-
8/9/2019 00396010
1/6
79
THE ANALYSIS OF PAPER DEGRADATION
BY-PRODUCTS AS A TOOL FOR MONITORING
FAULT
CONDITIONS IN OIL-FILLED ELECTRIC APPARATUS
P
J
Griffin*, L R Lewand* and B Pahlavanpour**
Doble Engineering Company
The National Grid Company plc
Cellulosic materials are
used
for electrical insulation
and mechanical support in apparatus such as
transformers and reactors. In order to evaluate the
condition of the insulation system of this type of
equipme nt it is necessary to determ ine if the
cellulosic materials are being degraded excessively.
Several experiments were performed to determine the
content of indicator by-products generated from the
degradation of cellulosic materials under electrical,
thermal, and oxidative stress. Several variables were
explored to determine their influence on by-product
generation and subsequent degradation. Case
histories from service-aged insulation provided
evidence to corroborate laboratory results.
Electrical discharges generate gaseous by-products
and may be identified from
gas
in oil sampling.
They do not result in the accumulation of furanic
compounds. Howe ver thermal and oxidative
degradation of cellulosic materials do generate furanic
compounds but these may then deteriorate under
some conditions; again furanic compounds may be
identified from an oil sample. Where there is access
to the winding degree of polymerization (DP) data
has been successfully employed in evaluating the
condition of cellulosic materials and for serving as
the standard to evaluate other methods of estimating
degradation.
DP results, when com bined with the analysis of gases
in oil, fbranic compounds in oil, and a complete
history of a unit, provides a good assessment of the
condition of a transformer's cellulosic insulation.
INTRODUCTlON
Cellulosic materials are used for electrical insulation
and mechanical support in apparatus such as
transform ers and reactors. In order to evaluate the
condition of the insulation system of this type of
equipmen t it is necessary t o determine t o what degree
the cellulosic materials are being degraded. Of
particular interest is the detection of oil-soluble by-
products which can be easily sampled and which
serve as indicators of the degradation of cellulosic
materials.
Analysis for dissolved gases in oil has been found to
be a useful technique for detecting incipient fault
conditions, but it does not have the desired specificity
for evaluating the cond ition of the cellulosic materials.
More recent techniques employing high performance
liquid chromatography (HPLC ) have been used to
detect furanic compounds which are formed
specifically from the degradation of the cellulosic
materials. The test for degree of polymeri zation (DP)
of the paper is used to provide a direct indication of
the average molecule size for cellulosic materials
which decreases with ageing from about 1000 for
processed, new in-service insulation to less than
200
for very aged materials.
This paper reports on the investigation of the nature
and quantities o f degradation by-products of cellulosic
materials formed under conditions of electrical stress
and thermal/ oxidative ageing. The roles of such
factors
as
temperature, water content, by-product
stability, and the presence of dicyandiamide (thermal-
upgrading agent) are explored.
In addition, two examples are given from testing
of
service-aged insulation (oil and paper) to determin e
the condition of cellulosic materials.
LABORATORY EXPERIMENTS
Electrical Stress
A test apparatus previously described' was built to
subject electrical insulating paper to partial discharges.
The test apparatus was set up with ten layers
of
dried
(
-
8/9/2019 00396010
2/6
80
or
no acetylene generated, th e concentration of carbon
dioxide and carbon monoxide is again low, and their
ratio can be less than 3.1 as shown by Baker'.
The results in Tables 1 and 2 show that, when the
paper is damaged, greater quantities of carbon
monoxide and probably carbon dioxide are generated.
However, the additional diagnostic criterion can be
seen in the ratio of carbon dioxide to carbon
monoxide which approaches 1:
1,
or when the carbon
monoxide value exceeds that of carbon dioxide.
Tests for furanic compounds were made immediately
after the test was terminated and again after storing
at 75C for up to
168
hours to allow for diffusion
from paper to the oil.
No
significant concentration of
furanic compou nds was detected in the oil. Visual
examin ation of the paper revealed clear evidence of
PD activ ity in the form o f burning, treein g and
punctures.
Data from failed transformers sometimes have
revealed a large increase in carbon monoxide and
sometim es in carbon dioxide. In cases where the
carbon dioxide increased much more than the carbon
monoxide, the generation
of
2-furfural was greatest.
A reasonable explanation for these results is that
when there is some overheating of the cellulosic
insulation in addition to the damage from the PD and
arcing, then significant quantities of carbon dioxide
and 2-furfural are generated with smaller quantities
of
carbon monoxide . When the damage is due only to
PD and arcing , then the sole reliable indicator appears
to be the ratio of the carbon oxide gases approaching
unity. Significa nt dama ge is required to generate
detectable quantities of any of the indicator
compounds.
Thermal and O xida t ive Age ing
Expe rim ent one. Samples of air saturated oil and
wet
or
dry Kraft paper in a 100 to
1
mass ratio were
aged in sealed ampules at 120 and 140 1C in
sealed glass containers in accordance with a Cigre
Task Force 15 01 03 protocol4. No copper catalyst
was employed.
The rate of by-product formation varied for each
compound. However, there were some similar trends.
The alcohol is apparently the most susceptible of the
comp ounds to degradation, at least during the ageing
of Kraft paper. Figure I which shows these results,
is denoted as accumulation as the total
concentration is the amount generated minus that
which is degraded. Evidence that furfural is degraded
is seen in that the 14OoC results are ab out equal t o or
lower than those at 120C, contrary to expected
values if degradation of the alcohol did not occur.
The trend for changes in concentration of 5
hydoxymethyl-2-furfural (HM F) and 2-furfural were
similar. The results for 2-furfural in Figure 2 show a
significant increase in concentration with increasing
temperature. However, values for the wet paper are
generally lower than those for ry paper at the same
temperature. Again this would sugges t that the water
which accelerates paper degradation is also increasing
the rate of HMF degradation rather than retarding its
rate of generation.
Acetyl furan was not generated in detectable quantities
for the duration o f the test at 120C. At 14OoC
detectable quantities were found and, in general, the
test results for wet and ry insulation were similar
after O days.
Experiments repeated with nitrogen-purged oil showed
a slower rate of 2-furfural generation probably due to
slower paper degradation and larger quantities of
furfural which is oxidatively unstable.
Exper iment
two.
There have been several indicators
which suggest that furanic compounds may not be
entirely stable. Further, evidence has been presented
which demonstrates that the amounts an d ratios of by-
products of ageing differ from Kraft and TU-Kraft
papers,6,7,8.The thermal upgrading compound used in
these cellulosic materials was dicyandiamide,
commonly referred to as di-cy . Questio ns then arise
as to what effects the di-cy has on the generation or
degradation of the by-products from cellulosic
materials and what happens in mixed insulation
systems, because TU insulation is generally not used
alone.
Another set of laboratory accelerated-ageing
experiments were designed to evaluate:
1. Susceptibility of furanic compounds to
thermal degradation.
2. Stability of furanic comp ounds in the
presence of the cellulose thermal-upgrading
agent dicyandiamide.
3 By-product accumulation during ageing of
mixed (TU-Kraft and Kraft) insulation
systems.
The apparatus to perform these tests and the protocol
were given earlier'. All the tests were performed at
120C with a copper catalyst, in sealed tubes.
Stability
of
By-products
Previous data from Unsworth and M itchell' indicated
that, at lower temperatures of 20 and 80C. the
-
8/9/2019 00396010
3/6
81
furanic compounds were quite stable in oil under
their test conditions. At O T , degradation effects
were evident.
Results are shown in Figure 3 from an experiment in
which furanic compounds were spiked in oil and then
aged at 120C in the absence of paper, in sealed
tubes with a copper catalyst. The rate of degradation
of furanic compounds in oil under these conditions is
quite significant and for the most part appears to be
a logarithmic decay as seen for 2-furfural. Samples
were ru in duplicate. The presence o f di-cy spiked
in some of the ageing tubes also appears to accelerate
the degradation process or effectively removes these
compounds from being detected by reacting with
them. This may partially explain why the amount of
2-furfural detected during the agein g of TU insulation
is less than that for Kraft paper. The di-cy also
appears to be altered during the ageing process as it
was not detected at any concentration even close to
the spiked quantity of 10 pg/mL, and eventually was
not detected.
Paper Ageing
in
Mixed Systems
Figure provides the results of paper ageing
experiments comparing Kraft, TU-Kraft, and mixed
insulation. The amount of 2-furfural detected was
highest for the Kraft, and lowest for the TU-Kraft as
expected. The mixed insulation system, typical of
many
US
transformers, yielded concentrations which
were between the two but closer to the TU-Kraft
results, suggesting that the di-cy may
be
reducing the
ageing of all the insulation and/or increasing the rate
of degradation of the 2-furfural generated.
IN-SERVICE INSULATION
Two examples of studies of in-service insulation are
presented for transformers from w hich oil and paper
samples were obtained for by-product and DP
measurem ents res pectively. In one case the 2-furfural
results appear to be a good indicator of the
degradation of cellulosic materials (in contrast to the
other).
Case 1 Table 3 . This case involved two banks of
General Electric of Canada rectifier transformers,
which had been in service for 10-13 years, which
showed accelerated ageing of the oil and cellulosic
insulation. Sludged oil and rust were observed in
several units.
The
oil service life was about
5-10
years. The very accelerated age ing occurred, in part,
because of air leakage into the units from poor seals
around the bus bars and from high loading.
Considering the short service time, the DP of the
insulation for some units was quite low. All of the
paper samples were acquired from the same location
from very similar transformers. One additional
sample, taken from unit #14 during the internal
investigation, came from a different location, the core
ground cable, because it appeared very degraded. The
DP of this sample was exceptionally low compared to
other samples from the LV winding . Sample s of oil
acquired with the paper were tested for furanic
compounds. The results given in Figure 5 show good
correlation between DP and 2-furfural. This could be
a useful tool for this family of transformers in
detecting degradation of cellulosic materials due to
incipient faults or general accelerated ageing. Note
that, for unit
14,
the 2-furfural content is higher than
for the other units with similar DPs for the top LV
winding samples. This may be due to an incipient
fault problem in this unit as indicated by the DP o f the
core ground cable insulation. The dissolved gas-in-oil
data was not relevant
as
it was apparent that much of
the gases were escaping through the leaky
containment.
Case
2
Table 4 . This
1956
Westinghouse
115
kV,
44 MVA transformer was inspected internally to check
the condition of the windings . The inspection
revealed globs of shredded paper and some other
materials.
To
determine if this shredding of insulation
was due to advanced ageing of cellulosic materials,
samples of paper and wood were taken from the high
voltage winding along with samples of the debris.
The results showed that there is significant remaining
life in the insulating materials with the one exception
of the wood wedge. If a DP of 400 is about half-life
to a DP of 200 as has been postulated , then it would
appear that the winding insulation has many years of
performance available. Some of the particles with the
shredded paper may have been glue which did not
retain its adhesive properties, thereby permitting some
paper to come loose. This was then forced through
pumps, etc, causing its disintegration. The 2-furfural
content was quite low and was not a good indicator of
the ageing of the cellulosic materials. This is not
unusual as many middle-aged US transform ers have
low concentrations of 2-furfural.
Samples Tested for Furanic Compounds and
Dissolved Gases in Oil
As part of our routine, dissolved gas-in-oil test
program for one
US
utility client, additional tests were
performed for furanic compounds in oil for samples
from transformers greater than or equal to 10 MVA
and 69 kV. The results agree with our previous
findings' fo r US transformer s which show that most
oil samples contained less than
100
ng/mL of 2-
furfural and much lower concentrations of the other
furanic compounds.
-
8/9/2019 00396010
4/6
82
CONCLUSION Degradation of Electrical Insulating Paper
Monitored with High Performance Liauid
In conclusio n, to detect cases where PD activity Chromatography , IEEE Transac tions on
involves cellulosic materials,
analysis of dissolved Electrical Insulation, Vol. 25, No 4. p 737-46.
gases in oil are most useful. For evaluation of
accelerated ageing of cellulosic materials from
IO .
Tutorial on Electrical-Grade Insulating Papers
thermal and oxidativ e stresses, analysis of dissolved in Power Transformers, Presented at the
gases, and furanic compounds in oil are the best
indicators.
To
confirm the true ageing of the
transformer, when there is access to the winding,
insulation tests such as DP should be performed on
the paper.
REFERENCES
Doble Planning Conference, Oct 1993.
1
2
3.
4.
5
6
7.
8.
9.
Gr i f f in , P . J . , Lewand , L .R . and
Pahlavan pour, B., 1994, Paper Degradation
By-Products Generated under Incipient-Fault
Conditions , Minute s of the 61st Annual Intl
Conf of D oble Clients, Sec. 10-5.1.
Baker, A.E., 1983 , Gassinp Characteristics
of Transformer Oils under Sustained Arcs ,
Minute s of the 50t h Annual Intl conf of
Doble Clients, Sec 10-801.
Baker, A.E., Gas Composition in Corona
Discharge , Minu tes of the 49th Annual Intl
Conf of Doble Clients, Sec.
10-701.
Minutes of CIGRE WG 15.01
Task
Force 03
Meeting
-
June 1993.
Griffin, P.J., Lewand, L.R. and Finnan,
E.,
1993, Measurement o f Cellulosic Insulation
Degradation Compounds in Oil , Minutes of
the 60th Annual Intl Conf of Doble Clients,
Sec. 10-3.1.
Oommen, T.V., Petrie, E.M. and Reckleff,
J.G., 1993, Furanic Com pounds Analysis by
GC-MS, and its Diagnostic Value for
Transfor mer Insulation Ageing . Minute s of
the 60th Annual Intl Conf of Doble Clients,
Sec 10-5.1.
Grant, D.H., 1992, A study of Furanic
Compounds Generated in Transformers
During Heat Runs , Minutes of the 59th
Doble Client Conf, Sec 10-5.1.
Azizian, H. and Massey, R.E., 1989,
Analysis of Trace By-products from
Overheated Paper Insulation in Power
Transformers , Canadian Elec Assoc,
Montreal, Quebec, Report 262T509.
Unsworth, 1 and Mitchell, F., 1990,
-
8/9/2019 00396010
5/6
83
- l a D c , o c l r a l a D c . m - l 4 c . D R Y * l 4 C . m
Figure 1: Furfural Accumulation
1
d
Figure 2: 2-Furfural Accumulation
- Figure 3: 2-Furfural Degradation
Figure 4 2-Furfural Accumulation
Figure
5:
2-Furfural Content
Versus
DP
-
8/9/2019 00396010
6/6
T A B L E
-
Dissolved Gases in Oil*
F r o m P D i n O i l P a p e r S ys te m
Hydrogen 6,356
Oxygen 25,120
Nitrogen 55,960
Carbon Monoxide 357
Carbon Dioxide 502
Methane 1,749
Acetylene 3,956
Ethylene 627
Ethane 814
ppm
vol/vol
at 25C and 760 Hg
TABLE 2 - Composition of Gas Space*
Hydrogen
Oxygen
Nitrogen
Carbon Monoxide
Carbon Dioxide
Methane
Acetylene
Ethylene
Ethane
59,176
199,587
774,277
1,450
1,079
2,366
2,745
33
1
288
*ppm vol/vol at 25C and 760mm Hg
TABLE
3
- D P of Service Aged Insulation
Sample ID
TR3 Coil Top LV
TR4 Coil Top LV
TR5 Coil Top LV
TR6 Coil Top LV
TRI
I
Coil Top LV
TR12 Coil Top LV
TR13 Coil Top LV
TR14 Coil Top LV
TR14 Core Gmd. Cable
TR16 Coil Top LV
DFV
230
297
384
52
740
580
685
405
156
846
TABLE
4
- Test of Service - Aged Insulation
Sample ID
I
D m
Top of Winding H , (green paper)
Top of W inding H I (brown paper)
Top of Wi nding H I (black paper)
Wedge Between H,-H, (wood)
Top of Wedge Between H,-H, (paper)
Top of Winding H, (paper)
No Load Tap Switch H I (debris)
Northeast Comer H, (debris)
Ledge South Side H I (debris)
South Side Ledge H, (debris)
H,
Around No Load Tap Switch (debris)
Shelf Near Tertiary Bushing (debris)
2-Furfural 5-Methy l 2-furfural
15
746
606
659
130
85 1
637
476
346
444
564
394
476
All other furanic compounds
< I
Furanic compounds: ng/mL at room temperature.