hydrogen gas generation due to moderately overheated...
TRANSCRIPT
1
1
Abstract: This paper first presents an overview of this phenomenon from its initial discovery to the stage of the factory / laboratory investigations performed to confirm this new mechanism of gas generation in power transformers. This is followed by an overview of the increased significance of developing accurate calculation of the core hot spot temperature. The paper then presents the detailed accurate diagnosis performed on a 600 MVA transformer that had this issue in the field, the improvement made, and the calculation of the predicted gas generation performance of its loading cycle throughout a year of operation. The paper then presents the basis for the additions / changes proposed to be implemented in the IEEE Standards as a result of the discovery of this new gas generation mechanism. The paper also provides the basis for recommending to the IEEE Standards the selection of 130°C as the allowed maximum core hot spot temperature under sustained worst conditions of load, core – excitation and ambient temperature.
Index Terms—hydrogen gas generation, core hot spot temperature, power transformers, transformer core
I. INTRODUCTION
A decade ago, one major utility in US reported moderate generation of hydrogen gas in six same design large power transformers with generation rates ranging from 0.5 to 3.5 parts per million (ppm) per day and an H2/CH4 ratio of 6 – 8. Other gases, mainly CO and CO2, were generated at a low rate. Dissolved Gas in Oil (DGA) interpretation suggested partial discharge (PD) activity but attempts to locate PD were fruitless. Degassing of the units had the effect of clearing the generated gases; however the gases would simply start rising again. Investigations on the source of this gassing started in 1996 and the results of these investigations were reported in References [1] and [2]. It was found that this gas generation phenomenon was caused by moderately overheated cores with core hot spot temperatures in the 120 – 160 ºC range. Since then, a number of other transformers with similar gas generation signature have been reported.
In this paper, the background of this newly discovered mechanism of H2 and CH4 generation is presented along with a short description of factory and laboratory investigations performed to explain this phenomenon. A more recent case of
core hydrogen gassing is examined in detail. For this case, calculated gassing rates are compared with actual field and factory measured rates to verify the core gassing phenomena and the developed relationship between core hot spot temperature and core gassing. This core gassing is shown to occur under a certain combination of core excitation, ambient temperature, and loading conditions. Alternatives for mitigation of the phenomena are discussed. The new emphasis on accurate calculation of the core hot spot is described. Finally, the impact on the transformer industry and Industry Standards is discussed.
II. DISCOVERY OF THE PHENOMENON
As mentioned above, the hydrogen generation was first observed in six large same – design power transformers. In particular, the one unit generated 4000 PPM of hydrogen over a period a little over 3 years with an average daily rate of 3.5 PPM. Other gases were generated but at lower rates (see Figure 1). Five other units of the same design showed a similar behavior but lower gas generation rates. The CH4/H2 ratio was between 0.045 and 0.12, and the limit of < 0.1 for corona was met in only two of the six sets of data.
0
500
1000
1500
2000
2500
3000
3500
4000
4500
1 2 3 4 5 6 7 8 9 10 11 12 13 14Sample
PPM
C2H2
C2H6
C2H4
CO2
CO
CH4
H2
Figure 1 – DGA of Transformer with moderately overheated core
In order to investigate the cause of the phenomena, one of
the units was transferred to a factory for testing. Neither extended heat – run tests nor extended core excitation tests at higher voltages at 180 Hz generated any gasses. However, extended 60 Hz core excitation tests at 90%, 100% and 110%
Hydrogen Gas Generation Due to Moderately Overheated Transformer Cores
Ramsis Girgis, Fellow, IEEE, and Ed G. teNyenhuis, Member, IEEE
_______________________________________________
2
2
voltage did generate gases similar to those generated in the field with the same H2/ CH4 ratios. Further tests demonstrated that the hydrogen gas generation was a function of the core hot spot temperature. The core hot spot temperatures for 90%, 100% and 110% excitation were 111°C, 125°C and 168°C respectively. It should be noted that the core hot spot temperature for this unit was greater than what would be normal in more recent designs.
These factory tests concluded that the hydrogen gassing was related to core over – excitation and not partial discharge. This was further confirmed in the field where the units had the core excitation reduced by changing the tap position. This resulted in a 10°C lower core hot spot temperature and a corresponding reduction of 60 % in the rate of generation of the hydrogen and methane, Ref [2].
Subsequently, laboratory investigations confirmed the phenomena where electrical steel bundles were immersed in transformer oil and aged in a stainless steel canister. The temperatures were elevated to a 100 – 200°C range and the hydrogen gas generation rates were measured. These hydrogen gas generation rates were very close to the field and factory test measurements. Further investigations showed that the thin oil film between the electrical steel sheets facilitates a chemical reaction where hydrogen atoms of the hydrocarbon chain of the oil are loosely aligned to the steel surface and then released as hydrogen molecules. This reaction occurs at relatively lower temperatures of 115°C to 160°C compared to when hydrogen is produced by bulk oil at temperatures of several hundred degrees.
0
2
4
6
8
10
100 110 120 130 140 150 160 170 180
Core Hot Spot Temperature [C]
H2
Gen
erat
ion
Rat
e PP
M/D
ay
Figure 2 – Hydrogen generation vs. core hot Spot Temperature
The final and most important result from the factory testing
and laboratory measurements is shown in Figure 2 above. The figure shows the relationship between hydrogen gas generation rate, in PPM per day and the core hot spot temperature. What is astonishing is that hydrogen gas generation begins to occur at a core hot spot temperature as low as 110°C and rises to levels of 4 PPM per day at 140°C.
From the above, core hot spot temperatures should be given an elevated importance. Prior to the discovery of this gas generation mechanism, the core hot spot temperature would not have been considered a critical design item. Moreover,
limits on this temperature used to be the same as that applied to metallic parts not exposed to Cellulose, i.e. 180°C.
III. CALCULATION OF CORE HOT SPOT TEMPERATURE The hottest spot in a 3 – phase, 3 – limb core is in the
geometrical center of the T – joint of the upper yoke as shown in Figure 3 (a) below. For other core types, the hottest spot is located in the upper part of the middle wound limb (s) as shown for a 1 – phase, 3 – limb core in Figure 3 (b). The location of the core hot spot is determined by both the loss density distribution and the cooling conditions in a core. The core hot-spot temperature rise is, therefore, a function of several design and operating parameters. These are core type; core dimensions, core material, operating induction, number of core cooling ducts, size of the cooling ducts, and finally the oil temperature.
Location of core hot-spot (a) 3 – phase, 3 – limb Core
(a) 1 – phase, 3 – limb Core
Figure 3 - Location of the Hot Spot in transformer cores
The absolute value of the core hot spot temperature is the
sum of the temperature of the oil around the location of the core hot spot and the core temperature rise at this location; caused by the core losses due to the main flux in the core. The maximum oil temperature occurs during full load current and maximum average ambient temperature. Thus, the hottest temperature value of the core hot spot is generally seen during full load, highest ambient temperature, and highest core
3
3
excitation. A description of the development and verification of an
accurate calculation of the core hot spot temperature in 3 – phase, 3 – limb cores is given in Ref. [3]. 2D FEM loss / thermal analysis was used to gain insight into the parameters that determines the magnitude of the core hot spot temperature rise and the contribution of each of these parameters. This analysis was then used to develop a simple method that has been used now for several years for everyday design calculations with an accuracy of 2°C in the majority of the designs.
IV. ANALYSIS OF A RECENT CASE OF H2 GENERATION Recent events with a 600 MVA transformer indicated the
hydrogen gassing phenomena while energized, over a period of approximately 5 months, at no load in a location of hot weather during the summer months with no fans running. This unit experienced a constant hydrogen gas generation of about 1 PPM / day, accumulating 150 ppm of hydrogen and about 20 ppm of methane over that period. Shown in Figure 4 below is the ratio of the generated hydrogen and methane gas. This ratio is in the range of 6 – 8, which is not a ratio that would indicate partial discharge activity. It also indicates hydrogen generation due to moderately overheated core.
0
2
4
6
8
10
100 110 120 130 140 150 160 170 180
Core Hot Spot Temperature [C]
H2
Gen
erat
ion
Rat
e PP
M/D
ay
Figure 4 – Measured Hydrogen and Methane Gas generation
in the 600MVA transformer
Obviously, it could not be attributed to PD because the rate of hydrogen generation was very low and almost constant over the five months period, it was also associated with a low ratio of H2 to CH4 of about 7, and no PD activity could be measured.
A. Diagnosis of the gassing In diagnosing this gas generation issue, the daily core hot
spot temperatures were calculated from daily ambient temperatures, oil temperatures, and core excitation. The calculated values are given in Figure 5. As can be seen from the figure, the core hot spot temperature was in the range where hydrogen gas would be generated. From this, the generated hydrogen gas (in daily PPM) corresponding to the calculated core hot spot temperatures was calculated as shown in Figure 6 below. Lastly, the calculated and actual measured
accumulated hydrogen gas (from DGA of the oil samples) over the 5 – month period of energization is compared in Figure 7. This figure demonstrates very clearly that the calculated hydrogen gas accumulation matches closely the corresponding measured hydrogen gas in the field. This confirms this hydrogen gas generation phenomena, the accuracy of the core hot spot calculation method, and the developed relationship, presented above in Figure 2, between hydrogen generation rate and core hot spot temperature.
90
95
100
105
110
115
120
125
130
1-Sep 26-Sep 21-Oct 15-Nov 10-Dec 4-Jan
MonthC
ore
Hot
Spo
t Tem
pera
ture
- D
eg C
Figure 5 – Calculated daily Core Hot Spot Temperatures for
the 600MVA Transformer
0.00.20.40.60.81.01.21.41.61.82.02.22.4
1-Sep 26-Sep 21-Oct 15-Nov 10-Dec 4-Jan
Date
H2
PPM
Gen
erat
ion
/ Day
Figure 6 – Calculated Daily Hydrogen Gassing Rate for the
600MVA Transformer
0
20
40
60
80
100
120
140
160
1/Sep 21/Sep 11/Oct 31/Oct 20/Nov 10/Dec 30/Dec 19/JanSample Date
Acc
umul
ated
Hyd
roge
n G
as [P
PM]
MeasuredCalculated
Figure 7 – Measured vs. Calculated Accumulated Hydrogen
Gas for the 600MVA transformer
A surprising note about this transformer was that it was
4
4
energized with minimal load and would thus not be expected to exhibit such high core hot spot temperatures. In fact, elevated top oil temperatures were caused by a combination of over-excitation (103.5%) and no load. This elevated top oil temperature led to an artificially high core hot spot temperature. Transformer losses, which were predominantly core loss at this condition, were not sufficient to drive the oil through the radiators and caused stagnate hot oil to accumulate near the top of tank. This stagnant oil situation was witnessed on another transformer as shown in the thermal scan in Figure . Actual top oil temperature measurements on the 600MVA transformer confirmed that the same phenomenon of oil stagnation occurred in this transformer.
Stagnate hot oil59 C
25 C
Figure 8 – Stagnate Hot Oil in the top of a transformer
B. Factory Testing This transformer was returned to the factory where it
underwent extended core excitation tests designed to simulate field conditions of core and oil temperatures. DGA samples were taken every 24 hours. Shown in Figure below is the measured and calculated Hydrogen gas generation during the tests. Again, the phenomenon was confirmed and the calculation method was proven to be accurate.
0
5
10
15
20
25
0 24 48 72 96 120 144 168
Test Time Duration [hrs]
Cum
ulat
ive
H2
PPM
Gen
erat
ion
CalculatedMeasured
Figure 9 - Measured and Calculated Hydrogen Gas generation
during Factory Tests on the 600MVA transformer
C. Improved Cooling and Predicted Performance
Cooling equivalent to a 5°C reduction in the average oil temperature was added to the transformer in order to reduce the # of days in the year that the transformer would produce hydrogen. An anticipated loading cycle for this transformer is shown in Figure 3 where the load is at the maximum rating of 600 MVA during the 3 summer months and down to 400 MVA the rest of the year.
Based on this, the daily hydrogen gassing rate was calculated by first calculating the corresponding core hot spot at the different rating conditions and accounting for the average daily temperatures during the year for the region where the transformer is located. From the calculated core hot spot temperature the corresponding daily hydrogen generation rate throughout the year was calculated using the relationship presented earlier in Figure 2. The result is presented in Figure 4 with and without the added cooling. The figure shows that the rate of generation of Hydrogen is in the range of 2.5 – 3.2 PPM / day during the summer months without the improved cooling which reduces this rate by 20 – 25 %. The rate of gas generation in the spring and fall is small and non existent the rest of the year.
0
100
200
300
400
500
600
700
J F M A M J J A S O N DMonth
MV
A
June/July/Aug 600 MVA Loading
400 MVA 400 MVA
Figure 30 – Predicted load cycle for the 600MVA transformer
Calculated Daily H2 PPM For a Year
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
J F M A M J J A S O N DMonth
H2
PP
M G
ener
atio
n / D
ay
Present CoolingPresent + Added cooling
Figure 41 – Calculated H2 gas generation throughout the year for the 600MVA transformer
Lastly, in Figure 5 below is shown the accumulated hydrogen gassing over a year, again calculated from Figure 11, with and without the added cooling. The accumulation starts
5
5
slow in June and increases at an almost constant rate through the summer months. Obviously, this is affected by the load on the transformer and the ambient noise in the off season.
Calculated Cumulative H2 PPM For a Year
0
50
100
150
200
250
300
J F M A M J J A S O N DMonth
Cum
ulat
ive
H2
PP
M G
ener
atio
n Present CoolingPresent + added cooling
Figure 52 - Calculated Cumulative H2 Gassing for the 600MVA transformer
V. IMPACT OF H2 GENERATION DUE TO CORE HEATING As this phenomenon is associated with core hot spot
temperatures greater than 110°C, hydrogen generation is only associated with transformers when operating under simultaneous worst case of high ambient, full load, and over-excitation. Under conditions of lower levels of any, or two, of theses parameters, very low hydrogen gassing would be generated which is the typical conditions for the large majority of power transformers. Figure 6 below demonstrates the impact of loading and over-voltage for a typical power transformer when the ambient temperature is 30°C ambient. Presented in this figure are calculated core hot spot temperatures at three levels of load and a 100 % – 115 % range of core excitation.
90
100
110
120
130
140
150
160
170
95% 100% 105% 110% 115% 120%
% of Rated Flux
Cor
e H
ot S
pot T
emp
[deg
C]
100% MVA
80% MVA
50% MVA
Figure 63 – Impact of Loading and Core excitation on Core
Hot Spot Temperature for a Power Transformer The figure demonstrates that the core hot spot temperature
would be 125°C when the transformer is operating at full MVA and 100% core excitation. But if the loading drops to
80%, hydrogen gas generation would not occur. Similarly, at 50% loading, the core would need to be overexcited by 110% before hydrogen gassing would occur. Generally, only generator step up transformers operate near full load and a few % over – excitation during some periods of the day. Most other transformers, especially substation and intertie transformers are typically mildly loaded and only rarely are over – excited. Moreover, the conditions that would cause hydrogen generation would have to be sustained for whole periods of the day and year to have significant accumulation. Obviously, only summer and transformers located in hot ambient temperature regions would experience this phenomenon. Therefore, realistically, it is only a small percentage of all transformers would experience this phenomenon.
Another aspect of this phenomenon is that the impact of hydrogen gas generation due to this phenomenon is not in itself harmful to the transformer. It is only that this hydrogen generation over a period of time could reach a high accumulation that could mask otherwise detrimental hydrogen generation due to other phenomena such as partial discharge and high temperature metal overheating of bulk oil. For this reason, it is advisable that when a transformer is experiencing hydrogen generation due to core overheating that the oil is de-gassed periodically and actions taken to change the core excitation (Tap position) and / or providing additional cooling of the transformer oil.
As mentioned earlier, manufacturers have not had in the past the need for an accurate calculation of the core hot spot due to the main flux in the core because: 1. This phenomenon or mechanism of hydrogen generation
due to core overheating was not known until a few years ago.
2. The limit on the core hot spot temperature was that of metallic parts not exposed to Cellulose.
VI. INDUSTRY STANDARDS AND CUSTOMER SPECIFICATIONS
This new H2 / CH4 gas generation mechanism has been presented to the industry in References [1] and [2] as well as in a tutorial given at the 2003 fall meeting of the IEEE Transformers Standards committee meeting. As a consequence, several power transformer users introduced requirements for verifiably accurate calculation of the core hot spot temperature, Ref. [3]. Additionally, and following the recommendations of the authors of this paper, users of power transformers included in their customer specifications a limit of 130°C on the maximum value of the core hot spot under worst conditions of load, ambient temperature, and core over – excitation.
Shortly after, a Task Force was formed as part of the IEEE Transformer Standards committee with the following objectives: (1) Define more accurately the operating conditions above rated voltage, or below rated frequency, stated in section 4.1.6 of C57.12.90. These conditions impact the flux density, the core hot spot temperature, and the hydrogen gassing. The Standards presently state that a
6
6
transformer must deliver rated MVA at 0.8 power factor at 105% continuous voltage on the secondary. This requirement typically means a potential core excitation up to 115 % for generator step – up transformers at full load when the LV voltage is calculated using the impedance voltage drop. However, generators would not be capable of delivering such a high voltage at full MVA. (2) Agree on, and add in the Standards, a limit on the maximum level of the core hot spot temperature allowed in a transformer design. Based on Figure 2 of this paper, a limit of 130°C was recommended for worst simultaneous conditions of core over – excitation, load, and ambient temperature. This ensures that under typical operating conditions through the year and life of a transformer, a maximum of 110°C for the core hot spot is not reached and consequently no gas generation would occur.
The Task Force recommended changes in the IEEE Standard C57.12.00 with regards to the core over-excitation by identifying that the primary voltage will be limited by the generator voltage or system voltage for the cases of a generator step transformer or system tie transformer, respectively, under full load conditions. The task force also recommended adding wording in C57.12.00 and C57.91 to limit the core hot spot temperature to 130ºC. As well, the Task Force recommended wording for the dissolved gas – oil – guide C57.104 to identify the gas signature seen with the core hot spot phenomena, namely generation of H2 and CH4 at a low PPM per day rate and a 6 – 8 ratio.
Finally, it should be noted here that the IEEE Loading Guide presently gives a maximum limit of 140°C for the core hot spot temperature under normal loading conditions which we know now would allow hydrogen gassing to occur within the limits of the present Standards.
VII. CONCLUSIONS
After providing an overview of the discovery of the mechanism of hydrogen gas generation due to elevated core hot spot temperatures, measured H2 generation of a 600 MVA transformer in the field was shown to be in an excellent agreement with the corresponding calculated values. This agreement again confirmed this hydrogen gas generation phenomena, the accuracy of the core hot spot calculation method, and the developed relationship presented in this paper between hydrogen generation rate and core hot spot temperature. Predicted gas generation and accumulation from this transformer with added cooling were calculated using actual load cycle of the transformer and typical average daily ambient temperatures throughout the year in the region the transformer is located. The study demonstrated how that significant accumulation of hydrogen is not common to transformers as it requires a combination of sustained long daily periods of high loading, core over – excitation, and ambient temperatures for many months or even years. Finally, the hydrogen gas accumulation in itself does not pose any risk to the transformer; however it may mask gassing problems that
are dangerous. Industry standards are being revised to recognize this phenomena and a 130°C limit for core hot spot temperature is recommended.
VIII. REFERENCES [1] T. V. Oommen, R. A. Ronnau and R. S. Girgis, “New Mechanism Of
Moderate Hydrogen Gas Generation In Oil-Filled Transformers”, CIGRE Conference Paper 12-206, Paris Meeting, Aug-Sep 1998.
[2] T.V. Oommen, R.S. Girgis, R.A. Ronnau, “Hydrogen Generation from Some Oil-Immersed Cores of Large Power Transformers”, Minutes of Sixty-Fifth International Conference of Doble Clients, 1998, Section 8-8
[3] Ed G. teNyenhuis, Günther F. Mechler, Ramsis S. Girgis and Gang Zhou, ”Calculation of Core Hot-Spot Temperature in Power and Distribution Transformers”, IEEE Transactions on Power Delivery, Volume 17, Issue 4, Oct. 2002 Pages 991 – 995.
IX. ACKNOLWLEDGMENT
The first author acknowledges the full support Commonwealth Edison gave to the initial work of investigating this phenomenon. Both authors would like to acknowledge the support they got from Lower Colorado River Authority for sharing of loading and DGA data.
X. BIOGRAPHIES Dr. Ramsis S. Girgis (F'93) is presently Technical Manager at the ABB Power Transformer factory in St. Louis, Missouri, USA. He is also the leader of the global ABB R&D activities in the Transformer Core Performance area and Technical SM activities for electrical steel. He is also co-leader of the global ABB R&D activities in the Transformer Noise and Vibrations area. Dr. Girgis received his Ph.D. degree from the University of Saskatchewan, Canada, in Electrical Power Engineering in 1978. He has over 40 years of R&D experience in the area of power, distribution, and high frequency transformers, rotating
machines, and pulse power components. He has published and presented over 60 scientific papers in IEEE, IEE, CIGRE, and other international journals. He is presently the chairman of the IEEE Transformers Sub-committee on Performance Characteristics. He is also a contributing member of several working groups and subcommittees in the IEEE Transformers Standards Committee. He co-authored chapters in two electrical engineering handbooks on transformer design and transformer noise. He is the past Technical Advisor representing the US National Committee in the IEC Power Transformer Technical Committee (14).
Ed G. teNyenhuis (M’97) is presently Technical Manager at ABB for Transformer Remanufacturing and Engineering Services in Brampton, Ontario. Ed was born in Barrie, Canada in 1966. Ed received his B.A.Sc. degree from the University of Waterloo, Canada, in 1990 and his M.Eng. Degree from North Carolina State University, USA, in 2000, all in electrical engineering. Ed has worked in the power transformer industry for 18 years. His past experience includes positions at ABB Power Transformers in Guelph Canada, Ludvika Sweden, and at ABB Electrical Systems
Technology Institute in Raleigh, NC, USA. Ed has published several technical papers in IEEE, SMM, and 2DM pertaining to power transformers, Magnetics, and electrical steel. He is presently Chairman of the IEEE Working Group on Loss Measurement and Tolerances of power and distribution transformers.
Table 1 - Summary DGA Status Table for O2/N2 Ratio ≤ 0.2
DGA Status Parameters H2 CH4 C2H6 C2H4 C2H2 CO CO2 Actions
1 (Low gas
levels and no indication of
gassing)
Age: ALL Gas Concentrations (µL/L) ≤
Continue Routine DGA and Normal
Transformer Operation
1-10 yrs. 75 45 30 20 1 900 5000
10-30 yrs. 75 90 90 50 1 900 10000
>30 yrs. 100 110 150 90 1 900 10000
Unknown 80 90 90 50 1 900 9000
Deltas: (AND) ALL Deltas (µL/L) ≤
40 30 25 20 (=0) 250 2500
Rates (3-6 pts): (AND) ALL Rates (µL/L/year) ≤
4-9 Months 50 15 15 10 (=0) 200 100
9-24 Months 20 10 9 7 (=0) 100 1000
2 (Intermediate
gas levels and/or
possible gassing)
Age: ANY Gas Concentrations (µL/L) > Status 1 AND ≤
Increase Transformer Surveillance and DGA
Frequency
1-10 yrs. 200 100 70 100 2 1100 7000
10-30 yrs. 200 150 175 95 2 1100 14000
>30 yrs. 200 200 250 95 4 1100 14000
Unknown 200 150 175 100 2 1100 12500
Rates (3-6 pts): (AND) ALL Rates (µL/L/year) ≤
4-9 Months 50 15 15 10 (=0) 200 100
9-24 Months 20 10 9 7 (=0) 100 1000
(OR)
Deltas: ANY Deltas (µL/L) >
40 30 25 20 (=0) 250 2500
Age: (AND) ALL Gas Concentrations (µL/L) ≤
1-10 yrs. 75 45 30 20 1 900 5000
10-30 yrs. 75 90 90 50 1 900 10000
>30 yrs. 100 110 150 90 1 900 10000
Unknown 80 90 90 50 1 900 9000
Rates (3-6 pts): (AND) ALL Rates (µL/L/year) ≤
4-9 Months 50 15 15 10 (=0) 200 100
9-24 Months 20 10 9 7 (=0) 100 1000
3 (High gas
levels and/or probable
active gassing)
Age: ANY Gas Concentration (µL/L) > Perform Fault Identification and
Transformer Assessment.
Take Appropriate Action based on
Transformer Assessment Results
and Company Policy.
1-10 yrs. 200 100 70 100 2 1100 7000
10-30 yrs. 200 150 175 40 2 1100 14000
>30 yrs. 200 200 250 95 4 1100 14000
Unknown 200 150 175 100 2 1100 12500
Rates (3-6 pts): (OR) ANY Rate (µL/L/year) >
4-9 Months 50 15 15 10 0 200 100
9-24 Months 20 10 9 7 0 100 1000
Table 2 - Summary DGA Status Table for O2/N2 Ratio > 0.2
DGA Status Parameters H2 CH4 C2H6 C2H4 C2H2 CO CO2 Actions
1 (Low gas
levels and no indication of
gassing)
Age: ALL Gas Concentrations (µL/L) ≤
Continue Routine DGA and Normal
Transformer Operation
1-10 yrs. 40 20 15 25 2 500 3500
10-30 yrs. 40 20 15 60 2 500 5500
>30 yrs. 40 20 15 60 2 500 5500
Unknown 40 20 15 50 2 500 5000
Deltas: (AND) ALL Deltas (µL/L) ≤
25 10 7 20 (=0) 175 1750
Rates (3-6 pts): (AND) ALL Rates (µL/L/year) ≤
4-9 Months 25 4 3 7 (=0) 100 1000
9-24 Months 10 3 2 5 (=0) 80 800
2 (Intermediate
gas levels and/or
possible gassing)
Age: ANY Gas Concentrations (µL/L) > Status 1 AND ≤
Increase Transformer Surveillance and DGA
Frequency
1-10 yrs. 90 60 30 80 7 600 5000
10-30 yrs. 90 60 40 125 7 600 8000
>30 yrs. 90 30 40 125 7 600 8000
Unknown 90 50 40 100 7 600 7000
Rates (3-6 pts): (AND) ALL Rates (µL/L/year) ≤
4-9 Months 25 4 3 7 (=0) 100 1000
9-24 Months 10 3 2 5 (=0) 80 800
(OR)
Deltas: ANY Deltas (µL/L) >
25 10 7 20 (=0) 175 1750
Age: (AND) ALL Gas Concentrations (µL/L) <
1-10 yrs. 40 20 15 25 2 500 3500
10-30 yrs. 40 20 15 60 2 500 5500
>30 yrs. 40 20 15 60 2 500 5500
Unknown 40 20 15 50 2 500 5000
Rates (3-6 pts): (AND) ALL Rates (µL/L/year) ≤
4-9 Months 25 4 3 7 (=0) 100 1000
9-24 Months 10 3 2 5 (=0) 80 800
3 (High gas
levels and/or probable
active gassing)
Age: ANY Gas Concentrations (µL/L) > Perform Fault Identification and
Transformer Assessment.
Take Appropriate Action based on
Transformer Assessment Results
and Company Policy.
1-10 yrs. 90 60 30 80 7 600 5000
10-30 yrs. 90 60 40 125 7 600 8000
>30 yrs. 90 60 40 125 7 600 8000
Unknown 90 60 40 100 7 600 7000
Rates (3-6 pts): (OR) ANY Rates (µL/L/year) >
4-9 Months 25 4 3 7 0 100 1000
9-24 Months 10 3 2 5 0 80 800
Comments for IEEE PC57.104D5 Negative Ballot
Jim Thompson
June 26, 2018
1. There is an omission of sampling interval guidance, in Clause 6.1.3 “DGA status norms” on pages 36‐37, in Table
4. The Omission is for less than 4 months sample intervals. This omission is estimated at about 23 percent of
the IEEE data set based on the following :
a) Table 4 does not address sample intervals less than 4 months, and
b) The 95th percentile of the complete data set from Slide 42 of F12 (Attachment A), and
c) The 95th percentile gassing rates as they correlate to sample intervals from slide 63 of the F12 Presentation
(Attachment B) that show for those rates the sample intervals are less than 120 days, and
d) Calculations from values for sampling intervals for the complete IEEE data base from Presentation slide 51
of F12 (Attachment C) which shows that 23 percent of the complete F12 data set is less than 4 months.
2. I propose the following to change my negative vote on the omission of sampling interval guidance:
a) Stay with the existing guide IEEE C57.104‐2008 and provide sampling guidance for sampling intervals for
days, weeks, and months based the following correlations for individual gassing rates:
Equation (1) Sample Interval (days) as a function of ppm/year of individual gases
Eq. 1 Sample Interval (days) = (K for individual gas) / (ppm/year)
KH2 = 5320
KCH4= 3784
KC2H6= 3068
KC2H4= 2547
KCO2= 37842
3. The Table 4 values, in Clause 6.1.3 “DGA status norms” on pages 36‐37, are so low as to create false positives
based on the following:
a) Percent of ”Suspicious Units” in the F12 Presentation
The data base with 521,700 records reported in the F12 Working Group presentation by Luis Cheim and Lin
Lan at F12.p7 (Attachment D) shows that 99.4 percent of the data represents units with a “Reason for DGA”
listed as “routine” or “no reason given.” Since this data presentation gives no other reasons for sampling
then there is an inference that as much as 9.4 percent of the data representing functional transformers
might fail the 90th percentile values for the H2, CH4, C2H6, C2H4, and CO values as shown in F12.p18
(Attachments E and F). Any hypothesis regarding levels of concern for this F12 IEEE larger data set below
98th creates too many false positives. It appears that the 98th percentile values on slide 42 of F12
(Attachment A) are more appropriate for Table 4 (Attachment G) in an Annex to the guide.
b) Gassing Rate Percentiles in the F12 Presentation
On slide 42 of the F12 presentation to the Working Group (Attachment A), the 90 percentile level for
hydrogen is 45 parts per million per year. This indicates that Table 4 would result in a caution for
approximately 10% of the complete data set. Similarly, the 90 percentile levels for methane, ethylene,
ethane, and carbon monoxide are 31, 15, 35, and 169 respectively, all resulting in a caution for 10% of the
units of that complete data set.
c) Correlations of Gassing Rates to Sample Intervals in the F13 Presentation for a Subset of Data
The graph on slide 42 of the F13 presentation (Attachment H) to the Working Group shows the 95 percentile
levels for hydrogen for sample intervals. It states Absolute Values for ppm but the graph gives rates as the X
axis is the time interval correlating to the ppm of the gas. It is clear that all 95 percentile values for 6 months
or less will exceed the Table 4 value of 50 parts per million per year. On slide 14 of the F13 Presentation
(Attachment I) 25 % of the sample intervals are 6 months or less. This indicates that Table 4 would result in
a caution for 25 % of the units with hydrogen gassing rates higher than Table 4. Similarly, from the graphs
on Slides 31, 33, 54, and 49 (Attachments J, K, L, and M), it is clear that all 95th percentile values for six
months or less will exceed Table 4 values of 15, 10, 15, 200 respectively for methane, ethylene, ethane, and
carbon monoxide all resulting in a caution for 10% of the units.
4. I propose the following to change my negative vote:
a) Stay with the existing guide and add the portions of the ballot document with 98th percentile values on slide
42 of the F12 Presentation (Attachment A) for Table 4 and remove Table 3.
5. The table 3 values, in Clause 6.1.3 “DGA status norms” on pages 36‐37, represent changes of gas ppm values for
indeterminate values of sample time intervals. Values used for short time intervals give high rates and the same
values used for long time intervals give much lower rates.
6. I propose the following to change my negative vote: Remove Table 3.
Attachment A‐ F12 Working Group Page 42
Attachment B ‐ F12 Working Group Page 63
Attachment C ‐ F12 Working Group Page 51
Attachment D ‐ F12 Working Group Page 7
Attachment E ‐ F12 Working Group Page 18a
Attachment F ‐ F12 Working Group Page 18b
Attachment G ‐ Clause 6.1.3 “DGA status norms” on pages 36‐37
PC57.104.D5
Attachment H ‐ F13 Working Group Page 42
Attachment I ‐ F13 Working Group Page 14
Attachment J ‐ F13 Working Group Page 31
Attachment K ‐ F13 Working Group Page 33
Attachment L ‐ F13 Working Group Page 54
Attachment M ‐ F13 Working Group Page 49
IEEE Trans. Electr. Insul, Vol EI-13 No 5, October 1978
IEEE AND IEC CODES TO INTERPRET INCIPIENT FAULTSIN TRANSFORMERS, USING GAS IN OIL ANALYSIS
R. R. RogersC.E.G.B. Transmission Division
Guildford, England
ABSTRACT
The detection of incipient faults in oil immersedtransformers by examination of the gases dissolved in theoil developed from the original Buchholz relay applica-tion. The gases produced during the deterioration ofmineral oil and cellulose were examined and techniqueswere established to assist in the interpretation of thetype of fault, incipient or growing, occurring withinthe transformer.
The major technique now employed is the study ofratios of pairs of certain deterioration gases. The con-cepts used in the development of this technique and themodifications made to enable the technique to be estab-lished as the present method of fault interpretation,recommended by the IEC are discussed.
INTRODUCTION
Attempts to diagnose the type of fault from the gasesevolved from the oil after the failure of mineral oilimmersed power transformers started 50 years ago [1],and had developed by 1956 [2], into a detailed assess-ment of the fault from the gases collected in theBuchholz relay. It was quickly realized that if gasesare evolved from the oil sufficient to operate aBuchholz relay, then slowly developing faults would alsobe producing decomposition gases which would be dis-solved in the oil and only appear in the Buchholz at theend of a complicated system of interchange between thegases contained in bubbles rising to the surface andthe less soluble atmospheric gases dissolved in the oil.It should thus be possible by analyzing the gases dis-solved in the oil or in a nitrogen cushion to detectany incipient faults which may be present in the trans-former [3,4]. Originally, indications of the type offault occurring were developed from past history, orfrom experimental work on oil filled samples wherethe gases collected were differentiated as a major com-ponent of, or a percentage of the total gas accumu-lated.
DEVELOPMENT OF THE INTERPRETATION THEORIESFrom 1968, the C.E.G.B. started regularly monitoring
by chromatographic analysis gas in oil samples fromincreasing numbers of EHV transformers on a routinebasis, and by 1970 over one thousand units of 132,275, and 400 kV voltage rating were checked at leastannually. From statistical assessment of the results,attempts were made to set gas levels whereby 90% ofall transformers being below these levels, the
assumption could be made that transformers with gaslevels above these "Norms" were under suspicion ofcontaining incipient faults likely to cause futuretrouble in service. However, study of the analysesshowed that all transformers in service, includinglightly loaded units known to be operating satis-factorily, evolved hydrogen and the other simplehydrocarbon gases, albeit in very small quantities,e.g. less than 0.1 ppm methane per month of service.Generally the total gas content depended on the timein service and the load conditions.
In 1970, Dornenburg [5] differentiated between faultsof thermal or electrical origin by comparing pairs ofgases with approximately equal solubilities and dif-fusion coefficients such as ethylene and acetylenewhere an increase in the ratio of ethylene/acetyleneabove unity indicated an electrical fault, and alsoused the ratio of methane/hydrogen to suggest an indi-cation of a thermal fault if the ratio was greater than0.1 or a corona discharge if the ratio was less than0.1. This method was recognized as promising, in thatit eliminated the effect of oil volume and it simpli-fied the choice of unit since one is measuring one gasproduced per unit of another. In particular, the plotof successive samples forms a basis which can be usedto deduce whether the fault is constant in nature.
0018-9367/78/1000-0349$00.75 (Q 1978 IEEE
349
IEEE Trans. Electr. Insul, Vol EI-13 No 5, October 1978
HydrogenH2
MethaneCH4
EthaneC2H6
EthyleneC2 H4
AcetyleneC2 H2
-a L-
a.=
6L°1CsCL /
Fig. 1: Comparative rates of evoZution of gases fromoil as a function of deccmrposition energy.
A C.E.R.L. report produced in 1970 by Halstead butnot published unti' 1973 [6], made a theoretical thermo-dynamic assessment of the formation of the simple de-composition gaseous hydrocarbons in mineral oil whichsuggested that on the basis of equilibrium pressures atvarious temperatures, the proportion of each gaseoushydrocarbon in comparison with each of the other hydro-carbon gases varied with the temperature of the pointof decomposition. This led to the assumption that therate of evolution of any particular gaseous hydrocarbonvaried with temperature, and that at a particulartemperature there would be a maximum rate of evolutionof that gas and that each gas would attain its maxi-mum rate at a different temperature. Study of theHalstead thermodynamic equilibria suggested that withincreasing temperature, the maxima would be in turnmethane, ethane, ethylene and acetylene, and Fig. 1shows a non-qualitative representation of this hypoth-esis. Hydrogen evolution is shown first, but this isrepresented initially by a large amount of hydrogenproduced by ionic bombardment under discharge condi-tions (i.e. "cold"), followed by a continual increaseof hydrogen representing the increasing rudimentaryfractionating of the long chain paraffin moleculeunder increasing temperature conditions.
THE FIRST RATIOS ICODE 1
The above considerations led to the choice of 4 ratiosfor fault diagnosis, based on the order given inFig. 1, i.e. methane/hydrogen, ethane/methane,ethylene/ethane and acetylene/ethylene [7]. The sig-nificant change point of each ratio was assumed arbi-trarily to be unity and Table 1 illustrates the firstelementary diagnoses used. In the tabulation, avalue above 1 is indicated by 1 and a value below 1 byzero. The ratios were chosen so that a series of fourzeros indicated satisfactory operation of the trans-former.
Statistical study of nearly ten thousand C.E.G.B.gas analyses suggested that certain types of faultconditions could be differentiated within more de-tailed ranges and combinations of gas ratios. This wasconfirmed by internal examination of a number of sus-pect transformers together with post mortems on faultyunits as well as by design studies of hot spots likelyto be found in transformers under operational condi-tions. The refined code of ranges of gas ratios isshown in Table 2 and a fault diagnosis table based onthe code is given in Table 3 [8]. The use of the Codefacilitates the programming of a computer to provide afault diagnosis directly from a chromatograph detectorrecord.
INTERNATIONAL STUDY OF THE RATIOS TECHNIQUEThe various interpretation schemes employed in Europe
were summarized in a CIGRE Study Paper in 1975 [9], anda trilinear scheme was described by Duval in 1974 [14].
In order to establish the identification of actualfaults, the CIGRE WG 15-01 assessed one hundred setsof analyses from transformers with known faults, contri-buted from the records of members of the Working Group.The results of considerable laboratory work on smallspecimens were provided by several members in order toassess the probable temperatures at which the ratiosindicated significant change. In the light of theseresults and further theoretical assessments, thesignificant changeover values of the ratios for bothelectrical and thermal faults were then modified.
Because the ratio ethane/methane only indicated alimited temperature range of decomposition, but did notassist in further identifying the fault, it wasdeleted. It was considered that the use of only threeratios would simplify interpretation. To assist under-standing of the technique, the tables were re-organizedto produce a more rational progression of faults fromminor incipient faults up to the major known faultsidentified in the above survey. The modified tableswere included in IEC Document 10A (Secretariat) 53 anddetailed at the 1977 Doble Conference [10] .
It should be emphasized that the tables were intendedto demonstrate the range of phenomena expected to belikely to occur in a transformer in service, from normalageing, through incipient electrical or thermal phenomenalikely to reduce transformer life expectancy by negli-gible amounts (but nevertheless important for under-standing of the reliability of a transformer underservice conditions), to phenomena likely to causeBuchholz relay operation within a short time and thusrequiring emergency action. In order to assist thisidentification, the tables were further modified at arecent IEC-10 meeting. The new table as shown inTable 4 is included in IEC Document 10A (Central Office)37.
350
Rogers: Interpretation of Incipient Faults
TABLE 1 - ORIGINAL FAULT DIAGNOSIS TABLE
CC4C2H6 C2H4 C2i 2 Percentage ofH2 CH C H 2H Diagnosis Sfled2 4 26 24 Sampled
CHo 0 0 0 If 4 or 0.1 Partial 2.0
H2discharge otherwise - normal deterioration 34.2
1 0 0 0 Slight overheating - below 150°C (?) 11.81 1 0 0 Slight overheating - 150 - 2000C (?) 9.00 1 0 0 Slight overheating - 200 - 3000C (?) 7.80 0 1 0 General conductor overheating 11.11 0 1 0 Circulating currents and/or overheated
joints 9.00 0 0 1 Flashover without power follow-through 2.10 1 0 1 Tapchanger selector breaking current 1.10 0 1 1 Arc with power follow-through - or
persistent sparking 9.7
TABLE 2 - REFINED CODE
Gas Ratio Range Code
CH4 Not greater than 0.1 ( 0.1) 5H_ Between 0.1 and 1.0 (> 0.1,< 1) 0H2 Between 1.0 and 3.0 ( 1, < 3) 1
Not less than 3.0 ( 3) 2
C H Less than 1.0 (< 1) 0Not less than 1.0 ( 1) 1
4
C2H4 Less than 1.0 (< 1) 02H Between 1.0 and 3.0 (¢ 1,< 3) 1
C H2 6 Not less than 3.0 ( 3) 2
C 2H Less than 0.5 (< 0.5) 0
C 2 Between 0.5 and 3.0 (¢ 0.5,< 3) 12 4 Not less than 3.0 ( 3) 2
TABLE 3 - FAULT DIAGNOSIS TABLE BASED ON CODE GIVEN IN TABLE 2
a4 2H6 2H4 C2H2H ai CH CH ianosis2 4 2 6 2 4 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
0 0 0 0 Normal Deterioration5 0 0 0 Partial Discharge
1/2 0 0 0 Slight Overheating - below 150 C (?)1/2 1 0 0 Overheating - 150 - 200°C (?)0 1 0 0 Overheating - 200- 300°C (?)0 0 1 0 General Conductor Overheating1 0 1 0 Winding Circulating Currents1 0 2 0 Core and Tank Circulating Currents, overheated joints0 0 0 1 Flashover without Power Follow Through0 0 1/2 1/2 Arc with Power Follow Through0 0 2 2 Continuous Sparking to Floating Potential5 0 0 1/2 Partial Discharge with Tracking (note CO)
351
IEEE Trans, Electr. Insul, Vol EI-13 No 5, October 1978
TABLE 4 - CODE FOR EXAMINING ANALYSIS OF GAS DISSOLVED IN MINERAL OIL
Ratios of CharacteristicGases
CRH CHi CH22.4 2 4
Code of range of ratios C2H4 H2 C2H6<0.1 0 1 00.1- 1 1 0 01-3 1 2 1>3 2 2 2
Case Characteristic Fault Typical Examples
0 No fault 0 0 0 Normal ageing
1 Partial discharges of 0 1 0 Discharges in gas filledlow energy density but not cavities resulting from
significant incomplete impregnation, orsuper saturation or cavitationor high humidity
2 Partial discharges of 1 1 0 As above, but leading to trackinghigh energy density or perforation of solid
insulation
3 Discharges of low l-12 0 1-e2 Continuous sparking in oilenergy (see Note 1) between bad connections of
different potential or tofloating potential. Breakdownof oil between solid materials
4 Discharges of high 1 0 2 Discharges with power followenergy through. Arcing-breakdown of
oil between windings or coilsor between coils to earth.Selector breaking current
5 Thermal fault of 0 0 1 Insulated Conductor overheatingtemperature <1500C(see Note 2)
6 Thermal fault of low 0 2 0 )Localised overheating of thetemgerature range )core due to concentrations of150 C to 300 C )flux. Increasing hot spot(see Note 3) )temperatures; varying from
)small hot spots in core,7 Thermal fault of 0 2 1 )shorting links in core, over-
medium temperature )heating of copper due to eddyrange 3000C.7000C )currents, bad contacts/joints
)(pyrolitic carbon formation)8 Thermal fault of high 0 2 2 )up to core and tank
temperature range )circulating>7000C (see Note 4) )currents
Note 1: For the purpose of this tabZe there wiZZ be a tendency for theratio C2H2/C2H4 to rise from 0.1 to above 3 and for the ratioC2H4/C2H6 to rise from 1-3 to above 3 as the spark developsin intensity. The characteristic code of the fauZt at anincipient stage wiZZ then be 1.0.1
Note 2: In this case the gases come mainZy from the decomposition of thesoZid insuZation, this explains the value of the ratio C2H4/C2H6.
Note 3: This fauZt condition is normaZZy indicated by increasing gasconcentrations. Ratio CH4/H2 is normaZZy about 1, the actuaZvaZue above or below unity is dependent on many factors such asdesign of oiZ preservation system, actuaZ Zevel of temperatureand oiZ quality.
352
Rogers: Interpretation of InciDient Faults
Note 4: An increasing vaZue of the amount of C2H2 may indicate thehot point is higher than 1000°C.
GeneraZ remarks (1) Significant vaZues quoted for ratios should beregarded as typicaZ onZy.
(2) Transformers fitted with in-tank on-Zoad tap-changers may indicate fauZts of type 202/102depending on seepage or transmission of arcdecomposition products in the diverter switchtank into the transformer tank oiZ.
(3) Combinations of the ratios not incZuded abovemay occur in practice. Consideration is beinggiven to the interpretation of such combinations.
CELLULOSE DETERIORATIONThe effect of cellulose deterioration in assisting in
the diagnosis of the phenomena should not be ignored.The relationship between carbon monoxide and carbondioxide under normal or abnormal conditions depends onthe amount of oxygen available to assist the thermaldecomposition of cellulose [8,11,12].
Partial discharges of high energy density (Table 4case 2) are generally accompanied by tracking ofsolid insulation and the production of measurable quan-tities of CO and CO2. Similarly for insulated con-ductor overheating (case 5), the increase of theunsaturated hydrocarbons such as ethylene comparedwith the saturated ethane should also be accompaniedby increased CO and CO2.
PROCEDURE
Any indication of abnormal deterioration shouldbe followed by a procedure such as detailed in Ref. 8or IEC-IOA (Central Office) 37. Similar flow chartscould be programmed into a computer which couldexamine previous records, request further samples andrecommend necessary action tothe maintenance engineer.The IEC document emphasizes that any resulting actionmust be undertaken only after proper engineeringjudgment.
DISCUSSIONAs most of the original statistical work was done on
high aromatic content, napthenic base oil and freebreathing conservator transformers, reservations weremade on the application of the ratios technique tosealed transformers filled with inhibited paraffinicbase oil.
International experience now indicates that theratios selected apply equally to inhibited paraffinicbase oil. With nitrogen sealed transformers thechanged equilibrium conditions across the oil/nitrogenor air surface may be adjusted if necessary by therelative diffusion and solubilities but can generallybe considered of limited significance. For diaphragmsealed units the ratios could be adjusted in accordancewith the diffusion coefficients [13]. Nevertheless,as the significant values of ratios quoted in Table 4should be regarded as typical only, with a large co-
efficient of variation, any adjustment can be neglected.The indication of changing values of ratios allied toincreased rate of gas production is of true signifi-cance, regardless of the type of oil or transformer.
The data in Table 4 is based on present day knowledgeand it is likely that with further international ex-perience, there may be modifications to the significantratios and new interpretations of other ratio combin-ations.
CONCLUSIONSThe application of the ratios technique to the in-terpretation of incipient faults in power transformersby dissolved gas analysis, has proved beneficial inthat forced outage and repair costs have been consider-ably reduced.
The application of the interpretation procedurerecommended in this paper should enable the smallerutility, without the need for extended statisticaland laboratory investigation, to apply the techniqueto reduce the overall costs for maintaining powertransformers in service.
Further application of this technique as a monitorduring the factory proving tests of power trans-formers, is being developed employing very muchgreater detection sensitivities than used in the field.This shows great promise in that a number of faultslikely to cause future trouble in service have beendetected, although in each case the transformer hadsatisfactorily passed the routine tests.
ACKNOWLEDGMENTThe author wishes to thank the Directors of the
Transmission Development and Construction Division ofthe Central Electricity Generating Board for permissionto publish this paper, and acknowledges the work ofcolleagues in IEC lOA WG.02.
353
IEEE Trans. Electr. Insul, Vol EI-13 No 5, October 1978
REFERENCES
[1] M. Buchholz "The Buchholz protection system andits application in practice" E.T.Z. 49 (1928) 34pp 1257-1262.
[2] V. H. Howe, L. Massey, A.C.M. Wilson. "Identityand significance of gases collected in Buchholzprotectors." M. V. Gazette May 1956.
[3] P. S. Pugh, H. H. Wagner. "Detection of incipientfaults by gas analysis." TAIEE 80 (1961) pp 189-195.
[4] L. C. Aicher, J. P. Vora. "Gas analysis - atransformer diagnostic tool." Allis ChalmersElectrical Review 28 (1963) 1, pp 22-24.
[5] B. Fallou, F. Viale, I. Davies, R. R. Rogers,E. Dornenburg. "Application of Physico-ChemicalMethods of analysis to the study of deteriorationin the insulation of electrical apparatus."CIGRE 1970 Report 15-07.
[6] W. D. Halstead. "A thermodynamic assessment ofthe formation of gaseous hydrocarbons in faultytransformers." J. Inst. Petroleum 59
Sept. 1973) 569 pp 239-241.
[7] B. Barraclough, E. Bayley, I. Davies, K.Robinson, R. R. Rogers and C. Shanks. "CEGBExperience of the analysis of dissolved gas intransformer oil for the detection of incipientfaults." IEE Conference on Diagnostic Testingof High Voltage Power Apparatus in Service 6-8March, 1973.
[8] R. R. Rogers. "UK Experience in the Interpretationof Incipient Faults in Power Transformers byDissolved Gas-in-Oil Chromatographic Analysis."Doble Client Conference 1975 Paper 42 AIC 75.
[9] CIGRE 15-01. "Detection of and Research for theCharacteristics of an Incipient Fault fromAnalysis of Dissolved Gases in the Oil of anInsulation." Electra 42 (October 1975) pp 31-52.
[10] R. R. Rogers. As Ref. 8. A Progress Report.Doble Client Conference 1977 Paper 44 AIC 77.
[11] M. Thibault, J. Raboud. "Application of dis-solved gas analyses to the maintenance oftransformers". Rev. Gen. Elec. 84 (Feb 1975) 2,pp 81-90.
[12] R. Muller, H. Schliesing, K. Soldner. "Assessmentof working condition of transformers by gas analy-sis." Elektrizitatswirtschaft 76 (1977) 11,pp 345-349.
[13] R. Andersson, U. Roderick, V. Jaakkola,N. Ostmann, "The transfer of fault gases intransformers and its effect upon the interpre-tation of gas analysis data." CIGRE 1976Report 12-02.
[14] M. Duval. "Fault gases in oil-filled breathingEHV power transformers. The interpretation of gasanalysis data". IEEE PES Conference paper No.C74 476-8 (1974).
Manuscript was received 11 November 1977, in finaZform 1 May 1978.
This paper was presented at the IEEE Power EnginerringSociety Winter meeting, January 1978 in Nsw York, N.Y.
354