Furfural Concentration in Transformer Oil

as an Indicator of Paper Ageing, Part 1: A Review

C.H. Zhang a and J.M.K. MacAlpine b
a

School of Electrical Engineering, Harbin Institute of Technology, China

b
Industry Center, Hong Kong Polytechnic University, Hong Kong

ABSTRACT The concentration of the chemical furfural

in transformer oil is well known to be an indicator of the
extent to which the paper insulation of the windings has
deteriorated. However, technical papers on the
relationship between the loss of mechanical strength of the
paper insulation in transformers and the increase in the
concentration of furfural in the oil are to be found
scattered over a variety of journals. The intention of this
communication is to bring together the main findings of
the laboratory investigations, and the rather fewer reports
of field experience, for the convenience of power engineers.
The linear relationship between the tensile strength and
the degree of polymerisation (DP) of paper insulation has
been demonstrated by several workers under laboratory
conditions; and so has the negative linear relationship
between the logarithm of the furfural concentration and
the DP. The concentration of furfural in transformer oil is
seen as a paper-specific indicator of degradation and a
method for measuring it using high performance liquid
chromatography has now been standardised (IEC 1198).
The presence of water increases the degradation rate of the
cellulose chains and so, because water is also a product of
the degradation process, this is an auto-catalytic (or
positive feedback) process. This makes the overall
process very much more complex. In addition, the
temperature varies across the transformer itself. However
a significant change in the FC over a period of months
would indicate the probable development of a fault. The
value of regular measurements of furfural concentrations
in transformers seems clear.
Index Terms Transformer, Furfural concentration,
Deterioration, Insulation paper..

I. INTRODUCTION
The windings of oil-filled transformers use paper
insulation and, if this becomes weakened, the
consequence of a fault current may be severe: the
inevitably high forces on the windings may result in
mechanical failure of the insulation and consequent
inter-turn short-circuits. Paper deteriorates slowly but
irreversibly with a rate, which is increased by high
temperatures and the presence of water. A major byproduct of this reaction is 2-furfuraldehyde or furfural
whose presence can be used as an indicator of the
progress of the deterioration. As an indicator, furfural

is paper-specific and thus complements the less specific

Dissolved Gas Analysis (IEC 567).
Technical papers on laboratory investigations into the
relationship between the loss of mechanical strength of
the paper insulation in transformers and the increase in
the concentration of furfural in the oil are to be found
scattered over a variety of journals; however, relatively
little has been reported regarding power companies
field experience of furfural measurements as indicators
of the extent of paper deterioration in actual
transformers. The intention of this communication is to
bring together the main findings of the laboratory
investigations, and the few reports of field experience,
for the convenience of power engineers. A companion
paper reports on the results of some tests on the furfural
content of transmission transformers belonging to an
electric power utility in the PR of China [1].
II. THE DEGREE OF POLYMERISATION OF
PAPER AND ITS TENSILE STRENGTH
Paper is made from vegetable cellulose fibres
dispersed in water and drained to form a mat. The
cellulose fibres are prepared from wood and form a
natural long-chain polymer. The monomer unit is shown
in Figure 1, and itself comprises three polymerised
glucose molecules.
The cellulose degrades by several routes but all
involve the breaking of the long cellulose molecules into
shorter lengths, usually at the oxygen link between the
glucose units. The strength of the paper is a function of
the average length of the cellulose molecules. This
average length is defined formally as the degree of
polymerisation (DP) which is the average number of
repeat units per chain and is measured indirectly by
determining the viscosity of a cellulose solution (ASTM,
D4243, IEC 450).

Figure 1. The monomer unit of cellulose

a

Corresponding author: zch852@gmail.com

142440178X/06/$20.002006IEEE

1088

PSCE2006

The classic work on the ageing of paper insulation in

power transformers was written by Shroff and Stannett
[2] and most of the work done since may be regarded as
a continuation or expansion of their work. They
demonstrated the linear relationship between the tensile
strength and the DP of paper, as shown in Figure 2:

Of the furans, the one occurring in the greatest

quantity and thus normally measured as the specific
indicator for paper degradation, is furfural. Spectrophotometric analysis of the reaction products between
furfural and aniline acetate has been suggested as a
faster means of determining the furfural concentration
(FC) in oil samples [11], but still requires the samples to
be sent off to a specialist test laboratory. Infra-red
spectroscopy has also been suggested as a faster method
and one which could possibly be developed into an online method of FC determination [12,13].
Under laboratory conditions a negative linear
relationship between the logarithm of the furfural
concentration and the DP was observed by Shroff and
Stannett [2]; and between the logarithm of the total
furan concentration and the DP by Hill et al. [3]. The
partitioning of furfural between oil and paper has been
found to be about 20:80 by Allan [14] and 15:85 by Kan
et al. [15] up to 100oC.
IV. THE PRESENCE OF WATER
Although the water content of oil and paper may be as
low as 0.5% on commissioning, it may rise to as much
as 2% during the transformers lifetime, even under
normal operating conditions, because, as mentioned
above, it is one of the by-products of the chemical
degradation of paper.

Figure 2. From Shroff and Stannetts paper [2]

which is reproduced from their paper. Other workers,

e.g. [3-5], have subsequently reported the same linear
relationship for various insulating papers.
Hill et al [6] demonstrated that the rate of bond
breaking was constant for a given temperature (at least
for the month or so of their tests) and varied, as for
many chemical reactions, with the (absolute)
temperature according to the Arrhenius equation:
k = A . exp(-B/T)
where A and B are constants. Figure 4 shows the effect
of temperature on the degradation rate (bond
scissions/time), which is plotted using data in [6].
III. FURFURAL MEASUREMENTS IN THE
LABORATORY
When cellulose degrades by the breaking of its long
molecules, the fragments produced are molecules of
water, oxides of carbon and a variety of carbohydrates,
mostly short-lived, which result from the opening of the
hexagonal glucose ring (Figure 1). Carbon oxides have
been considered for use as indicators of paper
degradation but experience worldwide has been mixed,
and interest is now concentrated on the furans [7 PW]
whose structures are shown in Figure 5. The
concentration of individual furans in transformer oil can
be separately measured by high performance liquid
chromatography (HPLC) [8,9]. The method is now
standardized (IEC 1198), but requires time-consuming
pre-extraction of the furans. A direct-injection HPLC
process has been proposed [10] which allows a fivefold
reduction in technician time.

Figure 3. From Shroff and Stannetts paper [2]

The presence of water was shown by Shroff and

Stannett [2] to increase the degradation rate. In Figure 3,
again reproduced from their paper, four sets of results
(for a water content of 0, 1, 2 and 4% in the insulating
paper) are shown. The rate at which the logarithm of DP
decreases with time d(log(DP))/dt appears to be
roughly proportional to the water content: compare the
gradients of lines (c), (d), (e) and (f) in Figure 4. If, as

1089

Degradation rate, chains/g/s

this implies, the presence of water increases the chainbreaking reaction rate, then, because water is also a
product of the degradation process, it is an auto-catalytic
(or positive feedback) reaction.
Indirect support for this interpretation was given by
Hill et al [6] in respect of their modelling of furfural
build-up in transformer oil and its under-estimate of the
actual rates.
If you have difficulties with the titles on your figures,
you can always elect to add in the titles as separate text
boxes, rather than importing the titles with the graph.
This is sometimes helpful in getting a lengthy verticallyoriented title to display correctly.
25
20
15
10

response to economic cycles (and to the lag-time of the

utility in responding to the economic cycle). In addition,
the temperature varies across the transformer itself.
General overheating between 150 and 200 oC, and
localised overheating between 500 and 900oC, have
been reported within operational transformers [12].
To sum up: the degradation of cellulose is the process
of breaking the long molecules and proceeds at slow but
steady rate in the absence of water. The reaction itself
produces water, which increases the rate; but water is
lost from the system at an unknown rate. Furthermore
the rate increases rapidly (an exponential function) with
temperature; the temperature of the transformer varies
with time in a complicated manner; and the temperature
varies across the transformer (due to the interactions of
hot spots, the insulation and the cooling system), so
the DP will also vary across the transformer.
Clearly no simple equation relating furan
concentration to time for an individual transformer is
possible.

5
0
100

VI. INDUSTRIAL OBSERVATIONS

120

140

160

180

Temperature, degrees C

Figure 4. The variation with temperature of the degradation

rate of insulation paper

Figure 5. The furan family from Hill et al [3]

V. PRACTICAL CONSIDERATIONS
The auto-catalytic effect mentioned in the previous
section leads to a far more complex process equation,
particularly as a transformer is not a closed system:
water can evaporate from the oil, particularly at high
temperatures. The load and hence oil temperature varies
with load on a daily and seasonal basis as well as in

1090

Relatively little has been published relating to real

industrial cases, probably because very few utilities
have made extensive HPLC measurements of FC over a
period of time and correlated them with failures; and
utilities and manufacturers naturally prefer not to expose
their performance data to public scrutiny, particularly
where it relates to equipment failures.
Murthys results for a faulty transformer [16] show
that as already suggested above the DP can vary
significantly across a transformer, due to local faults.
His results show good linearity between the breaking
strength and the DP of paper samples from the bluephase high-voltage lead, the red-phase high-voltage lead
and the tertiary coil. However the samples strengths
were in the ratio 5:4:3 respectively with the last of these
proving very weak indeed but the measured furfural
content was only 0.4 mg/l.
Allan reported on 40 aged transformers in
Queensland, Australia, and derived an insulation age
from the three key ageing indicators in the oil (CO,
CO2 and 2-furfural) [17]. He assumes a steady fall in
DP with respect to time so that, assuming a DP of 175 is
the effective end of a transformers life, a estimate can
be made of the remaining life in terms of a cumulative
failure probability as a function of time. This of course
neglects the possible changes to the average loadings of
individual transformers in the future.
Domun refers to 8 years of FC monitoring on a
particular 132-kV transformer with a persistent-thermal
fault [18]. The oil was changed when the FC level
reached 10 mg/l which took about 3.8 years, 3 years and
1.8 years on successive occasions, indicating a
deteriorating situation (These periods are estimated from
his graph). Another experiment was reported involving

of Dielectric Materials, Seoul, Korea, 25-30 May, 1997, pp.

256-258.
[5] M-C. Lessard, L. van Nifterik, M. Masse, J.F. Penneau and R.
Grob, Thermal ageing study of insulating papers used in
power transformers, Proc. Conf. on Electrical Insulation and
Dielectric Phenomena, San Francisco, USA, 20-23 Oct. 1996,
pp. 854-859.
[6] D.J.T. Hill, T.T. Le, M. Darveniza and T. Saha, A study of
the degradation of cellulosic insulation materials in a power
transformer. Part 1: Molecular weight study of cellulose
insulation paper, Polymer Degradation and Stability, vol.
48, pp. 79-87, 1995.
[7] B. Pahlavanpour and A. Wilson, Analysis of transformer oil
for transformer condition monitoring, IEE Colloquium
(Digest), vol. 603, pp. 1/1-1/5, 1997.
[8] P.J. Burton, J. Graham, A.C. Hall, J.A. Laver and A.J.
Oliver, Recent developments by CEGB to improve the
prediction and monitoring of transformer performance,
CIGRE Paper 12-09, 1984.
[9] J. Unsworth and F. Mitchell, Degradation of electrical
insulating paper monitored with high performance liquid
chromatography, IEEE Trans. EI, vol. 25, pp. 737-746,
1990.
[10] M-C. Lessard, C. Lamarre, A. Gendron, and M. Masse,
Determination of paper degradation by-products by direct
injection on an HPLC column, Proc. Conf. on Electrical
Insulation and Dielectric Phenomena, San Francisco, USA,
22-25 Oct. 1995, pp. 230-234.
[11] B. Pahlavanpour and G. Duffy, Development of a rapid
spectrophotometry method for analysis of furfuraldehyde in
transformer oil as an indication of paper ageing, Proc. Conf.
on Electrical Insulation and Dielectric Phenomena, Pocona
Manor, USA, pp. 493-498, 1993.
[12] R. Blue, D. Uttamchandani and O. Farish, Infrared
detection of transformer insulation degradation due to
accelerated ageing, IEEE Trans. DEI, vol. 5, pp. 165-168,
1998.

a 90-day high-load period for another 132 kV

transformer in which the temperature averaged 80oC and
the FC rose from 2 mg/l to 27 mg/l, at which point the
loading was removed. The DP was found to be under
200. He also indicated that a survey indicated norms of
2mg/l furfural over a very large number of transformers.
VII. CONCLUSION
Summarising, a considerable number of papers has
been published on the degradation of transformer paper
and the means for monitoring it, but most describe
small-scale laboratory tests involving accelerated or
short-term ageing experiments and relatively few refer
to long-term tests and measurements in the field.
Nevertheless it is clear that the furfural concentration
(FC) in the oil gives an indication that deterioration is
taking place specifically in the paper.
The paper deterioration rate varies across the
transformer and is particularly affected by high
temperatures, which can either be due to local hot-spots
or high loading. The latter can be said to increase the FC
due to gradual deterioration of a large percentage of the
paper; the former to increase the FC after a hot spot
occurs due to the development of a fault such as a
partial discharge. Clearly both effects can occur together
but it shows that a high FC will usually indicate the
paper has very low mechanical strength and that the
transformer is necessarily at risk under high-current
fault conditions; furthermore a relatively low FC may
not indicate freedom from concern as an area of paper
weakening may occur due to a local fault.
However a significant change in the FC over a period
of months would indicate the probable development of a
fault. The value of regular measurements of furfural
concentrations in transformers seems clear.

[13] R. Blue, P. Conlin, D. Uttamchandani and O.

Farish, Portable optoelectronic instrument for
measuring furfuraldehyde in transformer oil, IEE
Proc SMT, vol. 145, pp. 45-48, 1998.

ACKNOWLEDGEMENT
The authors wish to acknowledge the assistance and
support of the NSF and university grant.
REFERENCES
[1] J.M.K. MacAlpine and C.H. Zhang, Observations from
measurements of the furfural content of oil samples from
transmission transformers, Electrical Power Systems
Research, vol. 57, no. 3, pp. 173-179, 2001.
[2] D.H. Shroff and A.W. Stannett, A review of paper ageing
in power transformers, Proc IEE, vol. 132, pp. 312-319,
1985.
[3] D.J.T. Hill, T.T. Le, M. Darveniza and T. Saha, A study of
the degradation of cellulosic insulation materials in a power
transformer. Part III: Degradation products of cellulose
insulation paper, Polymer Degradation and Stability, vol. 51,
pp. 211-218, 1996.
[4] D. Zheng, B. Lu, L. Yuan and D. Zheng, Thermal ageing
behavior of the oil-impregnated paper at 160oC, Proc. of the
5th International Conference on Properties and Applications

1091

[14] D.M. Allan, Recent advances in the analysis and

interpretation of aged insulation from operating power
transformers, Proceedings of the 5th International
Conference on Properties and Applications of Dielectric
Materials, Seoul, Korea, 25-30 May, 1997.
[15] H. Kan et al., Absorption of CO2 and CO gases and
furfural in insulating oil into paper insulation in oilimmersed transformers, Proc. IEEE International
Symposium on Electrical insulation, Pittsburg, USA, 5-8
June, pp. 41-44, 1994.
[16] T.S.R. Murthy, Assessment of transformer insulation
condition by evaluation of paper-oil systems, Proc. IEEE
Conference on Electrical Insulation and Dielectric
Phenomena, San Francisco, USA, 20-23 October, 1996, pp.
332-335.
[17] D.M. Allan, Practical life-assessment technique for aged
transformer insulation, Proc.IEE Pt.A, vol. 140, pp. 404408, 1993.
[18] M.K. Domun, Condition monitoring of power transformers,
IEE Colloquium (Digest), vol. 675, pp. 2/1-2/3, 1994.