Sei sulla pagina 1di 7

21, rue dArtois, F-75008 PARIS

A2-102 CIGRE 2014


http : //www.cigre.org


Post-mortem Analysis of Transformer Insulating Paper and its Relationship to
the Determination of Chemical Markers

J. JALBERT*, M-C. LESSARD and B. GIRARD
Hydro-Qubec
Canada




SUMMARY
Evaluation of the state of insulating paper using chemical markers still presents many challenges.
Analysis of the oil allows the presence of chemical markers to be monitored enabling the condition of
the insulating paper to be assessed without any invasive interventions. Several laboratory models
involving the chemical markers and the degree of polymerization (DP
v
) have been proposed.
Nevertheless, for an accurate interpretation of these markers in a real transformer, it is imperative to
take into account some conceptual design parameters. Indeed, the comparison of equipments with the
same design (shell vs core) and the same type of cooling is essential. It has been shown that cooling
exerts an effect on the temperature distribution in the windings, provoking a distribution of the DP
v

values of the paper in the equipment. Moreover, similarly to water, the measured marker
concentrations in oil are temperature-dependent. Indeed, because of the partition phenomenon between
oil and the solid insulation, most of these markers are kept in the solid insulation and are equilibrated
with oil depending on the temperature and other physico-chemical parameters. In order to follow the
real trend of these species during the transformers service life, it is crucial to correct their
concentrations at a specific temperature, as is done for the water content. Recent publications have
proposed that correction factors be established to take into account this partition phenomenon of
alcohols and furanic derivatives.
Among its fleet of ageing transformers, Hydro-Qubec has the opportunity to access this equipment
before it is dismantled thus obtaining the occasion to propose a post-mortem study of insulating paper
in order to obtain a better relationship between DP
v
and the presence of chemical markers. It should be
mentioned that the transformers studied must include representative values of the oil quality together
with the content of the chemical markers.
Our approach consists in sampling a large amount of paper from each of the different sections of the
windings (LV, HV and neutral) for the most accurate representation of the transformer. One of our
goals is to establish a relationship between the different chemical markers and the state of the paper
insulation. This article discusses recent experience in this field together with some specific cases.


KEYWORDS
Transformer, chemical markers, post-mortem, methanol, furanic derivatives, degree of polymerization.



* jalbert.jocelyn@ireq.ca
1

INTRODUCTION
The interpretation of chemical markers for determining the state of paper insulation in a power
transformer is a subject that still presents many challenges. Analysis of the oil allows the alcohols and
furanic derivatives to be measured and the insulating paper condition to be assessed without any
invasive interventions. To date, several accelerated-ageing studies involving chemical markers and the
degree of polymerization (DP
v
) have been proposed but their application to real equipment is still
under investigation. Recently, a consensus about marker interpretation stipulates that it is imperative
to compared apparatus with the same configuration (shell vs. core) and the same type of cooling (e.g.
ONAF, ODAF) [1], [2]. In fact, the quantity of materials implied (oil and paper) is too different
between a shell and a core design to compare these types of apparatus. Cooling also has an effect on
the temperature distribution in the windings, thus provoking a distribution of DP
v
values of the paper
in the transformer. Similarly to water, the measured marker concentrations in oil are temperature-
dependent. Because of the partition phenomenon between oil and the solid insulation, most of these
markers are kept in the solid insulation and are equilibrated with the oil, depending on the temperature
and other physico-chemical parameters. In order to follow the real trend of these species during the
transformers service life, it is crucial to correct their concentrations at a specific temperature, as is
done for the water content [3]. Recent publications have suggested that correction factors be
established to take this partition phenomenon into account in the interpretation of alcohols and furanic
derivatives [4], [5]. Moreover, Hydro-Qubecs Research Institute (IREQ) is currently involved in two
related CIGRE working groups (A2.45 and A2.D1.46). Their common goals are to establish an
investigative protocol that includes the sampling of paper from the equipment and to determine a
relationship with the different chemical markers.
Among its fleet of ageing transformers that need replacing, Hydro-Qubec has the opportunity to
access these devices as they are dismantled and perform post-mortem studies of the insulating paper of
the scrapped transformers to obtain/validate the relationship between DP
v
and the presence of
chemical markers dissolved in oil. To minimize the associated costs of this initiative, the investigation
was jointly conducted by IREQ and the companies in charge of scrapping. Our approach consists in
sampling a large amount of paper from each of the different sections of the windings at different
places (top, middle and bottom) for the most accurate representation of the transformer DP
v

distribution. Prior to this action, representative oil samples need to be collected to assess the oil quality
together with the content of chemical markers. This article discusses six specific cases with complete
chemical data; where three of these transformers have been dismantled and papers have been sampled.

TRANSFORMER SCRAPPING
Oil characterization
Six open-breathing core-type power transformers built in 1958 coming from the same manufacturer
and placed in the same generation station were investigated (see Figure 1a). These transformers were
initially cooled with OFWF systems and then modified to OFAF cooling systems around the 1990s.
They were insulated with standard Kraft papers (0.11% nitrogen content) and filled with inhibited
naphthenic insulating oil. Oil analysis was performed on the six units before they were scrapped. The
oil acidities were in the range of 0.012 to 0.017 mg KOH/g of oil. Except for the high values of the
CO
2
gas, the DGA results of Table 1 indicate the transformers had no particular problem and
therefore, showed no obvious sign of abnormal or accelerated ageing of their solid insulation.
Paper sampling
For the three units investigated here (#1, #3, #5), the paper was sampled on both the high-voltage
(HV) and the low-voltage (LV) of the three-phase windings (A, B, C). When possible, a complete part
of the disc was selected (see Figure 1b) in order to measure the distribution of the degree of
polymerization (DP
v
). Moreover, the DP
v
measurement was performed on the first paper layer directly
in contact with the oil and on the paper layer directly in contact with the copper. Finally, samples from
the top, middle and bottom of each winding phase were selected. For each case, there were more than
100 DP
v
measurements per transformer. All the measurements (DP
v
, acidity, DGA, and furanic
2

derivatives) were assessed using normalized methods [6-9]; the alcohols were assessed with a method
published elsewhere [10].

a)
b)
Figure 1: Example of a) the transformer investigated and b) a cross section of the LV winding.
Table 1: Last DGA results taken before the final outage of the six transformers investigated.
Units H
2

(ppm)
CH
4

(ppm)
CO
(ppm)
CO
2

(ppm)
C
2
H
2

(ppm)
C
2
H
4

(ppm)
C
2
H
6

(ppm)
Post-
mortem
1 86 4 322 15947 < LD 34 3 2013
2 19 5 308 6900 < LD 26 3 2014
3 28 3 220 6340 < LD 10 3 2013
4 5 5 155 3307 < LD 17 3 2014
5 22 3 264 10540 < LD 28 4 2012
6 66 7 596 21461 < LD 55 4 N.A.
LD: Limit of detection; N.A.: Not available.

RESULTS AND DISCUSSION
DP
v
measurements
This section presents all the DP
v
statistic distributions observed for case #3. These distributions are
quite similar for the other two cases investigated. For a specific case, the total amount of compiled
data may differ from one distribution to another due to some missing description in the sample
identification.
Figure 2 shows a typical paper DP
v
distribution for the three phases of the LV and HV windings. In
this figure, DP
v
distribution of all the papers collected in the LV and the HV windings without any
distinction of the phase (A, B or C), the position (top, middle or bottom) or the paper layer (oil or
copper side) is taken into account. As expected, the DP
v
of the LV winding is lower than that the one
observed on the HV winding, probably due to their more restricted cooling channels. However, in this
case, the mean values of both windings are near their end of life, with mean DP
v
values of 177 and 306
for the LV and HV windings respectively.
3


50 100 150 200 250 300 350 400 450
0
2
4
6
8
10
12
14
16
Mean=306
n=73
C
o
u
n
t
DP
v
HV
LV
Mean=176
n=27

Figure 2: DP
v
distribution for case #3.
Knowing that this transformer operated with a forced-oil cooling system giving a more
homogeneous temperature distribution along the windings, we would expect the DP
v
distribution
between the top and the bottom to be similar in order of magnitude. Indeed, Figure 3 represents the
DP
v
distributions of three locations on the windings where a small difference is still observed. The top
shows the lowest mean values with an average of 246 compared to the middle, 277, and the bottom
with an average value of 335. Moreover, the dispersion seems greater for the data collected at the top.

100 200 300 400 500
0
1
2
3
4
5
6
7
8
9
10
Mean=335
n=17
Mean=277
n=37
Mean=246
n=48
C
o
u
n
t
DP
v
Top
Middle
Bottom

Figure 3: DP
v
distribution for case #3 at different winding locations.
Finally, Figure 4 presents the distribution for the measurement of the DP
v
of the paper directly in
contact with the oil and in contact with the copper conductor. The distributions are quite similar with
equivalent mean values (291 vs 290). This can be attributed to the good quality of the oil (acidity =
0.012 mgKOH/g oil) which had no detrimental effect on the winding papers directly in contact with
the oil. Tables 2 and 3 summarize all the results and statistics related to the three cases studied.
4


100 200 300 400 500
0
2
4
6
8
10
12
14
16
18
Mean=291
n=41
Mean=290
n=46
C
o
u
n
t
DP
v
Oil side
Copper side

Figure 4: DP
v
distribution in case #3 for different paper layers.
Table 2: Summary of the DP
v
mean values for the three cases investigated.
Units Global HV LV Top Middle Bottom Oil Copper
1 240 277 162 212 244 250 237 264
3 276 306 177 246 277 335 291 290
5 294 329 187 257 301 316 292 301
Table 3: Statistics observed for the three cases studied.
Units Global n StD Minimum Maximum
1 240 116 72 118 439
3 276 100 70 120 387
5 294 129 69 118 388
StD: Standard deviation.
These results confirmed the homogeneous ageing of this transformer. It is interesting to note that,
no matter where the samples are taken, unit #1 always shows the lowest values. However, a more
statistical evaluation could be performed; for example, it could be interesting to compare each winding
separately (A, B, C) to see if there is a difference between their positions. i.e. the paper located
between phases A and B or B and C compared to the paper winding facing tank sides.
Chemical marker measurements
It has been demonstrated that MeOH is directly related to the life of the cellulose by its
proportional generation after the opening of the 1.4- glycosidic bond [11-12]. Moreover, it has been
proposed that EtOH might be generated after high-temperature cellulose degradation [13]. On the
other hand, 2-FAL comes from chemical rearrangements of the degraded by-products of the cellulose
after the bond scissions [14] and exhibits an exponential trend [15]. It is important to note that this
marker is observed when paper or board manufactured by the standard Kraft process is used.
Table 4 shows concentrations of moisture and cellulose chemical markers together with their
corrected values at 20C. These corrections were obtained using equations published in IEC 60422 [3]
and in recent publications [4], [5] for moisture and chemical markers, respectively. For the raw
concentrations data, assuming that, when sampling, all chemical species are in equilibrium between
the oil and the paper insulation, a great discrepancy is noted. For example, of all the markers, case #6
seems to be the worst with concentration values of 7151, 1893 and 3928 ppb for MeOH, EtOH and
2-FAL respectively. Nevertheless, after correcting the concentrations at 20C, the worst-case
assignation changes, depending on the marker concentration. It is important to note here that cases #2
5

and #4 seem to exhibit better paper conditions but unfortunately we have no paper samples to confirm
this behaviour. Since a post-mortem of these transformers is planned in 2014, we will be able to
confirm the correlation of the paper deterioration with the chemical data observed.
Comparing the results obtained at 20C with the global DP
v
(Table 2) for the markers investigated,
a good relationship is observed for MeOH and 2-FAL. In the case of EtOH, as observed in lab ageing
experiments, the ratio MeOH/EtOH is always higher than one, which implies that no hot spot occurred
in these transformers. MeOH < EtOH was observed in some field equipment, probably related to a
high-temperature area in this apparatus [16].
Table 4: Moisture and cellulose chemical marker concentrations.
Units Oil temperature
(C)
H
2
O
(ppm)
MeOH
(ppb)
EtOH
(ppb)
2-FAL
(ppb)
Global
DP
v

40 18 3786 1210 1377
1
20
*
8 1582 616 1108 240
55 30 2190 1132 385
2
20
*
7 502 367 267 N.D.
40 16 2697 1040 1491
3
20
*
7 1127 529 1199 271
55 21 1473 863 400
4
20
*
5 338 280 277 N.D.
60 39 5203 1034 1185
5
20
*
8 977 289 780 294
60 57 7151 1893 3928
6
20
*
12 1343 529 2587 N.A
*
Corrected values, N.D. not determined, N.A. not available.

CONCLUSION
This paper has underlined the importance of normalizing the concentrations of chemical markers
with the oil temperature. This correction ensures a more realistic and accurate correlation between the
marker concentrations and the DP
v
of paper in a transformer. The paper also emphasizes that a post-
mortem analysis in order to draw the complete DP
v
profile of a transformer is crucial for the
applicability of any model. Finally, it opens the way to establishing concentration thresholds with
which to classify the condition of the paper insulation used in transformers in order to achieve more
accurate asset management.

ACKNOWLEDGMENTS
The authors would like to thank L. Jodoin and C. Vailles from Hydro-Qubec for their support in the
post-mortem project. They would also like to thank B. Morin, S. Duchesne and M-J. Vigeant of
Hydro-Qubec for their technical assistance.

BIBLIOGRAPHY
[1] L. Cheim, D. Platts, T. Prevost and S. Xu, Furan analysis for liquid power transformers, IEEE
Electr. Insul. Mag., Vol. 28, No. 2, pp. 38-51, 2012.
[2] CIGR Working Group D1.01.TF 13, Furanic compounds for diagnosis, CIGR Brochure No.
494, April 2012.
[3] IEC 60422, Mineral insulating oils in electrical equipment-supervision and maintenance guide, 3
rd

edition, 2005.
[4] J. Jalbert and M-C. Lessard, Cellulose chemical markers in transformer oil insulation. Part 1:
Temperature correction factors, IEEE Transactions on Dielectrics and Electrical Insulation, vol.
20, Issue 6, p. 2287-2291, December 2013.
6

7

[5] M-C. Lessard and J. Jalbert, Improvement of aging markers diagnostic of transformer solid
insulation, submitted to IEEE CEIDP Conference, Shenzhen, Peoples Republic of China,
October 20-23, 2013.
[6] ASTM D 4243, Standard Test Method for Measurement of Average Viscometric Degree of
Polymerization of New and Aged Electrical Papers and Boards, Vol. 10.02, 1999.
[7] ASTM D 974, Standard Test Method for Acid and Base Number by Color-Indicator Titration,
Vol. 5.01, 2012.
[8] ASTM D 3512, method C, Standard Test Method for Analysis of Gases Dissolved in Electrical
Insulating Oil by Gas Chromatography, Vol. 10.03, 2009.
[9] ASTM D 5837, Standard Test Method for Furanic Compounds in Electrical Insulating Liquids by
High-Performance Liquid Chromatography (HPLC), Vol. 10.03, 2012.
[10] J. Jalbert, S. Duchesne, E-M. Rodriguez-Celis, P. Ttreault, P. Colin, Robust and sensitive
analysis of methanol and ethanol from cellulose degradation in mineral oils, Journal of
Chromatography A 1256:240-245, 2012.
[11] R. Gilbert, J. Jalbert, S. Duchesne, P. Ttreault, B. Morin, Y. Denos Kinetics of the production
of chain-end groups and methanol from the depolymerization of cellulose during the ageing of
paper/oil systems. Part 2: Thermally upgraded insulating paper, Cellulose 17:253-269, 2010.
[12] R. Gilbert, J. Jalbert, P. Ttreault, B. Morin, Y. Denos, Kinetics of the production of chain-end
groups and methanol from the depolymerization of cellulose during the ageing of paper/oil
systems. Part 1: Standard wood Kraft insulation, Cellulose 16:327-338, 2009.
[13] E. Rodriguez-Celis, J. Jalbert, S. Duchesne, B. Noirhomme, and M-C. Lessard, Chemical
markers use for diagnosis and life estimation of power transformers, a preliminary study of their
origins, CIGRE Canada Conference, CIGR-083, 2012.
[14] A. M. Emsley and G. C. Stevens, Kinetics and mechanisms of the low-temperature degradation
of cellulose. Cellulose 1:26-56. 1994.
[15] J. Jalbert, R. Gilbert, P. Ttreault, B. Morin, D. Lessard-Dziel, Identification of a chemical
indicator of the rupture of 1,4--glycosidic bonds of cellulose in an oil-impregnated insulating
paper system, Cellulose 14:295-309, 2007.
[16] Ryadi M., Tanguy A., Jalbert J., Rajotte C., Alcohols based ageing chemical markers for the
diagnosis of transformer cellulosic insulation, CIGRE A2 & D1 Conference, PS2-O-4, Kyoto,
2011.

Potrebbero piacerti anche