Sunteți pe pagina 1din 5

The true meaning of CO & CO2 in power transformers

Learned through on-line DGA monitoring

Stan Lindgren
and
Jeff Golarz
Serveron Corporation

1.0 SUMMARY

Thermal faults involving paper are generally considered more serious than those in
insulating oil alone. A common perception is that such problems can be diagnosed
through periodic measurement of dissolved CO & CO2 however experiences with
multiple-gas on-line DGA increasingly show this to be problematic, for several reasons.
An overriding one is the little-known fact that CO2 and CO are both absorbed by the
paper insulation during decreasing temperature (ppm-in-oil decreases) and return to the
oil (ppm-in-oil increases) during increasing temperature. This causes highly-variable and
mystifying results in periodic DGA and correct interpretation is nearly impossible. This
typically-slow behavior is very apparent with on-line DGA and dramatic examples are
presented in the paper, consistent with published results for laboratory experiments in
Japan. Additional aspects of CO & CO2 dynamic behavior are presented along with their
diagnostic implications. Aging of paper produces CO2 and CO, particularly in heavily-
loaded transformers (e.g. GSU) however highly-soluble CO2 accumulates over the years
whereas low-solubility CO tends to escape. CO2/CO ratio is highly dependent upon how
tightly the transformer is sealed. Source of CO for a serious localized problem (e.g.
defective connection, over-insulated and/or inadequately-cooled joint) is typically limited
in size. Once the insulation involved burns generation of CO is greatly reduced (while
copious quantities of CH4 & C2H4 are generated from unlimited surrounding oil). Also,
CO gets diluted throughout the total oil volume so the ppm-increase tends to be small,
and possibly temporary since CO tends to escape unless transformer is very-tightly
sealed.

2.0 CO & CO2 DIAGNOSTIC LIMITATIONS

Source of CO for a serious localized problem (e.g. defective connection, over-


insulated and/or inadequately-cooled joint) is typically limited in size. Once the
insulation involved burns generation of CO is greatly reduced (while copious
quantities of CH4 & C2H4 are generated from unlimited surrounding oil). Also,
CO gets diluted throughout the total oil volume so the ppm-increase tends to be
small, and possibly temporary since CO tends to escape unless transformer is
very-tightly sealed.
As widely recognized, both CO and CO2 result from normal insulation-aging,
particularly in heavily-loaded transformers (e.g. GSU units). However, CO2 is
highly soluble and tends to accumulate over the years whereas, here again, low-
solubility CO tends to escape. CO2/CO ratio is highly dependent upon how tightly
the transformer is sealed ranging from 10 to 30 for transformers with sealed-
conservators (bladders) and can be >100 for a nitrogen-blanketed shell-form unit.
True, accelerated CO2 results from inadequate cooling (e.g. pump running
backwards or coolers plugged with cottonwood tree fluff) whereas severe thermal
problems accelerate CO. A CO2/CO ratio <3 can mean serious insulation
degradation. However, with high accumulated levels of CO2, any sudden increase
in either gas can be hard to detect either as ppm or CO2/CO ratio (although ppm
rate-of-change or incremental CO2/CO ratio can be meaningful provided
temperature effects described below are considered.) And, in some cases new
transformers show CO2/CO <3 from day-one in service (for unknown reasons).

3.0 CO & CO2 DYNAMIC BEHAVIOR

3.1 Absorption of CO & CO2 by insulation


Adding further confusion to above diagnostic concepts is the little-known fact that CO2
and CO are absorbed & desorbed by pressboard barriers & conductor paper causing huge
temperature effects. CO2 and CO are absorbed (ppm-in-oil decreases) with decreasing
temperature and return to the oil (ppm-in-oil increases) with increasing temperature - up
to about 80C. (Above 80C insulation no longer holds CO or CO2.) This is usually a
somewhat slow process but in general the lower the temperature goes the more these
gases get absorbed. And, if temperature stays low sufficiently long, most of the CO &
CO2 can disappear from the oil, particularly for shell-form transformers that typically
have a very low oil/pressboard ratio (2 to 3).

3.2 Diffusion of CO & CO2 from oil within insulation


At constant temperature, oil impregnating the insulation contains CO & CO2 (and other
gases) at same ppm as the main oil in the transformer. However, CO solubility-in-oil
increases slightly with temperature whereas CO2 solubility decreases substantially with
increasing oil temperature. Temperatures inevitably vary with daily & seasonal ambient
temperature, load and cooling operation, so gases slowly diffuse in and out of the oil
impregnating the insulation chasing equilibrium with the main oil.

3.3 Partitioning of CO & CO2 with a nitrogen blanket


Gases move back and forth between the surface of the oil and a headspace chasing
equilibrium (same relative-saturation). As mentioned above, solubility of CO in oil is low
- for equilibrium at 30C ppm-in-gas is about 8 x ppm-in-oil. So, a typical nitrogen-
blanket with 10% of the oil-volume can hold nearly half the total CO in the transformer.
This enhances the chances of CO escaping through pressure-regulator venting, gaskets,
leaks etc. As mentioned above, CO solubility increases only slightly with oil
temperature. Solubility of CO2 on the other hand is high for equilibrium at 30C ppm-in-
gas is nearly the same as ppm-in-oil. So, a 10% nitrogen-blanket holds less than 1/10th of
the total CO2 in the transformer. However, CO2 solubility decreases substantially with
temperature so that at 80C ppm-in-gas is nearly 1 x ppm-in-oil. Also, volume of the
nitrogen-blanket changes as oil expands & contracts. All of these factors contribute to the
variability of CO & CO2 ppm-in-oil during daily (or seasonal) load and ambient
temperature cycling.

2
3.4 Example Transformer A - Old 465 MVA 22/500 kV single-phase GSU transformer
(shell-form, nitrogen-blanketed) - Dramatic variation of CO & CO2 ppm-in-oil with
temperature.

Figure 1 Note variation in CO2 ppm (turquoise, linear-scale) with load & ambient
temperature (both green, sensor-value) for 12 months after shutdown in winter
(refueling). Oil temperature (blue) is measured at the gas analyzer therefore lower than
inside the transformer.

Figure 2 Same transformer before & after shutdown (next refueling) plus degassing of
the oil [1]. Note CO2/CO (summer) = 5000/30 ppm before & 4630/23 ppm after (log
scale). Very little CO2 was removed by degassing the oil. It was retained within the
pressboard and impregnating-oil, returning to the oil once the transformer reached
summer temperatures. (Estimated oil/paper-pressboard ratio = 1.9 [2]). The CO2/CO ratio
>100 is probably due to CO escaping over the years through the nitrogen-blanket.
Meanwhile, CO2 accumulated.

3
3.6 Comparison with published experimental results in Japan [3]

Figure 3 Transformer As peak CO2 values, August 2004 & August 2005 (5000 ppm @
estimated 80 C oil temperature) and lowest CO2 value, after refueling shutdown 2/22/04
(450 ppm @ 0 C ambient/oil temperature) shows a similar slope when plotted on Kan &
Miyamoto Fig 4 (per experimental results with oil/paper-pressboard ratio = 3) [3].

3.7 Example Transformer B - Refurbished/uprated core-form 185/228 MVA 24/500 kV


single-phase generator step-up (GSU) with sealed-conservator [1] - Overall variation of
CO & CO2 ppm-in-oil with temperature A less dramatic example
ZF2 - Gas in Oil

1000.00 180.00

160.00

140.00

100.00
120.00

100% LoadGuide Hydrogen (H2)


100.00
Oxygen (O2)

80.00 Methane (CH4)


10.00 Carbon Monoxide (CO)
60.00 Carbon Dioxide (CO2)
Ethane (C2H6)
40.00 Ethylene (C2H4)
Acetylene (C2H2)
20.00 TDCG
1.00
0% LoadGuide LoadGuide
0.00
Ambient Temp (C)
Moisture (%RS)
-20.00
Moisture (PPM)
Oil Temp (C)
0.10 -40.00
Jul `04 Aug Sep Oct Nov Dec Jan `05

Figure 4 CO2, CO, CH4 & C2H6 gases during 6-month period after unit back in service at
full load following refurbishment & oil processing.

The on-line gas analyzer was added during refurbishment however lab results before
shutdown were: CO2/CO 6900/647 ppm = 10.7 [4]. Six months (full load) after

4
reprocessing CO2/CO 710/64 ppm = 11.1. Apparently, high levels of CO2 and CO had
been produced due to insulation aging over the years in this GSU transformer however
the sealed conservator was very effective in preventing escape of CO. Also, absorption of
CO2 and CO by the insulation in this transformer (core-form) during summer shutdown
was much less than in Transformer A (shell-form).

5.0 CONCLUSI0NS

1. CO ppm by itself is not a reliable indicator of localized paper-insulation damage


because: a) level is usually reduced by dilution in a large quantity of oil, b) ppm
level is affected by oil-temperature (mainly absorption & de-sorption by paper
insulation) caused by load and/or ambient-temperature changes and c) its
tendency to escape depending upon type of oil expansion system and how tightly
transformer is sealed. (Relative CH4, C2H4 & C2H2 ppm from unlimited
surrounding oil is a much more reliable indicator for all types of thermal faults
including those that involve localized cellulose.)

2. CO2/CO ratios due to normal cellulose aging are highly dependent upon how
tightly sealed the transformer varying from 10 to >100 (and, though not included
here, some new transformers show ratios 3.0 & lower from initial installation for
unknown reasons.) With a high quantity of CO2, seeing a significant change in
CO2/CO ratio is close to impossible. However, incremental CO2/CO may be
helpful in identifying early paper degradation (provided potential temperature
effects are considered) as suggested in a 2004 Cigre paper. [5]

3. Additional factors affecting dynamic behavior of CO & CO2 are: a) diffusion in


and out of the oil impregnating the insulation chasing equilibrium and b)
movement back and forth between the oil surface and a nitrogen blanket chasing
equilibrium.

REFERENCES
[1] S. Lindgren, PJ de Klerk, Avoiding Failures Through On-Line DGA Lessons
Learned New Ballgame Phase II, PS2-18, Cigre SC A2 & D1 Colloquium,
October 7-12, 2007, Brugge, Belgium.
[2] Private conversation, M. Franchek, Weidmann Electrtical Technology Inc., St.
Johnsbury, Vermont.
[3] Hisao Kan, Teruo Miyamoto, Proposals for an Improvement in Transformer
Diagnosis Using Dissolved Gas Analysis (DGA), IEEE Electrical Insulation
Magazine, November/December 1995 Vol. 11 No. 6.
[4] L. Heath, GSU Transformer Upgrades, Proceedings of the 2005 International
Conference of Doble Clients, Boston, MA.
[5] S. R. Lindgren, Transformer Condition Assessment Experiences Using Automated
On-Line Dissolved gas Analysis, paper A2-202, Cigre 2004 Session, 29th August
3rd September, 2004, Paris, France.

S-ar putea să vă placă și