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

net/publication/287508470

Detection of Faults in Power Transformers Using an Expertise Method

Depending on DGA

Conference Paper · December 2012

CITATIONS READS

3 847

4 authors, including:

Sherif S. M. Ghoneim Sayed A. Ward

Taif University Delta University for Science and Technology
142 PUBLICATIONS 1,141 CITATIONS 72 PUBLICATIONS 341 CITATIONS

SEE PROFILE SEE PROFILE

Dalal Helmi
EEHC
8 PUBLICATIONS 19 CITATIONS

SEE PROFILE

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

Optimal grid size for precipitators using finite difference method based on full multi-grid method View project

Enhancement of Dielectric and Thermal Properties for Transformer Oil Using Nanoparticles View project

All content following this page was uploaded by Sherif S. M. Ghoneim on 21 December 2015.

The user has requested enhancement of the downloaded file.

Proceedings of the 15th International Middle East Power Systems Conference (MEPCON’12), Alexandria University,
Egypt, December 23-25, 2012, Paper ID 153.

Detection of Faults in Power Transformers

Using an Expertise Method Depending on DGA
Sherif Ghoneim1, 2, Sayed A. Ward3, Dalal H. Helmi4, Usama S. Zidan1
1
Suez Canal University, 2Taif University, 3Benha University,4Egyptian Electricity Transmission
Company, Egypt.

Abstract- Dissolved gas analysis (DGA) is a reliability of the suggested method. The age of
diagnostic tool that used to detect the incipient faults the transformer is taken into account in
of power transformers through the correlation between calculations. In addition, some cases from
the content of gases dissolved in transformers oil and previous literatures were used to compare their
a particular malfunction. The early detection of
incipient faults in power transformers reduces costly
results with the proposed method’s results.
unplanned outages. Some classical methods that
depend on gases concentration in transformers oils are II. Classical Methods to Diagnose
used to interpret transformer faults such as Transformer Faults
Dornenburg, Rogers, Duval triangle and key gases Based on DGA, many interpretative
methods. These methods in some cases did not give methods have been introduced to diagnose the
the same results; therefore, an expertise method is nature of the incipient deterioration occurred in
developed to give the fault type according to the transformer.
dissolved gases concentration in oil. A software code Over the years, several techniques have been
is designed using logic functions to get the type of the
faults in transformers. The age of transformer is taken
developed to facilitate the diagnoses of fault
into account in our calculations. The results from the gases such as Dornenburg method [8], Roger's
software code illustrate the program reliability as an ratio method [9], Key gases method [8], and
early detection tool of transformer faults. Duval Triangle method [10] as well as the
Key words: Dissolved gases analysis- transformer oil- recently developed techniques such as neural
interpretation of transformer faults. network and fuzzy logic.

I. Introduction A. Key gases method

Insulation is an important part of a power The key gas method identities the key gas
transformer, in general, solid and liquid for each type of faults and uses the percent of
insulation are widely used. During the this gas to diagnose the fault [8]. Key gases
operational of a transformer, gasses may dissolve formed by degradation of oil and paper
in transformer oil; these gasses arise as results of insulation are hydrogen (H2), methane (CH4),
transformer faults such as arcing, corona (partial ethane (C2H6), ethylene (C2H4), acetylene
discharges), overheating of transformer oil or (C2H2), Carbon monoxide (CO) and oxygen
overheating of paper insulation (cellulose). (O2). Except for carbon monoxide and oxygen,
Among the dissolved gasses the combustible all other gases are formed from the degradation
gasses are the most dangerous since these gasses of the oil itself. Carbon monoxide, carbon
may cause the burning and/or explosion of the dioxide (CO2), and oxygen are formed from
transformer. The combustible gasses commonly degradation of cellulosic (paper) insulation. Gas
appear in transformer oil are H2 (Hydrogen), type and amounts are determined by where the
C2H6 (Ethane), C2H4 (Ethylene) and C2H2 fault occurs in the transformer and the severity
(Acetylene) [1-3]. and energy of the event. Events range from low
Several artificial intelligence methods such energy events such as partial discharge, which
as Fuzzy Logic and Artificial Neural Network produces hydrogen and trace amounts of
(ANN) were developed as a novel technique to methane and ethane, to very high energy
interpret the faults in transformer [4-7]. sustained arcing, capable of generating all the
In this paper a suggested method combining gases including acetylene, which requires the
the classic dissolved gases analysis (DGA) most energy. The key gas method interprets the
techniques for diagnosis of fault transformers incipient faults in transformer according to some
with logic function is developed. Based on the significant gases to assign four typical fault
interpretation of the classical techniques to the types. These gases are called “key gases” [8]
cause of transformer faults according to the which are shown in Fig. 1.
gases concentration in oil transformer, a system
is suggested to give the main cause of the B. Dornenburg ratio method
transformer fault with the aid of logic functions
that is used as in Fuzzy and Neural Network. The Dornenburg method utilizes four
A lot of real cases of analyzing the dissolved calculated gas ratios to indicate a single fault
gases were collected and used to illustrate the type from three general fault types. This
procedure requires significant levels of the gases A flow chart that describes step by step
to the present in order for the diagnosis to be procedure to identify the reason behind
valid. The four ratios and their diagnosis values transformer faults is found in [8].
are given [8]. Dornenburg method uses five
individual gases or four-key gas ratios, which C. Roger's ratio method
are:- It is an additional tool that may be used to
look at dissolved gases in transformer oil. The
Overheated Cellulose Rogers ratio method takes into consideration
100
industrial experiences, laboratory tests, and
80 further theoretical assessment. This method was
60 further modified into an IEC standard [9, 11].
%

40 The original Rogers ratio method uses four gas

20
ratios which are CH4/H2, C2H6/CH4, C2H4/C2H6
and C2H2/C2H4 for diagnosis. The refined Rogers
0
method uses two tables: one defined the code of
CO H2 CH4 C2H6 C2H2 C2H2
the ratio, and the other defined the diagnosis
Gases
rule. The ratio C2H6/CH4 only indicated a limited
(a) temperature range of decomposition, but did not
assist in further identification of fault. Therefore,
Overheated Oil in IEC standard 599, the further development of
100
Roger's ratio method was deleted. Roger's ratio
80
method and IEC 599 have gained popularity in
60 industrial practices. However, it may give no
%

40 conclusion in some cases. This is the "no

20 decision" problem.
0 A flow chart to describe the step by step
CO H2 CH4 C2H6 C2H2 C2H2 application to give the reason of transformer
Gases
faults is found in [8] and is also based on the
thermal degradation principles.
(b)
D. Duval triangle method
Arcing in Oil
100
The Duval Triangle was first developed in
80
1974 [8]. Three hydrocarbon gases only (CH4,
60
C2H4 and C2H2) are only used. These three gases
%

40 are generated as a result of increasing the level

20 of energy necessary to generate gases in
0 transformers in service. Figure 2 indicates the
CO H2 CH4 C2H6 C2H2 C2H2 Triangle method. In addition to the 6 zones of
individual faults (PD, D1, D2, T1, T2 or T3), an
Gases
intermediate zone DT has been attributed to
(c) mixtures of electrical and thermal faults in the
transformer.
Corona in Oil PD
100
T1
80
T2
%
4

60
H

C2
C

H4
%

40
D1 D2 DT
20 T3

0
CO H2 CH4 C2H6 C2H2 C2H2
Gases %C2H2

(d) Fig.2: Duval triangle as a diagnostic tool to detect the

incipient faults in transformer.
Fig. 1: Kay gas method and four typical faults.
(T1 the zone of low thermal fault <300oC, T2 the zone of
R1=CH4/H2, medium thermal fault 300oC<T<700oC, T3 the zone of high
R2=C2H2/C2H4, thermal fault >700oC, D1 discharge of low energy arcing, D2
R3=C2H2/CH4, discharge of high energy arcing, DT attributes to mixtures of
R4=C2H6/C2H2. electrical and thermal faults and PD indicates partial
discharge)
 III. Decision tree as an expertise method Figure 5 illustrates the final report and Figure 6
When four classical methods of explains the form of the excel sheet that is used
transformer fault diagnosis such as Key gas to explain the main fault in transformer.
method, Dronenburg method, Roger's method
and Duval triangle method are applied to Fault type

interpret the cause of the fault in transformer,

conflicts may occurred. To overcome this
No Fault Fault No Fault Identify
problem a decision fault tree is developed which 0 1 2

contains the information between different faults

types. Every fault type takes a number to help us Discharge Thermal
7 3
to get the main cause of the transformer fault.
This is shown in Figure 3. >300 and
Partial Arcing <300oC >700oC
A software code in excel sheet is developed 8 11 4 <700oC
5
6

using the logic function to get transformer fault

from the four classical method that mentioned Low High Low High Thermal Cellulose
14
9 10 12 13
before; the results depend on the combustible
Fig. 3: Decision fault tree
gases that arise when fault occurs in transformer.
After determining the fault type from these
methods, the program decides the incipient fault
type.
The procedures that used to carryout the Excel
program are:
In the first, the program determines the fault type
according to the classical methods; the second
step is choosing a code for each fault as in
decision fault tree. The third step is summing the
codes for the same faults then the general fault is
specified. At the end the program specifies the
specific fault from the general fault using if (a)
statement and logic functions.
The age of transformer is taken into account in
the calculations as the standard level of gases
that mentioned in a guide for the interpretation
of gas in oil analysis data (SMS-1101-05-T)[12]
as a substation maintenance standard. Table.1[12]
explains the recommendation guide for
maximum safe evolved gas levels in oil
immersed equipment.
Table 1
Gas Dissolved gas concentration (ppmV) (b)
H2 20n+50 Fig. 4: The effect of age of the transformer on the result by
CH4 20n+50 Dornenburg method.
C2H6 20n+50
C2H4 20n+50
C2H2 5n+10
CO 25n+500
CO2 100n+1500
TCG 110n+710

Where, ppmV= part per million, by volume

n= Number of years in service. Serious or danger
levels are approximately 5-10 times the above
values.
The number of years is effect on the Dornenburg
results. This fact is shown as in Figures 4 (a and
b). When the age of transformer increases, the
results will give different transformer faults in
case of Dornenburg method.
Fig. 5: The final report
 Fig. 6: The excel sheet that use to give the main cause of transform fault.

IV. Some cases to specify the fault type in V. Validation of the proposed technique
transformers A comparison between the proposed technique
Some oil samples are taken from real results and the other results in literatures is
transformers which are in operation to carry explained in this section to specify the
out the study. validation of the proposed method.

Case 1: (Transformer 66/11 kV)

Date H2 CH4 C2H6 C2H4 C2H2 CO

14.7.2010 19 3 3 56 0.001 1223

The age of the transformer is assumed 5 years

and the gases concentration appear in the result
sheet. The fault type from the four classical
methods is specified which refer to thermal
fault in this transformer and the decision from
tree fault is high thermal fault. It is seen from
the data that the ppm for Ethyline is high and
then the present of it leads to high thermal
fault. The result from the lab refers that the
fault in transformer is thermal fault >700oC.
Therefore, the result by the code is compatible Fig. 7: The result of case 1.
with that from the lab. The result is shown in
Fig. 7.

Case 2: (Transformer 220/66 kV)

Date H2 CH4 C2H6 C2H4 C2H2 CO

22.5.2011 154 11 14 8 3 487

The software code gives the transformer fault

as arcing discharge and the decision from tree
fault method defines the fault as High arcing
discharge as shown in Figure 8. The Lab result
refers that the transformer fault is discharge of
high energy. The result reveals that the
software code is reliable to determine the
transformer fault based on the gas
concentrations.
Fig. 8: The result of case 2.
Example 1 in [13]
The dissolved gases that produced from an
actual transformer are as follows;
Date H2 CH4 C2H6 C2H4 C2H2 CO
- 64 19 11 82 0.001 459
The analyzed transformer has a problem of
high-temperature thermal fault as in [13].
Using the proposed technique and designed
program the final result is shown in Fig. 9.

Fig. 11: The result of case III.

VI. Conclusions
The results from different cases under study
reveal that the proposed technique is reliable to
use as a diagnostic tools to detect the fault in
transformer in its early stage. The conclusions
from the real cases explain that the nature of
Fig. 9: The result of Example 1 [13]. the insulating materials involved in the fault
The final result using the decision tree fault and the nature of the fault itself affect on
give the same result as in [13]. distribution of dissolved gases. Based on The
Example 2 in [3] results from the software code and the lab
The dissolved gases that taken from another results, the software code is reliable to produce
transformer are as follows; the transformer fault based on the gas
Date H2 CH4 C2H6 C2H4 C2H2 CO concentrations.
- 360 492 54 305 1 580
The analyzed fault of sample 1 in [3] is VII. References
thermal fault reach to 700oC. [1] Suwarno, "Dissolved Gas Analysis of Transformer
Oils: Effects of Electric Arc", Proceedings of the 6th
The final result from the proposed technique is WSEAS International Conference on Power Systems,
shown in Figure 10. Lisbon, Portugal, September 22-24, 2006.
[3] Sayed A. Ward “Evaluating Transformer Condition
Using DGA Oil Analysis”, 2003 Annual Report
Conference on Electrical Insulation and Dielectric
Phenomena.
[4] Fábio R. Barbosa, Otacílio M. Almeida, Arthur P. S.
Braga, Cícero M. Tavares, Márcio A. B. Amora, Francisco
A. P., “Artificial Neural Network Application In
Estimation Of Dissolved Gases In Insulating Mineral Oil
From Physical-Chemical Datas For Incipient Fault
Diagnosis,” IEEE Trans. Power Del., vol. 6, no. 2, pp.
601-607, Apr. 1991.
[5] CS Chang, CW Lim, Q Su, “Fuzzy-Neural Approach
for Dissolved Gas Analysis of Power Transformer
Incipient Faults", Australasian Universities Power
Engineering Conference (AUPEC 2004) 26-29 September
Fig. 10: The result of sample 1 [3]. 2004, Brisbane, Australia.
[6] Rahmat-Allah Hooshmand, Mahdi Banejad, “Fuzzy
Also here the proposed technique is able to Logic Application in Fault Diagnosis of Transformers
determine the type of fault as the conclusion of Using Dissolved Gases", Journal of Electrical Engineering
[3]. & Technology, Vol. 3, No. 3, pp. 293~299, 2008.
Example 3 in [14] [7] N.A. Muhamad, B.T. Phung, T.R. Blackburn, K.X
Lai," Comparative Study and Analysis of DGA Methods
The third example to satisfy the validation of for Transformer Mineral Oil", Journal of Electrical
the proposed technique will be taken from [14] Engineering & Technology, Vol. 2, No. 2, pp. 157~164,
case III. 2007.
Date H2 CH4 C2H6 C2H4 C2H2 CO [8] ANSI/IEEE Std C57.104-1991, IEEE Guide for The
Interpretation of Gases Generated in Oil-Immersed
- 127 24 0.001 32 81 0.001
Transformers, IEEE Power Engineering Society, 1992.
The analyzed fault as in [14] is arcing not [9] R. R. Rogers, “IEEE and IEC Codes to Interpret
involved cellulose. The proposed technique Incipient Faults in Transformers, Using Gas in Oil
gives the main cause of the fault as in Fig. 11. Analysis,” IEEE Trans. on Electrical Insulation, vol. 13,
no. 5, pp. 349-354, 1978.
[10] M. Duval, “Dissolved Gas Analysis: It Can Save
Your Transformer,” IEEE Electrical Insulation Magazine,
vol. 5, no. 6, pp. 22-27, 1989.
 [11] IEC Publication 60599, Interpretation of the analysis [13] J. BILBAO, et. al. "Expertise Method to Diagnose
of gases in transformer and other oil med electrical Transformer Conditions", WSEAS/IASME Conferences,
equipment in &, Geneva, Switzerland, 1999. Corfu, Greece, August 17-19, 2004.
[12] Substation Maintenance standard "Guide for [14] Joseph B. DiGiorgio, "Dissolved Gas Analysis of
Interpretation of Gas In Oil Analysis Data" Index Number Mineral Oil Insulating Fluids" , Northern Technology and
SMS-1101-05-T. testing,http://www.nttworldwide.com.

View publication stats