Sunteți pe pagina 1din 14


Furan Analysis for Liquid Power

Key words: transformer paper insulation, furans, aging, transformer life assessment, degree of

Introduction Luiz Cheim

The purpose and scope of this article is to discuss the his-
tory and development of furan testing as one of several diagnos- ABB TRES North America, 4350 Semple
tic tools to evaluate the health of transformer paper insulation. Ave., St. Louis, MO 63120
Much of the information on this subject has been derived from
laboratory studies of materials. The topic has been extensively Donald Platts
studied, and many technical papers have been written on the for- PPL Electric Utilities, 2 North Ninth St.,
mation of furanic compounds, which are dissolved in oil and
analyzed through normal transformer oil sampling. Allentown, PA 18101
Furans are one of several chemical compounds that are
produced as the paper insulation in a transformer ages and de- Thomas Prevost
grades. This article discusses how furans are produced, how they OMICRON Electronics USA, 230 Third Ave.,
dissolve in oil, the correlation of known furan data with the de-
gree of polymerization (DP), the parameters that affect the rate
Waltham, MA 02451
of furan production, and the limitations on the use of furans for Shuzhen Xu
diagnostic analysis.
This article attempts to capture and summarize the known FM Global, 1151 Boston Providence
technical information and recommendations for analysis of fu- Turnpike, Norwood, MA 02062
ran test results. It will also provide the technical basis for con-
tinued gathering and evaluation of furan data for liquid power
transformers, and provide a recommended structure for collect-
ing that data. The article reviews what is known
History of Furan Analysis about furans and provides a techni-
Paper insulation is a critical component in an oil-filled trans- cal basis for the continued collection
former. To estimate the degradation of insulation, different diag- and evaluation of furan data and
nostic tools have been used such as dissolved-gas analysis and
estimates of aging from loading history. Furans have received suggests a structure for collecting
increased attention in the last 20 years because they offer the that data.
hope of measuring a specific chemical that will translate directly
into an indicator of the extent of the aging of paper insulation.
The research on furans in transformers started in the 1980s.
In one of the early papers documenting their studies, Burton et Since those early studies, furan analysis has attracted many
al. [1] reported that thermal degradation of cellulosic insulation researchers from the electrical and chemical fields, leading to a
material produces 2-furaldehyde, 5-hydroxymethyl-2-furalde- large volume of publications. The key points that promoted this
hyde, 2-acetylfuran, 5-methyl-2-furaldehyde, and 2-furfurol. research are as follows:
When they are produced, these compounds partition themselves
between the paper and the insulating oil. As a result, their con- • Furans are specific to the degradation of insulation pa-
centration in the oil can be measured by means of high-perfor- per inside transformers, i.e., they are only generated
mance liquid chromatography. when insulation paper degrades [2], [3].

8 0883-7554/12/$31/©2012/IEEE IEEE Electrical Insulation Magazine

• The measurement of furans is not intrusive. One only Paper Insulation Degradation
needs an oil sample to perform the test, which can be The mechanism of cellulose degradation depends on the
obtained from outside of the transformer. The proce- conditions to which the paper insulation is subjected. There are
dure for sampling the oil is the same as that for dis- four major factors that promote degradation, namely, exposure
solved-gas analysis. to elevated temperature, oxygen, acid, and moisture. The deg-
radation processes associated with these causes are thermal,
How Furans Are Formed oxidative, and hydrolytic, which are shown in Figure 2. These
Paper Insulation Background three degradation processes result in chain scission or depo-
Kraft insulation paper is usually made by the delignifica- lymerization, yielding glucose or degraded forms of glucose.
tion of wood pulp by the KRAFT process, in which wood is Glucose will further degrade, depending on the conditions, to
treated with a mixture of sodium hydroxide and sodium sulfide. form furans and other chemical products such as water and
The main chemical component of insulation paper is cellulose, gases [2], [4].
which is a natural polymer of glucose. The cellulose fibers con-
Furan Chemistry
sist of a bundle of molecules of different lengths, positioned side
Furans are part of the degradation products of cellulose in-
by side. They are held together by hydrogen bonds involving
sulation paper in transformers, and they are partially soluble
the hydroxyl (OH) groups on the adjacent molecules, which are
in the insulation fluid. When the cellulose chain breaks down
linked together through the glycosidic bond as shown in Figure
during paper degradation, each splitting of the chain liberates a
glucose monomer unit that undergoes further chemical reactions
Thermally upgraded cellulose insulation was developed in
to become furan compounds and other products such as water
the late 1950s by the major manufacturers of that time. There
and gases. Most often, the following five furan compounds are
are basically two types of thermal upgrading processes:
(a) Modification of the cellulose chains specifically at OH
• 2-furaldehyde (2FAL)
groups by cyanoethylation and acetylation and
• 5-methyl-2-furaldehyde (5M2F)
(b) Addition of chemicals to protect the cellulose from
• 5-hydroxymethyl-2-furaldehyde (5H2F)
acidic by-products of aging; this is primarily achieved
• 2-acetyl furan (2ACF)
with nitrous compounds such as urea, melamine, dicy-
• 2-furfurol (2FOL)
andiamide, and polyacrylamide.
The chemical structures of these five furans are shown in Fig-
In cyanoethylation, the cellulose is chemically modified with
ure 3. The concentrations of these furans in the transformer oil
some of the less-stable water forming hydroxyl groups in the
can be identified and quantified by using high-performance liq-
cellulose chain being replaced by more stable cyanethyl groups.
uid chromatography. Formal test methods for measuring furan
This process must be done in the pulping stage of paper manu-
compounds, such as ASTM D5837 and IEC 61198, have been
facturing. The replacement of some of the hydroxyl groups also
published to ensure consistent results. Among these furanic
reduces the number of hydrogen bridges between the molecules.
compounds, 2FAL is always considered as the main compound
This reduces mechanical strength; however, it does lower water
in the analysis because of its relatively higher generation rate
absorption and shrinkage.
and stability inside a transformer.
In amine addition, nitrous compounds such as dicyandiamide
are added to the paper to act as stabilizing agents. The addi-
tion of stabilizers suppresses the self-catalyzing character of the The Stability of Furan Analysis
aging process by a chemical reaction with the aging products To evaluate the current technology and to make effective use
during which the additives are consumed. The stabilizing agents of the furan measurement process, it is necessary to examine
consume water by reacting chemically with it. They also con- the stability of furans in the environment found in an operating
tain organic bases that partially neutralize the acids, which are transformer. As stated by Unsworth [2], “It is pointless to cor-
also a by-product of aging. The changes in the initial mechanical relate furan concentrations with mechanical and dielectric prop-
properties of the cellulose product are negligible. The stabilizing erties if furans are not stable over time.” So for the purpose of
agents can be applied to the finished paper or as a component in evaluating or making better use of this technology, it is very im-
the final stages of the papermaking process, called sizing. portant to understand the stability of these various compounds in
the transformer operating environment.
Laboratory tests under nitrogen, or without the presence of
O2, show that below 100°C, all five of these furans are quite
stable without significant losses. When the temperature is above
100°C, and up to 160°C, 2FAL, 5M2F, and 2ACF are found to
be relatively stable. However, it was found that 2FOL is very
unstable and decomposes further when exposed to such high
Figure 1. The structure of insulation paper; n = degree of po- temperature [2], [5], [6]. Therefore, furans in the transformer oil
lymerization. should be quite stable because the normal operating temperature

March/April — Vol. 28, No. 2 9

Figure 2. The degradation processes of cellulose paper.

for the top oil (the hottest oil) rarely exceeds 100°C and should results obtained by Griffin et al. [8] show that in the presence
be limited as per C57.91 to less than 110°C. The temperature of dicy, 5H2F was depleted to a great degree, whereas 2FOL,
limit for the average winding temperature at rated condition is 5M2F, 2FAL, and 2ACF appeared to have a small decrease (see
95°C, with a hot-spot limit of 110°C. So again, during normal Figure 4).
operation, the furan compound in the insulation should remain One theory for the observation of lower levels of 2FAL is that
stable without significant loss. dicy prevents a buildup of acids (acid capture), which prevents
Test results from the experiments carried out by Allan in 1995, the formation of 2FAL. Hydrolysis of cellulose is always acid
in which the furans-in-oil solutions were open to atmosphere air, catalyzed [10].
show that when the temperature increases from 70°C to 110°C,
the oxidative stability of furans gets increasingly worse, espe- Furan Analysis Techniques
cially that of 2FOL and 5H2F. These two furans are almost de- Although furans are produced by the deterioration of the cel-
pleted by the oxidation reactions when the temperature is above lulose paper insulation, it is the concentration of furans in the oil
90°C. So Allan suggested that the usefulness of 2FOL and 5H2F that is measured. Oil samples are drawn and sent to a laboratory
would be reduced in the transformers with free-breathing sys- for testing, and three different approaches to the interpretation of
tems, but the remaining three furans, 2FAL, 5M2F, and 2ACF, furan test results have been used, namely,
would be stable enough to be used as diagnostic indicators for
almost all transformers [7]. (1) Correlating the furan concentrations in the oil to the DP
Under oxidizing conditions found when a significant amount of the paper,
of oxygen is present in the oil, laboratory tests show the stability (2) Comparing the furan concentration with a threshold
of these five furans is lowered. It has been found that the stability value based on statistical analysis of a large population,
of furan compounds has the following order: 2ACF ≈ 5M2F > and
2FAL > 5H2F > 2FOL [2], [6], [8], meaning that 2ACF is about (3) Trending the furan concentration level over time.
the same as 5M2F, and both are more stable than 2FAL, and so
on. The correlation between the furan concentrations in the oil
Dicyandiamide, known as “dicy,” is used in thermally up- and the DP of the paper has been the object of much research.
graded insulation paper to help improve thermal performance, Because DP has been accepted as a reliable indicator of the de-
control the rate of aging, and for stability. The effect of dicy terioration of Kraft paper insulation, a direct correlation to the
on the stability of furan compounds remains a controversial is- measured furan content would prove to be the missing link to de-
sue. It was reported in 1999 that dicyandiamide appears to be termine the extent of the aging of the paper without the intrusive
involved in breaking down furan compounds [9]. Laboratory measures required to obtain a representative sample that could

10 IEEE Electrical Insulation Magazine

Figure 3. The chemical structures of furan compounds.

be tested for its DP value. The efforts to establish that correla- DP is calculated using experimentally established equations that
tion have been based on laboratory tests, and they are discussed can be found in the standard test method. For a well-performed
in this article. The last two approaches are mainly based on the DP measurement, repeatability would typically be 2% and re-
statistical analysis on actual data collected from laboratories or producibility about 10% [11].
from utilities, and they will be discussed later in the article. Viscometric degree of polymerization measurement is an
analysis method that has long been recognized as a viable mea-
Correlation Between Determination sure of paper insulation aging. It gives the average number of
of the DP and Furan glucose units per cellulose chain. The length of the cellulose
Degree of polymerization is determined in accordance with molecule is measured in terms of the DP. New insulation pa-
ASTM D4243 - 99(2009), Standard Test Method for Measure- per has an average chain length of 1,100 to 1,500. Drying the
ment of Average Viscometric Degree of Polymerization of New transformer after winding reduces the DP of the paper further to
and Aged Electrical Papers and Boards. Standard IEC 60450 is around 1,000 to 1,100 [2], [12], [13].
a similar test method. The viscosity of a solution of macromol- When the insulation paper degrades to the point that the DP
ecules is dependent on the molecular weight at a given low con- is approximately 200, many would consider the paper to have
centration. This phenomenon is used for the determination of the reached the end of its useful life. It is no longer capable of re-
DP of cellulosic insulating materials. covering from the mechanical stresses expected in a normally
For each test a small amount of fluffed paper or board, of operated transformer. Although many transformers remain in
mass 50 to 100 mg, that has the oil removed is dissolved in cu- service with DP levels below 200, selecting a DP value as the
priethylenediamine. The viscosity of the paper solution is deter- point of the insulation end of life is a matter of engineering
mined using a capillary viscometer. From this result the average judgment.

March/April — Vol. 28, No. 2 11

Figure 5 shows one example of this relationship found in labora-
tory tests. To denote this relationship, several different equations
have been proposed by the various authors, and their results are
summarized below.

Kraft Paper
Several of the widely recognized technical papers that have
proposed correlation equations for the relationship of the furan
concentration to DP are summarized in the following:

A. Chengdong’s curve [14] is as follows: log10(2FAL) =

1.51 − 0.0035 × DP, where 2FAL is the concentration
Figure 4. Stability of furanic compounds in new oil with dicy- of furfural in the oil expressed in milligrams per liter
andiamide (dicy) at 120°C as found by Griffin et al. [8]. HMF and DP is the average degree of polymerization. In the
= 5H2F, FOL = 2FOL, FAL = 2FAL, AF = 2ACF, and MF = laboratory, the experiment simulated the oil–paper in-
5M2F. New oil spiked with 1,000 mg/mL of each of five furan sulation system of a transformer.
compounds and 100 mg/mL of dicy. All of these samples were B. De Pablo and Pahlvanpour [12] presented five different
sparged with nitrogen in valved, stainless-steel cylinders for equations to correlate 2FAL to DP in oil. These five
20 minutes and then sealed under nitrogen. No information on equations were provided by different laboratories that
type of oil used was given. Reprinted with permission from P. J. participated in a project to establish the relationship of
Griffin and L. R. Lewand, “Case studies,” in Proceedings of the furan concentration to DP. In all of these laboratories,
Sixty-Fourth Annual International Conference of Doble Clients, the paper used complied with IEC publication 60554-1,
1997, Sec 5-7. and a ratio of 1 g of paper to 100 mL of oil was used.
The oil used complied with the specifications described
To obtain the relationship between furan concentrations in in IEC publication 60296. The five correlation equa-
the oil and DP of the paper, accelerated aging tests of paper tions and their correlation coefficients (R) were as fol-
samples were carried out in different laboratories [1], [6], [12], lows (2FAL is expressed in mg/kg of paper):
[14]–[17]. Analysis of the test data from sealed tube test cells by
almost all of the laboratories suggests that there is an approxi- Lab 1: log10(2FAL) = 3.41 − 0.00264 × DP, R = 0.945;
mately linear relationship between the logarithm of the concen- Lab 2: log10(2FAL) = 3.57 − 0.00355 × DP, R = 0.915;
tration level of 2FAL in oil and the DP of standard Kraft paper. Lab 3: log10(2FAL) = 1.82 − 0.00166 × DP, R = 0.736;
Lab 4: log10(2FAL) = 3.61 − 0.00365 × DP, R = 0.835;
Lab 5: log10(2FAL) = 3.40 − 0.00287 × DP, R = 0.951.

All of the laboratories in the CIGRE investigation used

the same type of paper and the same type of oil, and
all studies were performed under strict laboratory con-
ditions. Despite that, discrepancies in the results are
significant, even if laboratory 3 is discarded as an out-
lier. The agreement between results seems to improve
toward end of life (DP <250); however, in insulation
with higher levels of DP (~800), the variation between
laboratories is significant (from as low as 8 to above
approximately 20 ppb). As demonstrated in [12], there
is no single model that applies to all test conditions or
to all transformers.
Figure 5. Correlation of degree of polymerization to C. In 2003, Cheim et al. proposed a mathematical cor-
log10(2FAL) for Kraft paper aged in oil at different tempera- rection to the Arrhenius model for transformer loss of
tures: ▲ = 120°C; ■ = 140°C; ○ = 160°C. The setting of the life in order to incorporate the effect of temperature
experiment was 4 g of Kraft paper in 40 to 50 mL of oil with gradients inside transformers. His analysis used that
moisture less than 10 mg/L and 1 ± 0.1% oxygen. The ratio of formula, evaluated the 2FAL data from a large number
oil to paper was around 10:1. No information was given on the of transformers (~400) installed in about 20 different
type of oil used [6]. Reprinted with permission from A. M. Ems- utilities in Brazil, and considered laboratory experi-
ley, X. Xiao, R. J. Heywood, and M. Ali, “Degradation of cel- ments of accelerated transformer aging. Based on that
lulosic insulation in power transformers. Part 2: Formation of analysis, he proposed the following model [18]:
furan products in insulating oil,” in IEE Proceedings Science,
Measurement and Technology, 2000, published by the IET.

12 IEEE Electrical Insulation Magazine

 2FAL ψd
DP =   ,
 λ 

where 2FAL is expressed in ppm and parameters λ, ψ,

and d depend on the type of paper used and also on the
winding longitudinal temperature gradient. Note that
the model is fundamentally transformer design depen-
dent, including the types of paper, and the reference
describes the methodology to derive the parameters.
D. The De Pablo equation [1] (De Pablo 1 in this article) is
as follows:

7, 100 Figure 6. A comparison of the various correlation equations

DP = , between degree of polymerization (DP) and 2-furaldehyde
8.88 + (2FAL)
where 2FAL is expressed in milligrams per kilogram.
This equation was developed by De Pablo assuming 7,337 ppb (by weight), whereas the furan level for up-
that the initial DP of paper is 800, molecular weight graded paper was 2,843 ppb (by weight). The reference
of 2FAL is 96, three cellulose chain scissions give one did not indicate which furans were used in the correla-
2FAL molecule, and the ratio of oil to paper is 25. Also, tion.
the paper is assumed to be uniformly degraded. B. In 2004 Lundgaard et al. published a paper where they
De Pablo also developed another equation, which is tested Kraft and thermally upgraded Kraft papers to ex-
based on the assumption that 20% of the paper de- amine various aging factors [21]. The Insuldur process
grades faster. The equation (De Pablo 2 in this article) was used to thermally upgrade the Kraft paper. The
is as follows: data on furan testing are represented in Figure 8. Note
that an increase in the number of chain scissions is pro-
800 portional to a decrease in the DP. Observations about
DP = .
0.186 × 2FAL + 1 this testing and the graphs are as follows:
• For this study, the Kraft paper aged significantly more
than the thermally upgraded Kraft, and the level of fu-
These equations are theoretically derived from chemis-
rans produced during this aging of the Kraft is very sig-
try and not from test data.
E. The Burton equation [1] is as follows: log(2FAL) = 2.5
• For the same exposure, the thermally upgraded Kraft
− 0.005DP.
experienced much less aging, fewer chain scissions,
F. The Paul Vuarchex equation [1] is as follows: log(2FAL)
and little furan production, as demonstrated by the
= 2.6 − 0.0049DP.
furan concentration, although furans were produced
throughout the study.
With the exception of the equation proposed by Cheim et al.,
Also note in Figure 8 that although it might appear that
the above five equations are plotted in Figure 6 to illustrate the
Insuldur insulation does not produce 2FAL at all, Insul-
variation in the results.
dur did not age to the same degree as Kraft, as demon-
strated by the number of chain scissions.
Thermally Upgraded Paper

A. In 2003, Myers et al. published a paper to state their

prediction of DP from total furan compounds (total fu-
ran means the summation of the five furan compounds
found resolved in the oil) for transformers with ther-
mally upgraded paper [19], which is as follows: DP =
[log10(total furan) − 4.0355]/−0.002908.
In a “Tech Corner” bulletin published in 2009 on the
Myers website [20], a direct comparison of nonupgrad-
ed with thermally upgraded paper was offered (Figure
7). In this comparison, the furan level corresponding to
a DP of 200, which is generally considered end of life Figure 7. Myers comparison of correlation of degree of polym-
for paper insulation, was compared. For a DP of 200, erization (DP) to furan level between thermally upgraded and
the non-thermally upgraded paper had a furan level of non-thermally upgraded paper [20].

March/April — Vol. 28, No. 2 13

Figure 9. Plot of degree of polymerization (DP) to log 2-fural-
dehyde from aging tests on Kraft and thermally upgraded paper
with both of the best-fit correlation equations [22].

sults. For thermally upgraded Kraft, at tested furan levels of 2.42

to 2.46 ppm, the range of the tested DP measurements extends
from 340 to 600. Similar variations exist with the Kraft paper
In Figure 10, the data from Figure 9 have been replotted with
the addition of error bounds. The error bounds of ±35% of the
calculated PD used with the derived correlation have been cho-
sen because they bound most of the data. The resulting conclu-
sion from this analysis of that data set is that we can state with
a fairly high degree of certainty that the equation on this graph
Figure 8. The concentration of 2-furaldehyde (2FAL) for Kraft will calculate the corresponding DP value of the samples aged
paper and Insuldur insulation [21 [AU6: Indicate which insula- and tested in this study with an accuracy of ±35%.
tion is shown is which graph (a and b). ]]. After reviewing the 11 correlation equations of furan to DP
that have been presented in this article and analyzing the graphs
shown in Figures 6, 9, and 10, the authors find that they cannot
C. In 2008, Wicks and Prevost presented data on DP ver- endorse any one of these proposed correlations. When reviewing
sus 2FAL as part of thermal aging tests for Kraft and the variability of the data in Figure 9, and the equations pre-
thermally upgraded paper, “C57.100 – IEEE Standard sented in the CIGRE round-robin testing, we find that a given
Test Procedure for Thermal Evaluation of Liquid Im- furan concentration may be matched with a given DP measure-
mersed Distribution and Power Transformers” [22]. ment in one researcher’s report and then in another report be
Their work on C57-100 resulted in thermal aging tests matched with another DP measurement that is several hundred
on both Kraft and thermally upgraded paper using dual units different. Even under the best of conditions, when a single
temperature aging cells. Each test sample was tested technician working in a single test laboratory tests multiple sam-
for DP of the paper and furans in oil content. The re-
sulting data are shown in Figure 9.

The dual temperature aging experiment was significant in two

ways. First, the scale of the experiment was significantly large so
that it allowed one to study the variability in furan results. Sec-
ond, the model more closely represented an actual transformer
insulation system in which there are hotter and cooler regions
within the insulation system and the oil is generally cooler than
the hot-spot insulation. Most important, only the insulating pa-
per on the conductor is thermally upgraded and the remainder of
the cellulose insulation, which accounts for more than 90% of
the insulation by weight, is nonupgraded. Sealed tube aging ex-
periments subject the entire insulation system to the same tem-
perature and typically use 100% Kraft or thermally upgraded
paper in their comparative studies. Figure 10. Data from Figure 9, with error bands plotted to in-
Figure 9 shows that even when the large-scale testing is done dicate ±35% of the calculated degree of polymerization (DP).
in one laboratory, there is often great variability in the test re- 2FAL = 2-furaldehyde.

14 IEEE Electrical Insulation Magazine

Figure 11. The furan distribution of 77 step-up transformers [14].

ples that have been aged at different rates for different times and tain threshold values of furans for diagnostic purposes. However
different temperatures using a standard aging test protocol and most of the survey results are from transformers using standard
when it is presumed he followed the identical test procedures, Kraft paper that were designed according to IEC standards. The
the results show little precision (Figure 9). After reviewing, and results may have limited application to transformers using ther-
analyzing, this incredibly wide variation in the test data, and the mally upgraded paper designed according to IEEE standards.
correlation equations, the authors cannot conclude that the pro-
cedure for these furan tests and the conversion into an equivalent A. Dong et al. [25] collected 750 furan test records of dif-
DP is valid today. ferent transformers from China. No detailed informa-
tion on these transformers was given. Through a sta-
Studies of Data From Operating tistical approach, a significant effect of service age of
step-up transformer on 2FAL level in oil was observed,
Transformers but the effects of rated voltage and capacity were con-
The correlation between the logarithm of 2FAL and DP in
sidered insignificant. They also claimed that the pro-
operating transformers has been difficult to establish—consider-
duction of 2FAL in oil would speed up, especially
able scatter was found on the correlation between the collected
after 15 years or so of service. This phenomenon of
DP value and the furan levels on the real transformers [23], [24].
accelerated generation was also observed in a labora-
This uncertainty may be caused by many factors such as a dif-
tory experiment conducted by Emsley et al. [17], who
ferent paper-to-oil ratio, partitioning of furan in paper and oil,
reported that the most significant rise in 2FAL level oc-
which may depend on the temperature, oil quality, and water
curred when DP had deteriorated to below 400.
concentration in oil, or possible decomposition of furans in the
B. Xue [14] collected 2FAL test results on 77 step-up
presence of dicy, oxygen and high temperature, and oil process-
transformers in power plants from northern China, with
rated voltages ranging from 100 to 500 kV. Through re-
Also, because the measured furan level in oil is an indicator
gression analysis of these data, the following equation
of the average paper degradation, it cannot indicate the degrada-
to calculate the mean values of 2FAL for transformers
tion at the worst-case location. In addition, the pattern of aver-
with different service years was obtained: log10(2FAL)
age paper degradation will vary from one transformer design to
= −1.83 + 0.058 × t, where 2FAL is expressed in milli-
another. A transformer with moderate levels of overall degrada-
grams per liter and t is the number of years in operation
tion can have the same 2FAL concentration in the oil as another
of the transformer. The upper and lower 99.5% confi-
unit with generally low overall levels but also a small area with
dence bounds of 2FAL level were obtained from these
severe deterioration. Similar transformers may be operated at
two equations: log10(2FAL) = −1.29 + 0.058 × t and
different average and different peak loading throughout their
log10(2FAL) = −2.37 + 0.058 × t. The results are plotted
lifetimes. It should be expected that the insulation would age
in Figure 11.
differently in these units, and therefore the furan content would
C. Researchers [19], [26] published several papers shar-
be different. Thus, an accurate correlation between DP and the
ing their experience with furan analysis in field trans-
furan concentrations in oil for field transformers is difficult to
formers. In 1998, a survey was conducted on the fu-
ran test results over the three-year period from 1994
through 1997 [26]. All small pole-top transformers
Statistical Analysis of Furan Data were excluded from the survey. In the survey, most of
Many have attempted to conduct statistical analyses on furan the samples were taken from mineral oil. However, the
data sampled from field transformers. The objective is to ob- sources of the transformers were mixed, including the

March/April — Vol. 28, No. 2 15

United States, Argentina, Australia, Brazil, Canada, In- spot is the critical area for furan production due to the fact that
donesia, Mexico, Peru, Saudi Arabia, and South Africa. it is the location of the highest winding temperature and thus the
The data are the total furan values, which include all of most favorable region regarding thermal degradation. Having a
the five commonly used furans: 2FAL, 5H2F, 5M2F, good approximation of the hot-spot temperature and knowing
2FOL, and 2ACF. They found the following distribu- the transformer typical load and load profile may shed new light
tion: 75% of the data were 100 ppb (weight), 87% were onto the interpretation of furans production and may also help in
250 ppb, 96.5% were 1,000 ppb, and 98.5% were 2,500 clarifying some differences that may occur in practice.
ppb. Another point that should be taken into account is the varia-
D. Statistical studies reported by Lewand [27] showed that tion in transformer design. Given two transformers operat-
90 to 95% of samples from transformers with thermally ing under the same conditions, side by side, but with different
upgraded insulation had less than about 100 μg/L of manufacturers and designs, it is common that these units would
2FAL. This applied to both mineral oil and silicone behave differently regarding thermal aspects. Most often indi-
transformers. vidual windings in a given transformer will show different longi-
E. A statistical survey reported by De Pablo and Pahla- tudinal temperature gradients, thus producing different average
vanpour [12] with the furan samples taken from 5,005 winding temperatures under the same loading condition. These
transformers in service in five European countries variations in temperature gradient may be even more significant
showed that there was a dependence of the furan level when comparing transformers with different manufacturers and
on rated capacity, rated voltage, and age, which may be designs. Considering that furans are produced by the thermal
used to explain the uncertainty of correlation between aging of the cellulosic insulation, it is then expected that dif-
DP and 2FAL for field transformers. The 95 percentile ferent windings would contribute differently to the occurrence
of the samples is located in the range of 1,000 to 5,000 of furans. Put simply, a cooler winding should be expected to
ppb depending on the source of the samples. produce less furan than would a warmer winding.
F. A survey of 2FAL values on 127 power transform- Similarly, and as another example to illustrate the issue of
ers at generating power plants located in Colombia, design dependency, a great number of small to medium power
South America, was conducted in 1997. Most of the transformers were built in the USA during the 1960s and 1970s
units were free breathing with rated voltage levels with enamel insulation on the low-voltage winding and paper
from 13.8 to 230 kV, and the age of the units ranged on the HV (this has always been typical for distribution trans-
from 1 to 42 years. The 90 percentile value of the sur- formers). If they are compared with similarly rated transform-
veyed samples was 1,706 μg/L and the 95 percentile ers built with paper on both windings, one would expect that
was 2,057 μg/L. The survey also concluded that the the furan concentrations in the enamel-insulated winding to be
average annual generation rate of 2FAL for 90% of significantly less, because there may be only one-half to about
the surveyed units was 64 μg/L per year. Another ob- two-thirds as much paper to age and produce the furans. Yet
servation from this survey was that the concentration with similar loading histories, the DP of the paper insulation
of 2FAL in oil was found to be somewhat dependent winding should be similar. Thus, a comparative analysis of the
on the service age [28]. two types of transformers may well indicate similar DP but
very different amounts of furans produced over a given operat-
Notice that the European studies of operating units indicated ing time.
a clear furan dependence of voltage and power ratings, along Typical ambient temperature is also a matter of consideration.
with the application of the transformer, whereas the laboratory Some investigators believe, based on experience and significant
models would supposedly apply to any given transformer. There databases, that transformers operating in warm environments,
is no consistency in the published results that could lead to a like those in tropical countries, would produce more furans that
recommendation to use any one of these methods. transformers operating in subtropical conditions. This is still
a matter of investigation, and it is brought up here as food for
Technical Issues and Limitations: The rate of furan production also seems to be dependent on
Parameters That May Affect Furan the operating temperature and on the aging itself, as has been
Production indicated by laboratory studies [6]. Those results showed faster
Researchers have been trying, for many years now, to estab- furan production at higher temperatures than at lower tempera-
lish a reliable correlation between furan production and the ac- tures for new paper insulation. In reference [6], the experiment
tual transformer age or, at least, the correlation between furan results show at more advanced aging (DP lower than 500), the
and the thermal degradation of the solid insulation measured by rate of furan production increases and then decreases when DP
its DP on actual transformers. Although several papers have re- approaches 200.
ported that it is applicable to laboratory accelerated aging tests, Other parameters such as insulation type, moisture content
several difficulties are hindering attempts to achieve that corre- of the insulation, amount of oxygen in contact with the insula-
lation for operating transformers. tion, presence of acids or other contaminants, furans partition
The most obvious parameter is the typical hot-spot tempera- between paper and oil, oil treatment such as in degassing and
ture for any given transformer. It is widely accepted that the hot drying out, as well as oil reclamation may play a significant role

16 IEEE Electrical Insulation Magazine

in the rate of the furan production. These parameters need to be • Is there any fundamental design or manufacturing dif-
analyzed and factored into future evaluation of furan analysis. ference between these units? Might it be relevant to fu-
The use of furfural in the oil-refining process can leave some ran production?
residues in the oil [3], [6]. The ASTM standard for new mineral • How similarly are the levels of 2FAL changing with
oil (D3487) does not have a limit for the concentration of furanic time? (This allows us to ignore for the moment the dis-
compounds in the new oil. It does include a note indicating that cussion of “normal” and “abnormal” levels.)
the maximum allowable levels should be specified by agreement • Is there any deviation from the reference or baseline
between the user and supplier. Therefore, if the transformer user measurements?
wants to monitor the furan concentration in the oil, this baseline • When a change in 2FAL concentration is noticed:
value should be measured and taken into account. (Note: IEC ○ Has any operational change been implemented
60296-2003 requires that unused insulating oils have a low level that might justify the differences, for example, a
of 2FAL and related compounds, and they are limited to a maxi- change in operating temperature or load?
mum 0.1 mg/kg.) ○ Has there been any failure issue (such as dis-
solved-gas analysis evidence), any oil treatment,
How to Use Results From Furan Testing or degassing?
As extensively discussed above, it is clear that although it re-
mains a very promising tool, so far the application of furans for After responding to all of these questions, one has much
the estimation of power transformer loss of life or actual aging more knowledge and information to proceed to a broader in-
is limited to extreme values. In general, units with high levels of vestigation, using the limited data available, and compare it, for
furans have experienced significant insulation degradation, and example, with statistical limits and references produced by any
those with low levels have little deterioration. Analyzing the rate other investigative work (e.g., Certified Laboratories, IEEE ref-
of furan production will guide the investigator as to the degrada- erences, and CIGRE). It is only then that conclusions can be
tion rate. This can be compared with the load profile and loss-of- made on a preliminary indication about the “normality” or “ab-
life estimations to determine whether the insulation degradation normality” of the test results at hand.
is exceeding predicted levels, which could indicate a thermal Finally, elevated levels of 2FAL or any of the other furan
problem involving the insulation system. compounds should not be used as the sole parameter when
As values become less extreme, it becomes much more dif- evaluating or analyzing a transformer oil sample. Other param-
ficult to estimate the actual relationship between furan levels and eters from the oil screen test and dissolved-gas analysis should
the actual degradation of the solid insulation in actual transform- be evaluated along with the presence of furans when trending a
ers. However, it is possible to compare the test results of two transformer’s condition or making a decision whether to replace
transformers and draw conclusions about the relative extent of a transformer.
their aging. The concentration of the furans in the oil of a particular trans-
Common sense and experimental investigations must guide former at any point in time is dependent on the content at a prior
us throughout the analytical process of interpreting the levels of date, the furans produced by the deterioration of the cellulose
furans found in a given transformer. For example, a transformer insulation during the intervening time period, the partitioning
that is known to have been manufactured with thermally upgrad- of the furans between the oil and the paper, degradation of some
ed paper is not expected to produce the same amount of furans as furan molecules (due to the stability issues) during that time in-
a similar transformer that was manufactured with normal Kraft terval, and probably several other factors.
paper and subjected to the same load and ambient conditions. Two actions that may affect the furan concentration of a
Transformers that operate routinely at normal high tempera- transformer are oil reclamation (treatment with Fuller’s earth)
tures (typically transformers that operate at or above nameplate and oil treatment such as in degassing and drying out. It is well
ratings) should be expected to contain relatively higher furan known that reclamation will strip all of the furans from the treat-
levels than similar units that operate almost constantly at light ed oil. When the oil with zero furan content is reintroduced to a
loads. Therefore, any comparison must take operating tempera- transformer, the partitioning of the furans in the insulation and
ture into account. oil will attempt to reach equilibrium, and then after a period of
Whatever the statistical references or mathematical models time, the observed concentration in the transformer oil will in-
used as reference on the furan levels, all of the differentiating crease. Therefore, when attempting to trend the furan concentra-
factors mentioned earlier must be considered. One way that tion and total the furans produced over time, an owner will need
might reduce the limitation in the applicability of the state-of- to attempt to make some adjustments in the data to account for
the-art models is continuous and systematic follow-up of the the furans removed during the oil reclamation.
evolution of furans for a particular fleet of transformers, when Clearly, additional industry-wide accumulation and evalua-
possible. tion of test result data on furans, transformer operating condi-
Take the example of 2FAL only. Suppose one is measuring tions, and paper sample test results are necessary before furan
2FAL every six months for a group of three similar units under testing can be used as the final determination of the condition
similar operating conditions and identical maintenance proce- and age of the insulation paper of a transformer.
dures. Consider the following questions:

March/April — Vol. 28, No. 2 17

Recommendations for Data Collection Table 1. Suggested Transformer Information Form.
From all the discussion and technical material presented in
this article, the word “consistency” is the one that is strikingly
absent. This IEEE Transformers Committee Furan Task Force S/N  
recognizes that there is still a need for data collection—particu- Maximum MVA  
larly better quality data that includes the possible influencing HV RATING (KV)  
factors on 2FAL and other furan production. Thus, a reliable da-
tabase should include the following simultaneous information:
• Test results for furans (indicate mg/kg or μL/mL), Year of manufacture  
• Test results for moisture in oil, Transformer type CF = core form
• Test results for O2, CO, and CO2, and   SF = shell form
• The typical operating temperature of the transformers
Preservation system type S = sealed
(at least top oil indication).
  O = open
The database should also provide information about the °C rise 55 65 IEC
transformer design and construction (see Table 1 for a suggested Transformer use GSU
comprehensive information list). Loading profile (Figure 12)
  T = transmission
and application should also be provided. The user should indi-
cate the typical loading profile for this transformer by providing   D = distribution
the two-letter code from the examples below. Choose the option   I = industrial
that best approximates the typical loading cycle for this trans- Insulating liquid MO-I = mineral oil inhibited
  MO-UI = mineral oil uninhibited
  NE = natural ester
Summary and Conclusions
This article has attempted to capture the history and latest   SE = synthetic ester
technical information regarding the formation of furans as trans-   S = silicone
former paper ages and degrades over time and the analysis of Liquid volume Gallons
testing. 2-Furaldehyde is always considered the main compo- LTC None
nent of the five furan compounds because of its stability and
relatively higher rate of formation in transformers.
A great deal of research has been performed and data gath-   Noncommunicating
ered to evaluate the correlation of 2FAL to DP as a measure Loading profile GB
to predict the remaining life of transformer insulating paper.   GC
Several equations describing DP as a function of log(2FAL) for
Kraft paper have been developed. Other equations for thermally
upgraded paper exist. There is considerable variability among   TDS
the data, and therefore, we find great differences in the proposed   Other
correlation equations. Operating age at the time of sampling  
There are many variables that can affect the rate of forma- Sample date (mm/dd/yyyy)  
tion of 2FAL including temperature, transformer design, type
Top oil temperature at sampling °C
of paper, presence of varying amounts of moisture and oxygen,
partitioning of furans between oil and paper, and maintenance   °F
activities including oil reconditioning or reclamation. Moisture concentration in oil ppm
Because of the wide range of variability among the furan and Oxygen concentration in oil ppm
DP data, the authors of this article cannot endorse any of the
Units mg/kg
equations proposed by various researchers or any of the meth-
ods for correlating 2FAL to DP. As demonstrated throughout this   μg/L
article, there is no single model that applies to all transform-   2-Furaldehyde (2FAL)  
ers. Transformer owners are warned to exercise caution in the   5-Methyl-2-furaldehyde (5M2F)  
application of any of these furan to DP correlation models and   5-Hydroxymethyl-2-furaldehyde (5H2F)  
to weigh any conclusions that might result from that analysis
  2-Acetyl furan (2ACF)  
against all other available technical information and testing re-
sults. It would not be prudent to take any actions based primarily   2-Furfurol (2FOL)  
on the results of furan tests. Historical oil processing Reconditioning
Although laboratory testing has provided a fairly consistent   Reclamation
correlation between aging and furan production, data sampling

18 IEEE Electrical Insulation Magazine

Figure 12. Examples of loading profiles with corresponding codes. (a) Generation or industrial base
load (coded as GB), (b) generator or industrial cyclical load (coded as GC), (c) T&D or industrial
cyclical load (coded as TDC), and (d) T&D or industrial cyclical with spike (coded as TDS).

from actual in-service transformers has not provided data that transformers with the furan test results, and then with tests of
will correlate the two. Again because of the variability of furan actual insulation removed from these units, are essential. Only
test results due to differences in design and application, none of after establishing that correlation can the transformer industry
the methods reviewed have provided a set of threshold values gain the experience and history required to allow furan testing to
or ranges of results that will allow classification of the extent of become a reliable and valuable tool for determining the degree
aging that apply to all transformers. of paper insulation deterioration in a particular transformer.
The best available analysis technique appears to be compar- As a final remark, it may be of interest to also look into other
ing relative furan test results between transformers that have insulation aging traces and possible correlations with furans
similar design and construction and similar service histories. as well as the actual deterioration of the solid insulation. One
The industry expectations must be managed to recognize that chemical indicator of aging that seems to be a serious candi-
perhaps the only valid conclusion could be that one of the units date is methanol [29], [30], although further investigation is still
in this comparison appears to be aging at a faster rate than the needed to clarify a number of issues, such as the effect of oil ag-
other. ing and nitrogenous substances on the occurrence of methanol,
Continued thermal aging tests while controlling the factors in order to provide a reliable indicator.
that affect the aging rate, and gathering of paper samples, along
with furan data, should help the industry to establish a more de- Acknowledgments
finitive relationship between furans, DP, and overall transformer The authors would like to thank the IEEE Transformer Com-
paper aging characteristics. Laboratory tests to examine furan mittee for its support of the Task Force of Furan Analysis for
behavior in a full range of degradation states, from around 1,000 Liquid Power Transformers and the following individuals for
DP up to 200 DP level, for 100% thermally upgraded insula- their valuable suggestions: Brian Penny, Jocelyn Jalbert, Jos
tion paper would enhance the understanding of this technology. Veen, Joseph Kelly, Juan Castellanos Gonzalez, Mike Lau, and
However, data to correlate the operating conditions of in-service Paul Boman.

March/April — Vol. 28, No. 2 19

References Procedure for Thermal Evaluation of Liquid Immersed Distribution and
[1] A. De Pablo, “Furfural and ageing: How are they related,” in IEE Col- Power Transformers,’” IEEE Transformer C57-100 Task Force Meeting,
loquium on Insulating Liquids, 1999, pp. 5/1–5/4. Mar. 18, 2008.
[2] J. Unsworth and F. Mitchell, “Degradation of electrical insulating paper [23] N. Dominelli, J. V. Landa, and E. A. Hall, “The analysis of furanic and
monitored with high performance liquid chromatography,” IEEE Trans. non-furanic compounds as a transformer diagnostic technique,” presented
Electr. Insul., vol. 25 no. 4, pp. 737–746, 1990. at the EPRI Substation Equipment Diagnostic Conference II, Palo Alto,
[3] P. J. Griffin, E. Finnan, and R. Lewand, “Case studies,” in Proceedings of CA, 1994.
the 1996 International Conference of Doble Clients, 1996, Sec 5-4. [24] J. T. Holboll and T. R. Blackburn, “General report for group 15,” Cigre
[4] D. H. Shroff and A.W. Stannett, “A review of paper aging in power trans- 1999 Report, 1999.
formers,” in IEE Proceedings C: Generation, Transmission and Distribu- [25] M. Dong, Y. Shang, and Z. Zheng, “Aging diagnosis of solid insulation
tion, vol. 132, no. 6, pp. 312–319, 1985. for large oil-immersed power transformers,” in 2002 Annual Report
[5] P. J. Griffin and L. R. Lewand, “Paper degradation by-products generated Conference on Electrical Insulation and Dielectric Phenomena, 2002, pp.
under incipient-fault conditions,” in Proceedings of the 1994 Interna- 400–403.
tional Conference of Doble Clients, 1994, Sec 10-5. [26] J. R. Sans, K. M. Bilgin, and J. J. Kelly, “Large scale survey of furanic
[6] A. M. Emsley, X. Xiao, R. J. Heywood, and M. Ali, “Degradation of compounds in operating transformers and implication for estimating
cellulosic insulation in power transformers. Part 2: Formation of furan service life,” in Conference Record of the 1998 IEEE International Sym-
products in insulating oil,” in IEE Proceedings Science, Measurement posium on Electrical Insulation, Jun. 1998, pp. 543–553.
Technology, vol. 147, no. 3, pp.110–114, 2000. [27] L. R. Lewand, “Practical experience gained from furanic compound
[7] D. M. Allan and C. F. Jones, “Thermal-oxidative stability and oil-paper analysis,” presented at the Seventy-Third Annual International Doble Cli-
partition coefficients of selected model furan compounds at practical ent Conference, MA, USA, 2006.
temperature,” presented at the Ninth International Symposium on High [28] E. Finnan, P. J. Griffin, E. Zuleta, A. Matamoros, A. Ciendua, and M. P.
Voltage Engineering, Graz, Austria, 1995. Diaz, “A report on the assessment of insulation aging and condition by
[8] P. J. Griffin and L. R. Lewand, “Cases studies,” in Proceedings of the means of laboratory tests,” in Proceedings of the Sixty-Fourth Annual
Sixty-Fourth Annual International Conference of Doble Clients, 1997, International Conference of Doble Clients, 1997, Sec 5-5.
Sec 5-7. [29] J. Jalbert, R. Gilbert, P. Tetreault, B. Morin, and D. Lessard-Deziel,
[9] A. B. Shkolnik, K. M. Bilgin, and J. J. Kelly, “Creating a preliminary “Identification of a chemical indicator of the rupture of 1,4-β-glycosidic
model for estimating degree of polymerization of thermally upgraded bonds of cellulose in an oil-impregnated insulating paper system,” Cel-
insulation paper based on furan concentration in transformer oil,” in lulose, vol. 14, no. 4, pp. 295–309, 2007.
Proceeding of the Sixty-Sixth Annual International Conference of Doble [30] R. Gilbert, J. Jalbert, S. Duchesne, P. Tetreault, B. Morin, and Y. Denos,
Clients, 1999, Sec 5-8. “Kinetics of the production of chain-end groups and methanol from the
[10] I. Hohlein and A. J. Kachler, “Aging of cellulose at transformer service depolymerization of cellulose during the ageing of paper/oil systems, Part
temperature. Part 2: Influence of moisture and temperature on degree of 2: Thermally-upgraded insulating papers,” Cellulose, vol. 16, no. 2, pp.
polymerization and formation of furanic compounds in free-breathing 327–328, 2009.
systems,” IEEE Electr. Insul. Mag., vol. 21, no. 5, pp. 20–24, 2005.
[11] Ageing of Cellulose in Mineral-Oil Insulated Transformers. Report
prepared by CIGRE TF D1.01.10. Thomas A. Prevost is primary manager
[12] A. De Pablo and B. Pahlavanpour, “Furanic compounds analysis: A tool for OMICRON Electronics USA. Prior
for predictive maintenance of oil-filled electrical equipment,” Electra,
to joining the OMICRON team, Prevost
vol. 175, pp. 9–18, 1997.
[13] A. M. Emsley and G. C. Stevens, “Review of chemical indicators of deg- worked for 25 years at Weidmann Electri-
radation of cellulosic electrical paper insulating in oil-filled transform- cal Technologies, a manufacturer of insu-
ers,” in IEE Proceedings: Science, Measurement and Technology, vol. lation materials and provider of diagnostic
141, no. 5, pp. 324–334, 1994.
services for the transformer industry. Ear-
[14] C. D. Xue, “Monitoring paper insulation aging by measuring furfural
contents in oil,” in 7th International Symposium on High Voltage Engi- ly in his career Prevost worked at Tampa
neering, 1991, pp.139–142. Electric Company as an engineer in distri-
[15] R. Heywood and B. Pahlavanpour, “Analysis of formaldehyde content of bution and production. Prevost received his
the oil and its diagnostics value for transformer aging,” in 14th Interna-
BSEE from Virginia Polytechnic Institute. He is a senior mem-
tional Power System Conference, 1999, pp. 99–109.
[16] B. Pahlavanpour, M. A. Martins, and A. De Pablo, “Experimental inves- ber of IEEE. He is vice-chair of the P2030 “Smart Grid” work-
tigation into the thermal-ageing of Kraft paper and mineral insulation ing group. He is the past-chair of the IEEE PES Transformers
oil,” in Conference Record of the 2002 IEEE International Symposium on Committee. He is a member-at-large of the IEEE PES Board of
Electrical Insulation, 2002, pp. 341–345.
Governors. Prevost is also active in ASTM committee D27 on
[17] T. K. Saha, “Review of modern diagnostic techniques for assessing
insulation condition in aged transformers,” IEEE Trans. Dielectr. Electr. Insulating Fluids and IEC TC10 on Insulating Fluids. He has
Insul., vol. 10, no. 5, pp. 903–916, 2003. written many technical papers on the subject of electrical insula-
[18] L. Cheim and C. DuPont, “A new transformer aging model and its cor- tion materials, transformer diagnostics, and condition monitor-
relation to 2FAL,” presented at the International Cigre SCA2 Colloquium,
Merida, Mexico, 2003.
[19] R. D. Stebbins, S. D. Myers, and A. B. Shkolnik, “Furanic compounds in
dielectric liquid samples: Review and update of diagnostic interpretation
and estimation of insulation aging,” presented at the 7th International
Conference on Properties and Applications of Dielectric Materials, Na-
Shuzhen Xu works as a senior research
goya, Japan, Jun. 2003.
[20] S. D. Myers, Tech Corner, 2009. Available: specialist at FM Global. Her responsibili-
corner/tech-2009-06-19-furanic.html. ties include risk and reliability assessment
[21] L. E. Lundgaard, W. Hansen, D. Linhjell, and T. J. Painter, “Aging of oil- for HV equipment such as transformers,
impregnated paper in power transformers,” IEEE Trans. Power Del., vol.
generators, and industrial motors. She has
19, no. 1, pp. 230–239, 2004.
[22] R. Wicks and T. Prevost, “DP vs. 2FAL as part of thermal aging tests for more than seven years of experience in fail-
Kraft and thermally upgraded paper for C57.100—‘IEEE Standard Test ure root cause analysis, diagnostic testing

20 IEEE Electrical Insulation Magazine

data analysis, and condition evaluation and reliability assess- Luiz Cheim joined ABB TRES North
ment for HV equipment. She has published numerous technical America in August 2009 as a consulting
papers on the subject of HV equipment reliability assessment. R&D engineer to support the Transformer
Xu is an active member of the IEEE Transformer Committee Condition Assessment program through
and is the chair of the task force on furan analysis for liquid advanced dissolved-gas analysis, trans-
power transformers. She received her PhD from Shanghai Jiao- former aging analytical tools, as well as
tong University in China. online monitoring and diagnostic systems.
Cheim is part of an ABB global R&D or-
ganization that works on the development
of those tools to support the customer’s efforts in maintaining
Donald Platts is presently a senior staff critical assets through the new Smart Grid Initiatives.
engineer in Transmission and Substation Before joining ABB, Cheim spent 10 years with Siemens
Standards Engineering with PPL Electric Transformers, later working in the United States with the Sie-
Utilities. He is responsible for most engi- mens Service Solutions Division (SSD/USA) based in Wendell,
neering activities involving transformers: NC. Cheim was the mentor for the development of the Sie-
preliminary studies and selection; purchase mens Transformer Online Monitoring and Diagnostics System
projects including writing specifications, (TMDS-L) for legacy transformers in the Americas. He was also
performing inspections, and witnessing part of the Siemens Asset Management Services team under the
testing; maintenance and operation sup- Siemens Smart Grid Programs.
port; dissolved-gas analysis; loading stud- Previously Cheim spent 18 years with the Center for Electric
ies; failure analysis; and equipment repair processes. His experi- Power Research (CEPEL) in Rio de Janeiro, Brazil, where he
ence with PPL includes 21 years in various engineering groups: started as a high voltage/high power test field engineer at the
Standards Engineering, System Maintenance Engineering, beginning of his career and achieved the position of senior R&D
Substation Component Engineering, and so on. In each of these and project coordinator with large utilities in South America.
positions, he has held similar responsibilities for transformer-re- During that period Cheim had the opportunity to work in co-
lated issues. For 16 years prior, he worked in the Substation En- operation with NIST/USA in Gaithersburg, MD, (Dr. Richard
gineering design group, where he was responsible for substation Van Brunt on “Memory Propagation Effects of PD”) and also
design projects and transformer procurement activities. He is an spent time as a visiting researcher at the Los Alamos National
active member of the IEEE Transformers Committee, where he Lab (New Mexico) in the Superconductivity Laboratory.
presently serves as the secretary of the committee. Prior to as- Cheim has been an active member of Cigre Paris since 1984,
suming his present duties, among other activities, he served for having acted as chairman of the Study Committee A2-Trans-
10 years as the chair of the Insulation Life Subcommittee. He formers in Brazil from 2000 to 2006. Cheim was awarded the ti-
has a BSEE degree from Lafayette College and is a Registered tle Cigre Distinguished Member and also given the Outstanding
Professional Engineer in Pennsylvania. Contribution Award by the Cigre Technical Committee, Paris,
in 2006. Cheim is an active member of the IEEE Transformer
Committee and holds a PhD in electrical electronic engineering
from the University of Nottingham, UK, as well as an MSc and
BSc in electrical electronic engineering from the Federal Uni-
versity of Rio de Janeiro, Brazil.

March/April — Vol. 28, No. 2 21