1 3

Cellulose

ISSN 0969-0239
Volume 21
Number 1

Cellulose (2014) 21:823-833
DOI 10.1007/s10570-013-0087-0
Thermal lifetime of cellulose insulation
material evaluated by an activation energy
based method
R.Setnescu, L.V.Badicu,
L.M.Dumitran, P.V.Notingher &
T.Setnescu
1 3
Your article is protected by copyright and all
rights are held exclusively by Springer Science
+Business Media Dordrecht. This e-offprint
is for personal use only and shall not be self-
archived in electronic repositories. If you wish
to self-archive your article, please use the
accepted manuscript version for posting on
your own website. You may further deposit
the accepted manuscript version in any
repository, provided it is only made publicly
available 12 months after official publication
or later and provided acknowledgement is
given to the original source of publication
and a link is inserted to the published article
on Springer's website. The link must be
accompanied by the following text: "The final
publication is available at link.springer.com.
ORI GI NAL PAPER
Thermal lifetime of cellulose insulation material evaluated
by an activation energy based method
R. Setnescu

L. V. Badicu

L. M. Dumitran

P. V. Notingher

T. Setnescu
Received: 10 December 2012 / Accepted: 14 October 2013 / Published online: 22 October 2013
Springer Science+Business Media Dordrecht 2013
Abstract A rapid method, based on a logarithmic
degradation model of insulation material, is proposed
to reduce the test duration in lifetime assessment of
cellulose paper insulating materials. This method
proposes the determination of the activation energy
from a non-isothermal measurement made by differ-
ential scanning calorimetry or another thermal ana-
lysis technique and an aging test at a single elevated
temperature. The use of the onset temperature of the
exothermal peak at ca. 300 C is proposed for
evaluation of the activation energy of degradation.
For comparison, the thermal aging of Kraft cellulose
paper for power transformer insulation was performed
according to the general standard IEC60216-1/2001 at
three different temperatures: 155, 135 and 115 C, and
subsequently, the lifetimes at different service tem-
peratures were estimated. The experimental data
proved to have good agreement between the applied
methods, the differences being\10 % in terms of the
estimated lifetime across the range of service temper-
atures. The novel proposed method is effective in
terms of both energy and manpower costs as compared
to the current method: a factor of around 10 in the case
of reducing the aging time, a factor of 3 for the time
needed for measurements, and a factor of 10 for the
reduction of power intake.
Keywords Lifetime (durability) estimation
Insulation materials Differential scanning
calorimetry Cellulose paper
Introduction
It is well known that under the service conditions in
power transformers, cellulose paper is subjected to
thermal and electrical stress in the presence of traces
of air and water, resulting in different types of
degradation (depolymerization, oxidation, decompo-
sition) of the macromolecular chains. As a result, the
mechanical properties, especially the mechanical
strength, of the cellulose material are worsened,
leading to both local mechanical failures and contam-
ination of oil by paper fragments. The mentioned
mechanical failures will consequently lead both to
electrical failures and to clogging of the oil ducts. The
latter process is a main cause of transformers thermal
failures (Emsley and Stevens 1994). Therefore, the
R. Setnescu T. Setnescu
Valahia University of Targoviste, 2 Bld. Carol,
Targoviste 130082, Romania
R. Setnescu (&) T. Setnescu
National R&DInstitute for Electrical Engineering (ICPE-CA),
313 Splaiul Unirii, Bucharest 030138, Romania
e-mail: Radu.Setnescu@yahoo.com; rsetnescu@yahoo.com
L. V. Badicu L. M. Dumitran P. V. Notingher (&)
University Politehnica of Bucharest, 313 Splaiul
Independentei Str., Bucharest 060042, Romania
e-mail: petrunot@elmat.pub.ro
1 3
Cellulose (2014) 21:823833
DOI 10.1007/s10570-013-0087-0
Author's personal copy
service life of a transformer is critically limited by the
lifetime of its solid insulation (Prevost 2005).
Basically, the evaluation of a materials lifetime
requires three elements (Ding and Wang 2011),
namely: a critical degradation parameter, a degrada-
tion variable or indicator, i.e. a parameter unequivo-
cally related to the degradation state of material; a
kinetic model, an equation describing the relationship
between the degradation parameter and the aging time;
and a method for accelerated testing, based on the
kinetic model, that will provide data for lifetime
evaluation.
Both cellulose degradation and durability have
largely been studied in relation either to conservation
of various historical documents (Whitmore and Bog-
aard 1994; Rychly et al. 2002; Blumich et al. 2003;
ojewski et al. 2011) or to application as electrical
insulating material in high voltage cables and trans-
formers (Emsley and Stevens 1994; Liland et al. 2008;
Gilbert et al. 2009; Ding and Wang 2011). The
pyrolysis of cellulose has been intensively studied in
relation to the use of vegetal materials in different
processes of bio-fuels and bio-chemicals production
(Emsley 2008; Sanchez-Jimenez et al. 2011). At
ambient temperature, hydrolysis is the main process
that occurs (if no light or weathering are present),
while at elevated temperatures, oxidation and pyroly-
sis become dominant, leading to both carbonyl groups
and glycosidic bond scission (Lojewski et al. 2011).
The cellulose paper used for common electrical
insulation in transformers is obtained by the Kraft
process and contains ca. 90 % cellulose, 67 % lignin
and the rest, up to 100 %, consists mainly of pentosans
(Emsley and Stevens 1994).
The normal value of the average polymerization
degree (DP) for natural wood cellulose ranges between
3,000 and 10,000 anhydroglucose units (OSullivan
1997; Wang 2009), being strongly dependent on
source, acquiring processes and further treatments
(Wang 2009). The degradation of cellulose is an
irreversible process occurring by chain scission. The
variation of DP is a direct measure of the degradation
level, DP value hence being considered a degradation
parameter. The DP values of freshly produced paper
for insulation purposes usually range between 1,000
and 1,500 (Saha 2003). The initial drying process in
the transformer results in a decrease of DP up to ca.
950. Subsequently, service conditions progressively
reduce the DP value. A tensile strength decrease
accompanies the decrease of DP, in the range 850250
(anhydroglucose units), then an abrupt drop in tensile
strength values was reported (Shroff and Stannett
1985). The quality of insulation paper with DP values
of around 200 is considered poor enough to take this as
the end of life criterion (Shroff and Stannett 1985;
Saha 2003; Glomm Ese et al. 2010), but it was noted
that many transformers continued to operate at lower
DP levels (McNutt 1992).
The values of activation energies for the loss of
mechanical strength were found to be similar to those
of the depolymerization process (Emsley and Stevens
1994). The correlation between the average polymer-
ization degree and the tensile strength of insulation
paper was discussed in earlier papers by Emsley et al.
(1997, 2000).
The mass loss rate as measured fromthermal analysis
(TG, DTG) has been widely used for kinetic character-
ization of the degradation processes under either inert or
oxidative atmosphere (Kashiwagi and Nambu 1992;
Capart et al. 2004). Other thermal analysis techniques,
such as chemiluminescence (CL) (Strlic et al. 2000;
Rychly et al. 2002) or differential scanning calorimetry
(DSC) (Soares et al. 1995) have been used as well to
characterize the mechanisms and kinetics of cellulose
paper degradation. In some cases, the thermal analysis
techniques were coupled to MS, GC, FTIR or other
analysis techniques (Price et al. 2000; Statheropoulos
and Kyriakou 2000) in order to analyze the gases
resulting from the degradation processes.
It must also be noted that the normal humidity of the
paper used as transformer insulation is 45 %, but
after winding, the paper insulation dries to a humidity
of\0.5 % (Emsley and Stevens 1994). Afterward, the
dried paper is submitted to an oil-impregnation
process, which imparts increased dielectric strength
and enables a better cooling of the windings (Emsley
and Stevens 1994). Therefore, the water amounts
detected in aged transformers originate mainly from
cellulose degradation rather than from the initial
humidity of the insulating paper. The water from
cellulose degradation during the thermal aging will
induce increased electrical conductivity and bubble
gas formation. Hence, the presence of degradation
products of cellulose will damage the electrical
properties of the insulation system, especially under
overload conditions (Emsley and Stevens 1994).
A study of different types of cellulose papers for
transformer insulation subjected to thermal aging in
824 Cellulose (2014) 21:823833
1 3
Author's personal copy
the range of 110140 C revealed that aging is
strongly dependent on temperature and on the pre-
sence of water and oxygen (Shroff and Stannett 1985).
However, in the case of thermally upgraded paper, no
signicant inuence of water and oxygen on the aging
process was observed.
Methanol production during the aging process has
been proposed as a universal indicator for cellulose
degradation in power transformers (Gilbert et al. 2009,
2010). As the initial rate constants for depolymeriza-
tion and methanol formation in thermally upgraded
paper were found to be near those for standard
cellulose, it was concluded that they result from the
aging patterns just before the retardant action occurred
(Gilbert et al. 2010). Since the values of the frequency
factors for methanol and chain-end groups are similar,
it was concluded that the rate of CH
3
OH production
from chain scissions is much higher than the depoly-
merization rate. The latter feature becomes the rate-
determining step of the general degradation process
(Gilbert et al. 2010).
For thermally upgraded paper, it was found that the
amount of stabilizer in the brous structure is enough
to suppress the self-catalyzing and external catalyst
supply effects (Gilbert et al. 2010). Therefore, the
extended lifetime of the thermally upgraded papers
was assumed to decrease the frequency of chain
scissions with time.
Different accelerated aging tests have been pro-
posed and some have been standardized for durability
estimation of materials in different conditions (Brown
1991; Song et al. 1998). The actual standardized
method used for the evaluation of the paper insulation
lifetime in power transformers is based on the general
standard IEC 60216-1/2001 (IEC 60216 2001).
According to this standard, the insulation material
shall be submitted to aging at three different temper-
atures. Electrical (e.g. resistivity or conductivity) or
mechanical (tensile strength) measurements are to be
performed on samples taken out at established aging
periods. Plots of the characteristic electrical or
mechanical parameter against the aging time shall
be obtained for each aging temperature. For each
aging temperature chosen, the maximum aging time
shall be extended beyond the time required to reach
the end of life criterion, for the tensile strength, for
example, beyond a decrease to 50 % of the initial
value. The activation energy of material degradation,
calculated for the end of life criterion, is used for
subsequent estimation of the lifetime at lower service
temperatures. It is clear that such a procedure is time-
and materials-consumptive, and a more economical
method is desirable, at least for the primary evalu-
ation of different materials subjected to qualication
tests of cellulose insulation materials for power
transformers.
Therefore, the aimof this paper is the elaboration of
a simpler, more rapid and hence more cost-effective
method for lifetime evaluation based on thermal
analysis and electrical measurements. Roughly, the
proposed procedure is based on a simple exponential
equation [ln (D(T) = a ? b/T)] (Notingher 2005),
where D(T) is the lifetime at a given temperature T.
The coefcients a and b are determined from DSC
measurements (which provide the activation energy of
the degradation process) and an accelerated aging test
at a single elevated temperature (as for example
155 C). This simplication is based on the assump-
tion that the overall degradation processes involved in
long term aging and in short term thermal degradation
tests would be characterized by similar values of the
activation energies: the alteration of electrical prop-
erties during the long-term aging test is the result of
different simultaneous mechanisms (oxidation, depo-
lymerization, hydrolysis) the occur, as well the
accelerated thermo-oxidative conditions of DSC mea-
surement. In this work, we checked the applicability of
the apparent activation energy calculated from DSC as
a substitute for the corresponding value determined
from standardized long-term aging tests for Kraft
electro-insulating paper.
Experimental
Materials
The insulation paper samples subjected to accelerated
aging experiments at elevated temperatures were
square sheets of Kraft type paper (Weidmann AG)
with dimensions of 100 9 100 9 0.24 mm
3
. For
thermal analysis measurements (DSC, STA, CL), as
well as for the ATR-FTIR records before and after
oven thermal stress at 185 C, samples of Kraft paper
of 0.08 mm thickness were used.
Cellulose (2014) 21:823833 825
1 3
Author's personal copy
Instruments and methods
Differential scanning calorimetry (DSC)
DSCmeasurements were performed in non-isothermal
mode on a Setaram 131 EVO (Setaram Instrumenta-
tion, France) instrument using square sheet paper
samples of about 3 9 3 mm
2
. The measurements were
carried out in the presence of air (air ow 50 ml/min),
in the temperature range 30380 C at 4 different
heating rates, namely 2, 4, 6 and 10 C/min. The
exothermic peak at ca. 300 C, roughly related to
oxidation, was used for the activation energy. The
oxidation onset temperature (OOT) was determined,
according to (ASTM E2009-08 2008) as the crossing
point of the recorded baseline and the slope of the
oxidation exotherm (Fig. 1), using the specic func-
tion (Temperature determination) of Calisto Data
Processing (CDP) software (Setaram/AKTS).
Simultaneous thermal analysis: DSC, TG, DTG
Simultaneous thermal analysis (STA), i.e. DSC,
thermogravimetry (TG) and derivate thermogravime-
try (DTG), measurements were performed in a mixture
of oxygen and nitrogen atmosphere (oxygen ow
20 mL/min, nitrogen ow 80 mL/min), on samples of
about 2 mg, using Netzsch STA 449 (Netzsch, Ger-
many) equipment. The temperature range was
20500 C and the heating rates were 2, 4, 6 and
10 K/min. The temperature of the decomposition start
was evaluated from TG and DTG measurements.
FTIR spectroscopy
The Fourier transform infra-red (FTIR) spectra were
recorded directly on the paper sample using the
attenuated total reectance (ATR) technique, before
and after thermal stress at 185 C. A Jasco 4200 FTIR
spectrometer (Jasco Inc., Japon) coupled to an ATR
accessory (with a diamond crystal) was used. The
following measuring conditions were applied for all
spectra: the spectral range, 4,000400 cm
-1
; scans
number, 48/spectrum; resolution, 4 cm
-1
. The exper-
imental data were processed with the Spectra Pro-
cessing software (Jasco).
Oven thermal stress at 185 C
This treatment was carried out in an oven with air
circulation of Memmert UNE 400 type (Memmert,
Schwabach, Germany). The cellulose paper samples,
in the form of squares of 2 9 2 cm were accurately
weighed (Mettler-Toledo XS 105 type (Mettler-
Toledo, Switzerland) before the treatment and after
1, 2, 3 and 4 h of thermal treatment at 185 C. After
the treatment, the samples were kept in a desiccator
with solid CaCl
2
, at room temperature.
Fig. 1 DSC curve from
Kraft paper recorded at
10 K/min. Under air
atmosphere (air ow:
50 mL/min)
826 Cellulose (2014) 21:823833
1 3
Author's personal copy
Long-term aging at elevated temperature
Long-term aging tests were carried out at three
different temperatures chosen in concordance with
the recommendations of IEC 60216-1 (IEC 60216-1
2001). Considering the temperature index for cellulose
as IT = 90 C, (i.e. the operating time at 90 Cshould
be at least 20,000 h) resulted in values of the testing
temperatures of T
1
= 155 C, T
2
= 135 C, and
T
3
= 115 C, with a test duration for the highest
temperature (155 C) of more than 100 h but \500 h,
as required by the mentioned standard. For the lowest
temperature, the recommended conditions concerning
the aging time of at least 5,000 h and the difference
between the test temperature and the IT being no more
than 25 K were met as well.
Aging at elevated temperature was performed in a
forced air circulation oven of Raypa (Raypa, Spain).
The relative humidity of air entering the oven was
around 45 % at room temperature. All samples were
thermally conditioned before aging at 90 C during
48 h. The aged samples were taken out after different
established aging periods, in order to measure their
electrical properties.
Electrical properties measurements
The volume resistivity q
v
(t) was calculated, according
to Eq. (1) (Notingher et al. 2008), from the absorption/
resorption currents measured with a Keithley 6517
Electrometer (Keitley, USA) coupled to a special test
cell equipped with a guard ring; a DCvoltage of 300 V
was applied during 3,600 s:
q
v
t
S
l

U
0
i
a
t i
r
t
1
where i
a
(t) and i
r
(t) are respectively the absorption and
the resorption currents (measured at a certain moment
t), U
0
is the applied DC voltage, S is the active
electrodes surface area and l is the thickness of the
tested sample.
Relative humidity conditions
As it was already mentioned (see the Sect. 2.2.5), the
relative humidity in the laboratory was 45 % at 24 C.
It didnt presented important variations during the
experiments due to relatively stable weather and
stabilizing inuence of an air conditioning apparatus.
Results and discussion
Life-time evaluation of Kraft paper following IEC
60811
The lifetime of an insulation system is the time
elapsed to failure. For an electrical insulation system,
it is the time necessary to attain the electrical failure
of a critical element (material) in that system. The
lifetime prediction requires the denition of a
(critical) diagnostic parameter describing the state
of the material and the limit of the critical value of
this parameter (end of life criterion), related to the
safe function of the material and consequently of the
equipment as a whole. Either the electrical resistivity
or the conductivity is usually chosen as the critical
parameter for insulation materials. The lifetime can
be dened as the necessary time to produce a change
in the value of the selected parameter beyond an
acceptable limit (end of life criterion). The correct
prediction of material lifetime requires knowing the
deterioration kinetics corresponding to a certain
degradation model of the established parameter (the
degradation parameter).
For practical use, in the case of the studied Kraft
paper, it can be assumed that the material deterioration
rate (v
R
) depends on temperature according to an
Arrhenius type equation, as given in Eq. (2):
v
R
v
0
exp
W
a
R
g
T

; 2
where W
a
, is the apparent activation energy of the
degradation reaction, T is the test temperature and R
g
is the general gas constant (8.314 J/mol K).
In Fig. 2 the degradation curves are presented for
the studied cellulose paper at three different temper-
atures: T
1
= 155 C, T
2
= 135 C, and T
3
= 115 C.
Assuming a value of 80 TX m for the volume
resistivity (q
v
) as the end of life criterion, a straight
line of the lifetime as a function of temperature in
coordinates ln D(T) = f(1/T) is obtained (see Fig. 3).
The equation of this straight line was found to be ln
D(T) = -25.49 ? 13,114.16/T; the corresponding
value of the activation energy is W
a1
= 109 kJ/mol.
This gure concurs with similar data obtained on
cellulose paper degradation, for example by Lundg-
aard et al. (2004).
The lifetime values at different temperatures were
calculated applying the linear regression for
Cellulose (2014) 21:823833 827
1 3
Author's personal copy
temperature in the range of transformer service
conditions (Table 1). The lifetime values of 43.73
and 5.58 years were found respectively for 70 and
90 C, which are in reasonable agreement to the
durability required for power transformers as well as
to the IT value for Kraft cellulose paper.
From the standpoint of materials, energy and
manpower consumption, the above method, based on
the IEC 60216 standard, is economically ineffective,
especially when many materials are to be tested and
selected. A rough evaluation indicates that applying
this method involved 200 samples (100 9 100 9
0.24 mm
3
), more than 6,000 h of aging and 90 h of
testing for only one material. In terms of electrical
power, around 6,000 kWh were consumed for this kind
of testing. Although the standardized method provides
reliable data, as generally recognized, the above
evaluation illustrates the practical need for a more
rapid and more cost-effective.
Life-time evaluation of Kraft paper following
the proposed rapid procedure
It has been shown that the lifetime of an insulation
material exposed to a temperature stress T can be
described by an exponential equation as given in Eq. 3
(Notingher 2005):
ln DT a
b
T
; 3
where a is a material constant depending on both the
limit value and the rate constant of the degradation
process, b = W
a
/R
g
and D(T) is the aging time up to
reaching the limit value (end of life) of the diagnostic
parameter, and W
a
and R
g
stand for the activation
energy and general gas constant respectively.
As can be readily observed, plotting ln D(T) versus
1/T yields a straight line with the slope equal to b (i.e.
W
a
/R
g
), enabling the calculation of W
a
.
With the new proposed simpler method, the acti-
vation energy can be calculated from a rapid thermal
analysis test using the high-temperature exothermic
peak from DSC measurements. Only a single aging
test carried out at an elevated temperature is required
to provide the data for lifetime evaluation using the
Eq. (3).
The most evident effect observed in thermal
analysis of cellulose paper is the exothermal peak at
around 300 C (see the kinetic parameters derived
from this peak in Table 2). It was assumed that the
lifetime of the material under the DSC test is related to
this exothermic peak, which is due mainly to thermally
induced oxidation. Depolymerization, hydrolysis and
gas evolution processes occur simultaneously in long-
term aging tests as well as in thermal analysis
Fig. 2 Accelerated thermal aging curves of the Kraft paper
tested according IEC 60216-1/2001 [11]
Fig. 3 The lnD(T) versus 1/T plotting for activation energy
determination (ln D(T) = -25.49 ? 13,114.16/T)
Table 1 The values of the lifetime (in years) of Kraft paper
calculated for different transformer operation temperatures
Method Temperature (C)
70 80 90 100 110 155
IEC 60216 43.73 15.17 5.58 2.16 0.88 0.013
Proposed in the
present paper
48.24 15.86 5.54 2.05 0.80 0.02
828 Cellulose (2014) 21:823833
1 3
Author's personal copy
measurement and accompany or overlap the oxidation
mechanism. The onset of this peak, called below the
oxidation onset temperature (OOT) even though other
processes than oxidation are involved, was chosen for
activation energy calculations. The low degradation
level in the temperature region of OOT has been
shown by TGA measurements (Fig. 4), which indi-
cated a weight-loss of \5 % in the temperature range
30 COOT.
Besides the main exothermic peak, the DSC curves
from the studied material also presented some small
peaks at temperatures under 200 C (see Fig. 1, the
encircled region), but the kinetic parameters of these
peaks could be not tted to the heating rate in air. After
a thermal treatment of up to 4 h at 185 C, a
temperature within the range of these small peaks,
no changes were observed in the position or absorption
of the characteristic cellulose peaks (Ali et al. 2001;
Lojewska et al. 2005), suggesting that no signicant
degradation is related to these peaks (Fig. 5). Only a
mass decrease as the duration of the thermal treatment
increased was observed. This process can be assigned
to moisture desorption (Scheirs et al. 2001). It is
known that the elimination of chemically bound water
starts at higher temperatures (220 C) and occurs up to
500 C, its thermal effect (endotherm) being over-
lapped with other degradation processes in the main
peak, which starts at ca. 300 C (Ramiah 1970;
Scheirs et al. 2001). In these conditions, the use of
the main DSC peak for calculation the activation
energy of the aging processes looks more reasonable.
The straight lines obtained for each method,
presented in Table 2, enable the calculate of activation
energy W
a
from their slopes, which are equal or
proportional toW
a
/R
g
. Figure 6 is presented an
example for the determination of W
a
according to
the Kissinger method. The onset temperature of the
main exothermic peak has been used in these calcu-
lations. The parameters of the straight lines (a
0
, b
0
and
r) for the different methods applied are presented in
Table 3, with the values of the calculated activation
energies. It can be seen that there is a good t between
the experimental data for the different equations used
for activation energy calculation, and the resulting W
a
values are close to each other.
The average value of the activation energies
summarized in Table 3, noted as W
a2
(112 kJ/mol),
was subsequently used in lifetime estimation by the
proposed method.
The value of W
a2
calculated from OOT is in good
agreement with previously reported values for differ-
ent processes related to cellulose degradation and
Table 2 Kinetic parameters for Kraft paper oxidation calcu-
lated from the non-isothermal DSC curves (the peak at ca.
300 C)
b (C/min) OOT (C) T
m
(C) DHox (J/g)
2 269.95 310.8 -1,616
4 275.60 306.1 -1,444
6 288.96 330.9 -1,722
10 302.05 329.2 -764
Fig. 4 STA
(DSC ? TG ? DTG)
curves of Kraft paper
recorded in presence of a
mixture O
2
? N
2
(20 and
80 % respectively) at a
heating rate of 4 K/min
Cellulose (2014) 21:823833 829
1 3
Author's personal copy
subsequent failure of cellulose based materials,
namely:
hydrolytic depolymerization of cellulose, 110 kJ/mol
(Emsley and Stevens 1994; Liland et al. 2008);
COand CO
2
evolution, 105150 kJ/mol (Hino and
Suganuma 1972);
aging (air and humidity presence), 110114 kJ/mol
(Lundgaard et al. 2008) or 111 kJ/mol (Emsley
and Stevens 1994).
However, it should be mentioned that the reported
activation energy values found for pure oxidation
processes in cellulose were considerably lower (Liland
et al. 2008). This disagreement can be explained by a
more complex mechanism of the processes related to
the exothermic peak, which involve not only oxida-
tion, but also depolymerization and gas formation.
Using the average value of the activation energies
calculated from DSC measurements (W
a2
), and the
lifetime value for Kraft paper obtained for T = T
1
=
155 C (Fig. 2), the parameters a and b for Eq. (3)
were calculated, resulting in the following values:
a = -26.21 and b = 13,471.72 K
-1
. The calculated
lifetime values for different operating temperatures of
the Kraft paper are summarized in Table 1.
According to the present data, in the case of
operation at a constant and moderate temperature
(T = 80 C), the lifetime of the insulation material is
D = 1516 years, but when the temperature increases
to 90 C, the lifetime of the studied Kraft paper
becomes three times lower. When hot spots occur, the
Kraft paper lifetime can be reduced in those areas down
to a few 100 h (as for example, D = 0.0463 years for
T = 155 C). On the other hand, good concordance
can be observed between the lifetime values estimated
by the applied methods (standardized and proposed).
The experiment related to the proposed method
required roughly 600 h of laboratory aging, 30 h of
DSCtesting, data processing, andinterpretation. Interms
of electrical power, around 600 kWh were consumed.
Even though the above experimental data provide a
rather low degradation level at the onset temperature
of the exothermic peak, the use of the activation
energy value calculated on this basis appears some-
what strange. The considerably higher temperatures
relative to the aging process could result in different
reaction pathways. However, the reasonably close
values of W
a1
found by the standardized method and
W
a2
from DSC (Table 2), as well as the similar
lifetime values determined by both methods (Table 3),
support the applicability of the proposed method. In
actuality, the W
a2
value includes the reactions that
occur during heating up to the main peak rather those
responsible for the main peak itself. The isothermal
onset time of the main peak can be calculated using the
equation proposed by Gimzewski (1992) for hydro-
carbon oxidation, as given in Eq. (4):
e

Wa
RT
iso
Z
t
ind
0
dt
1
b
Z
T
ind
0
e

Wa
RT
dT 4
Fig. 5 ATR-FTIR spectra of the Kraft paper submitted to
thermal treatment at 185 C for different time periods: 1 initial,
untreated; 2 1 h; 3 3 h
Fig. 6 Activation energy W
a
determination from OOT by
Kissinger method for the studied Kraft paper
830 Cellulose (2014) 21:823833
1 3
Author's personal copy
where W
a
is the activation energy, T is the thermody-
namic temperature; T
iso
is the temperature of an
isothermal experiment; T
ind
is the oxidation induction
temperature (OOT); R is the gas constant; t is the time;
b is the heating rate; and t
ind
is the induction time. The
denite integral
R
t
ind
0
dt is the oxidation induction time
(OIT) in isothermal conditions. In our case, the
process is more complicated than in the case of
hydrocarbon oxidation because of other additional
reactions different than oxidation: depolymerization,
hydrolysis, and decomposition. Hence, the term oxi-
dation used here has a more complex meaning,
including all other reactions that lead to the onset of
the exothermic peak. In the same manner, the oxida-
tion induction time is, in fact, the time up to the
starting of exothermal degradation (called below
induction time). There are some data in the literature
indicating that the degradation kinetics of cellulose
(described by mass loss in TGA) has a sigmoidal
prole specic to an auto-accelerated process (Capart
et al. 2004), similar to the polyolens oxidation
process. Generally, the oxidation induction time from
isothermal measurements has the meaning of the
materials lifetime under the experimental conditions.
Assuming the same meaning in our case, a value of
around 10.8 min was calculated for 265 C (near to
the onset temperature at the lowest heating rate) and
336 min at 200 C. However, it is interesting to
observe that the value of the time to onset (112 h)
calculated for 155 C was practically the same as in
the case of the lifetime determined from electrical
measurements at the same temperature following the
IEC method (114 h). At lower temperatures, the
values of the induction time calculated with Eq. (4)
were lower than those found by electrical measure-
ments, but nonetheless comparable. Fig. 7 presents the
values of the induction times calculated with Eq. (4),
i.e. using the activation energy W
a2
from the onset of
the main exothermic peak of cellulose degradation. As
Table 3 Activation energy values for oxidation process and tting parameters of the linear relationships (y = a
0
? b
0
x) corre-
sponding to different methods applied for calculations (T
f
= OOT)
Method Equation Plot Parameters W
a
(kJ/mol) W
a2
(kJ/mol)
Kissinger (1957), Starink (1996)
ln
b
T
2
f

Wa
R
g
Tf
C
1
ln
T
2
f
b
f 1=T
f

r = -0.96109 109.4 112
a
0
= 12.54
b
0
= -13,170.6
Ozawa (1992)
ln b 1; 0518
W
a
R
g
T
f
C
2
lnb f 1=T
f
r = 0.96622 113.4
a
0
= 27.31
b
0
= -14,352.81
Starink (1996), Boswell (1980)
ln
b
T
f

W
a
R
g
T
f
C
3
ln
b
T
f
f 1=T
f

r = -0.9635 114.6
a
0
= 19.98
b
0
= -13,791.58
Starink (1996)
ln
b
T
1:8
f

W
a
R
g
T
f
C
4
ln
b
T
1:8
f
f 1=T
f

r = -0.96082 110.4
a
0
= 14.01
b
0
= -13,212.10
Fig. 7 Lifetime values versus 1/T for the Kraft cellulose paper
submitted to thermal degradation in presence of air in different
conditions: square box induction time of the exothermic peak
from DSC measurements (DSC lifetime); circle lifetime
evaluated from thermal aging and electrical measurements
according to IEC-60216-1/2001; triangle lifetime evaluated
according to the proposed method
Cellulose (2014) 21:823833 831
1 3
Author's personal copy
the lifetime values evaluated by standardized method
and those estimated by the proposed method fall
acceptably close to the straight line of the calculated
induction time values (DSC lifetimes), it can be
concluded that W
a2
calculated from the main exother-
mic peak can be reasonably used for aging evaluation
following the proposed method.
Conclusions
Considering that the lifetime of the Kraft paper
insulation material D(T) presents an exponential
(Arrhenius type) dependence on temperature T, the
use of the proposed method involves the achievement
of two important steps: rst is the determination of the
activation energy (W
a
) by DSC or other thermal
analysis method, and then, the value of the parameter
b in Eq. (1) is calculated. The second step is the
conduct of an accelerated aging test of the Kraft paper
material at an elevated temperature, as for example
T = 155 C, and nding the lifetime D under these
conditions. Subsequently, using the calculated param-
eter b, the parameter a in Eq. (3) for the lifetime curve
is calculated. With parameters a and b, the evaluation
of the Kraft paper lifetime for different operating
temperature is possible, using the simple linear
equation, Eq. (3).
Our results are in good agreement with the data
obtained by the classical method (IEC 60216) per-
formed in a parallel experiment, the differences in the
estimated lifetime being \10 % over the whole the
range of service temperatures. The proposed method
appears convenient in terms of manpower, materials
and energy costs.
The use of the exothermal peak for OOT and
subsequent W
a
calculation hold in the case of the
studied material because no signicant changes were
observed during the sample heating at lower temper-
atures. Moreover, the processes involved in the exo-
thermic peak are more or less similar to those occurring
in the case of oven aging at elevated temperature and
hence in the real case of aging under service conditions.
Further work is planned to check the applicability of
the proposed method in the case of other cellulose
materials as well as for different conditions, including
the inuence of specic impurities.
The novel method proposed in this work requires
shorter periods of both aging and testing as compared
to the IEC 60216 standard: a factor of around 10 in the
case of the aging experiments, while, in the case of
testing, the measurement time is reduced by a factor of
around 3; the energy consumption was 10 times lower.
Acknowledgments This work has been supported by the
Romanian Ministry of Labor, Family and Social Protection
through the Financial Agreements, POSDRU/6/1.5/S/16 5159
and POSDRU/89/1.5/S/62557, as well as by the Romanian
Ministry of Education, Research and Youth, Projects MIDMIT
22 080/2008 and Grant 10EU/2011. R. S. acknowledges
Dr. Traian Zaharescu, senior researcher at R&D Institute of
Electrical Engineering in Bucharest (ICPE-CA), for his kind
help and very useful discussions.
References
Ali M, Emsley AM, Herman H, Heywood RJ (2001) Spectro-
scopic studies of the aging of cellulosic paper. Polymer
42(7):28932900. doi:10.1016/S0032-3861(00)00691-1
***ASTM E 2009-08 (2008) Standard test method for oxidation
onset temperature of hydrocarbons by differential scanning
calorimetry. doi:10.1520/E2009-08
Blumich B, Anferova S, Sharma S, Segre AL, Federici C (2003)
Degradation of historical paper: nondestructive analysis by
the NMR-MOUSE. J Magn Reson 161(2):204209. doi:10.
1016/S1090-7807(03)00034-X
Boswell PG (1980) On the calculation of activation energies
using a modied Kissinger method. J Therm Anal
18(2):353358. doi:10.1007/BF02055820
Brown RP (1991) Survey of status of test methods for acceler-
ated durability testing. Polym Test 10(1):330. doi:10.
1016/0142-9418(91)90038-Y
Capart R, Khezamia L, Burnhamb AK (2004) Assessment of
various kinetic models for the pyrolysis of a microgranular
cellulose. Thermochim Acta 417(1):7989. doi:10.1016/j.
tca.2004.01.029
Ding H-Z, Wang ZD (2011) On the degradation evolution
equations of cellulose. Cellulose 15(1):205224. doi:10.
1007/s10570-007-9166-4
Emsley AM (2008) Cellulosic ethanol re-ignites the re of
cellulose degradation. Cellulose 15:187192. doi:10.1007/
s10570-007-9191-3
Emsley AM, Stevens GC (1994) Review of chemical indicators
of degradation of cellulosic electrical paper insulation in
oil-lled transformers. IEEE Proc Sci Meas Technol
141(5):324334. doi:10.1049/ip-smt:19949957
Emsley AM, Heywood RJ, Ali M, Eley CM (1997) On the
kinetics of degradation of cellulose. Cellulose 4(1):15.
doi:10.1023/A:1018408515574
Emsley AM, Heywood RJ, Ali M, Xiao X(2000) Degradation of
cellulosic insulation in power transformers. Part 4: effects
of aging on the tensile strength of paper. IEEE Proc Sci
Meas Technol 147(6):285290. doi:10.1049/ip-smt:
20000644
Gilbert R, Jalbert J, Duchesne S, Tetreault P, Morin B, Denos Y
(2009) Kinetics of the production of chain-end groups and
methanol from the depolymerization of cellulose during
832 Cellulose (2014) 21:823833
1 3
Author's personal copy
the aging of paper/oil systems. Part 1: standard wood Kraft
insulation. Cellulose 16(2):327338. doi:10.1007/s10570-
008-9261-1
Gilbert R, Jalbert J, Duchesne S, Tetreault P, Morin B, Denos Y
(2010) Kinetics of the production of chain-end groups and
methanol from the depolymerization of cellulose during
the aging of paper/oil systems. Part 2: thermally-upgraded
insulating papers. Cellulose 17(2):253269. doi:10.1007/
s10570-009-9365-2
Gimzewski E (1992) The relationship between oxidation
induction temperatures and times for petroleum products.
Thermochim Acta 198(1):133140
Glomm Ese M-H, Liland KB, Lundgaard LE (2010) Oxidation
of paper insulation in transformers. IEEE Trans Dielectr
Electr Insul 17(3):939946. doi:10.1109/TDEI.2010.
5492270
Hino T, Suganuma T (1972) Rapid measurement of deteriora-
tion of oil-immersed paper. IEEE Trans Electr Insul EI-
7(3):122126. doi:10.1109/TEI.1972.299194
***IEC 60216-1 (2001) Guide for the determination of thermal
endurance properties of electrical insulating materials.
Aging procedures and evaluation of test results, 5 edn.
Geneva
Kashiwagi T, Nambu H (1992) Global kinetic constants for
thermal oxidative degradation of a cellulosic paper. Com-
bust Flame 88(34):345368. doi:10.1016/0010-2180(92)
90039-R
Kissinger HE (1957) Reaction kinetics in differential thermal
analysis. Anal Chem 29(11):17021706. doi:10.1021/
ac60131a045
Liland KB, Glomm Ese M-H, Lundgaard LE, Kes M (2008)
Oxidation of cellulose. IEEE Int Symp Electr
2008:304307. doi:10.1109/ELINSL.2008.4570334
ojewska J, Miskowiec P, ojewski T, Proniewicz LM (2005)
Cellulose oxidative and hydrolytic degradation: in situ
FTIR approach. Polym Degrad Stab 88(3):512520.
doi:10.1016/j.polymdegradstab.2004.12.012
ojewski T, Zieba K, Koodziej A, ojewska J (2011) Follow-
ing cellulose depolymerization in paper: comparison of
size exclusion chromatography techniques. Cellulose
18(5):13491363. doi:10.1007/s10570-011-9562-7
Lundgaard LE, Hansen W, Ingebrigtsen S (2008) Aging of
mineral oil impregnated cellulose by acid catalysis. IEEE
Trans Dielectr Electr Insul 15(2):540546. doi:10.1109/
TDEI.2008.4483475
Lundgaard LE, Hansen W, Linhjell D, Painter TJ (2004) Aging
of oil-impregnated paper in power transformers. IEEE
Trans Power Deliv 19(1):230239. doi:10.1109/TPWRD.
2003.820175
McNutt WJ (1992) Insulation thermal life considerations for
transformer loading guides. IEEE Trans Power Deliv
7(1):392401. doi:10.1109/61.108933
Notingher PV (2005) Materials for electrotechnics. Politehnica
Press, Bucharest
Notingher PV, Stancu C, Dumitran LM, Notingher P Jr, Rak-
owska A, Siodla K (2008) Inuence of the Aging State of
Insulation Systems on Absorption/Resorption Currents.
Revue Roum Sci Tech Electr Energ 53(2):163178
OSullivan AC (1997) Cellulose: the structure slowly unravels.
Cellulose 4(3):173207. doi:10.1023/A:1018431705579
Ozawa T (1992) Estimation of activation energy by isoconver-
sion methods. Thermochim Acta 203(1):159165. doi:10.
1016/0040-6031(92)85192-X
Prevost TA (2005) Thermally upgraded insulation in trans-
formers. In: Proceedings of the electrical insulation con-
ference and electrical manufacturing expo, 2005. doi:10.
1109/EEIC.2005.1566272
Price DM, Hourston DJ, Douglas J, Dumont F (2000) Ther-
mogravimetry of polymers. Encycl Anal Chem. doi:10.
1002/9780470027318.a2037
Ramiah MV (1970) Thermogravimetric and differential thermal
analysis of cellulose, hemicellulose, and lignin. J Appl Polym
Sci 14(5):13231337. doi:10.1002/app.1970.070140518
Rychly J, Strlic M, Matisova-Rychla L, Kolar J (2002)
Chemiluminescence from paper I. Kinetic analysis of
thermal oxidation of cellulose. Polym Degrad Stab 78(2):
357367. doi:10.1016/S0141-3910(02)00187-8
Saha TK (2003) Review of modern diagnostic techniques for
assessing insulation condition in aged transformers. IEEE
Trans Dielectr Electr Insul 10(5):903917. doi:10.1109/
TDEI.2003.1237337
Sanchez-Jimenez PE, Perez-Maqueda LA, Perejon A, Pascual-
Cosp J, Ben tez-Guerrero M, Criado JM (2011) An
improved model for the kinetic description of the thermal
degradation of cellulose. Cellulose 18(6):14871498.
doi:10.1007/s10570-011-9602-3
Scheirs J, Camino G, Tumiatti W (2001) Overview of water
evolution during the thermal degradation of cellulose. Eur
Polym J 37(5):933942. doi:10.1016/S0014-3057(00)
00211-1
Shroff DH, Stannett AW (1985) A review of paper aging in
power transformers. IEEE Proc C 132(6):312319. doi:10.
1049/ip-c:19850051
Soares S, Camino G, Levchik S (1995) Comparative study of the
thermal decomposition of pure cellulose and pulp paper.
Polym Degrad Stab 49(2):275283. doi:10.1016/0141-
3910(95)87009-1
Song MS, Hu GX, Hu LJ (1998) Prediction of long-term
mechanical behaviour and lifetime of polymeric materials.
Polym Test 17(5):311332. doi:10.1016/S0142-9418(97)
00060-3
Starink MJ (1996) Anewmethod for the derivation of activation
energies from experiments performed at constant heating
rate. Thermochim Acta 288(12):97104. doi:10.1016/
S0040-6031(96)03053-5
Statheropoulos M, Kyriakou SA (2000) Quantitative thermo-
gravimetric-mass spectrometric analysis for monitoring the
effects of re retardants on cellulose pyrolysis. Anal Chim
Acta 409(12):203214. doi:10.1016/S0003-2670(99)
00859-4
Strlic M, Kolar J, Pihlar B, Rychly J, Matisova-Rychla L (2000)
Chemiluminescence during thermal and thermo-oxidative
degradation of cellulose. Eur PolymJ 36(2000):23512358
Wang HH (2009) Cellulose and Pulp. In: Owens JN, Lund GH
(eds) Forest and forest plants, vol II. EOLS Publishers, UK,
pp 158178
Whitmore PM, Bogaard J (1994) Determination of the cellulose
scission route in the hydrolytic and oxidative degradation
of paper. Restaurator 15(1):2645. doi:10.1515/rest.1994.
15.1.26
Cellulose (2014) 21:823833 833
1 3
Author's personal copy