Modification of Duval Triangle for Diagnostic

Transformer Fault through a Procedure of

Dissolved Gases Analysis
Sobhy S. Dessouky A, Ahmed E.Kalas B, R.A.Abd El-Aal C, Abdel Moneim M. Hassan D
A, B, C
Electrical Engineering Dept. Faculty of Engineering, Port-Said University. Port Said, Egypt
D
Abo-Sultan Steam Power Plant, Ismailia. Egypt

 by the individual gases is difficult in practice since each

Abstract— Dissolved gas-in-oil analysis (DGA) is a sensitive and incipient fault produces traces of other gases in addition to the
dependable technique for the detection of incipient fault condition key gas of such fault. The ratio methods for fault diagnosis use
within oil-immersed transformers. When the mineral oil is certain ratios of dissolved gas concentrations according to
subjected to high thermal or/and electrical stresses, it decomposes combinations of codes [4, 5]. An incipient fault is detected
and, as a result, gases are generated. when a code combination matches with the code pattern of the
This paper presents modification of Duval triangle DGA fault. The most widely used ratio methods are the Doernenburg
diagnostic graph to numerical method that is easy to use for Ratio Method, Rogers Ratio Method, and IEC standard. Six gas
diagnosing and a Matlab program. To study such as the following ratios have been used by different methods. The major
evaluation. This evaluation is carried out on DGA data obtained drawback of ratio methods is the “no decision” problem
from three different groups of transformers each group are two
associated with some cases that lie out of the specified codes.
identical transformers. A Matlab program was developed to
In recent years, many researchers have studied the
automate the evaluation of Duval Triangle graph to numerical
modification, Also the fault gases can be generated due to oil application of artificial intelligence based techniques for
decomposing effected by transformer over excitation which transformer fault diagnosis. These techniques include expert
increasing the transformer exciting current lead to rising the systems, fuzzy logic, artificial neural networks or mixed
temperature inside transformer core beside the other causes. techniques [6, 7]. However, these methods are too complicated
to be implemented practically on a wide range.
Index Terms— Dissolved Gas Analysis )DGA), mineral oil, This paper investigates the new aspects, accuracy and
decomposition, degradation, and transformer condition. consistency of these methods in interpreting the transformer
condition.
I. INTRODUCTION II. DGA TO DIAGNOSE TRANSFORMER FAULTS

D issolved gas analysis (DGA) is a popular diagnostic

technique that is used to detect incipient faults in oil-filled
power transformers [1]. By using DGA data, transformer
When an incipient fault occurs, either thermal or/and
electrical, a number of gases are generated and dissolved into
the oil. These gases are mainly H2, CH4, C2H2, C2H4 and C2H6.
criticality can be identified with proposing the proper In addition CO and CO2 will exist if cellulose degradation is
maintenance action [2]. involved, based on the type and amount of generated gases [1,
Several methods were proposed to diagnose incipient faults 8-9].
based on DGA. These methods are key gas method, Rogers's
ratio methods, Duval triangle method, Doernenburg Ratio A.Duval Triangle (DGA) Diagnostic Graph Method
method, Basic Gas Ratio, and artificial intelligence based M. Duval. Proposed another diagnostic method to
methods. overcome this limitation, well known as Duval triangle. This
The key gas method identifies the key gas for each type of method is based on a triangle graphical representation to
fault and uses the percent of this gas to diagnose the fault as visualize the different cases for oil-insulated high-voltage
suggested by IEEE standard C57.104 [3]. The percent amount equipment (mainly transformers), Fig. (I) provides a graphical
of gas is obtained in terms of the total combustible gases (TCG). method of identifying a fault. It uses a three-axis coordinate
The main disadvantage of this method is that the interpretation

Sobhy S. Dessouky, Electrical Engineering Dept. Faculty of Engineering, R.A.Abd El-Aal, Electrical Engineering Dept. Faculty of Engineering, Port-
Said University. Port Said, Egypt (e-mail: [email protected]).
Port-Said University. Port Said, Egypt (e-mail: [email protected]).
Abdel Moneim M. Hassan, Abo-Sultan Steam Power Plant, Ismailia. Egypt
Ahmed E.Kalas , Electrical Engineering Dept. Faculty of Engineering, Port-
(e-mail: [email protected]).
Said University. Port Said, Egypt (e-mail: [email protected]).

TRANSACTIONS ON ENVIRONMENT AND ELECTRICAL ENGINEERING ISSN 2450-5730 Vol 1, No 3 (2016)

© Sobhy S. Dessouky, Ahmed E. Kalas, R.A.Abd El-Aal & Abdel Moneim M. Hassan
system, where concentrations of CH4, C2H4 and C2H2 are used modification. Table (II) shows the Modification of Duval
as coordinates, and the likely fault falls within one of the fault triangle DGA diagnostic graph to numerical method.
regions of the triangle. The various regions within the Duval For example if C2H2 = 0.1, CH4 = 0.3 and C2H4 = 0.6. We
Triangle are given in Table (I) [10-13]. can use table (II) easy to determine the fault Diagnostic
For example if C2H2 = 0.07, CH4 = 0.2 and C2H4 = 0.73. (Thermal fault t > 700 °C), and if C2H2 = 0.36, CH4 = 0.32
The fault diagnostic is T3 (Thermal fault t > 700 °C), and if and C2H4 = 0.32, the fault diagnostic is (High-energy electrical
C2H2 = 0.36, CH4 = 0.32 and C2H4 = 0.32, the fault diagnostic discharge), the same results as in the previous example.
is D2 (High-energy electrical discharge), as shown in fig (I).
TABLE II. MODIFICATION OF DUVAL TRIANGLE (DGA)
DIAGNOSTIC GRAPH TO NUMERICAL METHOD

C2H2% CH4% C2H4% Fault

Partial discharge
0.00 - 0.02 0.98 - 1.00 0.00 - 0.02 (electrical fault)
Thermal fault
0.46 - 0.80 0.20 - 0.50 300 < t < 700 °C
0.00 - 0.04
thermal fault
0.76 - 0.98 0.02 - 0.20 t < 300 °C
Thermal fault
0.00 - 0.15 0.00 - 0.50 0.50 - 1.00 t > 700 °C
0.04 - 0.13 0.47 - 0.96 0.00 - 0.40
0.13 - 0.29 0.21 - 0.56 0.40 - 0.50 Mixtures of thermal
and electrical faults
0.15 - 0.29 0.00 - 0.35 0.50 - 0.85
0.13 - 0.29 0.31 - 0.64 0.23 - 0.40 Discharge of high
energy (electrical
0.29 - 0.77 0.00 - 0.48 0.23 - 0.71 fault)

III. CASE STUDY DISSOLVED GAS ANALYSIS

The case study carried out from three different groups of
transformers each group are identical in Abu-Sultan steam
Fig. 1. Duval Triangle power plant. Fig. (2) Shows the schematic diagram
configuration for transformers under testing. The first group of
transformers are three single phase 192 MVA, 15/220 KV, Off
TABLE I. FAULT CODE
L.T.C. The Second group of transformers are three phase 16
PD Partial discharge MVA, 220/6.3KV, ON.L.T.C, and the third group of
transformers are three phase 16 MVA, 15/6.3/6.3 KV,
ON.L.T.C. The rating and (DGA) testing results for the above-
T1 Low-range thermal fault (below 300 °C) mentioned Power Transformer are shown in tables (III, IV).

T2 Medium-range thermal fault (300-700 °C)

T3 High-range thermal fault (above 700 °C)

D1 Low-energy electrical discharge

D2 High-energy electrical discharge

DT Indeterminate - thermal fault or electrical discharge.

A. Duval Triangle Graph to Numerical Method Fig. 2. Schematic Diagram for Transformers
In this paper, we developed A Matlab program to automate under Evaluation
the evaluation of Duval Triangle graph to numerical
TABLE III. TABLE IV.
RATING OF POWER TRANSFORMER UNDER TESTING (DGA) TESTING RESULTS

Main transformer unit no. 2 ph (B)

Rated Voltage
Transformer

Main transformer unit no. 1 ph (B)

from 08/05//2013 to 27/11/2013

from 08/05/2013 to 05/11/2014

Rated Power

from 07/04/2013 to 27/11/2013

From 07/04/2013 to 02/04/2014

Phases

from 08/05/2013 to 06/05/2014

from 08/05/2013 to 29/03/2015

Number Of
Name

Transformer & Samples date

Operating

Aux. transformer unit no. 1

Aux. transformer unit no. 2

KV
MVA

Oil Type

Start Up transformer A

Start Up transformer B
Date
Main transformer
Unit no. 1 ( TR1) 19/3/1983
192 15/220

3 single
Main transformer

Phase
Unit no. 2 ( TR2) 15/8/1983

Start Up

Mineral Oil Naphthenic

transformer A 19/3/1983
( TR3) 16 220/6.3

without C3H6 & C3H8

3- Phases
Start Up

Total combustible
gases ( T.C.G )
transformer B 15/10/1984
( TR4) 274 164 98 219 246 193
477 592 249 426 429 400
Aux. transformer
unit no. 1 ( TR5) 19/3/1983 16 15/6.3/6.3
3- Phases

Aux. transformer

Hydrogen
Unit no. 2 ( TR6) 15/8/1983 9 3 1 5 14 7
H2
7 16 19 6 28 35

25 15 2 19 48 9
IV. DIAGNOSTIC METHOD USED BY CH4
48 37 4 61 49 12
MODIFICATION SYSTEM. C2H2
0 0 0 0 0 0
Combustible gases
0 0 0 0 0 0
The diagnostic methods for DGA are used by a numerical
Hydrocarbons

5 1 2 5 3 2
C2H4
method, The Matlab program diagnoses output for the under 2 12 8 6 10 3
testing transformers. Table (V) shows application of the faults 12 10 1 57 28 2
C2H6
diagnosed by various methods, which indicate that all 29 50 3 142 45 3
transformers are thermal faults. C3H6 14 2
& 14 5 2 30
- 3
C3H8 26 - 3 81
V. RESULTS AND DISCUSSION
Comparison of various methods as shown in the table (V),
a thermal fault in oil within all transformers is diagnosed for all
CO 223 135 91 132
Carbon Oxides

five methods. Where winding temperature do not exceed 95°C 154 173
392 477 215 212 297 348
and oil temperature do not exceed 85°C for all transformers
during normal operation. Moreover, the possible collapse of
cooling system during operation in this case is too small and 439
CO2 2877 775 482 848 1632 2581
there is no increase in the viscosity of the oil, as it is clear in the 6052 4854 1324 1772 3787
results of chemical analysis of samples oil and no wax
materials. However, there is an important factor is the
increased over excitation due to reduction of generator speed 2042 1633 3432 991 1420
Non-Combustible gases

O2 1118
when some of the generating units from the network goes out 2664 3758 5766 1911 13615 3300
during normal operation or the frequency disturbances that
or atmospheric gases

occur when large loads are connected to the electrical network

Non-fault

system.
Over-excitation or/and under frequency protection may be \
or may be not operate depends on the response of power system 82762 30606
control. The under frequency relay operate at 47.5 Hz with time N2 31551 45633 39302 74493 137375 119152
38801 90526 56161 88856
lag 0.5 sec and over excitation relay operate at V/Hz = 1.1pu
for 45 sec time lag or V/Hz =1.18 pu for 2 sec time lag at
generators.
TABLEpractices
V. or operator error without a disruption to operations. The
resultingBY
APPLICATION OF THE FAULT DIAGNOSED thermal faultsMETHODS
VARIOUS lead to oil decomposing to generate fault

Duval's triangle
Transformer

numerical modified Basic gas ratio Doernenburg ratio Rogers Ratio Kay gas
P(96/4) P(77/8) P(71/3) P(62/5) P(42/58)
no.

TR1 thermal fault thermal fault slight overheating

t < 300°C t < 300°C thermal t <150 °C pyrolysis in cellulose
decomposition
TR2 Thermal fault thermal fault slight overheating
300 < t < 700 °C t < 300°C thermal 150-200 °C pyrolysis in cellulose
decomposition
TR3 Thermal fault thermal fault of low general conductor
t > 700 °C temperature t <150°C Cannot be applicable overheating pyrolysis in cellulose
TR4 thermal fault thermal fault slight overheating
t < 300 °C t < 300°C thermal 150-200 °C pyrolysis in cellulose
decomposition
TR5 Thermal fault thermal fault
t > 700 °C t < 300°C thermal Cannot be applicable pyrolysis in cellulose
decomposition
TR6 Thermal fault general conductor
300 < t < 700 °C Cannot be applicable Cannot be applicable overheating pyrolysis in cellulose

If frequency decreases and the voltage is constant, the The practices or operator error without a disruption to
transformer core is heated. Fig. (3) Shown voltage, current and operations. The resulting thermal faults lead to oil decomposing to
frequency of generating unit transformer number one at Abu- generate fault gases H2, CH4 at temperature 120°C, C2H6 at
temperature 150°C, C2H4 at temperature 300°C, and C2H2 at
sultan steam power plant from 17/5/2015 to18/5/2015, which
temperature 700°C.
indicate that frequency, reduced to 49.2 Hz at voltage 14.85. KV. In addition, degradation of insulating material is cumulative.
The rated generator voltage and frequency is 15 KV and A transformer or generator that survives a serious over excitation
50Hz respectively. So generator is over excitation =1.0061 Pu. event or many small events may fail because of a moderate event
At unit, start up the voltage may be built to 15KV at generator during normal service as all transformers under study.
frequency 48 Hz then 1.042 Pu over-excitations. Disturbance in In addition, if voltage increased, at rated frequency, the
frequency is repeated from 18/5/2015 to 20/5/2015 in power exciting current increases, as shown in Fig. (6). So Tr1 through Tr6
system as shown in Fig. (4) and affect all network transformers are effected by over excitation due to network normal operation but
in this moment and there is an instantaneous decrease in power Tr1, Tr2,Tr5, Tr6 are effected by Over excitation damage usually
system frequency to 45.36 Hz without operate under frequency occurs during periods of off-frequency operation such as start up
or/and over-excitation relays because disturbance duration less or shut down for unit transformer as shown in Fig.(2) and table
than 0.5 sec as shown in Fig. (5). (VI).
Transformers require an internal magnetic field to operate. The
core of a transformer is designed to provide the magnetic flux
Necessary for rated load.
An over-excitation condition occurs when this equipment is
operated such that flux levels exceed design values. The voltage
output of a transformer is a function of the rate of change of the
flux and the number of turns in the output winding. e = N dφ/dt
during normal power system operation.
The voltage is sinusoidal and the rate of change is determined
by the frequency, which is in turn determined by generator speed
[14].
The equation shows core flux to be directly proportional to
voltage and inversely proportional to frequency φ α V/f. The actual
magnitude of flux in transformer core is can be quantified in terms
of per unit volts / Hertz.
A generator or transformer operating at no load with rated
voltage and frequency would have one per unit excitation. The
same equipment operating at rated voltage and 95% frequency
would have 1.0/0.95 = 1.05 Pu flux or 1.05 Pu excitation.
Over-excitation will result from high voltage at rated frequency
and from rated voltage with low frequency.
Because over excitation is a function of voltage and frequency,
it can occur without notice. Transformers and generators can be
subject to repeated over excitation by inappropriate operating.
Generator voltage

System frequency

Generator current

Fig 3. Voltage Current and Frequency for Unit No.1 Generator

Generator voltage

Generator current

System frequency
Voltage/Hertz increased
Frequency reduction

Fig 4. Repeating Disturbances in Power System Frequency

System frequency

Fig 5. Instantaneous Decrease in Power System Frequency

Fig 6. Voltage Increased, at Rated Frequency Exciting Current Increase

TABLE VI.
CAUSES OF THERMAL FAULTS, NORMAL AND ACCELERATED AGING

Tr1 Tr2 Tr3 Tr4 Tr5 Tr6

Fault currents
Temperature

Overload &Unbalanced load

Normal aging due to dynamic load cycle

Over

Cooling system failure

Increased Oil viscosity

maintain the set point voltage at

Unit startup low frequency Χ Χ Χ Χ
Over excitation

field breaker fails to open when

Unit shutdown the generator trips Χ Χ Χ Χ
The charging current for a high-
Over Voltage At rated frequency voltage transmission line. Χ Χ Χ Χ Χ Χ
Loss of some units
Power system During operation or suddenly Χ Χ Χ Χ Χ Χ
disturbance heavy load
Oxidation of the insulation and
Temperature 80 - 120 °C.

Moisture oils forms acids, Acid attacks Χ

normal operating

cellulose and accelerates

Accelerating aging

insulation degradation, with

Oxygen moisture (PD) Electrical stress Χ
can occur and more insulation
degradation
Acidity Χ

VI. CONCLUSION. Transformers and generators can be subject to repeated over

excitation by inappropriate operating practices or operator error
Modification of Duval triangle DGA diagnostic graph to
without a disruption to operations. It's can be concluded also,
numerical method is easy to use for diagnoses and a Matlab
the resulting thermal faults lead to oil decomposing to generate
program. Transformer thermal faults during dynamic load cycle
fault gases H2, CH4 at temperature 120°C, C2H6 at
due to temperature increase from over load, cooling system
temperature 150°C, C2H4 at temperature 300°C, and C2H2 at
failure or trouble, fault currents and /or over excitation
temperature 700°C.
condition.
The gas type and gas quantity depends on the intensity and
Over excitation, damage usually occurs during periods of
duration of Over-excitation. Transformer diagnostic thereby
off-frequency operation such as start up or shut down for unit
results depends on the events inside evaluation interval or
transformer. In addition, the fault gases can be generated due
before evaluation time.
to oil decomposing effected by transformer over excitation.
REFERENCES
Ahmed E.Kalas received the B.Sc. degree in
[1] T. K. Saha, “Review of modern diagnostic techniques for
electrical engineering from the Suez Canal
assessing insulation condition in aged transformers”, IEEE
Transactions on Dielectrics and Electrical Insulation, Vol. 10, University with honor first rank in EGYPT 1982,
pp. 903-917, 2003. M.Sc. degree (Power electronic and electrical
[2] A. Abu-Siada and S. Islam, “A new approach to identify power Drives), from the Suez canal university, EGYPT
transformer criticality and asset management decision based on 1987, ph. D. degree (Power electronic and
dissolved gas-in-oil analysis”, IEEE Transactions on Dielectrics electrical Drives) from Gdansk university,
and Electrical Insulation, Vol. 19, pp. 1007-1012, 2012. POLAND 1994 From 1994 up to 2010 he worked
[3] “IEEE guide for the interpretation of gases generated in oil- as a lecture in electrical engineering at Suez canal
immersed transformers”, IEEE Standard C57.104-2008, 2009. university, from 2010 up to now he worked as a
[4] M. J. Heathcote, the J & P Transformer Book, Twelfth Edition, lecture in electrical engineering at port said university research contributions,
Reed Educational and Professional Publishing Ltd, 1998. as well as his on-going efforts/investigations in the area of AC drives and power
[5] S. M. Islam, T. Wu and G. Ledwich, “A novel fuzzy logic electronics, can be classified into the following topics:
approach to transformer fault diagnosis”, IEEE Transactions on
Dielectrics and Electrical Insulation, Vol. 7, pp. 177-186, 2000.
Control of electric machines; Vector control, nonlinear control, adaptive
[6] M. A. Izzularab, G. E. M. Aly and D. A. Mansour, “On-line
control, model predictive control, double feed induction motors ,DTC -Power
diagnosis of incipient faults and cellulose degradation based on
artificial intelligence methods”, IEEE International Conference electronic converters, two-level and multilevel, matrix converter, ZS-Artificial
on Solid Dielectrics (ICSD), pp. 767-770, 2004. Intelligence in machines and power electronics control, Fuzzy logic, neural
[7] Md Umar Farooque, Shufali Awani,Shakeb akan "Artificial networks -Renewable energy conversion for PV and wind systems, maximum
neural network (ANN) based implementation of Duval power point tracking -Fault detection Diagnosis in electrical machines and
pentagon"2015 International Conference on condition assesment drives.
techniques in electrical systems (CATCON) pp 46-50, 2015.
[8] Diaa-ELdin A.Monsour "-Development of a new graphical
technique for dissolved gas analysis in power transformers based R.A.Abd El-Aal was born in Egypt, 1971. He
on the five combustible gases"IEEE Transactions on Dielectrics received the B.Sc. degree (1996) and M.Sc. (2002)
and Electrical Insulation, Vol. 22, pp. 2507 - 2512, 2015. in electrical engineering from Suez Canal
[9] Alamuru Vani and Pessapaty Sree Rama Chandra Murthy" University. He received the Ph. D. degree in H.V
Hybrid diagnosing techniques for analyzing dissolved Engineering from Port Said University in 2008.
gases in power transformers " ISSN 2006 - 9790, pp 33-34, He works as lecture in electrical engineering
2015. Dept., Port Said University, Egypt. His research
[10] M. Duval, “A review of faults detectable by gas-in-oil analysis interests are H.V Engineering and power system
in transformers”, IEEE Electrical Insulation Magazine, Vol. 18, protection.
pp. 8-17, 2002.
[11] Nitin K. Dhote1 and Jagdish B. Helonde" Fuzzy Algorithm for
Abdel Moneim M. Hassan. was born in
Power Transformer Diagnostics" Academic Editor: M. Onder
Ismailia of Egypt in 1963. He received the B.Sc.
Efe, pp 1-2, 2013.
degree (1986) in electrical engineering from Hel-
[12] Stefan Tenbohlen , Sebastian Coenen , Mohammad Djamali " wan University. He works as General Manager in
Andreas Müller Diagnostic Measurements for Power Abu Sultan steam power plant 4*150 MW, from
Transformers" Academic Editor: Issouf Fofana, Energies, pp 2, 1988 to 1998, He worked in operation department
2016. as operation engineer in Abu Sultan power plant,
[13] Sherif S.M.Ghoneim , Ibrahim B.M.Taha , Nagy I.Elkalashy from 1998 to 2014, He worked as electrical
"Integrated ANN-Based Proactive Fault Diagnostic Scheme for maintenance, measuring and protection engineer
Power Transformers Using Dissolved Gas Analysis" IEEE in the same plant.
Transactions on Dielectrics and Electrical Insulation, Vol.23,No
3 , pp1838-1845, 2016
[14] L.G Hewitson “Protective Relaying for Power Generation
Systems” Book Taylor & Francis Group, Publishing Ltd 2006.

Sobhy Serry Dessouky was born in Dakahlie of Egypt

in 1946. He received the B.Sc. degree (1970) and M.Sc.
(1977) in electrical engineering from Suez Canal
University in Helwan University respectively. Dr.
Dessouky received the Ph. D. degree from TU, Dresden,
German in 1982. From Oct. 1970 to 1975, he was Joined
Faculty of Engineering, Suez Canal University, as
Demonstrator. He worked as Demonstrator from 1975-
1977 in Faculty of Engineering, Helwan University. In
1977, he worked as lecturer assistant in Electrical Engineering Department,
Faculty of Engineering, Suez Canal University. From 1983 to 1987, he worked
as Assistant Professor (Lecturer), in Electrical Engineering Department, faculty
of Engineering, Suez Canal University, Port Said Campus. In 1987, he
promoted as Associate Professor in the same Department. In 1991, Dr.
Dessouky became a full Professor of Electrical power and H.V Engineering. He
was a member in IEEE from 1996. In parallel, he worked as a department chair,
Vice Dean for Community Affairs and Environment, and Director of
Engineering Research Center for Developing and Technological Planning in
Suez Canal University.