You are on page 1of 11

D

o
b
l
e

E
n
g
i
n
e
e
r
i
n
g




A STUDY ON THE RELATIONSHIP BETWEEN PASSIVATOR AND GAS FORMATION


By: Annalie Lombard
Eskom Enterprises Division
and
John Osborne
Osborne Analytical


Corrosive sulphur problems in insulating oil have been experienced internationally and a number of
failures have been well documented. Passivation of the oil was offered as a quick solution for this
problem at a time when panic struck the electrical industry. The addition of passivator was proposed
based on limited experience with insulating oil, but based more on the experience with lubricating oils.

Immediately following the addition of passivator increased dissolved gas concentrations, particularly
hydrogen, have been observed in some transformer oils which were dosed with 100 ppm passivator.

Laboratory studies sought to establish any clear relationship between dissolved gas and passivator
content under conditions of temperature and exposure to copper (the metal passivated). Tests were
performed in ambient air as well as in sealed containers.

The distinct relationship of increased hydrogen with increased Irgamet 39 was observed. If a limited
amount of oxygen is available more hydrogen is produced. If oxygen is eliminated, very little of any
gas is produced. The copper was found to enhance or catalyse gas production.

This paper discusses the laboratory observations as well as passivator related experience with
transformers in the field.


1. BACKGROUND

The South African transformer oil or insulating oil market has been dominated by the usage of
uninhibited Naphthenic oil for several decades. This however is rapidly changing and may have
catastrophic consequences for uninformed users, regenerators and recyclers of transformer oil.

The bulk of electrical insulating oil used in Southern Africa is mineral oil mixtures of hydrocarbon
compounds originating as crude oil from different geographical areas around the world. As a result
oils currently in use comprise of slightly different chemical structures, most with higher amounts of
Naphthenic compounds than Paraffinic compounds, but also containing low amounts of Polyaromatic
hydrocarbons (PAH) as well as aromatic compounds. Paraffinic refers to carbon atoms bonded to one
another in straight chains or branched structures, whereas carbon atoms bonded to one another to
form 5, 6 or 7 carbon rings are referred to as Naphthenic.

The primary functions of transformer or insulating oil are to insulate the primary windings from the
secondary windings and ground, as well as to function as a heat transfer medium to transfer the heat
from the windings to external cooling equipment [1].

It is thus important that the oil posses good insulation and dielectric properties. This will include, but
are not limited to low moisture content, high dielectric breakdown voltage and good anti-ageing
properties thus a low tendency to form oxidation breakdown products such as acids and sludge. The
full detail regarding the required characteristics will normally be specified by the equipment
manufacturer as well as the end-user.

These requirements from the power industry as well as environmental legislation placed refiners
under pressure to control the refining process and deliver a non-corrosive Naphthenic oil at an
economically viable price. The end product conforming to corrosive sulphur, with negative gassing
characteristics and other requirements are defined by the crude oil chemistry that is obtainable as well
as the refining process utilized by a specific refiner. As a result only some refiners are capable of

D
o
b
l
e

E
n
g
i
n
e
e
r
i
n
g


producing uninhibited (class U) products whilst others can produce only inhibited (class I) and others
class U and I. [1]

Uninhibited oil must contain natural inhibitors, which are semi reactive and probably sulphur
containing. However, the identification of such unique compounds present in the oil are yet to be
achieved as the composition of oil, which comprises in excess of thousand different hydrocarbon
structures are currently not possible. Corrosive sulphur testing have become more stringent resulting
in the removal of natural inhibitors during more demanding refining processes to achieve stable oil.
This has reduced the supply of available uninhibited class U transformer oil in the marketplace.

1.1 CORROSIVE SULPHUR AND PASSIVATION

A number of processes are employed during refining following crude fractionation.
Processes such as solvent extraction, hydrotreating and hydrocracking are specifically
employed during transformer oil production, each resulting in an end product with different
chemical characteristics.

The amount of sulphur present in oil and thus possibly the corrosiveness of transformer oil
can be controlled by a refiners process. If the requirement is to eliminate or greatly reduce
the risk of corrosive sulphur then the user should look for severely refined transformer oil
which has corrosive sulphur compounds stabilised. It must be noted however that the more
sulphur is stabilised within the oil the less likely it will make a totally uninhibited class U oil,
this implies that inhibitors will be added resulting in class I products. The natural inhibiting
process of uninhibited oil depends on the remaining sulphur compounds or the natural
inhibitors in the oil during refining. If the oil choice is to use totally uninhibited oil then it is
advised to choose oil that has been fully certified as free of corrosive sulphur by specific
methods. The chemistry to support totally uninhibited oil is diminishing in the Naphthenic
crude supply, resulting in limited availability of class U oils. The primary crude sources that
have traditionally been rich in the natural inhibitors to make uninhibited class U oil have
become very hard to find. The process of refining to produce an inhibited class I transformer
oil lends itself to ensuring that the corrosive sulphur is stabilised. This also allows a refiner to
pull from a wider variety of Naphthenic crudes to keep up with demand.

As a result of the above gassing tendency, inhibitor and passivator content as well as
corrosive sulphur amongst other important properties needs to be specified during oil
purchases. With oil having a highly positive gassing tendency, the application of the oil
needs to be carefully considered as mentioned above. When oil contains passivators and
inhibitors, these need to be monitored during the life of the equipment to ensure that the
levels of depletion do not drop to less than 25% of the original added value. Whilst the
residual values of these products need to be monitored following purification and
regeneration. Certain in-line processing units are also known to remove these chemicals.

Corrosive sulphur problems in insulating oil have been experienced internationally and a
number of failures have been well documented. Passivation of the oil was offered as a quick
solution of this problem at a time when panic struck the electrical industry. The addition of
passivator was proposed based on limited experience with insulating oil, but based more on
the experience with lubricating oils.

Immediately following the addition of passivator increased dissolved gas concentrations,
particularly hydrogen, have been observed in some transformer oils which were dosed with
100 ppm passivator.

Irgamet 39 is the trade name of N-Bis (2-ethylhexyl)-aminomethyl-tolutriazole, and is
manufactured by the CIBA company. Its function in a mineral oil is as a metal deactivator or
passivator. The recommended concentration in oil is 100 ppm (mg.kg
-1
). The chemical
composition is C
24
H
43
N
4
which translates to 74.4% carbon, 11.1% hydrogen and 14.5%
nitrogen.

D
o
b
l
e

E
n
g
i
n
e
e
r
i
n
g


Irgamet 39 has a high viscosity (80 mm
2
/sec), low melting point (-30C) and solubility of
5% in oil and esters, and 0.01% in water and a density of 0.95 g/ml. The boiling point and
decomposition temperature is not documented. The chemical structure is given below.

C
4
H
9
-CH-CH
2
-
CH
3
CH
2
-CH
2
-CH-C
4
H
9
CH
3
CH
2
CH
3
N
N N
N
CH
3
C
4
H
9
-CH-CH
2
-
CH
3
CH
2
C
4
H
9
-CH-CH
2
-
CH
3
CH
2
-CH
2
-CH-C
4
H
9
CH
3
CH
2
-CH
2
-CH-C
4
H
9
CH
3
CH
2
CH
3
N
N N
N
CH
3


Oil samples have been received from transformers which have exhibited high levels of
dissolved hydrogen when a high concentration of passivator Irgamet 39 has been dosed.
Although the recommended dose is 100 mg.kg
-1
, concentrations up to 2000 mg.kg
-1
have
been present in some Southern African oils. The hydrogen introduced with the Irgamet 39
is covalently bound, as it is with the mineral oil component. Hydrogen gas is released in a
transformer when there is a certain fault (i.e. an electrical discharge). Hydrogen is released
by the breaking of hydrogen-carbon bonds in the hydrocarbon material of the oil. If
increased hydrogen is present due to excess passivator, this probably indicates that
hydrogen is more easily released from the Irgamet 39 molecule than from the oil material.

At high levels of hydrogen in a transformer oil sample, partial discharges are indicated. If
significant concentrations of hydrogen are introduced through the addition of excess
passivator, electrical faults will be falsely diagnosed, and a unit may be taken out of service
for a repair which turns out not to be necessary. If a distinct correlation between dissolved
hydrogen gas and Irgamet 39 concentration can be found, a maximum dose of the
passivator must be enforced. By performing tests with increasing concentrations of Irgamet
39 under different conditions, any trends should indicate levels at which unacceptable
hydrogen concentrations are produced.

2. OBJECTIVES

Laboratory studies sought to establish any clear relationship between dissolved gas and passivator
content under conditions of temperature and exposure to copper (the metal passivated). Tests were
performed in ambient air, elevated temperatures as well as in sealed containers, with argon and air
respectively.

The distinct relationship of increased hydrogen with increased Irgamet 39 had to be observed. The
effect of oxygen on the reaction needed to be monitored and the absence of oxygen versus the
amount of gas produced had to be observed. The effect of the copper on the whole reaction had to be
established in order to see if copper reacts as a catalyst to the production of gas. A few key questions
had to be considered, these are:

What effect do various conditions have on the production of gas?
How much hydrogen is produced by this process?
Is there a difference between new and regenerated oil with respect to types and rate of
gas production?
Which other gases are produced following passivation?
Will in-line processes such as Dry-Keep drying remove the passivator?

3. PROCEDURES FOLLOWED AND OBSERVATIONS

New uninhibited oil was obtained courtesy of Engen, of type Poweroil TO 1020 (60U) - uninhibited.
Irgamet 39 was obtained from Ciba (but with no certificate of purity). The purity was accepted to be
approximately 95%. A stock solution of 100 g of oil containing 1% Irgamet (10g.l
-1
) was prepared.
Further dilutions of this stock in the new oil were used in the experiments.

D
o
b
l
e

E
n
g
i
n
e
e
r
i
n
g


Each concentration was prepared by mass in the open laboratory (i.e. no measures were taken to
eliminate the elements of air). The main focus of the experiment was to determine any relationship
between the release of hydrogen and concentration of the passivator, under the test conditions. The
influence of dissolved ambient air on this experiment was expected to be negligible, but may play a
role in sealed transformers. Any variations in other gases (methane, carbon monoxide, ethane,
ethylene and acetylene) would be indicated by the dissolved gas analysis (DGA).

All samples were prepared on the same day. 11 ml of the prepared oil was measured into a
headspace vial. After the exposure/reaction period, samples were analysed for dissolved gases by
headspace capillary Gas Chromatography (GC) and for residual passivator (Irgamet 39)
concentration by High Performance Liquid Chromatography (HPLC).

3.1 Experiment 1 - Dissolved Gas Production versus Passivator Concentration with New
Oil.

During this experiment new oil was passivated and heated in an oven at 100C. The rate
and type of gas was monitored.

The stock (1% Irgamet 39) oil was diluted in the pure oil to produce 6 different
concentrations of the passivator between 100 ppm (the recommended dose) and 2000 ppm
(the maximum encountered). A blank oil (0 ppm added Irgamet 39) was also used.

For each concentration, two vials containing the oil only and two with oil plus a copper strip
were prepared, one of each was placed in the oven, and the other was kept at ambient
temperature (22C 2C) for 1 week (168 Hours). After this the DGA was performed and the
remaining sample was analysed for Irgamet 39 content by HPLC. An empty sealed vial
was heated for a week to establish whether there was any contribution from vial, cap or
seal.

3.2 Experiment 2 Dissolved Gas Production Rate with 100 mg.l
-1
Passivator

with New
Oil.

Samples of oil containing 100 mg.l
-1
of Irgamet 39 was prepared and placed in an oven at
100C, 20 samples were left at ambient temperature. At intervals of 24 hours, one vial with a
copper strip and one without were removed from the oven and cooled. Together with one
vial with a labeled, these four were submitted for DGA and Irgamet 39 determination. This
test was continued for 1 week. An extra set was removed after 2 weeks for another data
point.

3.3 Experiment 3 - Dissolved Gas Production rate in Argon with New Oil.

After the first few samples were analysed, it was apparent that only about 70-80% of the air
had been eliminated by flushing. The flushing method was therefore only partially effective
or air was able to return to the vial before sealing. The vial of oil (and a cap) and the argon
outlet were enclosed in a light plastic bag. This was allowed to fill during a flush, then
squeezed out whilst the argon flowed. This was repeated several times during the (approx. 2
minute) flush after which the cap (inside the bag) was crimped on (with the crimping tool
outside the bag). This latter action punctured the bag, so a new bag was used for each test
sample. In this way >99.5% of air was eliminated.

This experiment was conducted for heated samples only, since gas concentrations were
very low for ambient tests in all previous experiments, and were not expected to reveal new
information in this experiment.

The data produced in this experiment gave some interesting observations. Firstly the
hydrogen produced in the atmosphere of less than 0.5% air was in concentrations between
100 and 200 vpm, about one tenth of that produced in the reduced (25%) air experiments.
Also the concentrations did not increase with time.

D
o
b
l
e

E
n
g
i
n
e
e
r
i
n
g


The oxygen was depleted within the first 24 hours of the experiment and the hydrogen
appeared within a few hours (the first sample was removed after 3 hours). After that time
the measurements fluctuated but showed no distinct trend. (The fluctuations are suspected
to be due to small amounts of air leaking past the seals on some of the vials.)

3.4 Experiment 4 Dissolved Gas Production versus Concentration in Ambient Air with
Regenerated Oil.

This was a repeat of Experiment 1, however during this experiment uninhibited regenerated
oil was used to conduct the experiments.

3.5 Experiment 5 Dissolved Gas Production versus Passivator Concentration in Argon
with Regenerated Oil.

Over 99% of the air was eliminated in most of these tests, by means of the Argon flush
procedure explained in Experiment 3.

3.6. Experiment 6 - Dissolved Gas Production Rate with 100 mg.l
-1
Passivator in Ambient
Air with Regenerated Oil.

500g of oil containing 100 mg.l
-1
of Irgamet 39 was prepared from the stock (1% oil) and
pure oil. 11 ml of this was accurately measured into each of 40 headspace vials. Into each
of 20 of these was placed a single standard copper strip. All vials were sealed (with
aluminium liners). 20 of these vials (10 with copper strip and 10 without) were placed in an
oven at 100C, and the remaining 20 were left at ambient temperature. At intervals of
approximately 24 hours one vial with copper strip and one without were removed from the
oven and cooled. Together with one vial with copper strip and one without which had been
at ambient, adequately labeled, these four were submitted for DGA and Irgamet 39
determination. This test was continued for 8 days.

3.7 Experiment 7 - Dissolved Gas Production Rate with 100 mg.l
-1
Passivator in Argon
with Regenerated oil.

The sample preparation was the same as experiment 6, except that the vials were
thoroughly argon-flushed before sealing.

Hydrogen production of up to 200 vpm with copper, and about 120 vpm without was
measured in the heated tests. After an initial increase, the concentrations did not increase
with time. This contrasted with the new oil in which hydrogen concentrations increased to
around 2000 vpm in the same time (with heating and with or without copper).

The oxygen was depleted within the first 48 hours of the heated experiments (slower than
with new oil which was totally consumed by 24 hours). There was only a slight deterioration
in the Irgamet 39 concentration with time.

3.8 Experiment 8 - Passivator removal by Molecular Sieve

In order to observe the possible removal of passivator by Molecular Sieve used in on-line
drying systems such as Dry-Keep an experiment was set up to establish if removal
occurs. A mass (approx. 100 g) of molecular sieve granules was placed in a dry extraction
column. Approx 200 g of the new oil containing Irgamet 39 was added to the top of the
Molecular Sieve. This was allowed to drain through under gravity (approx. 3 ml/min at first)
until it stopped dripping (about 2 hours). The oil collected was stirred for 10 minutes
(magnetic stirrer), then sampled for Irgamet 39 determination.





D
o
b
l
e

E
n
g
i
n
e
e
r
i
n
g


Concentration (mg/kg) 104.97 1052.1
Before 100.9 940.58
Mass Irgamet 39 (mg) 20.11 188.79
Mass of molsieve (g) 100.01 100.018
Mass of oil (g) 199.28 200.714
Concentration I39 (mg/l) 87.93 876.03
% Removal 12.85 6.86
Mass Irgamet 39 removed (mg) 2.58 12.96

3.9 Eskom Transformer Data Passivator addition versus Gas produced


% for SEALED
(COSERVATOR BAGS)
% for NOT SEALED
Amount of transformers 29 26
No gas produced 3 15
Hydrogen produced 97 85
Methane produced 61 39
CO/CO2 produced 64 62
Ethane produced 45 27
Ethylene produced 45 27
Acetylene produced 6 15


4. DISCUSSIONS

4.1. Experiment 1 and 2

There is a distinct correlation between passivator (Irgamet 39) concentration and
hydrogen generation in transformer oil, particularly at the elevated temperatures at which
transformers may operate. The relationship is proportional. Copper seems to reduce the
hydrogen production rate slightly. Hydrogen generation coincides with reduction in Irgamet
39 concentration indicating that it originates from the passivator and not from the oil.
When the passivator concentration had depleted, so did the hydrogen production. It may
have been consumed during normal oxidation reactions. Pahlavanpour et al reports that the
Irgamet 39 does not decompose at elevated temperatures (in sealed transformers),
except when the oil is oxidised [2]. That report also reports the observation that the
passivator remains irreversibly bound to the copper surface, so that maintaining an
excessive dose has no advantage.

Production of alkane gases (ethane and methane) seemed to be stimulated by the
presence of passivator, but was not proportional to its concentration. This suggests a mild
catalytic reaction, (i.e. Irgamet 39 participates in but is not consumed by this reaction).

Carbon monoxide was initially consumed in the presence of Irgamet 39, but then
increased gradually with Irgamet 39 concentration increase. In the time progression, the
carbon monoxide steadily increased when copper was present, even when the Irgamet 39
had depleted, whereas when copper was absent, carbon monoxide depleted after the
Irgamet 39 had been used up. This suggests the copper has a reducing effect, preventing
the conversion of carbon monoxide to carbon dioxide, which probably happened in the
absence of copper.

For the 100 mg.kg
-1
Irgamet 39 test, the passivator concentration depleted by 95% within
one week at the elevated temperatures, both with and without copper. This raises questions
of its lasting applicability, particularly in transformers that are not sealed and in which
circulation is slow. Regular (maintenance) dosing may be necessary, although this will be
coupled with unwanted gas production. However, once initially applied, the adhesion of the
passivation layer is permanent [2] thus only an initial excess of the Irgamet 39 is
necessary, after which no additions will be required. In the event of refurbishment of a
transformer, a fresh dose of Irgamet 39 may be appropriate to passivate any copper
surface that may be new or physically affected.

D
o
b
l
e

E
n
g
i
n
e
e
r
i
n
g



These experiments were prepared at ambient conditions i.e. without any preliminary
degassing or exclusion of the elements of air (particularly oxygen and carbon dioxide). The
tests were performed on new oil of one type from one manufacturer. It is possible that other
oils may have different effects when Irgamet 39 is used.

The effects reported may also involve a factor of incomplete mixing. If the circulation of
insulating fluid in the transformer is slow, it will take some time before the distribution of the
additive is relatively even throughout the oil volume. There may be localised hydrogen
production, which may show up depending on how close it is to the sample point. With
forced circulation, the time will be relatively short, and the existence of localised dissolved
hydrogen pockets will be less evident. Also if the passivator has not circulated well, parts
of the copper windings may not be passivated before the Irgamet deteriorated (particularly
for free breathing units). This aspect may also deserve study.

4.2 Comparative Observations Experiments 1-3

4.2.1 Hydrogen

The overall concentrations of hydrogen varied differently at the different air levels. When air
was limited the concentrations of hydrogen were highest and the hydrogen concentration
was less when no air limitation was attempted. This observation indicates that the reactions
involving hydrogen depend strongly on the air (oxygen) present. There appears to be two
types of reaction, those generating hydrogen and those consuming it. When air is absent,
no significant hydrogen is produced. With limited air, hydrogen is produced more effectively
but consumed slowly. When ample air is present, hydrogen is probably produced at an
optimal rate, but also consumed, (probably by some reaction during oxidation or perhaps
producing moisture with the oxygen). In all cases, the passivator has a role in accelerating
hydrogen production.

4.2.2 Hydrocarbon Gases

Average concentrations after one week exposure with 100 mg.l
-1
presence of Irgamet 39
are given in Table 4.1.1. (Not included is the enhanced ethylene production observed at
reduced air levels at ambient temperature, 60 vpm with copper present and 25 vpm
without). Acetylene was present in detection limit amounts or less.

Table 4.1 Hydrocarbon gas production for heated tests at different air levels

Methane concentration (vpm)
Argon flush Air concentration Heated with copper Heated only
Nil 100% 500 200
Partial 25% 150 50
Full <0.5% 15 15
Ethylene concentration (vpm)
Argon flush Air concentration Heated with copper Heated only
Nil 100% 9 6
Partial 25% 12 2
Full <0.5% 8 2
Ethane concentration (vpm)
Argon flush Air concentration Heated with copper Heated only
Nil 100% 300 250
Partial 25% 350 150
Full <0.5% 14 10

Clearly lower concentrations of air present (oxygen) resulted in lower levels of hydrocarbon
formation, and hydrocarbons were formed irrespective of the quantity of passivator present,
but were enhanced by the metal (copper windings). Ethylene formed rapidly and reacted
rapidly with full oxygen, i.e. it is intermediate in the formation of something else (ethane).
Formation was found to be faster than consumption when oxygen was limited.

D
o
b
l
e

E
n
g
i
n
e
e
r
i
n
g


4.2.3 Oxides of Carbon

Carbon oxide (both CO and CO
2
) production is affected by Irgamet 39 concentration, and
is limited by the amount of oxygen available. Trends are similar for both gases and the table
4.1.2 shows the 1-week gas concentrations at 100 mg.l
-1
concentration of the passivator at
different air levels. At ambient temperatures in full air the carbon dioxide was steady around
600 vpm and carbon monoxide <10 vpm for all tests. In reduced air the ambient
concentrations were CO
2
150 vpm and CO <1 vpm, and with full flush: 10 vpm and <1 vpm
respectively.

Table 4.1.2 Hydrocarbon gas production for heated tests at different air levels

Carbon Monoxide concentration (vpm)
Argon flush Air concentration Heated with copper Heated only
Nil 100% 1500 1500
Partial 25% 350 350
Full <0.5% 40 40
Carbon Dioxide concentration (vpm)
Argon flush Air concentration Heated with copper Heated only
Nil 100% 3600 1400
Partial 25% 2500 800
Full <0.5% 150 150

The relationship between copper and the monoxide concentrations is clear when air is
available.

Unless otherwise stated, the comparisons were made using data from the heated tests with
copper present. These tests provided higher gas concentrations and enabled better
trending than the other tests.

4.3 Comparative Observations Experiments 4-7

4.3.1 Hydrogen

There was a distinct correlation between Irgamet 39 concentration and hydrogen
production in experiments with both the new and the used oil. The presence of copper had
little influence on the hydrogen concentration. It is obvious that this reaction occurred far
more readily in the new oil.

4.3.2 Passivator Behaviour

The passivator deteriorated from 100 mg.l
-1
to below 20 mg.l
-1
only with the new oil in the
presence of air. In other series minimal deterioration was observed over the week i.e when
air is (almost) absent and in either case with regenerated oil. This was supported by
comparing different concentrations of passivator, the new oilin-air tests showed most
Irgamet 39 deterioration than the others (though the higher range of concentrations made
the effect look less distinct).

4.3.3 Hydrocarbon Gases

The production of hydrocarbon gases are best depicted by methane as the others (ethane,
ethylene and acetylene) were generally produced in minimal concentrations. This is not
related to the Irgamet 39 concentration. In new oil more methane is produced in air than
when air is limited, however in the regenerated oil slightly more is produced in the absence
of air.

Methane was produced increasingly with time and it was observed that the production in
new oil was the fastest. The production was significantly reduced in an argon atmosphere,
thus the reaction involves components in the new oil and oxygen.

Ethane was produced in concentrations of up to 500 vpm in new oil with air, and not related
to passivator concentration. Low concentrations were observed in argon atmosphere, and

D
o
b
l
e

E
n
g
i
n
e
e
r
i
n
g


ethylene concentrations always low, acetylene barely detectable. Small concentrations of
ethane were produced in the regenerated oil with argon atmosphere only, and not related to
passivator concentration. Traces of ethylene were produced, with acetylene barely
detectable.

4.3.4 Oxides of Carbon

In contrast to the trends observed in the new oil, the carbon monoxide production does not
appear to be significantly influenced by the passivator concentrations for regenerated oil. In
the new oil, more CO was produced in air, whereas in regenerated oil the concentrations in
argon atmosphere are larger than in air.

The production of carbon monoxide increases with time, and is far faster in the new oil than
in the regenerated oil.

4.2.4 Experiment 8 Passivator Removal by Molecular Sieve

Molecular Sieve has an effect of removing Irgamet 39 from the regenerated transformer
oil. Under the conditions used, the removal has a low efficiency. The removal rate is
dependent on concentration of Irgamet 39 in the oil. It is suspected that removal efficiency
depends on flow rate and probably temperature. The capacity for removal needs to be
determined by an experiment designed differently (e.g. extended soaking of Molecular
Sieve in an oil of excess Irgamet 39 concentration.)

4.2.5 Gas Formation Following Passivation

From actual transformer data extracted from the Eskom transformer oil analysis database it
was evident that some degree of hydrogen and hydrocarbon formation did take place
following passivation. In nearly all the cases there was an increase in the ethylene gas
content as well.

5. CONCLUSIONS AND RECOMMENDATIONS

5.1 Experimental Data

The observations of the first investigations (where air was not 100% eliminated) were
confirmed but additional tests showed that chemical behaviour in sealed conditions was
different in many ways.

There is a distinct correlation between passivator (Irgamet 39) concentration and
hydrogen generation in transformer oil, particularly at the elevated temperatures at which
transformers may operate, and even with reduced air levels.

Production of hydrogen is affected by the presence of oxygen (since nitrogen was not
detectably depleted in all experiments, but oxygen was totally consumed rapidly in the
sealed test vessels). For hydrogen, concentrations were higher compared to where the air
was reduced to 25%, but significantly reduced when <0.5% air was present. Air is initially
involved in generation of hydrogen, then reacts to consume hydrogen (in reactions to
oxidise oil, or addition reactions to saturate ethylene to ethane, or simply to combine to
produce moisture).

Irgamet 39 passivator is consumed significantly over time and with heat when oxygen is
present. When oxygen is completely eliminated, very little decomposition was observed. It
is the free or excess passivator in the oil phase that is oxidised, since the passivator bond
on the copper surface remains intact. The decomposition and hydrogen production will be
minimised by effective sealing of the transformer. Further protection can be made by
supplying a constant positive low pressure of nitrogen (or inert gas) to the unit, so that inert
gas exiting any leaking seal will prevent air (and thus oxygen) ingress. Employing the latter
may however assist in partially flushing dissolved gases out of the oil, which factor will need
to be considered for diagnostic interpretation of DGA.

D
o
b
l
e

E
n
g
i
n
e
e
r
i
n
g



As previously observed, production of alkane gases (ethane and methane) seemed to be
stimulated by the presence of passivator but not related to passivator concentration. In
reduced air, ethylene production was significant at ambient temperature, but not at 100C. It
is suspected that the ethylene is saturated to ethane more rapidly at higher temperature
and with copper present (i.e. ethylene concentration at any time is negligible in hot tests).
This reaction consumes hydrogen. Any acetylene initially formed will quickly be converted
to ethylene then ethane by the same process. Similar reactions may be expected in
unsaturated (carbon-carbon double or triple bond containing) compounds in the oil. For all
hydrocarbon gases the production/presence was minimal when air was 95.5% eliminated.

There is a definite correlation between passivator concentration and the generation of
carbon oxides. For this air is required, correlation was only observed in the presence of
sufficient air. In limited air, the correlation for CO was clear up to 600 mg.kg
-1
of passivator.
Above this the correlation was less clear, probably due to the reduced amount of oxygen
available. The CO concentrations in full and limited air were also higher when copper was
present, again suggesting catalytic action in its formation. The conversion of CO to CO
2
is
affected less than CO production.

This investigation indicates that the addition of Irgamet 39 in excess of its recommended
level (100 mg. l
-1
) has no benefit whatever, but has several detrimental effects:

increases gas generation
disturbs diagnostic trends
increases oxidation rate of the oil
excess Irgamet 39 itself is decomposed in a relatively short time (longer
for sealed systems).

The optimal passivation philosophy is probably to insure the correct dose in the oil when it is
initially added, and add small amounts after any maintenance has been performed.

The event of passivator addition should be flagged on the oil sample, so the analyst is alert
to a possible excursion in the dissolved gas trends, and will factor this in to any
recommendations made.

The comparison of the tests performed in Experiments 4-7 on the used transformer oil gave
different results than the new oil previously investigated. Trends were similar but gas
production of all gases was far less, and the deterioration of the passivator (Irgamet 39)
was minimal, in extreme contrast to the new oil.

It appears that new oil contains components that are readily oxidisable (and generally
reactive). After these components have reacted, remaining components are more stable, i.e.
less reactive in oxygen and less reactive toward the passivator. Thus it appears that the
initial passivator is consumed (in new oil) by initial reaction with these reactive components,
and not necessarily with passivation of the copper.

Generation of hydrogen again correlated with Irgamet 39 concentration, and significantly
reduced in the absence of air (in this respect the same as for the new oil), but the striking
effect was that the concentrations were one tenth of that in the new oil.

Methane concentrations were higher in minimal oxygen conditions, indicating carbon from
the oil is more amenable to reaction with hydrogen. Neither in the presence or (almost)
absence of oxygen is there any correlation with Irgamet 39 concentration, but a gradual
increase with time. Significant ethylene production was only observed in the argon-flushed
tests, gradually increasing with time but not relating to Irgamet 39 concentration. The data
for ethylene and acetylene were in too low a concentration to derive any meaningful
conclusions. The Irgamet 39 cannot therefore be related to the production of hydrocarbon
gases.

D
o
b
l
e

E
n
g
i
n
e
e
r
i
n
g


The trends in carbon dioxide and carbon monoxide production were erratic, but no distinct
correlation with passivator concentration. There were more carbon oxides in air as would be
expected. The monoxide reached 300-350 vpm in air (compared to 1500 with the new oil).
In both air and its (almost) absence when heated, carbon monoxide increased with time
then leveled off at about 100-120 hours. With lack of air it actually decreased after 120
hours. Carbon dioxide did the same in air, but kept increasing over the 200 hours without
air, suggesting that the carbon monoxide was slowly converted to the dioxide. In inert
atmosphere, concentrations reached 70 vpm for carbon monoxide and 700 vpm for the
dioxide, compared to 40 vpm and 300 vpm respectively for new oil.

When new oil is used the initial reaction between reactive components in the oil and
passivator produces hydrogen and reduces the passivator concentration. Analysis after a
determined time will indicate the more stable state of the oil. Any way of removing gas (inert
gas purging or vacuum treatment on the oil) after this period will remove initial hydrogen and
render the data from ongoing dissolved gas analysis more reliable. Also adjustment to the
passivator content should be made after the first period, when the initial deterioration is
complete.

This first period will probably depend on the oil type and the operating temperature (i.e. the
reactive components present and the reaction rate). It is expected to be in the range of 1 to
6 months (but this will need to be established by experiment or experience).

The production of other gases (hydrocarbons and oxides of carbon) appears to have little or
no correlation with passivator concentration, but is significantly greater during the tests with
new oil rather than the regenerated oil.

It is not suspected that the difference is derived from regeneration, but from the initial
reaction of new oil. This needs to be confirmed by testing oil before and after regeneration.

5.2 Real Transformer Data

The data from live plant supported all the laboratory findings.
There were a higher percentage of free breathing transformers that produced
hydrogen (15% vs 3%).
There was a definite relationship between increased gas production and the
addition of passivator.
The passivator is consumed over time with heat that is produced by the energized
transformers.
More sealed units produced hydrocarbon gas, but it can be concluded that sealed
and non-sealed units produced hydrocarbon gas following passivation.

6 REFERENCES

1 Jimmy M. Rasco; Petroleum refining production of Naphthenic transformer oil control of
corrosive sulphur. 2007, Doble Engineering Company - 74th Annual International Doble
Client Conference.

2 Mineral Insulating Oil Passivation Effectiveness of Passivation to Stop Copper Deposition.
Pahlavanpour, B (Prof.), Sundkvist, K. Nynas Naphthenics Ltd.