You are on page 1of 6

International Conference on Emerging Trends in Engineering Innovations and Technology Management (ICET: EITM-2017)

NIT Hamirpur, India. December 16-18, 2017

DGA Techniques: Recent Advancements


Kingshuk Chatterjee1, Vinay Kumar Jadoun2 and R.K. Jarial3
1,2,3
Department of Electrical Engineering,
National Institute of Technology, Hamirpur, India
E-mail: 1kingshukchatterjee94@gmail.com, 2vjadounmnit@gmail.com, 3jarial0@gmail.com

Abstract—Dissolved Gas Analysis is one of the most In 1978, IEEE published guidelines for the detection of
reliable and important established technique for gases in oil-immersed transformers, known as ANSI/
determination of incipient faults in power transformers. It is IEEEC57.104-1978 [5]. It covers sampling procedures,
also a fact that there is no crystal clear definite rule or exact instrumentation, methods for extraction and analyzing of
scientific method to interpret the collected DGA data to
gases, and data interpretation. In 1992, IEEE published
identify the fault type. The interpretation techniques used
previously have undergone further modifications and various guidelines (IEEE Std C57.104-1991) dealing mainly with
newer techniques have been also developed in due course of DGA data interpretation [6]. This standard was revised in
time, aiming to achieve greater accuracy and research work 2008 (IEEE Std C57.104-2008) [7]. In 1977, IEC published
in this field is ceaselessly continuing even till date. In this guidelines for the sampling of gases and oil in oil-filled
paper, the historical development of interpretation electrical equipment and the analysis of free and dissolved
techniques are briefly noted, after which the newer gases [10].
generations of graphical techniques, combined ratio methods Several methods used for detecting the gas
or advancements to eliminate stray gassing errors are concentration in transformer oil are Gas Chromatography
accounted for. Future scope for modification of the existing
(GC) [8] and [9], Hydrogen Online Monitoring [10] and
techniques is also discussed in the end of this paper.
Keywords: Dissolved Gas Analysis, Power Transformer, Photo Acoustic Spectroscopy (PAS) [11]. Among them,
Fault Prediction, Interpretation Techniques GC is the most reliable technique, but it is a complex
laboratory method. So, it is a routine test scheduled to be
I. INTRODUCTION performed once in a year. PAS is the best method for
continuous online monitoring [11].
Power transformers are one of the most costly and Now, to interpret the Dissolved Gas Analysis (DGA)
important investments in any power system network. Due data attained from these methods, various techniques have
to operation beyond their designed operating limits and been used. They are mainly based on comparative study
also the normal ageing process the health condition of the and ratio of gas concentration, like Key Gas Method,
transformers gradually deteriorates with time. Thus health Ratio Methods (Doernenburg, Roger’s & IEC), Duval
monitoring of a transformer is of great important to assess triangle and Duval Pentagon Methods [6], [12] and [13].
its present condition, and help predict catastrophic failures Since these methods are based on practical experience
in advance. rather than scientific theory, they might conclude separate
Due to continuous use, Insulation system of the faults for the same data set [1, 14]. For this reason, various
transformers might degrade due to impurities and Artificial Intelligence (AI) techniques are used to achieve
degradation of insulating oil, mechanical imperfections at the conclusion in a more reliable way. The AI techniques
joints and surface, overloading of the transformer etc. mostly used are Artificial Neural Network (ANN) &
These lead to the development of thermal and electrical Fuzzy Logic followed by Support Vector Machine (SVM)
stresses, causing arcing, partial discharge (PD), and Particle Swarm Optimization [4].
overheating and development of hotspots in the core and The reminder of paper is organized as follows:
winding. These faults result in generation of gases which Section 2 presents faults producing gas in oil; Section 3
get dissolved in the transformer oil itself, reducing its describes Historical development of interpretation
dielectric strength [1] and [2]. Typical gases that are found methods; the development of newer techniques is
in transformer oil during incipient faults are hydrogen presented in Section 4, and conclusion and future scope in
(H2), methane (CH4), ethane (C2H6), ethylene (C2H4), Section 5.
acetylene (C2H2), carbon mono-oxide (CO) and carbon di-
oxide (CO2) [3]. The analysis of the dissolved gas in the II. FAULTS PRODUCING GAS IN OIL
oil to detect the incipient fault developing in the
transformer is known as Dissolved Gas Analysis (DGA) Gas formation takes place inside the transformer due to
and is very effective and well-known method for health decomposition of the transformer oil and solid insulation.
assessment of transformer [4]. Transformer oil is a hydrocarbon. Due to thermal or

[193] ISBN: 978-93-86724-30-4


International Conference on Emerging Trends in Engineering Innovations & Technology Management

electrical faults, the Carbon to carbon or hydrogen bond Where, FG is the sum of concentration of H2, CH4,
breaks down resulting in formation of radicals, which C2H6, C2H4 C2H2, and CO in ppm. V is the volume of oil
further recombine to produce various gases which gets in transformer (liters (gallons)).
dissolved in the oil. Recombination occurs depending upon One of the very first techniques to interpret DGA data
the fault type and temperature [15, 16]. was Key Gas Method. It is based on the fact that for
The type of gas to be formed depends on the type of different types of faults different gas concentrations are
hydrocarbons present in the insulating oil, distribution of predominant. These gases are called ‘Key Gases’. The
energy and temperature in the nearby region of the fault, four key gases are C2H4, CO, H2 and C2H2 and the higher
duration for which the oil is under thermal or electrical percentage of which indicates thermal-oil, thermal-paper,
stress [6]. low energy partial discharge and high energy discharge
Types of gases found dissolved in the oil for various respectively [3, 6, 18].
types of faults are listed below [3]: Ratio methods have considered analysis of DGA data
Partial Discharge–H2, CH4, C2H2 (trace amount) based on values of gas ratios. The gas ratios considered are
Arching - H2, CH4, C2H2, C2H4 [8] CH4/H2 (R1), C2H2/C2H4 (R2), C2H2/CH4, (R3),
Thermal Fault (T1) - H2, CH4, C2H4 (trace C2H6/C2H2 (R4), C2H4/C2H6 (R5) and C2H4/CH4 (R6) [7].
amount), C2H6 The ratio methods were developed first by Doernenberg,
Thermal Fault (T2) - H2, CH4, C2H2 (trace considering 4 gas ratios R1, R2, R3 and R4. Roger’s Ratio
amount), C2H4, C2H6 Method (RRM) consist of 3 gas ratios R1, R2 and R5 while
Thermal Fault (T3) - H2, CH4, C2H2, C2H4, in the IEC Ratio Method there were further modification of
Leaks in oil expansion system, gaskets etc.–CO2, the RRM to predict a total of 6 types of fault [3, 6, 19, 20].
O2, H2O. From a general trend, it has been seen that DRM and
Mineral oil decomposition - H2, CH4, C2H2, IRM are most accurate interpretation tool where KGM is
C2H4, C2H6. the least accurate [4] and [18]. However, one drawback of
Cellulose Ageing–CO, CO2, H2O. all the Ratio Methods are that in some cases, where the
gas ratio combinations do not fit into any fault conditions
Thermal Faults (Cellulose) - H2, CO, CO2, CH4,
[20]. These unresolved cases decrease the accuracy of the
O2.
Some types of oil produce H2 & CH4 gases when they Ratio Methods. This problem ultimately made researchers
are heated for the first time in their lifetime. The develop the modern techniques of Graphical methods, as
concentrations vary within 5-250ppm depending upon the discussed in the next section.
type of oil. So, DGA performed at the very beginning of the
life of a new gas might have abnormal results indicating IV. DEVELOPMENT OF NEWER TECHNIQUES
fault, but it is perhaps only due to production of stray gases.
Investigation is going on to determine the stray gassing oil A. Duval Triangle Method
currently in use and the concentration levels of produced The Duval Triangle Method (DTM) pioneered the
stray gasses to avoid this confusion [17]. graphical methods and was developed by M. Duval. The
Duval triangle is an equilateral triangle where each side is
III. HISTORICAL DEVELOPMENT OF scaled with the relative percentage concentration of a key
INTERPRETATION METHODS gas in parts per million (ppm) [17]. The three key gases
After the DGA data has been arrived at hand, the considered are CH4, C2H4, and C2H2. Inside the triangle,
interpretation of the data is required. Various methods for there are six different fault zones and depending upon the
this purpose are discussed as they evolved in the concentrations of these gases, a point is plotted. The zone
chronological order. When a fault occurs in the transformer, in which the point lies indicates the type of fault.
thermal or electrical, the gases produced by oil The relative concentrations in percentage are
decomposition are given in the previous section. However, calculated as follows in [17]:
this inference is based more on practical results, experience % C2H2 = 100 x /(x + y + z);
and statistics than on definite science. Depending upon the %C2H4 = 100y /(x + y + z);
relation between faults occurred and gas formed, the IEEE %CH4 = 100z /(x + y + z),
guideline for monitoring of the insulation is given in IEEE Where x = (C2H2); y = (C2H4); z = (CH4), in ppm.
Std C57.104™-2008. It is based on analyzing the total The only fault n the triangle previously not discussed
volume of gas produced as indication of the magnitude of is DT, which refers to both thermal and electrical fault.
incipient fault. For finding out the Total Dissolved Duval Triangle is shown in Figure 1 as in [17].
Combustible Gas Concentration volume (TDCGV), the The Duval Triangle 2 is proposed for on load tap
formula used is given below and as in [7] changers of the oil type, where arcing in oil happens
TDCGV = FG(V)/(1000000) (1) during operating conditions. The Duval Triangle 3 is

[194]
DGA Techniques: Recent Advancements

proposed for natural or synthetic esters as insulating oil for The relative percentage of concentration of each gas
transformers [17]. is plotted in the axis of the individual gases (the red
DTM (Duval Triangle 1) proved to be a very squares in Figure 2). The centroid of the irregular polygon
successful tool for DGA interpretation. However, formed by joining those points gives the DGA result for
sometimes faults occurring in the boundaries of two zones the case [22].
become difficult to identify. The location of the five red squares (xi, yi) in the
Also, due to the occurrence of stray gassing in some plane made by the X and Y axes can be easily found out
oil, confusion occurs in the zones T1, T1 and PD [21]. with help of the angle between the co-ordinate axes and
Due to these problems, Duval Triangles 4 and 5 were gas concentration axis. The (x, y) coordinates of the
developed to take care of these problems. These triangles centroid of these five points are then calculated using the
4 and 5 are to be applied only when fault identified by equations indicated in [23]:
Duval Triangle 1 is either of PD, T1 or T2. Triangle 4 uses n 1
Cx 1
xi xi xi yi xi 1 yi (2)
three so called ‘low energy gases H2, CH4 & C2H6. It is 6A 1 1
i 0
developed by inspecting cases of transformer faults and
stray gassing test results performed in the laboratory [17]. Cy 1
6A
(yi yi 1 ) xi yi 1 xi 1 yi (3)
n 1
1 (4)
A (x i y i 1 x i 1yi )
2 i 0
Where A is the surface of the polygon Cx, Cy are the
coordinates of the centroid.
The fault zones in Pentagon 1, as shown in figure 3,
are same as that of Duval triangle. Apart from those six
faults stray gassing is also identified by the Duval
Pentagon as determined in the laboratory tests [24].

Fig. 1: The Duval Triangle

B. The Duval Pentagon


Duval Pentagon was developed based on the
philosophy of the Duval Triangle considering relative
concentration of five key gases scaled from 0 to 100%.
For each gas, the 0% concentration is plotted at the centre
of the pentagon and 100% concentration at one of the
vertices, with intermediate concentrations along the line
joining the centre and that corner [22]. Fig. 3: Fault Zones for Mineral oil in Duval Pentagon 1

Figure 4 shows Duval Pentagon 2, with fault zones


corresponding to the three electrical faults, PD, D1, and
D2, and the four advanced thermal faults used in Duval
Triangles 4 and 5 [26]:
T3-H: thermal faults T3-H in oil only,
C: thermal faults T3-C, T2-C, and T1-C with
carbonization of paper,
O: overheating T1-O <250°C, and
S: stray gassing S of mineral oil at 120 and
200°C in the laboratory
The Duval pentagon is a very useful tool in
determining the fault type by Dissolved Gas Analysis and
its precision is superior to all the previous methods [22],
Fig. 2: The Duval Pentagon [25] and [26].

[195]
International Conference on Emerging Trends in Engineering Innovations & Technology Management

The efficiency of this method is comparatively much


higher than all the ratio methods discussed earlier.
However, the location of the fault, whether in paper or oil,
cannot be determined by this method. The flow chart of
this method is shown in Figure 5 as in [27].
T ABLE 1: RESULT OF FAULT P ATTERN ANALYSIS
WITH GAS R ATIO COMBINATIONS
Gas-Ratio Distinction Distinction Distinction
Combination between Among Among
Electrical and Electrical Faults Thermal Faults
Thermal PD- PD- D1- T1- T1- T2-
Faults D1 D2 D2 T2 T3 T3
R1-R2
R1-R3 O X X X
R1-R4 X X X X
R1-R5 O
Fig. 4: Duval Pentagon 2, with Fault Zones R1-R6 O o
Corresponding to the Three Electrical Faults R2-R3 O
R2-R4 O
R2-R5 O 0
C. New Gas Ratio Combined Method & Relative Content R2-R6 O X O O
of Fault Gases R3-R4 X X X
R3-R5
This method is based on gas ratio method, but it is R3-R6 X o O
more precisely a combination of different gas ratios from R4-R5
R1 to R6 to clearly classify 6 types of faults, and since one R4-R6 O O
gas ratio is not enough to distinguish between all types of R5-R6 X O

faults, the selected ratios were reorganized into 15 types of Clear distinction, unclear distinction, O
gas-ratio combinations to analyze fault patterns as shown distinguishable, X undistinguishable
in table 1 [27].
D. Advanced Method to Eliminate Stray Gassing Error–
HS-TS Method
Another recently developed technique takes care of
the errors generated due to stray gassing in mineral oil.
Stray gassing experiment was performed by heating the
insulating oil to temperatures of 130 °C to 160° C with an
increment of 10° C for 16 hours each to get dissolved gas
data. This experiment was conducted six times over a six
month period according to CIGRE JTF Dl.OllA2.l1 (TF
11). Collected data are analyzed to identify any of the
existing 6 types of faults and any stray gassing
simultaneously [28].
Diagnosis can be done if, either of the concentration
of H2, CH4, C2H2, C2H4 or C2H6 generated is beyond the
dissolved gas concentration limit given in IEEE C57.104
and Fhigher. Table 2 shows the characteristics of each
fault based on the percentage concentration of the five
Fig. 5: Flow Chart for Fault Detection by Gas Ratio Method gases. From the table, it can be interpreted, that if
It is evident from the table, that R1–R2 combination is concentration of C2H2 is zero, and that of C2 H4 is below a
first used to distinguish between thermal and electrical certain value, the identification of fault can be performed
fault. Then, if it is thermal fault, R5–R6 combination using the graphical method HS-5 depicted in Figure 6.
differentiates T1 from T2 & T3. R2–R5 ratio will finally Also the entire process is presented in a flowchart as in
distinguish between T2 & T3. On the other hand, if it Figure 7. This method has very high accuracy for fault
comes out to be electrical fault, R2–R5 ratio alone is detection by DGA method. It reduces the error in
sufficient to distinguish between PD, D1 & D2 fault types. detecting faults, mainly thermal faults, due to stray

[196]
DGA Techniques: Recent Advancements

gassing, which ultimately enhanced greatly the overall A lot of work has been done and still the research is
percentage of correct analysis of DGA data. Around 90% going on in DGA technology. It is amongst the most
overall correct detection is achieved by this method [27], reliable condition monitoring tools for power transformer.
[28], [29] and [30]. However, since the accuracy is still not 100%, there is
always some scope for future development. The graphical
TABLE II: GAS VALUES (%) FOR E ACH FAULT TYPES methods can be further modified using AI techniques. Due
Fault % % % % % to the sharp boundary between fault zones closely lying
H2 CH4 C2H6 C2H4 C2H2
STRAY Type PD 75 or 15 or 7 or 5 or 0
points are interpreted to be different faults. With adequate
GAS higher lower lower lower training process using ANN, the boundaries between fault
Type T1 20 or 35–50 30–50 10 or 0 zones can be fuzzified. The entire area of the pentagon can
lower lower be divided into two types of regions, ‘region of certainty’
Types T1-T2 27–34 38–47 23–28 5 or 0 and ‘region of uncertainty’, depending upon the accuracy
lower
of prediction in various regions. Then, a criterion is set for
Type T2 38–65 20–33 10–35 10 or 0
lower cross checking with other methods, for points which lie in
Experiment 20 or 20–50 35–60 15 or 0 regions, where more than one faults are probable. Then,
lower lower we can have the idea, which results are reliable, and which
PD 87–98 1.8–12 2.5 or 0.1 or 0 of them are doubtful, and asks for further analysis. Better
lower lower
decision making can be arrived by firstly identifying the
T1 10–30 50–55 10–20 10 or 1 or
lower lower most accurate techniques and then combining the
T2 35 or 30–55 5–20 10 - 40 2 or collective interpretation achieved by these techniques.
lower lower Also, the investigations can be done to find out, if data
giving error in one technique can give correct
interpretation in another technique, and systematically
classifying them so as to develop an algorithm so that a
combined interpretation can be done using various
techniques, whereby each technique are fed with data type
with which its response is better.

REFERENCES
[1] Standard Test Method for Analysis of Gases Dissolved in Electrical
InsulatingOil by Gas Chromatography, ASTM D3612-02
(Reapproved 2009),2009.
[2] N.A. Muhamad, B.T. Phung, T.R. Blackburn, and K.X. Lai,
“Comparative study and analysis of DGA methods for transformer
mineral oil,” in 2007 IEEE Lausanne Power Tech, 2007, pp. 45–50.
Fig. 6: Identification of Fault Using the Graphical Method HS-5 [3] Duval, M.: ‘Dissolved gas analysis: it can save your transformer’,
IEEEElectr. Insul. Mag., 1989, 5, (6), pp. 22–27
[4] Norazhar Abu Bakar “ A Review of Dissolved Gas Analysis
V. CONCLUSION AND FUTURE SCOPE Measurementand Interpretation Techniques” DEIS, vol. 30, no. 3,
pp 39-49
[5] IEEE Guide for the Detection and Determination of Generated
Gases in Oil-Immersed Transformers and Their Relation to the
Serviceability of the Equipment, ANSI/IEEE Std C57.104-1978, p.
0_1, 1978.
[6] IEEE Guide for the Interpretation of Gases Generated in Oil-
ImmersedTransformers, IEEE Std C57.104-1991, p. 0_1, 1992.
[7] IEEE Guide for the Interpretation of Gases Generated in Oil-
Immersed Transformers, IEEE Std C57.104-2008 (Revision of
IEEE Std C57.104-1991), pp. C1–27, 2009.
[8] O.D. Sparkman, Z.E. Penton, and F.G. Kitson, “Chapter 2—Gas
chromatography,”in Gas Chromatography and Mass Spectrometry,
2nd ed. Amsterdam, the Netherlands: Academic Press, 2011, pp.
15–83.
[9] Standard Guide for Sampling, Test Methods, and Specifications for
Electrical Insulating Oils of Petroleum Origin, ASTM D117-10,
2010.
[10] J.L. Kirtley, W.H. Hagman, B.C. Lesieutre, M.J. Boyd, E.P. Warren,
H.P. Chou, and R.D. Tabors, “Monitoring the health of power
transformers,” IEEE Comp. Appl. Power, vol. 9, pp. 18–23, 1996.
[11] D. Skelly, “ Photo-acoustic spectroscopy for dissolved gas
analysis:Benefits and experience,” in 2012 International
Conference on Condition Monitoring and Diagnosis (CMD), 2012,
Fig. 7: Flow Chart of HS-5 Method pp. 29–43.

[197]
International Conference on Emerging Trends in Engineering Innovations & Technology Management

[12] R.R. Rogers, “IEEE and IEC codes to interpret incipient faults in [22] M. Duval & L. Lamarre, “The Duval Pentagon-A New
transformers, using gas in oil analysis,” IEEE Trans. Electr. Insul., Complimentary Tool for the Interpretation of Dissolved Gas
vol. EI-13, pp. 349–354, 1978. Analysis in Transformers”, IEEE Electr. Insul. Mag., Vol. 30 No.
[13] M. Duval, “ Use of Pentagons and Triangles for the Interpretation of 6, pp. 9-12, Nov/Dec. 2014.
[23] P. Bourke, Calculating the Area and Centroid of a Polygon.
DGA in Electrical Equipment”, Proc. Of TechCon North America
Available: http://paulbourke.net/geometry/polyarea.
Conf., Albuquerque, Feb. 2016.
[24] CIGRE Technical Brochure #296, “ Recent Developments in DGA
[14] A. Abu-Siada and S. Islam, “A new approach to identify power Interpretation,” Jun. 2006.
transformer criticality and asset management decision based on [25] M. Duval, “ Use of Pentagons and Triangles for the Interpretation of
dissolved gas-in-oil analysis,” IEEE Trans. Dielectr. Electr. DGA in Electrical Equipment” , Proc. Of TechCon North America
Insulation, vol. 19, pp. 1007–1012, 2012. Conf., Albuquerque, Feb. 2016.
[15] F. Jakob, P. Noble, J.J.Dukarm, “A Thermodynamic Approach to [26] M. Duval, “ State of the Art of Technical Diagnosis through
Evaluation of the Severity of Transformer Faults,” IEEE Transactions Intepretation of DGA” , Proc. Of My Transfo Conf., Torino, Italy,
on Power Delivery, vol.27, no.2, pp.554, 559, April 2012. Nov. 2012
[16] Lelekakis, N., Martin, D., Guo, W.: ‘Comparison of dissolved gas- [27] Sung-wook Kim, Sung-jik Kim, Hwang-dong Seo, Jae-ryong Jung,
in-oil analysis methods using a dissolved gas-in-oil standard’, IEEE Hang-jun Yang, Michel Duval, “ New Methods of DGA Diagnosis
using IEC TC 10 and Related Databases Part I : Application of
Electr. Insul. Mag., 2011, 27, (5), pp. 29–35
Gas-ratio Combinations,” IEEE Transactions on Dielectrics and
[17] M. Duval, “The duval triangle for load tap changers, non-mineral Electrical Insulation, vol. 20, no. 2, pp. 685-690, 2013.
oils and low temperature faults in transformers,” IEEE Electr. [28] Sungjik.Kim*, Hwangdong.Seo, Jaeryong.Jung, “ Advanced
Insul. Mag., vol. 24, pp. 22–29, 2008. Dissolved Gas Analysis Method with Stray Gassing Diagnosis
[18] S. Corporation, “Serveron White Paper: DGA Diagnostic Advanced Dissolved Gas Analysis Method with Stray Gassing
Methods,” 2007. Diagnosis” , in 2016 International Conference on Condition
[19] R.R. Rogers, “IEEE and IEC codes to interpret incipient faults in Monitoring and Diagnosis - Xi'an–China, pp. 522-525, 2016.
transformers, using gas in oil analysis,” IEEE Trans. Electr. Insul., [29] Soo-jin Lee, Young-min Kim, Hwang-dong Seo, Jae-ryong Jung,
vol. EI-13, pp. 349–354, 1978. Hangjun Yang, “New Methods of DGA Diagnosis using IEC TC
10 and Related Databases Part 2 : Application of Relative Content
[20] Jeff Golarz, “ Understanding Dissolved Gas Analysis (DGA)
of Fault Gases,” IEEE Transactions on Dielectrics and Electrical
Techniques and Interpretations” , International Conference on Insulation, vol. 20, No. 2, pp. 691-696, 2013.
Transmission and Distribution Conference and Exposition (T&D), [30] Sung-jik Kim, Hwang-Dong Seo, Sung-Wook Kim, Mr. Jae-Ryong
2016 IEEE/PES, 3-5 May 2016. Jung, Hang-Jun Yang “ Application of New DGA Diagnostic
[21] M. Duval et al., Joint Task force D1.01/ A2.11 of CIGRE, “Recent Method for Power Transformers” ICEE PSS&RP-0686, pp. 865-
developments in DGA interpretation,” CIGRE Brochure # 296, 2006. 870, June 2014

[198]

You might also like