Crawfish Creek Paper

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

COMPARISON OF TWO ON-LINE MONITORING SYSTEMS TO

LABORATORY DISSOLVED GAS ANALYSIS

Larry W. Weathington
Georgia Power Company

ABSTRACT

The transformer under study is a 230-46 kV General Electric unit which began gassing in 2001. Serveron
TM8 and Kelman Transfix DGA on-line monitors have been installed and operational since September
2005. Georgia Power performed DGA in our laboratory to compare with the two monitors. This paper
will provide the data retrieved from the monitors in comparison to the laboratory analysis. Also
addressed will be information concerning the installation and maintenance issues encountered for each
on-line monitor.

CRAWFISH CREEK 230/46 BANK

FIGURE 1.

1
BACKGROUND

This transformer is a 30/40/50 MVA OA/FA/FOA, 3-phase, 230/46 kV with an LRT-200-2 load tap
changer. The unit has an active nitrogen blanket and was built by General Electric at Rome, GA. in 1979.
The LTC failed in 1985. GE rebuilt the LTC and the transformer was returned to service. In 1986 the
LTC failed again due to water ingress. The DGA history shows gassing began in 2002 after replacing GE
type “U” bushings in November of 2001.

This transformer was selected for the evaluation of the Kelman Transfix and the Serveron TM8 for
several reasons. The geographic location of Crawfish Creek is in a remote area of northwest Georgia in
the foot hills of the Appalachian Mountains. At best, it’s a three hour drive from Atlanta requiring a route
through Chattanooga, Tennessee.

Additionally, this transformer contained all the gases generated by a transformer with a problem. The rate
of gas generation was sufficiently slow as to allow for the completion of the comparison before the
transformer would have to be removed from service.

Samples collected from each monitor were analyzed by the Georgia Power laboratory beginning on
September 29, 2005 and continuing through November 14, 2005. The on-line monitors were configured
for sample collection once every four hours providing six samples in each twenty-four hour period.

From January 2001 to April 2004 a significant increase in acetylene (C2H2) with a lesser increase in
ethylene (C2H4) and ethane (C2H6) was noted. In April 2004 a hot spot was discovered on the H1
bushing top terminal. A mobile unit was installed at that time and the transformer was removed from
service to investigate the heating of the bushing top terminals and to perform an internal inspection.

It was discovered that when the bushings were replaced in 2001, the nuts on the draw leads had not been
installed on the draw leads. This allowed the heating of the bushing top terminals. An internal inspection
at this time did not reveal any reason for the generation of gases. The transformer was degassed and
placed back in service. The transformer resumed gas generation after returning to service. A significant
increase in gas generation rate began in September of 2005.

8 YEARS OF DGA HISTORY

350
300
250
200
PPM

150
100
50
0
1/ 199

4/ 200
1/

3/ 00
11 200
6/ 200
3/ / 20
4/ 200
1/ / 20
2/ 200

6/ 00
10 200
10 / 20
10 200
10 20
10 1 /2
11 /20
12 200
4/

8/
28

28 0

10 2
17 03
20 4
25 04
16 5

22 5
/7 2

/6 5
/7 05
/1 5
/2 05
/3 00
/9 05
/6 5
2 8

2 5
/

/
/

/
/

/
5/

1 5
/
/2
00
5

C2H4 C2H6 C2H2 CH4 CO H2 T.C.G.

8 YEARS OF DGA HISTORY 1998 to 2006


FIGURE 2.

2
DYNAMIC BEHAVIOR OF GASES

To better understand gas trending over long periods of time the dynamic behavior of the various gases
needs to be considered, especially since this is a nitrogen-blanketed transformer. When looking at DGA
results over a long period, all of these factors have to be taken into account.
• All of the gases tend to move back and forth between the oil and the gas-space chasing
equilibrium or the same relative saturation. Equilibrium depends upon each specific gas, the gas-
space volume, pressures, temperature and operation of the nitrogen regulator as it exhausts gases
at high pressure and admits nitrogen at low pressure.
• Acetylene, ethylene, ethane and carbon-dioxide are relatively oil soluble. Concentration in the
gas space is similar to the concentration dissolved in the oil at equilibrium. These gases tend to
accumulate during a continuing problem, and disappear slowly between intermittent problems.
• Hydrogen is very insoluble. At equilibrium, the concentration in the gas space is over 20 times
the concentration in the oil. This means there can be more hydrogen contained in the gas-space
than there is in the total oil volume. Hydrogen tends to escape in a variety of ways and does not
accumulate, particularly in nitrogen-blanketed transformers.
• Carbon monoxide, while more soluble than hydrogen, is still relatively insoluble and also tends to
escape rather than accumulate.

.
Gassing was first noted in this unit in March 2002. The time period that we will be looking at in the life
of this transformer is the DGA history from April 2004 through November 2005. In April 2004, after
installing nuts on the draw leads where they had been omitted and performing an internal inspection, the
transformer was returned to service. It continued to gas with acetylene and hydrogen as the primary gases
generated. In Figure 3 you can see the acetylene (C2H2) and hydrogen (H2) levels increasing.

GAS EVALUATION SINCE JULY 2004

350

300

250

200
PPM

150

100

50

0
4/
5/ 00
6/ 00
7/ 00
8/ 0
9/ 00
10 00
11 /20
12 /20
1/ /20
2/ 00
3/ 00
4/ 0
5/ 00
6/ 0
7/ 200
8/ 00
9/ 0
10 00
11 /20
12 /20
1/ /20
1/
1/ 4
1/ 4
1/ 4
1/ 4
1/ 4

1/ 04
1/ 5
1/ 5
1/ 5
1/ 5
1/ 5
1/ 5
1/ 5
1/ 5

1/ 05
/1 4
/1 04
/1 04

/1 5
/1 05
/1 05
2
2
2
20
2
2

2
2
20
2
20

2
20
2

20
06

C2H4 C2H6 C2H2 CH4 CO H2 T.C.G.

DGA FROM MARCH 2001 TO JANUARY 2006

FIGURE 3.

3
SITE PREPARATION FOR THE SERVERON AND KELMAN MONITORS

Site preparation is the essentially same for both monitors and can be completed by utility personnel. Good
site preparation enables a smooth and quick installation and commission process. Below is an excerpt of
the site preparation checklist of actions:
1. A location for mounting the monitor has been identified.
2. The transformer oil supply and return ports have been selected.
3. Fixtures for bleed valve are available.
4. Oil supply and return port fittings determined and provided.
5. Power is present and available at the installation site for the monitor.
6. Communication protocol (RS232, RS485, Ethernet, or Modem) has been selected.
7. A cylinder of chromatography grade helium has been purchased and is on location for the
Serveron Monitor.
8. All shipped items and configurable accessories have been relocated to site.
9. Electrical conduit and enclosures have been installed (if required).

THE INSTALLATION PROCESS AT CRAWFISH CREEK

• The installation of both units was similar. In each the oil is circulated from the transformer, to the
monitor, and then returned to the transformer through ¼-inch O.D. stainless steel tubing.
Stainless steel tubing was used in conjunction with compression fittings to minimize the risk of
leakage. The oil circulation paths for the monitors are shown below (Figure 4 for the Serveron
TM8 and Figure 5 for the Kelman TRANSFIX).
• Installation was performed by Georgia Power personnel in conjunction with Serveron and
Kelman Technical Services.
• Performed jointly by Georgia Power, Serveron and Kelman personnel, the commissioning
verified that all equipment, communications and calibration was performing optimally. Some
training time was spent on software installation, calibration and the interpretation of results with
Georgia Power personnel.
• At Crawfish Creek, the oil was pulled from the bottom valve and returned to the top valve for
both monitors. Each monitor had its own sample valve. A 2-inch valve was installed for the
Serveron for its sample. The Kelman used the existing 2-inch drain valve for its sample. A
sample valve was installed on each monitor to allow sample collection for the laboratory
analysis.
• Gas connections for the Serveron consisted of connections for the Helium carrier gas and the
NIST traceable calibration bottle.
• Cable connections for the Serveron and Kelman were jointly installed and configured. A
junction box was installed in the front due to space restrictions closer to the transformer.
• A split core CT was installed on the CT circuit to obtain Load Guide data for both monitors.
• Alarm levels were configured in the software for both monitors.
• Complete installation, commissioning, and training at Crawfish Creek took 1½ to 2 days per
monitor.

4
SERVERON TM8

FIGURE 4.

KELMAN TRANSFIX

FIGURE 5.

5
SERVERON TM8 VALVE CONNECTION WITH MOISTURE SENSOR

FIGURE 6.

KELMAN TRANSFIX VALVE CONNECTION

FIGURE 7.

6
BLEEDER FIXTURE
A critical factor for both monitors during the installation process is the ability for each monitor to be a
closed loop, or sealed system, for oil flow and gas extraction. This was accomplished by using a bleeder
fixture to remove all air from the oil lines. The bleeder fixture at Crawfish Creek was installed at the top
valve.

OIL RETURNS FROM BOTH MONITORS THROUGH TOP FILL VALVE

FIGURE 8

AFTER INSTALLATION WAS COMPLETED

FIGURE 9

7
DESCRIPTION OF EQUIPMENT USED FOR THE COMPARISON

TRANSFIX

The Transfix unit is designed for continuous on-line monitoring of oil filled equipment.

Transfix uses Photoacoustic Spectroscopy to measure eight diagnostic gases (H2, CO2, CO, C2H6, C2H4,
CH4, C2H2 and O2) as well as dissolved moisture (H2O). An optional load sensor is also provided to trend
gas data to transformer load. PAS technology has proven to provide accurate and reliable results. It
should be noted, however, that because this instrument is sensitive to acoustics, it must not be attached to
the actual wall of the transformer.

Transfix takes a representative oil sample from the transformer into the gas extraction sub-system. Here
the dissolved gases are removed from the oil sample using a variation of the headspace method. The
extracted gas is then analyzed using a PAS measurement module. When this is complete the oil sample is
returned to the transformer, containing no bubbles or excess oxygen. By using this method of sampling
the Transfix does not have any averaging of the results over time, thereby ensuring each result is a
discrete and independent DGA sample and the time-response of a rise in gas values is optimized.

Transfix requires no consumable carrier or calibration gases and needs no regular calibration. Sampling
intervals are user settable from a maximum of once per hour up to once every 24 hours. The Transfix can
also be programmed to automatically to increase the sampling rate when a ‘caution’ or ‘warning’
condition is detected.

Transfix is supplied with Transcom, a comprehensive software tool for data storage and trending. This
provides a high degree of flexibility in a graphical interface to control how the data is viewed and
analyzed by the user. The software also provides operational control of the Transfix unit and its settings.

Transcom provides alarm options through six independent ‘pages’ which allows elaborate monitoring of
the transformer fault activity. This gives the user six individual sets of alarms based on concentration,
rates of change and gas ratios which are all user definable. Password protection along with alarm history
is also incorporated.

Communication with Transfix is possible using either local comms (USB) or remote comms over a
variety of options such as PSTM/Cell modem (SMS alerts available), internet access via LAN, Bluetooth,
Modbus or DNP3. There are also three programmable contact relays. Transfix has built in intelligence to
diagnose any problems in the system and alert the user.

SERVERON TM8

The Serveron measurement system has two components; gas extraction and gas measurement. The two
components are tightly integrated in order to properly extract, move and measure fault gases. The entire
measurement system is designed and tested to be leak-tight to avoid contamination by atmospheric gases
and not introduce any additional gas components that may alter the gas-in–oil content of the sample being
analyzed.

To extract a sample, gas first is allowed to equilibrate from the oil to a gas space across a semi-permeable
membrane ensuring no altering of oil temperature or gas pressure. The extracted gas is then transferred to
the gas measurement system, a laboratory grade gas chromatograph. The gas measurement system uses

8
chromatographic columns to separate the gases present in the sample into individual gas peaks that are
then applied to a thermal conductivity detector (TCD).

The TCD measures all eight fault gases from a single 0.5ml injection of sample gas. Helium is used as
carrier gas to move the sample through the detector to avoid contaminating the O2, CO and CO2
components in the gas sample. The TCD measurements are then converted to gas-in-oil results using
partitioning coefficients and the pressure and temperature readings measured at the point of extraction.
The system automatically checks its calibration against a NIST-traceable gas standard on a periodic basis
as part of its ongoing operation.

The monitor is configured to perform a sample DGA analysis once every four hours during normal
operating conditions. A complete GC analysis takes approximately 20 minutes. If any gas generates an
alarm, the TM8 will automatically begin sampling the oil on an hourly basis, until the alarm is cleared.

Once an analysis has been completed, the results can be viewed using the TM View or the SMS View
software either locally, or remotely from any internet connection. This allows the user to track the gas
concentration levels over time and monitor the gas levels against user defined caution and alarm values.

Additionally, the system included Ambient Temperature and load guide sensing devices. The load guide
sensor provides a relative measure of transformer load. These external sensors can be correlated with
DGA gas information to allow a complete diagnostic overview of the transformer’s operation. The Duval
Triangle and the Rogers Ratio graphics are automatically populated with real-time gas data for acetylene
(C2H2), ethylene (C2H4) and methane (CH4).

GEORGIA POWER LABORATORY (TOGA) SYSTEM.

Georgia Power uses a PerkinElmer-Arnel Transformer Oil Gas Analysis (TOGA) system to perform
laboratory dissolved gas analysis. This instrument was designed following ASTM method D 3612-C,
which is the headspace sampling technique for dissolved gas analysis by gas chromatography. This
technique consists of bringing an oil sample in contact with a gas phase (headspace) in a closed vessel
purged with argon. As a result, a concentration of dissolved oil is transferred to the headspace. At
equilibrium, the relationship between the remaining concentration of gas in the oil and its concentration in
the headspace may be deduced by mass equivalence.

In this technique, 20mL glass vials are crimp sealed, and purged with a continuous flow of argon for 3
minutes. A 10mL sample of oil is then introduced into the vial. The vial is placed in a headspace
autosampler where it is heated to 70°C and shaken for a period of 30 minutes.

The vial is then pressurized with argon, and an aliquot is transferred into the chromatograph sample loop
from where it is injected into the chromatograph. The sample is injected onto a carboxen column, and
then analyzed by a thermal couple detector at 40°C for 4 minutes to detect H2, O2 and N2. The oven
temperature then ramps up to 170°C from 4 to 11 minutes and passes onto a molecular sieve column for
analysis of CO2, C2H2, C2H4, and C2H6. The sample is then valved onto a catalytic converter where the
temperature is ramped to 250°C for the analysis of CH4 and CO by a flame ionization detector.

The headspace preparation time and the sample analysis require approximately one hour per sample. It
is possible to have one sample run the chromatograph while the next sample is going through the
headspace preparation thus reducing the time required when running multiple samples.

9
Raw gas data for combustible concentration is converted to parts per million (ppm) using calibration
standards prepared by Doble Engineering. Atmosphere is used to calibrate for oxygen and nitrogen.

MVA LOAD GRAPH

The split CT from the Kelman and Serveron was attached to the center phase of the low side bushing for
the load guide. The MVA load from September 29, 2005 through December 1, 2005 was calculated from
the Kelman split CT.

During this comparison the load on the bank from September until November was running 12 MVA. In
November the load increased to 18 MVA and increased again to 20 MVA in December as ambient
temperatures decreased. Summer load will normally be around 30 MVA.

There was a correlation between the load on the transformer to the amount of gas being generated.

LOAD FROM 9/29/2005 TO 12/1/2005

50
45
40
35
30
MVA

25
20
15
10
5
0
9/

10

10

10

10

11

11

11

11

12
29

/6

/1

/2

/2

/3

/1

/1

/2

/1
/2

/2

3/

0/

7/

/2

0/

7/

4/

/2
00

00

20

20

20

00

20

20

20

00
5

05

05

05

05

05

05

MVA LOAD FROM 9/29/2005 TO 12/1/2005

FIGURE 10

10
HYDROGEN RESULTS

The average Serveron Monitor results for hydrogen were 14 ppm (or 13%) less than its laboratory sample.
The Kelman Monitor average results were 41 ppm (or 39%) less than its laboratory sample. There was a
42 ppm (or 40%) difference in the average results between the two monitors for hydrogen.

HYDROGEN

250

200

150
(H2) PPM

100

50

0
5

5
05

05

05

05
00

00

00
20

20

20

20
/2

/2

/2
3/

0/

7/

0/
29

/6

/3
/1

/2

/2

/1
10

11
9/

10

10

10

11
Serveron Monitor (H2) Serveron Lab (H2)
Kelman Monitor (H2) Kelman Lab (H2)

FIGURE 11.

METHANE RESULTS

The average Serveron Monitor result for methane was 8 ppm (or 34%) less than its laboratory sample.
The Kelman Monitor average result was 8 ppm (or 18%) less than its laboratory sample. There was 6 ppm
(or 28%) difference between the two results monitors for methane. There was a 4 ppm (or 14%) different
in the two laboratory samples.

METHANE

50
45
40
35
30
(CH4) PPM

25
20
15
10
5
0
5

5
05

05

05

05
00

00

00
20

20

20

20
/2

/2

/2
3/

0/

7/

0/
29

/6

/3
/1

/2

/2

/1
10

11
9/

10

10

10

11

Serveron Monitor (CH4) Severon Lab (CH4)


Kelma n Monitor (CH4) Kelman La b (CH4)

FIGURE 12.

11
ETHYLENE RESULTS

The average Serveron Monitor result for ethylene was 5 ppm (or 18%) less than the laboratory sample.
The Kelman Monitor average results were 1 ppm (or 4%) less than the laboratory sample. There was a 3
ppm (or 12%) difference between the two monitors.

ETHYLENE

50
45
40
35
(C2H4) PPM

30
25
20
15
10
5
0
5

5
05

05

05

05
00

00

00
20

20

20

20
/2

/2

/2
3/

0/

7/

0/
29

/6

/3
/1

/2

/2

/1
10

11
9/

10

10

10

11
Serveron Monitor (C2H4) Serveron Lab (C2H4)
Kelman Monitor (C2H4) Kelman Lab (C2H4)

FIGURE 13.

ETHANE RESULTS

The average Serveron Monitor result for ethane was 5 ppm (or 52%) less than its laboratory sample. The
Kelman Monitor average results were 2 ppm (or 20%) less than its laboratory sample. There was 5 ppm
(or 40%) difference between the two monitors. The Serveron has a 5 ppm minimum detection limit for
ethane. The average concentration was 4.7 ppm. This is probably the reason its data is scattered.

ETHANE

30

25

20
(C2H6) PPM

15

10

0
5

5
05

05

05

05
00

00

00
20

20

20

20
/2

/2

/2
3/

0/

7/

0/
29

/6

/3
/1

/2

/2

/1
10

11
9/

10

10

10

11

Se rve ron Monitor (C2H6) Serveron Lab (C2H6)


Ke lman Monitor (C2H6) Kelman Lab (C2H6)

FIGURE 14.
ACETYLENE RESULTS

12
The Serveron Monitor average results were 8 ppm (or 4%) less than its laboratory sample. The Kelman
Monitor average results were 21 ppm (or 10%) less than its laboratory sample. There was a 13 ppm (or
8%) difference between the two monitors results.

ACETYLENE

200
180
160
140
(C2H2) PPM

120
100
80
60
40
20
0
5

5
05

05

05

05
00

00

00
20

20

20

20
/2

/2

/2
3/

0/

7/

0/
29

/6

/3
/1

/2

/2

/1
10

11
9/

10

10

10

11
Serveron Monitor (C2H2) Serveron Lab (C2H2)
Kelman Monitor (C2H2) Kelman Lab (C2H2)

FIGURE 15

CARBON DIOXIDE RESULTS

The Serveron Monitor average results were 19 ppm (or 4%) less than the laboratory sample. The Kelman
Monitor average results were 44 ppm (or 10%) less than the laboratory sample. There was an average 42
ppm (or 8%) difference between the two monitors.

CARBON DIOXIDE

700

600

500
(CO2) P P M

400

300

200

100

0
5

05

05
05

05

05

05
00

0
20

20
/2

/2

/2

/2

/2
/

/
29

/6

/3
3

0
/1

/2

/2

/1
10

11
9/

10

10

10

11

Serveron Lab (CO2) Serveron Monitor (CO2)


Kelman Monitor (CO2) Kelman Lab (CO2)

FIGURE 16

13
CARBON MONOXIDE RESULTS

The Serveron Monitor average results were 7 ppm (or 37%) less than the laboratory sample. The Kelman
Monitor average results were 9 ppm (or 42%) less than the laboratory sample. There was an average 2
ppm (or 10%) difference between the two monitors.

CARBON MONOXIDE

40

35

30

25
(CO) PPM

20

15

10

0
5

5
05

05

05

05
00

00

00
20

20

20

20
/2

/2

/2
3/

0/

7/

0/
29

/6

/3
/1

/2

/2

/1
10

11
9/

10

10

10

11
Serveron Monitor (CO) Serveron Lab (CO)
Kelman Monitor (CO) Kelman Lab (CO)

FIGURE 17

TOTAL COMBUSTIBLE RESULTS


The Serveron Monitor average results were 25 ppm (or 9%) less than its laboratory sample. The Kelman
Monitor averaged results were 19 ppm (or 20%) less than its laboratory sample. There was a 30 ppm (or
12%) difference between the two monitors results.

TOTAL COMBUSTIBLE GAS

300

250

200
TCG PPM

150

2
100

50

0
5

5
05

05

05

05
00

00

00
20

20

20

20
/2

/2

/2
3/

0/

7/

0/
29

/6

/3
/1

/2

/2

/1
10

11
9/

10

10

10

11

Serveron Monitor TCG Severon Lab TCG


Kelman Monitor TCG Kelman Lab TCG

FIGURE 18

14
MOISTURE

The Kelman average was 11 ppm. That was 4 ppm above the Karl Fisher analysis in the laboratory. The
Serveron averaged was 3 ppm, which was 4 ppm below the laboratory results.

MOISTURE PPM

20
18
16
14
12
PPM

10
8
6
4
2
0
5

5
05

05

05

05
00

00

00
20

20

20

20
/2

/2

/2
3/

0/

7/

0/
29

/6

/3
/1

/2

/2

/1
10

11
9/

10

10

10

11
Serveron Moisture (PPM) Kelman Lab Moisture (PPM) Kelman Moisture (PPM)

FIGURE 19

PERCENT SATURATION

The laboratory and Serveron both averaged 3.9% saturation. The Serveron used the bottom oil
temperature for its calculation and the laboratory uses the top oil temperature for its calculation.

PERCENT SATURATIONS

10
9
8
7
6
(%RS)

5
4
3
2
1
0
5

5
05

05

05

05
00

00

00
20

20

20

20
/2

/2

/2
3/

0/

7/

0/
29

/6

/3
/1

/2

/2

/1
10

11
9/

10

10

10

11

Serveron (%RS) Lab (%RS)

FIGURE 20

15
TRANSFIX GRAPH

GAS ANALYSIS FOR 4-MONTHS WITH GASES AND MOISTURE.

FIGURE 21

TRANSFIX LOG

LOG WITH GASES, WATER, LOAD GUIDE, TOP OIL TEMP

FIGURE 22

16
TRANSFIX GRAPH

TRANSFORMER LOAD GUIDE, TOP OIL, AMBIENT TEMPERATURE

FIGURE 23

SERVERON GRAPH

GRAPH WITH GASES, MOISTURE, % RS AND LOAD GUIDE.

17
FIGURE 24

SERVERON LOG

LOG WITH GASES, TOP OIL TEMPERATURE AND LOAD GUIDE.

FIGURE 25

18
DUVAL TRIANGLE FOR CRAWFISH CREEK (as provided by TM8 software)

FIGURE 27

FIGURE 26

DIAGNOSTIC FROM DUVAL TRIANGLE

All of the points on the Duval Triangle, representing relative amounts of acetylene, ethylene
and methane are clustered in the D1 zone indicating discharges of low energy. This points to
a “spitting or sparking” problem. Although commonly thought of as partial-discharges (PD)
these are complete discharges of low-energy without power follow-through. These are
commonly caused by a poorly grounded metal part such as a bolt, static-shield, flux shield,
tap-changer shaft or core/coil clamping member.

INTERNAL INSPECTION

On January 19, 2006 the transformer was taken out of service for power factor and FRA testing. At this
time an internal inspection was performed. The core and coils were in good condition, but one of eight
wedges between the frame and the tank wall showed signs of heating. This wedge was found to be loose
as well. Upon removing the wedge, there appeared to have been arcing between the wedge and frame.
The wedge was removed, cleaned and retightened. The other wedges appeared to be normal.

19
WEDGE APPEARED TO BE HEATING

FIGURE 27

WEDGE APPEARED TO BE HEATING

FIGURE 28

20
While it’s not unusual to see where the frame has made this kind of indentation, what was unusual was
black, carboned looking oil on the wedge. Also unusual were the pitted marks on the wedge that appeared
to be the results of spitting or sparking.

PITTED MARKS ON THE WEDGE

FIGURE 29

CONCLUSION

Both monitors agreed very well with the laboratory results. The comparison of the plotted data shows
each monitor and the laboratory fluctuating in the same manner.

On-line DGA is effective in showing gas generated problems that periodic sampling can miss. Periodic
sampling is only a snap-shot in time. Many things could happen that would lead to failure which a
periodic sample can miss.

A good laboratory is one that can give reliable and concise results that can be used to determine the
condition of the equipment. This also applies to On-line DGA monitors. Both of these monitors meet
this definition.

REFERENCES

1. Abel J. Pereira, “On-Line DGA for 500 KV Autotransformer Correlates Gassing Severity With
Load”, Proceedings of the Annual International Conference of Doble Clients, Boston, MA, 2004

2. Michael P. Anderson, John V. Hinshaw, “Comparison of Laboratory Dissolved Gas Analysis


Versus an On-Line Monitoring System”, Proceedings of the Annual International Conference of
Doble Clients, Boston, MA, 2004

21
ACKNOWLEDGEMENT

The author would like to acknowledge the following for their assistance, without which this paper would
not have been possible: Rick Cantrell, Vice President of Southeast Sales for Serveron Corporation, and
Roland Weir, Customer Support Manager of Kelman.

BIOGRAPHY

Larry Weathington has been with Georgia Power for thirty-eight years. He started in Substation
Maintenance in 1969. In 1975 he took a position in Transmission Support as a Doble Technician.
In 1976 he became involved in oil quality assessment where he has been instrumental in setting up the
DGA and oil quality testing program for Georgia Power.
He is currently the team leader for the Georgia Power Laboratory and field technicians for diagnostic
testing.
lwweathi@sounthernco.com

22
23
24
25
26

You might also like