Vous êtes sur la page 1sur 7

1

Hydrogen Gas Generation Due to Moderately


Overheated Transformer Cores
Ramsis Girgis, Fellow, IEEE, and Ed G. teNyenhuis, Member, IEEE

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.

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 130C as the allowed maximum core hot
spot temperature under sustained worst conditions of load, core
excitation and ambient temperature.

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.

Index Termshydrogen 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 C 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

4500
C2H2

PPM

4000

C2H6

3500

C2H4

3000

CO2
CO

2500

CH4

2000

H2

1500
1000
500
0
1

7
8
Sample

10

11

12

13

14

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,
978-1-4244-4241-6/09/$25.00 2009 IEEE

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. 180C.

extended 60 Hz core excitation tests at 90%, 100% and 110%


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
111C, 125C and 168C 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 10C 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 200C 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 115C to 160C compared to
when hydrogen is produced by bulk oil at temperatures of
several hundred degrees.

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

H2 Generation Rate PPM/Day

10
8
6
4
2
0
100

110

120

130

140

150

160

170

180

Core Hot Spot Temperature [C]

Figure 2 Hydrogen generation vs. core hot Spot


Temperature
(a) 1 phase, 3 limb Core
Figure 3 - Location of the Hot Spot in transformer cores

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 110C and rises to levels of 4 PPM per day at 140C.
From the above, core hot spot temperatures should be given
an elevated importance. Prior to the discovery of this gas

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
2

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.
Core Hot Spot Temperature - Deg C

hottest temperature value of the core hot spot is generally


seen during full load, highest ambient temperature, and
highest core 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 2C 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.

125
120
115
110
105
100
95
90
1-Sep

H2 PPM Generation / Day

6
4
2

1-Sep

120

130

140

150

160

170

15-Nov

10-Dec

4-Jan

2.4
2.2
2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0

0
110

21-Oct

Figure 5 Calculated daily Core Hot Spot Temperatures for


the 600MVA Transformer

100

26-Sep

Month

10
H2 Generation Rate PPM/Day

130

26-Sep

21-Oct

15-Nov

10-Dec

4-Jan

Date

180

Figure 6 Calculated Daily Hydrogen Gassing Rate for the


600MVA Transformer

Core Hot Spot Temperature [C]

Accumulated Hydrogen Gas [PPM]

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

160
140

Measured
Calculated

120
100
80
60
40
20
0
1/Sep

21/Sep

11/Oct

31/Oct 20/Nov
Sample Date

10/Dec

30/Dec

19/Jan

Figure 7 Measured vs. Calculated Accumulated Hydrogen


Gas for the 600MVA transformer
3

A surprising note about this transformer was that it was


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.

C. Improved Cooling and Predicted Performance


Cooling equivalent to a 5C 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.

Stagnate hot oil

59 C

700

25 C

June/July/Aug 600 MVA Loading


600
500

400 MVA
MVA

Figure 8 Stagnate Hot Oil in the top of a transformer

400 MVA

400
300

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.

200
100
0
J

J
Month

Figure 30 Predicted load cycle for the 600MVA transformer

25
Calculated
Measured

20

H2 PPM Generation / Day

Cumulative H2 PPM Generation

Calculated Daily H2 PPM For a Year


4.0

15
10

Present Cooling

3.5

Present + Added cooling

3.0
2.5
2.0
1.5
1.0
0.5

0.0
J

0
0

24

48

72

96

120

144

168

J
Month

Figure 41 Calculated H2 gas generation throughout the year


for the 600MVA transformer

Test Time Duration [hrs]

Figure 9 - Measured and Calculated Hydrogen Gas generation


during Factory Tests on the 600MVA transformer
4

Hot Spot Temperature for a Power Transformer


The figure demonstrates that the core hot spot temperature
would be 125C 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 degassed 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.

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 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
Cumulative H 2 PPM Generation

300
Present Cooling
Present + added cooling

250
200
150
100
50
0
J

J
Month

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 110C, hydrogen generation is only
associated with transformers when operating under
simultaneous worst case of high ambient, full load, and overexcitation. 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 30C ambient.
Presented in this figure are calculated core hot spot
temperatures at three levels of load and a 100 % 115 %
range of core excitation.

VI.

Core Hot Spot Temp [deg C]

170
160
150

100% MVA

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 130C 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

80% MVA
50% MVA

140
130
120
110
100
90
95%

100%

105%

110%

115%

INDUSTRY STANDARDS AND CUSTOMER SPECIFICATIONS

120%

% of Rated Flux

Figure 63 Impact of Loading and Core excitation on Core


5

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 130C limit for core hot spot
temperature is recommended.

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
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
130C 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 110C 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 130C. 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 140C 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.

VIII. REFERENCES
[1]
[2]
[3]

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.
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
Ed G. teNyenhuis, Gnther 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).

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

Ed G. teNyenhuis (M97) 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

7
electrical steel. He is presently Chairman of the IEEE Working Group on Loss
Measurement and Tolerances of power and distribution transformers.

Vous aimerez peut-être aussi