You are on page 1of 7

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

net/publication/265911414

Investigations Into the Stray Gassing of Oils in the Fault Diagnosis of


Transformers

Article in IEEE Transactions on Power Delivery · October 2014


DOI: 10.1109/TPWRD.2014.2316501

CITATIONS READS

12 1,788

5 authors, including:

Daniel Martin Nick Lelekakis


The University of Queensland Monash University (Australia)
80 PUBLICATIONS 800 CITATIONS 28 PUBLICATIONS 433 CITATIONS

SEE PROFILE SEE PROFILE

Jaury Wijaya Tapan Kumar Saha


Monash University (Australia) The University of Queensland
21 PUBLICATIONS 349 CITATIONS 551 PUBLICATIONS 6,073 CITATIONS

SEE PROFILE SEE PROFILE

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

Peer-to-Peer (P2P) Energy Trading under Network Constraints View project

Intelligent fault diagnosis of power transformer using big data analytics View project

All content following this page was uploaded by Nick Lelekakis on 28 April 2015.

The user has requested enhancement of the downloaded file.


This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

IEEE TRANSACTIONS ON POWER DELIVERY 1

Investigations Into the Stray Gassing of Oils in the


Fault Diagnosis of Transformers
Daniel Martin, Member, IEEE, Nick Lelekakis, Jaury Wijaya, Michel Duval, Life Fellow, IEEE, and
Tapan Saha, Senior Member, IEEE

Abstract—In this paper, we present an analysis of the gases We found that for the tested mineral oil in the presence of high
generated by mineral oil and cellulose transformer insulation, at oxygen levels, ethylene can form at temperatures well below
temperatures lower than those expected to be caused by an actual those expected [2]. This finding is similar to that observed by
thermal fault. Gassing at relatively low temperatures, 100 C to
200 C, often presents a problem to the asset manager since it can Besner [6].
be difficult to distinguish this event from that caused by an actual For a transformer, Besner found ethylene being produced
fault. Previous stray gassing tests were only performed for 164 h, below its typical generation temperature and, thus, set out
at two temperatures, and two oxygen contents. We tested a mineral to reproduce ethylene generation in a laboratory. They con-
oil under 20 different conditions of oxygen and water content, for cluded that ethylene can be produced at unexpectedly low
up to nearly 100 days. We found that at low-temperature levels
and that by using high oxygen levels, the generation of ethylene temperatures.
can be more significant when compared to the other hydrocarbon
gases than first thought.
II. BACKGROUND
Index Terms—Dissolved gas analysis, oil insulation, paper insu-
lation, power transformer insulation, power transformers. In 2006, a CIGRE task force reported on their efforts to de-
termine both levels and rates of change of gases, dissolved in
mineral oil, for normally operating transformers [4]. As part of
I. INTRODUCTION their endeavors, thermal stray gassing of various mineral oils
was investigated. This involved heating oil to 120 C and 200
C for a specific period, and measuring the changes in dissolved

D ISSOLVED gas analysis (DGA) has been used to diag-


nose faults for decades. Ratios of certain gases are used
to determine the energy signature of various faults. Three com-
gas content. They found that although stray gas levels differ for
each brand of mineral oil, there are general trends: stray gassing
is related to oxygen content, and metals (even stainless steel) af-
monly used methods are those by the IEC [1], IEEE [2], and fect gas formation.
Duval triangle [3]. At 120 C, the main gas produced, in general, is hydrogen
A drawback to using DGA is that stray gassing may occur, followed by methane, while at 200 C, the main gases formed
which can lead to an incorrect fault determination being made. are methane and ethane [4]. Based on the results of the CIGRE
One definition for stray gassing is unexpected gas formation study, gas ratios were developed to differentiate between faults
at temperatures between 80 and 250 C [4]. Another, given and stray gassing.
by the IEC [5], is the production of gases at low temperatures ASTM D7150–05 defines a test method to determine gassing
120 . This standard notes that inhibited oils “typically characteristics of insulating fluids [7]. It is similar to method
produce less stray gassing than uninhibited ones.” used by the CIGRE task force, where oil is heated to 120 C
Approximate thresholds are given by Halstead’s thermal for 164 h. The oil is sparged with either air or nitrogen gas.
equilibrium partial pressures as a function of temperature [2]. However, the oxygen consumed by chemical reactions is not
If there are significant quantities of any gas forming below its replenished. This standard notes that oils sparged with air pro-
threshold temperature, gassing would be considered stray. In duce hydrogen, whereas when sparged with nitrogen, they pro-
this paper, we use data from the accelerated aging of insulation duce more hydrocarbons in relation to hydrogen.
to analyze this stray gassing by using more comprehensive
aging conditions. III. KINETIC APPROACH TO ANALYZING GAS PRODUCTION
Various compounds are created and consumed by chemical
Manuscript received September 01, 2013; revised November 25, 2013; ac-
cepted March 31, 2014. Paper no. TPWRD-01010-2013. reactions occurring in the oil, some of which involve oxygen.
D. Martin and T. Saha are with the University of Queensland, St. Lucia, QLD Some of these compounds are the gases detected by DGA [8].
4072, Australia (e-mail: D.Martin6@itee.uq.edu.au; saha@itee.uq.edu.au).
The generation of a gas is dependent on the fault energy present
N. Lelekakis and J. Wijaya are with TransformerLIFE Centre, Monash Uni-
versity, Clayton, VIC 3800 Australia (e-mail: Nick.lelekakis@monash.edu). [9].
M. Duval is with Quebec Electricity Research Institute, Varennes J3X 1S1 The rate law describes how fast a reaction proceeds and
Canada (e-mail: duval.michel@ireq.ca).
how the reaction rate depends on the concentration of chemical
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. species involved [8]. The appearance of a product can be used
Digital Object Identifier 10.1109/TPWRD.2014.2316501 to determine the rate similar to monitoring the consumption

0885-8977 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2 IEEE TRANSACTIONS ON POWER DELIVERY

of reactants. The relationship between the rate of reaction and TABLE I


temperature is determined by the Arrhenius (1)[10] RESULTS OF STRAY GASSING FROM HEATING SHELL DIALA AX, AT 120 C.
GAS CONCENTRATION IN PPM [4]

(1)

where is the rate, is the gas constant, is temperature in


Kelvin, is the pre-exponential factor, and is the activa-
tion energy. The A-value depends on the frequency with which
reacting molecules collide and the chance of collision in the cor-
rect orientation. Not all molecular collisions will result in the re-
action occurring because kinetic energy that is higher than that
prescribed by the activation energy of the reaction is required.
The proportion of molecules with kinetic energy higher than that
specified by the activation energy is calculated from the term
.
Rate can be estimated by assuming that the reaction follows
a zero-order kinetic equation, where the change in concentration
of a product is linear with time (2)[11], [12]. In practice, once
the concentration of reactants decreases, the rate falls. In this
study, the initial rate of change was investigated, where is
the linear region of the change in concentration of a gas

(2)

One problem with using gas ratios is that, according to rate


(1), gases will form at any temperature. A practical problem that
this presents is that gases are created even in the absence of a
transformer fault, which has been termed “stray gassing.”

IV. INVESTIGATION
In our previous investigation on the aging of paper insula-
tion used in power transformer [13], [14] frequent dissolved gas
measurements were made. Mineral oil (Shell Diala AX) and
paper insulation (Kraft, Kraft Thermally Upgraded and Grade
3 presspaper) were investigated. They were heated to tempera-
Fig. 1. Rig to age insulation.
tures of 100, 120, 130, 140, and 160 C for various durations.
This provided an opportunity to analyze the gas formation de-
fined to be stray. The insulation media were heated in various
different levels of oxygen and water content (since they are both temperature. A 2-l conical flask with a 1-l rubber bag was used
known to affect aging). as a conservator for when the oil expanded. The conservator
The CIGRE working group [4] had studied stray gassing in was placed outside the oven. Its rubber bag was purged with
this mineral oil, heating their sample to 120 C for up to 180 dry argon. Twenty-one grams of paper strip (Kraft, Thermally
h. They found, as given in Table I, that mostly hydrogen was Upgraded Kraft and Grade 3) were placed in the vessel, along
generated with some methane, and even less ethylene. with 270 g of 1-mm-thick type K pressboard. The pressboard
Our investigation furthers the CIGRE study because we col- was formed into a coil with enamel coated copper wire (0.8 mm
lected dissolved gas data for both far longer durations and for 20 core), used to separate the turns and allow oil to flow through.
different conditions of temperature, oxygen, and water level. We The paper plus pressboard to oil ratio was 1:14.
set out to determine whether these factors affect stray gassing Twenty different conditions of temperature, paper water
levels. content and dissolved oxygen content were studied. The con-
ditions studied were of paper with a water content of 0.5%,
1.6% and 2.7%, and of oil with an oxygen content within the
V. EXPERIMENT SETUP AND METHOD
range of 0–6000 ppm (low), 7000–14 000 ppm (medium) and
The paper samples were aged in 5-l glass vessels shown 16 500–25 000 ppm (high).
in Fig. 1. The lid was manufactured from stainless steel and Samples of oil were regularly extracted for DGA, determined
included paper sample holders, a temperature and moisture in accordance with the headspace gas chromatograph method
probe, and ports for adding water and oxygen. A magnetic ASTM D3612-C [15]. The gases measured were acetylene,
stirrer was used to circulate the oil within the vessel. The carbon dioxide, carbon monoxide, ethane, ethylene, hydrogen,
vessel was placed in an air circulating oven set to the desired methane, nitrogen, and oxygen.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

MARTIN et al.: INVESTIGATIONS INTO THE STRAY GASSING OF OILS IN THE FAULT DIAGNOSIS OF TRANSFORMERS 3

Fig. 2. Ethylene generation in 16 500–25 000 ppm levels of dissolved oxygen,


Fig. 3. Ethane generation in 16 500–25 000 ppm levels of oxygen.
plotted in terms of paper water content and temperature.

VI. RESULTS AND DISCUSSION

A. Results
In total, more than 400 dissolved gas measurements were
made during the investigation. Periodically the aging vessels
were opened so that samples of paper could be removed for DP
measurement, some gas would have thus undoubtedly escaped.
Gas loss was reduced by minimizing the duration that the vessel
was open during paper extraction. The pressboard was periodi-
cally replaced with dryboard to control the water content of the
system.

B. Effects of Oxygen and Water Content


Stray gassing of this oil was only investigated by CIGRE
using two cases, low and high oxygen levels, at one temperature.
Our experiment was more thorough as more cases of oxygen,
water, and temperature were investigated.
The initial rate of gas generation was analyzed by plotting Fig. 4. Methane generation in 16 500–25 000 ppm levels of oxygen.
the dissolved gas content, in ppm, against the aging time in
hours. The gas levels would eventually plateau, so only the ini-
tial change was plotted. Figs. 2–5 show the gas generation in and water on individual gas generation did not follow a general
high oxygen levels. The trend lines are fairly linear, thus al- trend. For instance, under conditions of high oxygen the rate of
lowing for rate to be determined from their gradient (2). As change of ethylene and hydrogen both increased, although it fell
evident in Fig. 5, the concentrations of hydrogen fell in two of for ethane. For methane there seemed to be some temperature
the cases, which were those for 0.5% 120 C and 1.6% 120 C. dependency because its rate of generation fell at temperatures
This seemed to have occurred at around 400–500 h. Points after less than 120 C.
the fall-in concentration were excluded when determining the There appears to be a general trend that under high levels of
reaction rate constants. Methane and hydrogen levels were of oxygen, the rate of gas generation reduces as the paper water
the same order of magnitude (Figs. 4 and 5) often in the level of content increases at 120 C and 140 C, which is not apparent
100s of ppm as opposed to ethylene and ethane levels which for samples aged in low oxygen. At 100 C, there does not ap-
were less. This observation is consistent for the mineral oils pear to be any correlation between water content and gas gener-
tested by the CIGRE working group, since they tested oils under ation. At 120 C gas generation levels fell by about three quar-
low and high levels of oxygen. ters for ethylene, four fifths for methane and about by a half for
The rate of gas generation increases as the temperature rises ethane. The generation of hydrogen also fell. At 140 C, with
(Table II) as expected because according to (1), temperature in- high oxygen, the generation of three of these gases has fallen
creases the rate constant of the reaction. The effect of oxygen (ethylene, methane and ethane).
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4 IEEE TRANSACTIONS ON POWER DELIVERY

TABLE II
RATE OF CHANGE OF GAS

Fig. 6. Plot of reaction rate constant, k, for ethylene, methane, hydrogen, and
ethane, against the inverse of temperature in Kelvin for the system aged with
0.5% paper water content in low oxygen.

a straight line, which has a y-intercept of and a gradient


Fig. 5. Hydrogen generation in 16 500–25 000 ppm levels of oxygen. of .

This could mean that in order to apply the outcomes of studies (3)
on stray gassing to a transformer, there is a need to take into ac-
count the availability of water in addition to the effect of oxygen. Figs. 6–11 show the comparative relationships between gas
generation under different, specified, conditions. The CIGRE
C. Comparing Rate of Change of Ethylene to That of Ethane, DGA working group noted that, in their tests on mineral oil, the
Hydrogen and Methane generation of ethylene was negligible in comparison to the rates
of generation of the other gases. In our study, for low oxygen
Table II shows the initial rate of change given as functions (Figs. 6 and 7), this general observation held. The rate of ethy-
of temperature, oxygen content and paper water content. The lene production is low in comparison to that of the other gases.
effect of temperature on gas generation is taken into account by However, this was not observed for high oxygen levels at the
rearranging (1) into (3). Plotting against will provide lower temperature of 100 C.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

MARTIN et al.: INVESTIGATIONS INTO THE STRAY GASSING OF OILS IN THE FAULT DIAGNOSIS OF TRANSFORMERS 5

Fig. 7. Plot of reaction rate constant, k, for ethylene, methane, hydrogen, and
ethane, against the inverse of temperature in Kelvin for the system aged with
2.7% paper water content in low oxygen.
Fig. 10. Plot of reaction rate constant, k, for ethylene, methane, hydrogen, and
ethane, against the inverse of temperature in Kelvin for the system aged with
0.5% paper water content. This shows the effect of different oxygen levels on
aging.

Fig. 8. Plot of reaction rate constant, k, for ethylene, methane, hydrogen, and
ethane, against the inverse of temperature in Kelvin for the system aged with
0.5% paper water content in high oxygen.

Fig. 11. Plot of reaction rate constant, k, for ethylene, methane, hydrogen, and
ethane, against the inverse of temperature in Kelvin for the system aged with
2.7% paper water content. This shows the effect of different oxygen levels on
aging.

of the second generated gas (methane). The observation of ethy-


lene being the second produced gas is inconsistent with that of
the CIGRE study. Possible reasons are that the CIGRE study
used 120 C, a higher temperature, and that their aging times,
164 h, were shorter than that of this study. Note in Figs. 6–11,
hydrogen values are likely to be higher than those indicated
Fig. 9. Plot of reaction rate constant, k, for ethylene, methane, hydrogen, and because of hydrogen losses during the tests, as mentioned in
ethane, against the inverse of temperature in Kelvin for the system aged with Section VI-B. Gas losses, however, only marginally affected
2.7% paper water content in high oxygen.
the relative proportions of the other gases: ethylene, ethane, and
methane.

As shown in Table II, at the lower temperature end using high


VII. CONCLUSIONS
oxygen levels, the rate of generation of ethylene was second to
that of hydrogen for the experiments performed at 0.5% (Fig. 8) During the accelerated aging investigations, the generation
and 1.6% water content of paper. For the 2.7% case, shown in of hydrocarbon gas was observed to be related to the water con-
Fig. 9, the rate of generation of ethylene was very close to that tent of paper, which is thus related to the water content of oil
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6 IEEE TRANSACTIONS ON POWER DELIVERY

(although it could not be determined whether the water was to the University of Queensland, Brisbane, Australia, working on a transformer
causal). condition monitoring project funded by the local utilities.
Ethylene was found to be a major stray gas under certain con-
ditions, similar to a finding by Besner [6], and future work is re-
quired to reveal the mechanisms behind this gassing. The design Nick Lelekakis received the B.Sc. degree in chem-
of the test cell should also be verified as optimal at minimizing istry (Hons.) from Monash University, Clayton, Aus-
tralia, in 1995.
gas leaks . He then moved to electrical and computer systems
For future work, the specific additives should be taken into engineering where he studied moisture migration
account when investigating stray gassing, as these may affect in transformer insulation systems. He conducted
many experiments by investigating the effect of
gas production. temperature and moisture on the risk of bubbles and
water droplets formed in power transformers. Be-
REFERENCES tween 2004 and 2008, he conducted a large research
project, investigating the effects of temperature,
[1] Mineral Oil-Impregnated Electrical Equipment in Service—Guide to moisture, and oxygen on the degradation of paper insulation for developing
the Interpretation of Dissolved and Free Gases Analysis, IEC Standard a life estimation tool for transformers. Between 2009 and 2012, he was the
60599, 2007, IEC Switzerland. Project Leader for a number of field research projects involving Australian
[2] IEEE Guide for the Interpretation of Gases Generated in Oil-Immersed utilities on dryout, aging, and failure investigations of power transformers. In
Transformers, IEEE Standard C57.104–2008, 2008. 2013, he is leading the Transformer Research Centre at Monash University,
[3] M. Duval, “A review of faults detectable by gas-in-oil analysis in trans- Clayton, Australia, to develop a transformer lifetool to access moisture, dryout,
formers,” IEEE Elect. Insul. Mag., vol. 18, no. 3, pp. 8–17, May/Jun. and insulation aging.
2002.
[4] CIGRE Joint Task Force D1.01/A2.11, “Recent developments in DGA
interpretation,” Paper no. 296, 2006.
[5] Fluids for Electrotechnical Applications—Unused Mineral Insulating
Jaury Wijaya was born in Baubau, Indonesia.
Oils for Transformers and Switchgear, IEC Standard 60296, 2012, IEC
He received the B.Eng. degree in electrical engi-
Switzerland.
neering from Petra Christian University, Surabaya,
[6] S. Besner, J. Jalbert, and B. Noirhomme, “Unusual ethylene production
Indonesia, in 1999 and the M.Eng.Sc. degree in
of in-service transformer oil at low temperature,” IEEE Trans. Dielectr.
electrical engineering from Monash University,
Electr. Insul., vol. 19, no. 6, pp. 1901–1907, Dec. 2012.
Clayton, Australia, where he is currently pursuing
[7] Standard Test Method for Determination of Gassing Characteristics of
the Ph.D. degree in electrical engineering.
Insulating Liquids under Thermal Stress at Low Temperature, ASTM
He was a Technical Officer in the Department
D7150–05, 2005, ASTM International USA.
of Electrical and Computer Systems Engineering at
[8] C. Heghes, “C1–C4 hydrocarbon oxidation mechanism,” Ph.D. dis-
Monash University, Clayton, Australia, from 2004
sertation, Dept. Chemistry, Rupertus Carola Univ. Heidelberg, Heidel-
to 2009.
berg, Germany, 2006.
[9] F. Jakob, P. Noble, and J. Dukarm, “A thermodynamic approach to
evaluation of the severity of transformer faults,” IEEE Trans. Power
Del., vol. 27, no. 2, pp. 554–559, Apr. 2012.
[10] J. Daintith, Dictionary of Chemistry, 6th ed. London, U.K.: Oxford Michel Duval (LF’12) received the B.Sc. degree
University Press, 2008. in chemical engineering from the University of
[11] H. Mottola, Kinetic Aspects of Analytical Chemistry, ser. Chemical Toulouse, Toulouse, France, in 1966 and the Ph.D.
analysis. Hoboken, NJ, USA: Wiley, 1988, vol. 96, . degree in chemical engineering from the University
[12] P. Houston, Chemical Kinetics and Reaction Dynamics. New York, of Paris, Paris, France, in 1970.
USA: Dover, 2001. He has worked for IREQ (Hydro-Quebec, Canada)
[13] N. Lelekakis, D. Martin, and J. Wijaya, “Ageing rate of paper insulation since 1970. In the field of dissolved gas analysis
used in power transformers. Part 1: oil/paper system with low oxygen (DGA), he is well known for his Triangle method
concentration,” IEEE Trans. Dielectr. Electr. Insul., vol. 19, no. 6, pp. of interpretation used worldwide, and for his other
1999–2008, Dec. 2012. contributions to DGA (gas-in-oil standards, gas
[14] N. Lelekakis, D. Martin, and J. Wijaya, “Ageing rate of paper insulation levels in service, online gas monitors). He has also
used in power transformers. Part 2: oil/paper system with medium and been active in the field of electrical insulating oils and polymers.
high oxygen concentration,” IEEE Trans. Dielectr. Electr. Insul., vol. Dr. Duval is the convenor of several CIGRE, IEC, and IEEE working groups.
19, no. 6, pp. 2009–2018, Dec. 2012. He holds 16 patents and is the author of more than 90 scientific papers and inter-
[15] Standard Test Method for Analysis of Gases Dissolved in Electrical national standards. He is a Fellow of the Chemical Institute of Canada and the
Insulating Oil by Gas Chromatography, ASTM D3612–02, ASTM In- recipient of the IEEE Herman Halperin Electric Transmission and Distribution
ternational USA, 2002, . Award for 2012.

Daniel Martin (M’12) received the B.Eng. degree


in electrical and electronic engineering (with study
abroad in Germany) from the University of Brighton, Tapan Saha (SM’97) was born in Bangladesh and
Brighton, U.K., in 2000 and is currently pursuing the immigrated to Australia in 1989.
Ph.D. degree in electrical insulation at the University Currently, he is Professor of Electrical Engi-
of Manchester, Manchester, U.K. neering with the School of Information Technology
He joined Racal Electronics, London, U.K., and Electrical Engineering, University of Queens-
which became the international electronics company land, Brisbane, Australia. Before joining the Uni-
Thales, working on communication and aircraft versity of Queensland, he taught at the Bangladesh
systems. He left Thales in 2004. He investigated University of Engineering and Technology, Dhaka,
the suitability of using vegetable oils and synthetic for three-and-a-half years and then at James Cook
esters as substitutes for mineral oil within large power transformers, and University, Townsville, Australia, for two-and-a-half
graduated in 2008. He joined Monash University, Clayton, Australia, working years. His research interests include power systems,
on transformer condition monitoring, quickly assuming the directorship of the power quality, and condition monitoring of electrical plants.
Centre for Power Transformer Monitoring. At the beginning of 2013, he moved Prof. Saha is a Fellow of the Institution of Engineers, Australia.

View publication stats

You might also like