounts as
for the aging (see section 5.1). The containers were closed by a
separate lid and a weight (about 1 kg) was kept on the top of the
container. For electrical faults, adjustable pointer was used with
the lid. Schematic diagrams for thermal and electrical faults are
shown in Figures 11 and 12.
Under thermal faults two sets of samples were aged under
sealed condition at 160 0C for 12 h and 200 0C for 1 h to
represent two different intensities of T1 [29]. In addition, the
samples aged at 120 0C for 7 weeks were also used as another
case of T1. The details of the samples are given in Table 9.

Interestingly, the breakdown voltages of coconut oil

increased with the aging duration and were well above the
recommended levels [23]. In general, the breakdown voltages
showed a positive impact of coconut oil based on the selected
worse coconut oil (i.e. copra type) despite other negative
effects such as acidity, IFT and conductivities.
CA

TD

RL

60.7

69.3

70
60

31.3

31.7
18.3

13.5

20

14.5

34
18.6

40
30

39

50
18.6

Breakdownvoltage(kV)

90
80

Figure 11.Test setup for thermal faults.

77.9

TW

10
0
0

Time(weeks)

Figure 10. Variation of the breakdown voltage values for both coconut oil
and mineral oil samples with respect of the aging period.

Figure 12.Test setup for electrical faults

6.2 RESULTS
Figures 13 and 14 show the comparison of DGA results of
coconut and Mineral oil samples for thermal and electrical faults
respectively. The common dissolved gasses are CH4, C2H2, C2H4,
C2H6, H2, CO and CO2 [18, 30]. Natural esters which have double
bonds between carbon atoms generate dissolved hydrocarbon
gases during aging and faults [30]. However, DGA analysis
shows that the content of the hydrocarbons in coconut oil were
insignificant during aging and faults. On the other hand, the tested
mineral oil samples contained considerable amount of gases such
as C2H6 and CH4 showing the formation of gasses. Considering
coconut oil, when subjected to thermal or electrical faults,
formation of hydrocarbons by scission of C-C and C-H bonds
seems to be less effective as the C-O bonds of the triglycerides
are weaker. Thus formations of hydrocarbon gasses are
negligible. At the same time less solubility of hydrocarbon gases
in coconut oil may also have become a reason to observe less
hydrocarbon gases in coconut oil.
When considering other common gases such as H2, CO and
CO2, in electrical faults, coconut oil had shown 10 to 20 times
more CO than mineral oil. However, the gasses such as CO2
and H2 contents were in the similar range and for the latter
case, H2 appears in the breakdowns only.
On the other hand for thermal faults the contents of those
gases were in a similar range for coconut and mineral oils. The
above observations could be explained as follows: Usually
low energy breakdown releases more energy than partial
discharges [7] so that H2 was released in low energy
breakdowns. According to [28] decomposition of solid
cellulose insulation, moisture and oxygen are the major reason
for formation of CO and CO2 gases.

79

204

631
573
223

795

CTF120
TTF120
CTF200

100

30
26
27

Gasamount/(ppm)

1000

10

1
H2

CH4

CO

CO2

C2H4

C2H6

C2H2

Typeofgas

887
760
839
739

1000

285

385

Figure 13. Variation of dissolve gases in coconut oil and mineral oil under
simulated thermal faults.

CBD

TBD

CPD

TPD

C2H6

C2H2

28

100

16

18
20
10

Under electrical faults two sets of samples were subjected to

partial discharges for four hours and 20 electrical breakdowns to
represent the PD and D1 respectively [29] (see Table 9). The ac
voltage was applied to the oil samples through a rod-plain gap
with gap distance of 15 mm by 60 kVA, 60 kV 50 Hz,
transformer. First corona inception voltage and breakdown
voltage was obtained by increasing the voltage gradually. Then
the samples (CPD, TPD) were exposed to PD at corona inception
voltage. Afterwards, the other samples (CBD, TBD) were
subjected to low density discharges by increasing to breakdown
voltage. One minute was kept between two breakdowns.
After the faults were applied the samples were tested by
dissolved gas analysis (DGA) and FDS measurements. The
DGA was done using Myrkos Transformer Fault Gas
Analyzer which uses the Gas chromatography method. The
FDS was measured by IDA 200 using the same procedure
described in section 3.2. The conductivities were obtained by
fitting the FDS curves with models.

10000

Gasamount/(ppm)

Table 9. Details of the coconut oil and mineral oil samples prepared for
thermal and electrical faults.
Coconut oil Mineral oil Description
C
T
New samples without aging
CTF120
TTF120
Aged at 120 0C for 7 weeks
CTF160
TTF160
Aged at 160 0C for 12 hours
CTF200
TTF200
Aged at 200 0C for 1hour
CPD
TPD
Partial discharge for 4 hours
CBD
TBD
20 electrical breakdowns
TF- Thermal fault, PD Partial discharge, BD - Breakdown

3003
2784

B. S. H. M. S. Y. Matharage et al.: Performance of Coconut Oil as an Alternative Transformer Liquid Insulation

1
H2

CH4

CO

CO2

C2H4

Typeofgas

Figure 14. Variation of dissolve gases in coconut oil and mineral oil under
simulated electrical faults.

The inclusion of pressboard pieces during faults could be

the cause for formation of such gases recorded in both coconut
oil and mineral oil samples. In opposition to hydrocarbon
gases, CO and CO2 might be more highly soluble in coconut
oil than in mineral oil which may be the reason for the higher
amount of dissolved CO and CO2 gases in coconut oil for
faults over that of mineral oil.
Figures 15 to 18 show the FDS results obtained for coconut
oil and mineral oil with thermal and electrical faults
respectively. Table 10 shows the estimated conductivity
values.
Newtransformeroilmeasured
TTF120measured
TTF160measured
TTF200measured
Newcoconutoilmeasured
CTF120measured
CTF160measured
CTF200measured

50

Permittivitye'

894

0.5
0.001

0.01

0.1

1
Frequency[Hz]

10

100

1000

Figure 15. Variation of measured and modeled real part of permittivity with
respect of frequency for mineral oil and coconut oil under different thermal
faults.

IEEE Transactions on Dielectrics and Electrical Insulation

100000

Newtransformeroilmeasured
TTF120measured
TTF160measured
TTF200measured
Newcoconutoilmeasured
CTF120measured
CTF160measured

10000
1000
100

Permittivitye"

Vol. 20, No. 3; June 2013

10
1
0.1
0.01

0.01

0.1

1
Frequency[Hz]

10

100

1000

Figure 16.Variation of measured and modeled imaginary part of permittivity

with respect of frequency for mineral oil and coconut oil under different
thermal faults.

Permittivitye'

50

Newtransformeroilmeasured
TPDmeasured
TBDmeasured
Newcoconutoilmeasured
CPDmeasured
CBDmeasured

0.5
0.001

0.01

0.1

1
Frequency[Hz]

10

100

1000

Figure 17. Variation of measured and modeled real part of permittivity with
respect of frequency for mineral oil and coconut oil under different electrical
faults.
100000
Newtransformeroilmeasured
TPDmeasured

10000

TBDmeasured
1000
Permittivitye"

Table 10. Estimated conductivity values for coconut oil and mineral oil
samples under different faults.
Sample No. dc (pS/m) Sample No. dc (pS/m)
C
845
T
54
CTF120
877
TTF120
4.1
CTF160
59.5
TTF160
8.9
CTF200
161
TTF200
15.1
CPD
1498
TPD
15.1
CBD
1063
TBD
32
TF- Thermal fault, PD Partial discharge, BD - Breakdown

7 FIELD AGING

0.001
0.0001
0.001

895

Newcoocnutoilmeasured
CPDmeasured

100

CBDmeasured

10

A coconut oil (purified RBD) filled sealed type distribution

transformer had been installed in the Wathara area in
Kesbewa, Sri Lanka in 2001 January. Its name plate
parameters were three phase, 160 kV, 33 kV/400 V, 50 Hz,
etc. This transformer has been supplying a 35 kVA bulk
consumer (rubber factory) through a 400 V feeder and
domestic consumers through another two 400 V feeders for
the last 11 years. During its service, the transformer has been
feeding an average load of about 40% and exposed to outdoor
tropical weather conditions. The transformer worked well
without having any reported failures. Figure 19 shows a photo
of this transformer.
On 9th October 2012, an oil sample of volume 300 ml was
taken from this transformer. The transformer was first
interrupted and then the oil sample was taken from the top of
the transformer tank by opening the pressure valve. The oil
sample was collected in a glass bottle and kept sealed and
brought to the University of Peradeniya for chemical and
electrical analysis. The analysis included color, acidity
viscosity and FDS measurement as described in section 5.1.
The viscosity was measured with a Redwood viscosity meter
which measures the time consumed to flow 50 ml of oil out
from a specially designed container in constant temperature.
Then the results were converted from redwood seconds to
centistokes. From the FDS measurements the conductivity was
estimated.
Figures 20 and 21 show the obtained permittivity and the
loss tangent results from FDS measurements on RBD type
field aged coconut oil sample. For comparison, the laboratory
aged copra type oil samples (section 5) and neutralized and
deodorized RBD type oil sample (section 4) are also included.

1
0.1
0.01
0.001

0.01

0.1

1
Frequency[Hz]

10

100

1000

Figure 18. Variation of measured and modeled imaginary part of permittivity

with respect of frequency for mineral oil and coconut oil under different
electrical faults.

In general, coconut oil had higher values of dielectric losses

as well as permittivitys for both thermal and electrical faults.
This increase was significant for electrical faults. The
estimated conductivities also showed similar behavior by
having higher conductivity values for electrical faults. Once
the coconut oil was subjected to faults, with the hydrolysis,
acids were formed. Thus the FDS results of coconut oil
showed the increased losses and conductivity values.

Figure 19. Coconut oil filled 160 kVA, 33/0.4 kV distribution transformer
installed at Wathara, Kesbewa Sri Lanka in 2001.

896

B. S. H. M. S. Y. Matharage et al.: Performance of Coconut Oil as an Alternative Transformer Liquid Insulation

Tandelta

Permittivitye'

Table 11 shows the summary of the test results for the field laboratory aged sample CA5. However, greenish color could
aged and laboratory tested samples. The color, acidity, be seen in the sample. Based on all the results of the tested and
conductivity, loss tangent at 50 Hz and permittivity at infinite estimated parameters, the oil condition of the field aged
are included.
transformer is still in good condition.
According to the FDS measurements, the field-aged sample
Table 11. Summary of test results.
follows a similar pattern as the laboratory aged samples for
both in permittivity and loss tangent plots. It is interesting to Sample
Acidity (mg
dc
Tan delta
Permittivity
Color
No.
KOH/g oil)
(pS/m)
@ 50Hz
at infinite
note that 11 years of field aging did not show any significant
Pale
aging effect compared to laboratory aged samples even with
CA2
yellow
2.2
368
0.055
2.80
the shortest duration of 2 weeks. However, compared to a new
(1)
sample, the field aged sample showed higher permittivity and
Yellow
CA5
2.3
512
0.081
2.79
(2)
losses. The other tested parameters also showed a similar
Brown
behavior. The acidity value for the field aged samples was
CA7
2.4
877
0.121
2.88
(3)
higher than the recommended value (0.5 mg KOH/g oil) for
Green
CFA
0.5
218
0.031
2.70
mineral oil in working transformers [31]. However, the value
yellow
was well below the laboratory aged samples (see Table 11).
The permittivity, loss tangent value and the estimated
conductivity value for the field aged sample was below than
8 CONCLUSIONS
the laboratory aged samples. The loss tangent and permittivity
values are also below the laboratory aged samples.
The work summarizes the following results:
Transformers in the field may not have undergone any thermal
1. As far as dielectric properties are concerned, refined,
aging of the oil due to the light load. (The temperature might
bleached and deodorized (RBD) coconut oil shows the
have been below the threshold by a good margin)
best insulation properties whereas copra coconut oil
shows the worst.
100
Newneutrlizedanddemoisturized
2.
The conductivity of RBD coconut oil can be improved
Laboratoryaged(2weeks)
further by dehydration and neutralization. The sample
Laboratoryaged(5weeks)
which was neutralized, heated and cooled under low
Laboratoryaged(7weeks)
10
Fieldaged(140months)
humid conditions showed the maximum reduction of its
conductivity at 75% compared to non-treated oil
samples.
1
3. When copra type coconut oil is thermally aged; its
interfacial tension reduces, acidity level increases,
dielectric loss increases, and conductivity level
increases. The formation of fatty acids by hydrolysis
0.1
0.0001
0.001
0.01
0.1
1
10
100
1000
might have caused those parameter changes. On the
Frequency[Hz]
contrary, the breakdown voltage remains above the
recommended level showing the positive aspects of
Figure 20. Variation of measured permittivity with respect of frequency for
insulation properties of the coconut oil.
field aged coconut oil and thermally aged coconut and mineral oil.
4. When copra type coconut oil is subjected to thermal and
1000
Newneutrlizedanddemoisturized
electrical faults, its dielectric losses and conductivity
Laboratoryaged(2weeks)
levels increase due to the formation of fatty acids.
Laboratoryaged(5weeks)
100
However, the obtained dissolved hydrocarbon gasses are
Laboratoryaged(7weeks)
Fieldaged(140months)
negligible compared to mineral oil.
10
5. The coconut oil filled sealed type distribution
1
transformer works well during 11 years of service
without having any insulation failures. The condition
0.1
of the field-aged coconut oil after 11 years was
relatively good compared to laboratory-aged oil
0.01
samples.
0.001
6. The frequency dielectric spectroscopy (FDS)
0.001
0.01
0.1
1
10
100
1000
measurement provides useful information about the
Frequency[Hz]
condition of the tested oil samples. The FDS results
Figure 21. Variation of measured loss tangent with respect of frequency for
were in good agreement with the chemical tests.
field aged coconut oil and thermally aged coconut and mineral oil.
7. The FDS can be modeled by Cole-Cole and inverse
The viscosity of the CFA sample was 29.3 cSt at 40 0C
power dependence models. The estimated conductivity
which was same as the viscosity of the new coconut oil sample
values obtained from those models can be used to
used in [13]. The color of the CFA sample was similar to the
interpret the condition of the insulation.

IEEE Transactions on Dielectrics and Electrical Insulation

Vol. 20, No. 3; June 2013

Based on the above observations gathered from the

laboratory investigations and field performance, it can be
concluded that, coconut oil can be a good choice as an
alternative to the conventional mineral oil for power
transformers. As far as the long-term performances of
power transformers are concerned, further studies are
suggested to analyze the aging aspects of treated coconut
oil together with transformer solid insulation.

ACKNOWLEDGMENT
The authors would like to express their deep gratitude to
Ceylon Electricity Board Assets management branch for
conducting the tests, Lanka Transformers PLC for
providing material samples, Department of Chemical and
Process Engineering of University of Peradeniya for aging
the samples and Department of Electrical and Electronic
Engineering and Department of Mechanical Engineering of
University of Peradeniya for conducting the tests. A special
thank goes to Mr. W.M.D. Wijesundara, Ceylon Electricity
Board for assisting the field investigations.

REFERENCES
[1]
[2]
[3]
[4]
[5]

[6]

[7]
[8]

[9]
[10]
[11]

[12]

[13]

J. Singh, Y. R. Sood, R. K. Jarial, and P. Verma, Condition

Monitoring of Power Transformers-Bibliography Survey, IEEE
Electr. Insul. Mag., Vol. 24, No. 3, pp. 11-25, 2008.
S. P. Balaji, I. P. M. Sheema, G. Krithika, and S. Usa, Effect of
Repeated Impulses on Transformer Insulation, IEEE Trans.
Dielectr. Electr. Insul., Vol. 18, No. 6, pp. 2069-2073, 2011.
A. A. Pollitt, Mineral Oils for Transformers and Switchgear,
Journal of the Institution of Electrical Engineers - Part I: General,
Vol. 89, No. 20, pp. 366- 367, 1942.
T. V. Oommen, Vegetable Oils for Liquid Filled Transformers,
IEEE Electr. Insul. Mag.,Vol. 18. No. 1, pp. 6-11, 2002.
Q. Liu, Z. D. Wang, and F. Perrot, Impulse Breakdown Voltages
of Ester-based Transformer Oils Determined by Using Different
Test Methods, IEEE Conf. Electr. Insul. Dielectr. Phenomena
(CEIDP), pp. 608-612, Oct. 2009.
A. Guushaa, E. W. Westenbrink, C. Rogers, A. S.Vaughan,and S.
G. Swingler, Aging of Biodegradable Oils and Assessment of
Their Suitability for High Voltage Applications, IEEE Trans.
Dielectr. Electr. Insul., Vol. 18, No. 3, pp. 728-738, 2011.
Imad-U-Khan, Z.D. Wang, I. Cotton and S. Northcote, Dissolved
Gas Analysis of Alternative Fluids for Power Transformers,
IEEE Electr. Insul. Mag., Vol. 23, No. 5, pp. 5-14, 2007.
J.-I. Jeong, Jung-SikAn and C.-S. Huh, Accelerated Aging
Effects of Mineral and Vegetable Transformer Oils on Medium
Voltage Power Transformers, IEEE Trans. Dielectr. Electr.
Insul., Vol. 19, No. 1, pp. 156-161, 2012.
S. Tenbohlen and M. Koch, Aging Performance and Moisture
Solubility of Vegetable Oil for Power Transformers, IEEE Trans.
Power Delivery, Vol. 25, No. 2, pp. 825-830, 2010.
D. Martin, N. Lelekakis, W. Guo, and Y. Odarenko, Further
Studies of a Vegetable-Oil-Filled Power Transformer, IEEE
Electr. Insul. Mag., Vol 27, No. 5, pp. 6-13, 2011.
G .K. Frimpong, T.V. Oommen and R. Asano, A Survey of Aging
Characteristics of Cellulose Insulation in Natural Ester and
Mineral Oil, IEEE Electr. Insul. Mag., Vol. 27, No. 5, pp. 36-48,
2011.
J. Li, Z. Zhang, P. Zou, S. Grzybowski, and M. Zahn, Preparation
of a Vegetable Oil-based Nanofluid and Investigation of Its
Breakdown and Dielectric Properties, IEEE Electr. Insul. Mag.,
Vol. 28, No. 5, pp. 43-50, 2012.
J.R. Lucas, D.C. Abeysundara, C. Weerakoon, K.B.M.I. Perera,
K.C. Obadage and K.A.I. Gunatunga, Coconut Oil Insulated
Distribution Transformer, Annual Conf. of IEE Sri Lanka,
Colombo, Sri Lanka, pp. 1-5, 2001.

897

[14] S. Ranawana, C.M.B. Ekanayaka, N.A.S.A. Kurera, M.A.R.M.

Fernando and K.A.R. Perera, Analysis of Insulation
Characteristics of Coconut Oil as an Alternative to the Liquid
Insulation of Power Transformers, IEEE 3 rd Intl. Conf. Industrial
and Information Systems (ICIIS), pp 1-5, 2008.
[15] CODEX STAN 210, Codex standard for named vegetable oil,
1999.
[16] D. Hinduja, Gayathri, C.S. Kalpage and M.A.R.M. Fernando,
Laboratory Investigation of Treated Coconut Oil as Transformer
Liquid Insulation, IEEE 6 th Intl. Conf. Industrial and
Information Systems (ICIIS2011), Kandy, Sri Lanka, pp. 108-113,
2011.
[17] I. L. Hosier, A. Guushaa, E. W. Westenbrink, C. Rogers, A. S.
Vaughan, and S. G. Swingler, Aging of Biodegradable Oils and
Assessment of Their Suitability for High Voltage Applications,
IEEE Trans. Dielectr. Electr. Insul., Vol. 18, No. 3, pp. 728-738,
2011.
[18] N. Lelekakis, D. Martin, W. Guo and J. Wijaya, Comparison of
Dissolved Gas-in-Oil Analysis Methods Using a Dissolved Gas-inOil Standard, IEEE Electr. Insul. Mag., Vol. 27, No. 5, pp. 29-35,
2011.
[19] U. Gafvert, L. Adeen, M. Tapper, P. Ghasemi and B. Jonsson,
Dielectric Spectroscopy in Time and Frequency Domain Applied
to Diagnostics of Power Transformers, Intl. 6th Conf. Properties
and Application of Dielectric Materials (ICPADM), Xian,
China, Vol. 2, pp. 825-830, 2000.
[20] T.K. Saha, Review of Modern Diagnostic Techniques for
Assessing Insulation Condition in Aged Transformers, IEEE
Trans. Dielectr. Electr. Insul., Vol. 10, No. 5, pp. 903-917, 2003.
[21] C. Ekanayake, S.M. Gubanski, K.D. Mularachchi and M.A.R.M.
Fernando, Diagnostic of Power Transformers in Sri Lanka:
Application of
Dielectric Spectroscopy in Frequency Domain,
IEEE Conf. Electr. Insul., Cincinnati, USA, pp. 593-596, 2001.
[22] F.M. Dayrit, O.E.M. Buenafe, E.T. Chainani, I.M.S. de Vera,
I.K.D. Dimzon, E.G. Gonzales and J.E.R. Santos, Standards for
Essential Composition and Quality Factors of Commercial Virgin
Coconut Oil and its Differentiation from RBD Coconut Oil and
Copra Oil, Philippine J. Sci., Vol. 136, No. 2, pp. 119-129, 2007.
[23] IEC 60296, Ed. 2,Specification for unused mineral insulating
oils for transformers and switchgear,1982.
[24] G. Ajithkumar, Analysis, Modification and Evaluation of the Cold
Flow Properties of Vegetable Oils as Base Oils for Industrial
Lubricants, Ph.D. Thesis, Cochin University of Science and
Technology, India, 2009.
[25] C.F. Ten, M.A.R.M. Fernando and Z.D. Wang, Dielectric
Properties Measurements of Transformer Oil, Paper and
Pressboard with the Effect of Moisture and Ageing, IEEE Conf.
Electr. Insul. Dielectr. Phenomena (CEIDP), Vancouver, Canada,
pp. 727-730, 2007.
[26] B.S.H.M.S.Y. Matharage, M.A.A.P. Bandara, M.A.R.M.
Fernando, G.A. Jayantha, and C.S. Kalpage, Aging Effect of
Coconut Oil as Transformer Liquid Insulation- Comparison with
Mineral Oil, IEEE 7th Intl. Conf. Industrial and Information
Systems (ICIIS), Chennai, India, pp. 1-6, 2012.
[27] M. H. Abderrazzaq, and F. Hijazi, Impact of Multi-filtration
Process on the Properties of Olive Oil as a Liquid Dielectric,
IEEE Trans. Dielectr. Electr. Insul., Vol. 19, No. 5, pp. 16371680, 2012.
[28] IEC 60599, Ed. 2, Mineral Oil-impregnated Electrical Equipment
in Service-Guide to the Interpretation of Dissolved and Free Gases
Analysis, 1993.
[29] B.S.H.M.S.Y. Matharage, M.A.R.M. Fernando, E. Tuncer,
M.A.A.P. Bandara, and C.S. Kalpage, Coconut Oil as
Transformer Liquid Insulation- Ageing and Simulated Thermal
and Electrical Faults, IEEE Conf. Electr. Insul. Dielectr.
Phenomena (CEIDP), Montreal Canada, pp. 839-842, 2012.
[30] C. Perrier, M. Marugan, and A. Beroual, DGA Comparison
between Ester and Mineral Oils, IEEE Trans. Dielectr. Electr.
Insul, Vol. 19, No.5, pp. 1609-1614, 2012.
[31] IEC 60422, Ed. 2, Supervision and Maintenance Guide for
Mineral Insulating Oils in Electrical Equipment, 1989.

898

B. S. H. M. S. Y. Matharage et al.: Performance of Coconut Oil as an Alternative Transformer Liquid Insulation

M. A. R. M. Fernando (M07-SM10) was born in

Colombo, Sri Lanka in 1966. He received the B.Sc. Eng.
degree from the University of Peradeniya, Sri Lanka in
1993, the Tech. Lic., degree from the Royal Institute of
Technology, Stockholm Sweden in 1997 and the Ph.D.
degree from the Chalmers University, Gothenburg,
Sweden in 1999. He started his career as an instructor in
the University of Peradeniya in 1993 and at present, he is
a Professor. He was a guest researcher at HTWS, Zitau Germany in 1997, a
guest researcher in Chalmers University of Technology SWEDEN in 2000
and an academic fellow in the Manchester University UK in 2006/2007. He is
a chartered Engineer and an International Professional Engineer. He was the
founder chair of IEEE Sri Lanka Power and Energy Society Chapter in 2010,
the general chair of IEEE fourth International Conference on Industrial and
Information Systems in 2009 and the chair of IEEE Sri Lanka central region
subsection in 2009/2010. His research interests include condition monitoring,
alternative insulation, problems related to outdoor insulation.
B. S. H. M. S. Y. Matharage was born in Kandy, Sri
Lanka in 1987. He received the B.Sc. Eng. degree from
the University of Peradeniya, Sri Lanka in 2011. He is
currently working as a Ph.D. student in the University of
Manchester, UK. His main research interests are
Alternative insulation materials, aging effects and
dielectric modeling.

M. A. A. P. Bandara was born in Ratnapura, Sri Lanka in

1986. He received the B.Sc. Eng. degree from University of
Peradeniya, Sri Lanka in 2010. He started his career as an
instructor in the University of Peradeniya from 2010 to
2011. He is currently working as an operation engineer in
the Kotmale Power Station, Sri Lanka. His main research
interests are condition monitoring and identification of faults
of power and current transformers using different methods.

G. A. Jayantha was born in Matara, Sri Lanka in

1959. He received the B.Sc. Eng. degree from
University of Peradeniya, Sri Lanka in 1983. He
started his career as the engineer in-charge of
Udawalawe power station Sri Lanka and later
worked as the senior electrical engineer of
Samanalawewa Hydro Power project (120 MW),
Kukuleganga Hydro power project (70 MW). From
1993-2004 he worked as the senior electrical engineer of Mahaweli
Hydro Power Complex in Sri Lanka. From 2004 -2011 he worked as the
Chief engineer in-charge of condition monitoring of generators,
transformers and HV circuit breakers of all the hydro power stations in
Sri Lanka. From 1993 he was involved in fault diagnosing, major
repairs and overhaul work of hydro power stations in Sri Lanka. He has
served in many technical committees of hydro power projects in Sri
Lanka. Since 2011 he has been working as the Deputy General Manager
of Generation projects where he is the in-charge of refurbishment and
modernization of hydro power stations in Sri Lanka. He has more than
28 years of service as a hydro power engineer. He is a charted engineer.
Condition monitoring and fault diagnosing of Generators and
Transformers are his main research interest.

C. S. Kalpage was born in Kandy, Sri Lanka in 1968.

He received the B.Sc. Eng. degree from the University
of Moratuwa, Sri Lanka in 1994, and the Ph.D. degree
from the University of Birmingham, UK in 2005. He
served in industry at capacities of Production Engineer
and Chemical Engineer for nearly four years before
joining the University in 1998. At present, he is the
Head of the Department of Chemical and Process
Engineering. His research interests include bio-diesel, bio-ethanol and
environmental engineering related research particularly of wastewater
treatment by low cost materials.