Sie sind auf Seite 1von 71

494

Furanic Compounds for Diagnosis

Working Group
D1.01 (TF13)

April 2012
Furanic compounds for diagnosis

Working Group
D1.01 (TF 13)

Members

Marie-Claude Lessard (CA) (Convener), Ivanka Atanasova Hhlein (DE), Teresa Buchacz (PL),
Luiz Cheim (US), Steve Eeckhoudt (BE), Barry Finlay (AU), Maja Koncan-Gradnik (SI),
Harald E. Haehre (NO), Anne Marie Haug (NO), Richard Heywood (GB), Hans Josef Knab (CH),
Riccardo Maina (IT), Maria Augusta Martins (PT), Tom A. Prevost (US), Yongyuth Vachiratapadom (TH),
Julie Van Peteghem (BE)

Copyright 2012
Ownership of a CIGRE publication, whether in paper form or on electronic support only infers right of use for
personal purposes. Unless explicitly agreed by CIGRE in writing, total or partial reproduction of the publication
and/or transfer to a third party is prohibited other than for personal use by CIGRE Individual Members or for use
within CIGRE Collective Member organisations. Circulation on any intranet or other company network is forbidden
for all persons. As an exception, CIGRE Collective Members only are allowed to reproduce the publication.

Disclaimer notice

CIGRE gives no warranty or assurance about the contents of this publication, nor does it accept any
responsibility, as to the accuracy or exhaustiveness of the information. All implied warranties and
conditions are excluded to the maximum extent permitted by law.

ISBN: 978-2-85873-186-2
TABLE OF CONTENTS
Page

1 Summary ................................................................................................................................ 4
2 Introduction ........................................................................................................................... 5
2.1 Background and ambition of work .................................................................................. 5
2.2 Origin of the furanic compounds ..................................................................................... 5
2.3 Furanic compounds detection .......................................................................................... 7
3 Analytical concerns ............................................................................................................... 8
3.1 Introduction...................................................................................................................... 8
3.2 Participant laboratories .................................................................................................... 8
3.3 Description of the RRT samples ...................................................................................... 9
3.4 Results of RRT .............................................................................................................. 10
3.4.1 Furanic compounds analysis ........................................................................... 10
3.4.2 Degree of polymerisation of insulating paper ................................................. 13
3.5 Results and Recommendations ...................................................................................... 15
3.5.1 Furanic compounds analysis ........................................................................... 15
3.5.2 Measurement of Average DPv of insulating paper ......................................... 16
4 Factors influencing furanic compounds concentration ................................................... 17
4.1 Introduction.................................................................................................................... 17
4.2 Partition effect................................................................................................................ 17
4.3 Factors influencing the 2-FAL formation ...................................................................... 19
4.3.1 Temperature .................................................................................................... 19
4.3.2 Moisture .......................................................................................................... 20
4.3.3 Oil type and oil condition ............................................................................... 21
4.3.4 Design: free breathing or closed type transformers ........................................ 21
4.3.5 Insulation paper type: plain Kraft paper or thermally stabilized Kraft
paper ....................................................................................................................... 21
4.4 Stability .......................................................................................................................... 22
4.5 Influence of maintenance ............................................................................................... 24
5 Diagnosis methods ............................................................................................................... 27
5.1 Correlation with degree of polymerization .................................................................... 27
5.2 Presence of the other related furanic compounds .......................................................... 28
5.3 Critical thresholds .......................................................................................................... 29
5.4 Ratios of different furanic compounds .......................................................................... 30
5.5 Residual life estimation ................................................................................................. 30
6 Statistical study .................................................................................................................... 38
6.1 Universal values: are they possible? .............................................................................. 38
6.2 Typicals values .............................................................................................................. 39
6.3 Conclusions from statistical study ................................................................................. 41
7 Diagnostic experience .......................................................................................................... 42
7.1 Survey ............................................................................................................................ 42
7.1.1 Introduction ..................................................................................................... 42
7.1.2 Methodology ................................................................................................... 42
7.1.3 Results ............................................................................................................. 43
7.1.4 Conclusion ...................................................................................................... 47

Page 2 of 70
7.2 Case Studies ................................................................................................................... 48
7.2.1 Case 1: Local hot spot wrapped by cellulose insulation ............................. 48
7.2.2 Case 2: General ageing of cellulose insulation caused by insufficient
cooling .................................................................................................................... 49
7.2.3 Case 3: As complimentary test to DGA in paper overheating ........................ 50
7.2.4 Case 4a & b: Different 2-FAL results in two transformers with similar
internal faults assessed by DGA............................................................................. 52
7.2.5 Case 5: Restoring of 2-FAL after oil changing; accelerated process of
cellulose degradation mutually confirmed by DGA and 2-FAL results ................ 54
7.2.6 Case 6: Restoring of 2-FAL after repair ......................................................... 55
7.2.7 Case 7: Restoring of 2-FAL after oil reclaiming; mutually confirming
CO2, CO2/CO and 2-FAL results ........................................................................... 57
8 Conclusion ............................................................................................................................ 59
8.1 Conclusions.................................................................................................................... 59
8.2 Recommendations.......................................................................................................... 60
Annex 1: Summary of methods used to detect furanic compounds in oil .................................... 61
Annex 2: Round robin test additional results ................................................................................62
Annex 3: Literature review of the statistical populations retained for the study of 2-FAL in
transformers .................................................................................................................................. 63
Annex 4: CIGRE WG D1.01 (TF 13) TRANSFORMER Oil Lab Survey ...................................66
References .................................................................................................................................... 68

Page 3 of 70
1 SUMMARY

The analysis of furanic compounds in the insulating oil of power transformers is used to
characterize the thermal degradation of the solid paper insulation in transformers. In this
Brochure, different aspects (analysis, factors of influence, diagnosis methods, statistical
approach and users experiences) have been considered to make the interpretation easier. The
formation, detection and behavior of these components in a transformer are described. A
survey was conducted to present how these compounds are currently being used by utilities.
Finally, real cases are given which demonstrate the usefulness and limitation of the
technique.

Page 4 of 70
2 INTRODUCTION

2.1 BACKGROUND AND AMBITION OF WORK

Since the mid-1980s, the analysis of furanic compounds in the insulating oil of power
transformers has been used to characterize the thermal degradation of the solid paper insulation.
The production of these compounds, particularly the 2-furfural, is known to depend on several
factors such as moisture content, temperature, type and quantity of paper involved, etc.).
Interpretation rules are needed to ensure the accuracy of the diagnosis and to determine whether it
is possible to extend the technique to general condition assessment.

This report discusses the use of furanic compounds analysis in insulating oil as a diagnostic tool
for solid insulation of power transformers. Different aspects have been considered to make the
interpretation easier. The formation, detection and behavior of these components in a transformer
are described. We also, by means of a survey, present how these compounds are currently being
used by utilities.

Different interpretation techniques are presented and discussed. Finally, real cases are given
which demonstrate the usefulness and the limitation of the technique.

2.2 ORIGIN OF THE FURANIC COMPOUNDS

Furanic compounds are generated during the thermal degradation of the insulating paper used in
electrical equipment such as transformers. Solely related to the degradation of the paper and
detectable in oil, furanic compounds are able to provide information on the condition of the
insulation without requiring invasive work or de-energizing of equipment.

In 1984, Burton et al. [1] suggested that the analysis of the six furanic compounds:

- 2-furfural (2-FAL).
- 2-acetylfuran (2-ACF).
- 2-furoic acid.
- 5-methyl-2-furfural (5-MEF).
- 2-furfurylalcohol (2-FOL).
- 5-hydroxymethyl-2-furfural (5-HMF).

detected in oil could be used to characterize the thermal decomposition of the cellulosic
insulating paper at temperatures as low as 100C. The chemical structure of the six furanic
compounds observed at the time under various ageing conditions is presented in Figure 1.

Page 5 of 70
2-FURFURAL (2-FAL)
2-ACETYLFURAN (2-ACF)
2-FURALDEHYDE
2-FURYL METHYL KETONE
2-FURFURALDEHYDE

5-METHYL-2-FURFURAL (5-MEF)
2-FUROIC ACID
5-METHYL-2-FURALDEHYDE

2-FURFURYLALCOHOL (2-FOL) 5-HYDROXYMETHYL-2-FURFURAL (5-


FURFURYLALCOHOL HMF)
2-FURFUROL 5-HYDROXYMETHYL-2-FURALDEHYDE

Figure 1: Chemical structure of furanic compounds.

It has been proposed that furanic compounds may primarily originate from the degradation of
hemicellulose rather than cellulose. According to several authors [2],[3], this could decrease their
potential for use given that hemicellulose is the most fragile part of the paper and is not really
representative of insulating paper (it accounts for 10-20% of the latters composition). However,
the degradation mechanisms proposed at temperatures between 100 and 200C show that 2-
furfural (2-FAL) should also be generated through the 5-hydroxymethyl-2-furfural (5-HMF)
resulting from the cellulose.

Several mechanisms for the thermal degradation of cellulose have been proposed, depending on
whether the predominant step is pyrolysis or hydrolysis of the glucose units [4]. In the first case,
pyrolysis of glucose, the reaction proceeds via levoglucosan (1,6-anhydro--D-glucopyranose), a
dehydrated sugar which, after a series of bond rearrangements and further losses of water and
formaldehyde (or hydrogen plus carbon monoxide), yields 2-FAL, among other products. The
hydrolysis of glucose units proceeds through the formation of an epoxide or an enol followed by
internal re-arrangements and further dehydration and elimination of formaldehyde to yield
furfuraldehyde (2-FAL) and other products. Figure 2 shows the one of the most probable
mechanism.

Page 6 of 70
Figure 2: One of the most probable cellulose thermal degradation mechanism.

2.3 FURANIC COMPOUNDS DETECTION

Even though many laboratories around the word are reporting the presence of 5 of the 6
compounds described in Burtons first work (excluding the furoic acid), it is the 2-furfural (2-
FAL) compound that is detected in a higher concentration [5],[6].

Table 1 presented in Annex 1 lists the main detection techniques. The ones applying liquid
chromatography analysis [7] are most often used. Less complex and costly than gas or liquid
chromatography coupled with mass spectrometry, these methods allow furanic compounds to be
detected at thresholds close to the ppb range.

Page 7 of 70
3 ANALYTICAL CONCERNS

3.1 INTRODUCTION

As it will be presented later, in Chapter 7, it has been proposed by many authors that the degree
of polymerization of the cellulose can be related to the 2-furfural (2-FAL) detected in the oil. To
verify the reliability of the furanic compound analysis in mineral oil and the average degree of
polymerization of insulating paper (DPv), WG D1.01 (TF 13) prepared two different
interlaboratory tests.

3.2 PARTICIPANT LABORATORIES

13 laboratories participated to each Round Robin Test (RRT). Table 1 presents the list of
participant laboratories. The goal of these interlaboratory tests was to assess the precision, the
accuracy and the repeatability of those methods which are an integral part of a further general
diagnosis of the transformer paper insulation.

Table 1: RRT participant laboratories list.

Laboratory Countr y
ABB AS Division Power Products Sweden
Laborelec Belgium
Laboratoire Jeanne dArc Hydro Canada
Qubec
FKH Versuchsstation Switzerland
SIEMENS AG, Trafowerk Nuernberg Germany
Service LCIE France
Jeff Male Laboratories South Africa1
Elektroinstitut Milan Vidmar Slovenia
Labelec EDP Group Portugal
IREQ Hydro Quebec Canada
Aurecon Australia
Sea Marconi Technologies Italy2
Doble Power Test Ltd England
ZPBE Energopomiar-Elektryka Poland
1 2
Only DPv samples Only Furanic compounds samples

Page 8 of 70
3.3 DESCRIPTION OF THE RRT SAMPLES

The samples for the RRT on furanic compounds were prepared in February 2006 in Ljubljana
(Slovenia). Each participating lab received 50 ml glass bottles of sample A, B and C.

Sample A contained 0,30 mg/kg of each furanic compound (being 2-FAL, 5-HMF, 2-FOL, 2-
ACF and 5-MEF). These were prepared by using 300 l of stock solution with a concentration of
1000 mg/l mix of 5 furanic derivates (Gravimetric Cerificate, Dr. Ehrensdorfer, Germany). This
was diluted to 1 kg by adding new uninhibited transformer oil (Nynas Nytro3000) by use of
100_l HPLC Syringe (dilution error 1%). This solution had a RSD of 3%.

Sample B contained approximately 0,03 mg/kg 2-FAL. This was an oil sample from a
31,5_MVA, 110/20 kV transmission transformer (29 years in service) with free breathing
conservator containing 14000 kg naphtenic inhibited oil. Such transformers are found in
transmission with low loading, working at low winding temperatures (parallel operation of two
transformers in a station, reserve transformers etc). According to Slovenian experience, values
from 0,01-0,05 mg/kg 2-FAL are very common in an old but non-aged population.

Sample C contained approximately 1,5 mg/kg 2-FAL. This oil sample was taken from a
300_MVA, 400/115/31,5 kV transmission transformer (27 years in service) with free breathing
conservator, containing 75000 kg of Technol Y3000 oil. This inhibited oil was degraded because
of overheating in this OFAF (oil forced air forced cooling) transformer. The DPv on LV winding
exit sample was measured at 430 in October 2002 when 2-FAL was 1,03 mg/kg.

The specific characteristics of the furanic compound samples B and C are given in the Table 2.

Table 2: Characteristics of samples B and C .

Property Method Unit Measured value


Sample B Sample C
____________________________________________________________________________________________
Appearance IEC 60296 - Clear, free of sludge and sediments
Color ISO 2049 - 1 5
Density at 20oC ISO 3675 kg/dm3 0,896 0,875
o
Flash point (closed cup) ISO 2719 C 158 144
o
Refr. index at 20 C DIN 51423 - 1,4780 1,4810
Acidity IEC 62021 mg KOH/g oil 0,01 0,14
Interfacial tension at 25oC ASTM D971 mN/m 33 9
Water content in oil IEC 60814 ppm 5 13
o
Top oil temp at sampling C 24 38
Anti-oxidant additives IEC 60666 mass % 0,13 0,11
Breakdown voltage IEC 60156 kV 71 68
___________________________________________________________________________

Concerning the insulating paper samples for determining DPv-value, three (3) samples of
insulating Kraft paper have been sent for DPv with low, medium and medium-high values. The
samples were delivered by IREQ.

Page 9 of 70
The Samples 1 and 2 were traditional Kraft paper, dried and aged in oil in the laboratory (IREQ,
Hydro-Qubec, Canada). Homogeneity of the samples prepared were verified (DPv (average
value, n = 6) 522 and 722).The samples were sent impregnated with mineral naphtenic oil (Nynas
Nytro10CX, inhibited oil) to the participating labs.

Sample 3 came from a 19 years old 400 MVA, 400 kV generator transformer in a power plant in
Slovenia which was opened and renovated after being diagnosed as no more reliable. DGA
history was normal, 2-FAL had been increasing continuously (0,25 mg/kg). Early critical aging
(after 19 years top windings paper DP 300) was found to be the consequence of insufficient
cooling conditions (inadequate cooling settings), related to reduced top-oil temperature indication
in an OFAF cooled transformer. During opening of the transformer a paper sample from the 3rd
turn from the top of outer windings was collected and prepared (n-pentane washed, dried and
grinded in a coffee grinder) for the international RRT by Elektroinstitut Milan Vidmar, Slovenia.

3.4 RESULTS OF RRT

3.4.1 Furanic compounds analysis

The High Performance Liquid Chromatography IEC 61198-method (Mineral Insulating Oils -
Methods for the Determination of 2-Furfural and Related Compounds) is most often used (10/13
labs). This HPLC method is combined with liquid-liquid or liquid-solid extraction. Most labs
(8/10 labs) preferred the solid-liquid extraction (solid phase extraction-SPE technique) which is
described in method B of IEC 61198.

The method according to ASTM D5837 (Standard Test Method for Furanic Compounds in
Electrical Insulating Liquids by High-Performance Liquid Chromatography (HPLC)) was also
used (2/13 labs). This method was performed with direct oil injection and showed good
repeatability (according to the participating laboratories). One laboratory (1/13) is performing
GC/MS; this method is described in the ASTM D5837 as an alternative method.

The reported limit of detection varies from 1 g/kg to 50 g/kg. The reported repeatability of the
methods by the RRT-participants varies from 5% to 15% depending on the type of furanic
compound being measured.

The average, the standard deviation (STD), the relative standard deviation (RSD which is the
standard deviation/average) and the 95% Interval of Confidence of the expected value (IC95%)1
of all the furanic compounds detected in the 3 samples are given in the Table 3. The
reproducibility criterion according to IEC (ratio IC95%/average) is also presented in this table.

STD
1
IC 95% tn;0,95
n
where tn;0,95 is Student parameter for n values and 95% level of confidence (n is the number of results taken into account
after subtracting 0, 1 or 2 results of the excluded values as presented in Table 3)

Page 10 of 70
Table 3: Furanic compounds detected in the RRT oil samples.
N=13 Reference Lab. Results STD RSD (%) IC95% IC95% Excluded
material average mg/kg (mg/kg) /average* values**
mg/kg mg/kg (%)

2-FAL A (0,30) 0,27 0,03 9,7 0,02 6,3 1

2-FAL B 0,04 0,02 62,5 0,01 37,8 0

2-FAL C 1,54 0,06 3,9 0,04 2,5 1

5-HMF A (0,30) 0,25 0,03 12,0 0,02 8,6 2

2-FOL A (0,30) 0,21 0,06 25,6 0,04 19,3 1

2-ACF A (0,30) 0,28 0,03 9,3 0,02 6,6 2

5-MEF A (0,30) 0,25 0,02 6,8 0.01 4,5 2


* IEC reproducibility criteria ** Values < or > that average 2STD

For each sample, the average, standard deviation (1STD and 2STD) for the 2-FAL compound and
reproducibility according IEC 61198 (IC95%/average) (which is 15% for compound under
1 mg/kg and 10% over) for each sample are indicated in the graphs (see Figure 3).

Comparing the results for furanic compound analysis of the 3 samples, allows the following
observations to be made:

- All the methods used give reliable results except for the GC-MS method.
- Most of the labs are within the limits of one standard deviation and the IEC
reproducibility limits for the measurement of the 3 samples, except for lab 11 which
presented better results in a second trial. Lab 3 showed an important precision problem
particularly with the other furanic compounds. The five components were analysed with
about the same performance except for the 2-FOL (IC95%/average=19,3%). This
phenomenon is probably related to the fact that this compound usually shows a great
instability and interference with oil degradation by-products. Extended table results are
presented in Annex 2.
- The results for sample A, the only one where the 5 furanic compounds were present, are
very good. Lab 10, which is the only lab performing the analysis by Gas Chromatography
coupled with a Mass Spectrometer (GC-MS) experienced a few problem of precision.
- The results for sample B , which was inhibited oil with a low acidity, are much more
spread. Three labs were within the IEC reproducibility limits and most of the labs within
the 2 standard deviations. In fact the calculate performance (37,8%) does not comply with
the IEC criteria of 15%. It can easily be explained by the very low concentration of 2-
FAL present in this oil (~ 40 g/kg) which is very near or even under the detection limit
of some laboratories (as reported by Lab 1, 2, 13 and 14).
- The results for sample C , which was an aged inhibited oil with high acidity
(0,14 mg KOH/g) and a higher concentration of 2-FAL, are generally very good. Most of
the labs are situated between the standard deviations limit.

Page 11 of 70
Sample A
2-FAL
0,44
0,42
0,4
0,38
0,36
0,34
0,32
0,3
Mean Value
0,28
mg/kg

0,26
0,24
0,22
0,2
0,18 CEI Reproducibility (15%)

0,16 One standard deviation


Two standard deviations excluded value
0,14
0,12
0,1
LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB No.
No.1 No.2 No. 3 No. 5 No. 6 No. 7 No. 8 No. 9 No.10 No.12 No. 13 No. 14 No.11 11
(2th)

Sample B
2-FAL

0,12

0,11 CEI Reproducibility (15%)


One standard deviation
0,1
Two standard deviations
0,09

0,08
mg/kg

0,07

0,06

0,05

0,04 Mean Value

0,03

0,02

0,01
LDM LDM
0
LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB No.
No.1 No.2 No. 3 No. 5 No. 6 No. 7 No. 8 No. 9 No.10 No.12 No. 13 No. 14 No.11 11
(2th)

Sample C
2-FAL
2,00
1,95
1,90
1,85
1,80
1,75
1,70
1,65
1,60
Mean Value
1,55
mg/kg

1,50
1,45
1,40
1,35
1,30
1,25 CEI Reproducibility (10%)
1,20 One standard deviation
1,15 Two standard deviations
1,10
1,05
1,00
0,95 Excluded Value
0,90
0,85
0,80
LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB No.
No.1 No.2 No. 3 No. 5 No. 6 No. 7 No. 8 No. 9 No.10 No.12 No. 13 No. 14 No.11 11
(2th)

Figure 3: Detection of 2-furfural in the 3 oil samples.

Page 12 of 70
3.4.2 Degree of polymerisation of insulating paper

The vast majority of the labs (10/13) determined the degree of polymerisation in accordance with
the IEC 60450-method (Measurement of the Average Viscometric Degree of Polymerization of
New and Aged Cellulosic Electrically Insulating Materials). The 3 other labs used the ASTM D
4243-method (Standard Test Method for Measurement of Average Viscometric Degree of
Polymerization of New and Aged Electrical Papers and Boards).

These test methods describe procedures for determining the average viscometric degree of
polymerization (DPv) of new or aged electrical papers. The determination is made by measuring
the intrinsic viscosity of a solution of the paper in an appropriate solvent. The degree of
polymerization of a particular cellulose molecule is the number of anhydro--glucose monomers,
C6H10O5, in the cellulose molecule as shown in Figure 4. Within a sample of paper, not all the
cellulose molecules have the same degree of polymerization so that values measured by
viscometric method is an average value and is not necessarily the same throughout all of the
insulation of the equipment. It is well known that the normal gradient temperature observed in
equipment can easily produce differences in aging. Accordingly DPv will vary along the
windings and also from layer to layer at each location. The paper insulation will then present an
average DPv of 1000 (for a new transformer) to 200 (for a very aged transformer).

CH2OH OH CH2OH
O HO O
HO
HO OH
O O HO
O
OH CH2OH OH
n-2

Figure 4: Structural formula of cellulose.

The average, standard deviation (STD), relative standard deviation (RSD), 95% Interval of
Confidence (IC95%) and ratio IC95%/average of each paper sample of this RRT are given in the
Table 4.

Table 4: DPv measured on the RRT paper samples.


N=13 Lab. Results STD RSD IC95% IC95 % Excluded
average (%) / average value**
(%)
Sample 1 522 28 5,4 17,8 3,4 1
Sample 2 722 32 4,4 20,1 2,8 1
Sample 3 303 17 5,8 10,6 3,3 1
** Values < or > that average 2STD

All the labs presented results within the limits of two standard deviation with an RSD under 10%.
It is important to mention that only the second series of results coming from lab 11 were taken
into account here because the first one presented erratic values. These results including average,
standard deviation (1STD and 2STD) and reproducibility (IC95%) are presented in Figure 5.

Page 13 of 70
Paper DPv
Sample No.1

750
725
700
675
650
625
600
575
550
Mean Value
525
DPv

500
475
450
425
400 Excluded value
Reproducibility 95% confidence level
375 One standard deviation
350
Two standard deviations
325
300
275
250
LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB
No.1 No. 2 No. 3 No. 4 No.5 No. 6 No.7 No. 8 No. 9 No.10 No. 12 No.13 No.14 No.5 No. 11 No. 11
(2th)

Paper DPv
Sample No.2
900
875
850
825
800
775
750
725 Mean Value
700
675
650
625
DPv

600
575
550
Excluded value
525
500
475
450
425
Reproducibility 95% confidence level
400
One standard deviation
375
350 Two standard deviations
325
300
LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB
No.1 No. 2 No. 3 No. 4 No.5 No. 6 No.7 No. 8 No. 9 No.10 No. 12 No.13 No.14 No.5 No. 11 No. 11
(2th)

Paper DPv
Sample No. 3

440
420
400
380
360
340
320
Mean Value
300
280
DPv

260
Excluded value
240
220
200
180 Reproducibility 95% confidence level
One standard deviation
160
140 Two standard deviations

120
100
LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB
No.1 No. 2 No. 3 No. 4 No.5 No. 6 No.7 No. 8 No. 9 No.10 No. 12 No.13 No.14 No.5 No. 11 No. 11
(2th)

Figure 5: DPv results for the 3 paper samples.

Page 14 of 70
Comparing the results of the DPv values of the 3 samples, allows the following observations to
be made:

- The 3 samples give a good result with an acceptable standard deviation (RSD between
4,4% and 5,8%).
- There is no difference of performance between ASTM and the IEC methods.
- The dissolution time can range from 1 to 18 hours but this did not seem to interfere with
the quality of the results.
- There is no difference in the trend for the results of paper aged in the lab or collected in
the field.
- Majority of the laboratories (11/13) failed to comply with the IEC recommendation to
increase the concentration of paper when the DPv is medium or low, but it did not seem
to have a bad impact on the results.

3.5 RESULTS AND RECOMMENDATIONS

3.5.1 Furanic compounds analysis

- The RRT identified that the 2-FAL analysis seems to be well controlled by the labs using
an HPLC method (RSD < 10%). The very low concentration of 2-FAL (40 ppb in
sample B) was measured with less accuracy (RSD: 62,5%) only because the detection
limit of some laboratories were not low enough to allow a good estimation. As these
concentrations can already be useful, the detection limits of the methods should be
lowered from 50 to 10 g/kg (0,05 to 0,01 mg/kg) as the actual methods satisfy this very
easily.
- The reliability of the 2-furfurol analysis is not very good. It was observed, as expected
that the 2-FOL is not very stable in time, quantitative interpretation should be done with
great care.
- The determination of 2-ACF and 5-HMF are acceptable. We can mention then 5-HMF is
the one presenting the poorer RSD (12%). Risk of elution in the dead volume or coelution
with oil by-products should also be mentioned.
- The efficiency of the extraction/clean-up processes can be influenced by the quality
(acidity) of the oil. Polar compounds present in aged mineral oil samples can interfere
with the furanic compounds and can compete on Solid Phase Extraction (SPE) cartridges.
- If needed, the use of a buffer in the water part of the eluant can stabilize the retention
times during HPLC analysis. The IEC 61198 standard suggests to use acetic acid but
sulphuric acid can also be used.
- The presence of the metal passivator can interfere with other furanic derivates. This
should be taken into account when analysing passivated insulating oils. The method
should be designed so that the time of elution is different from these of the furanic
compounds.
- The use of a diode array detector is not a requirement but gives a very helpful device to
avoid misinterpretation (interferences or coelution).
- The fact that some labs are using the library matching to identify formally compound
with their diode array system, should be clarified in the standard methods. Its important
to emphasise the fact that the spectra are dependent on the instrument used and the

Page 15 of 70
concentration present in the oil. A library with different spectra for different
concentrations is a solution to this problem.
- The preparation of the stock solution should be made in toluene, especially for 5-HMF.

3.5.2 Measurement of Average DPv of insulating paper

- The RRT identified that the degree of polymerisation seems to be well controlled by all
the 13 labs and that the results are not affected by the method, the dissolution time or the
concentration of paper used.

- The use of very short dissolution time (1 to 2 hours) with mechanical stirring at low
temperature, which gives very good reproducibility and precision, should be considered
particularly when this test is done on a routine base.

Page 16 of 70
4 FACTORS INFLUENCING FURANIC COMPOUNDS CONCENTRATION

4.1 INTRODUCTION

Under the influence of temperature, oxygen and moisture, Kraft cellulose decomposes into
various low molecular products, which include gaseous products (mainly carbon oxides), water
and oil soluble products such as furanic compounds. These parameters (temperature, oxygen and
moisture) which the transformer is normally exposed to over its service life, lead to an inevitable
decrease of the mechanical strength of the solid insulation. As mentioned in Chapter 1, the major
furanic derivative, formed from thermally non stabilized Kraft paper is 2-furfuraldehyde (2-FAL).
A lot of research around the world is devoted to the correlation between 2-FAL concentration and
ageing of the solid insulation in terms of remaining life evaluation. It will become clear, that the
2-FAL formation is also influenced by the presence of other aggressive agents that will be
presented in this chapter. This leads to result interpretation being more difficult than was
expected. Recently [8] the effects of water, oxygen and acidity upon ageing (furanic production)
and the type of paper have been studied.

It is important to mention here that even though we focus only on the behavior of 2-FAL in this
chapter, the four other furanic compounds, will also be influenced more or less in the same way
(similar chemical structure) as 2-FAL.

4.2 PARTITION EFFECT

Firstly, it is imperative to make a difference between the quantity of furanic compounds formed
in a transformer and the concentration that can be detected in the oil. As discussed before, the
formation of the 2-FAL and other furanic compounds will be influenced by the ambient
conditions (humidity, temperature, oxygen etc.). Their concentration measured in oil will be
influenced by their ability to remain in the paper after their formation or to dissolve in the oil.

Furanic compounds present in impregnated insulation system are primarily found in the paper.
Allan [9] proposes partition coefficients for a few furanic compounds which would, at least in the
case of 2-furfural, not be dependent on the temperature (between 25 and 90C). For example, the
author suggests a ratio of 2-FAL concentration in oil to the concentration in paper of about 0,2 at
25oC. This ratio, with advantage of the 2-FAL concentration in the paper (5 in paper against 1 in
oil) allows one to suppose that even after the oil is regenerated (which would eliminate its furanic
compounds), the balance of the furanic compounds in the paper-oil complex would once again re-
establish the content in the oil, thereby restoring the information concerning the condition of the
cellulose insulation. The partition coefficients (concentration of furanic compound in oil/
concentration of furanic compound in paper) found by Allan for 5-HMF, 2-FOL, 2-ACF and 5 -
MEF at 25C are respectively 0, 0,03, 0,61 and 0,49 which means by example that for the 5-HMF,
with the value of 0 for coefficient, he did not detect 5-HMF in oil because it had remained totally
in the paper. The author also suggests that these coefficients are dependent to a greater extent on
temperature.

Myers and Sans [10] suggest that furanic compounds will migrate back and forth between the
paper and oil insulation, as they are going to follow the moisture equilibrium. They also suggest
that partition between oil and paper will follow the oil moisture saturation: the concentration of
furanic compounds in the oil will decrease as the moisture saturation will increase.
Griffin [11] observed roughly the same phenomenon, and proposed that the division of furanic
compounds in the transformer is fairly complex even if it is not dependent on temperature. He

Page 17 of 70
also noted that at 200C, a damp paper retains most of the furanic compounds it generates, which
is not the case for paper heated in the same way but which has been previously dried. However,
the furanic compounds would appear to migrate quickly in the oil. At 80C, equilibrium seems to
be attained in less than 48 hours.

Pahlavanpour et al [12],[13] also observed that the furanic compounds are absorbed more in the
paper than in the oil. They also noted that the correlation between these specific compounds
detected in oil and the DP value of the insulating paper is not easy to establish since they depend
on a lot of factors such as temperature, type of oil, type of paper, transformer design, and so on.

To refine the ageing models using the furanic compound measurements, it will be crucial to find
the partitioning of furanic compounds between paper and oil as well as the parameters that could
influence them. Some recent experimental work [14] showed that furanic compound paper/oil
ratios are

- greatly influenced by the humidity of the paper (which is related to the water content
in oil),
- less influenced by temperature unless it is in conjunction with an acidic oil,
- different for each furanic compound (particularly for 5-HMF and 2-FAL),
- also influenced by the type of paper.

Related to this specific experimental work, Figure 6 shows how the 2-FAL could partition
between paper and oil in different conditions (i.e. ratio of 2-FAL concentration in paper to the
concentration in oil versus temperature).

1,8

1,6

in new oil with dry paper


1,4
oil/ paper 2-FAL partition

in acidic oil with dry paper


1,2

in new oil with wet paper ( 5% weight)


1,0
in acidic oil with wet paper (5% weight)

0,8

0,6

0,4

0,2

0,0
20 30 40 50 60 70 80 90 100

Temperature ( degree C)

Figure 6: Example of the ability of 2-FAL to partition between paper and oil
(acidic oil: 0,16 mg KOH/ g oil, dry paper: < 0,5% weight).

Page 18 of 70
4.3 FACTORS INFLUENCING THE 2-FAL FORMATION

4.3.1 Temperature

The temperature is undoubtedly a very important factor, influencing the 2-FAL formation. It has
been proven through many research groups, that temperatures over 120C accelerate 2-FAL
formation. Therefore, a slow but continuous increase of 2-FAL could point out to a failure in the
cooling system, whereas a strong immediate increase could point out to an acute thermal failure.

Diagnostic procedures like dissolved gas analysis (DGA) or oil quality control analysis and
determination of oil values are not sensitive enough to indicate that there might be moderate
overheating issues up to 150C that might result from defective cooling. These temperatures are,
however, high enough to lead to an increase in 2-FAL content significantly. The unusual increase
in 2-FAL values in one transformer in comparison to other transformers of the same family
(generation step up (GSU) at similar service conditions) detected in 1992, have been an
unambiguous indication of a thermal irregularity see example given in Figure 7 [15],[16]. In
2002 defective thermo-replicas have been detected. No discrepancies in the DGA have been
noticed. Elevated oil values e.g. color numbers (3-3,5) have been measured in the case of the
three transformers with defective thermo-replicas (WTI).

0,9

0,8

0,7

0,6
FAL (mg/kg oil)

0,5

0,4

0,3

0,2

0,1

0
Oktober
October March
Mrz 97 Juli 98
July Dezember
December April
April01 September
95
95 97 98 99 01 02
Sampling date

Figure 7: 2-FAL analysis of four GSU transformers, commissioned in 1992.

Page 19 of 70
4.3.2 Moisture
Practical experience shows that significant 2-FAL concentrations in oil can be generated at
service temperatures in wet transformer insulation (see Figure 8).

50 1
Breakdown voltage
45 0,9

40 0,8
Breakdown voltage 2,5 m (kV)

2-FAL (mg/kg oil)


35 0,7
Water Content (mg/kg)

30 0,6

25 0,5

20 0,4

2-FAL
15 Water content 0,3

10 0,2

5 0,1

0 0
31.12.2002 10.04.2003 19.07.2003 27.10.2003 04.02.2004 14.05.2004 22.08.2004

Sampling Date

Figure 8: Variation of the breakdown voltage, humidity and 2-FAL in a 75 MVA


transformers, manufactured in 1981.

However, as mentioned in the former section, the increase of 2-FAL can be caused by a change in
its distribution between the oil and insulation paper [17], which will be influenced by the
presence of moisture, in addition to additional 2-FAL being generated. In Figure 9, we can
observe that the 2-FAL concentration increases as the water content increases in the equipment.
The phenomenon here is probably more related to the variation of the partitioning of the 2-FAL
between the oil and paper when moisture is present than to a continuous generation of 2-FAL.
The distribution of 2-FAL between paper and oil at different humidity and temperature suggests
that transformers with a higher humidity would exhibit a higher concentration of 2-FAL in oil
[14],[18]. Figure 9 shows a similar phenomenon when the equipment is experiencing an increase
in loading. This suggests again a temperature/moisture distribution dependence of 2-FAL
between oil and solid insulation [19].

Page 20 of 70
Figure 9: Increased 2-FAL concentrations in oil in times of increased loading for two
different transformers.

4.3.3 Oil type and oil condition


There is considerable evidence from laboratory studies and practice that clearly demonstrates that
the oil type has a major influence on the 2-FAL generation. The 2-FAL content of inhibited and
passivated oils is significantly lower (near 10 times) than the 2-FAL content of non-inhibited oils
with the same degree of polymerisation (DP) of paper [20],[21]. Nevertheless it is not clear if this
is due to the presence of a lower kinetic of formation of 2-FAL or just another effect of partition.
It is interesting to mention that WG A2.40 (Corrosive Sulfur) is presently investigating the effect
of passivator on the partition of the furanic derivatives between oil and paper.

The ageing condition of the oil (e.g. the presence of acids or sludge) influences not only the
absolute amount of 2-FAL in the oil, but also its stability and again the partitioning between oil
and solid insulation. A rapid deterioration in the oil properties such as acidity or loss factor is
often connected to an increase of 2-FAL content in the oil.

4.3.4 Design: free breathing or closed type transformers


The presence of air promotes the formation of furanic derivatives from cellulose. Therefore the 2-
FAL analysis may be a better tool for ageing evaluation in the case of free breathing than closed
type transformers. Generally lower furanic concentrations are to be expected in the case of a
transformer with a rubber bag or nitrogen cushion [22]. The same is valid for high voltage
bushings with a nitrogen cushion.

4.3.5 Insulation paper type: plain Kraft paper or thermally stabilized Kraft paper
The thermally stabilized paper (being used more by the electrical equipment industry) guarantees
higher reliability of the solid insulation in thermally stressed regions. Investigations have shown
that thermally stabilized Kraft paper develops much less furanic compounds in oil than plain
Kraft paper [22], [23], [24], [25]. Laboratory results (Figure 10) [26] demonstrate a dependence
of 2-FAL generation or partitioning (this has not been totally clear so far) on the nitrogen content

Page 21 of 70
in paper. Both Kraft and thermally stabilized Kraft types of paper are being used in modern
transformer design.

100000
Kraft (0,08% N2)
55 800 ppb
Kraft CE (1,2% N2)

10000 manning 250 (2,3% N2)

2200 ppb
log fur-2 (ppb)

2300 ppb
1000

100
Aging at 130C
paper humidity content: 0,5%

10
0,0 1,0 2,0 3,0 4,0 5,0 6,0
Degradation factor (1000*(1/DP-1/DP
1000 * (1/DP0-1/DP) 0))

Figure 10: Dependence of 2-FAL formation in oil with nitrogen content in the insulating
paper.

4.4 STABILITY

The amount of oxygen and moisture in the oil during dynamic loading are also factors of
fundamental importance to the ageing assessment. The rupture of cellulose chains has been
known for some time to be heavily influenced by these contaminants. Emsley et al. [27]
investigated this statement in laboratory experiments (paper in oil tubes in the oven) under very
different conditions of oxygen and moisture. Figure 11 presents the production of 2-FAL per
gram of paper on different moisture and oxygen conditions. It is important to notice that the
strong instability observed in this figure is a function of the high levels of concentration of
oxygen and moisture (red curve to the left).

Figure 11: 2-FAL production from paper in oil experiment at 140C.

Page 22 of 70
In real operating conditions, there is still no consensus on the stability of the furanic compounds
above 110C. For instance, Griffin [28] believes that the presence of water in the transformer
could degrade 5-HMF. Whereas Allan [9] suggests that furanic compounds are stable at operating
temperatures in the absence of any oxygen. The author demonstrated however that after 8 weeks
at 110C when oxygen is present, a non-negligible loss of 2-furfural (20 to 40%) was observed
along with a substantial loss of the corresponding 2-furfurylalcohol (40 to 100%). Lutke et al.
[22] observed the same kind of instability in the presence of oxygen, even at operating
temperatures as low as 60oC.

In 2000, Emsley et al. [29] made roughly the same observations with the exception of 2-FOL,
the furanic compounds are stable up to about 140C. However, in his opinion the lack of stability
of 2-FOL would be an asset since its presence could only be attributed to a very recent fault. The
team headed by Myers and Sans [10] even suggests that the four furanic compounds normally
found at very small concentration levels compared to 2-furfural appear to be transient species and
would not be found during the normal transformer ageing process.

Figure 12 shows the depletion of 2-FAL over a 7 year period in a 275 kV transformer [19]. Note
that 2-FAL was introduced into this transformer unfortunately in 1993 when a batch of oil
containing 2-FAL was used in an oil-change during maintenance.

14.00
275kV Transform er
Furaldehyde Content in Oil

12.00

10.00
(pp )

8.00

6.00

4.00

2.00

0.00
15/09/93 15/09/94 15/09/95 15/09/96 15/09/97 15/09/98 15/09/99 15/09/00
D ate

Figure 12: Decrease in 2-FAL Transformer contaminated with 2-FAL during oil change.

As part of another study, Emsley [29] suggests that in the case of transformers undergoing normal
ageing (i.e. with no thermal faults), the presence of furanic compounds could mainly be attributed
to hydrolytic degradation. This leads one to assume that at low temperatures their development
could be attributed to a greater extent to the acidity or water content in the oil rather than the
thermal degradation of the cellulose insulation. Scheirs [3], [30] made the same observations
when he noted that furanic compound levels were much higher in parts of the transformer where
water content tends to accumulate.

More recently [31], a U.S. engineering firm claimed that the progression of furanic compounds in
electrical equipment often competes with their degradation and thus makes them even more
difficult to interpret.

Page 23 of 70
4.5 INFLUENCE OF MAINTENANCE

Another possible factor complicating interpretation of 2-FAL concentrations is the effect of oil
processing on the 2-FAL concentration in the oil. Some evidence for this is provided by results
presented in Figure 13 [19] for two similarly and consistently loaded transformers after oil
changes. It is then clear that oil maintenance can influence the diagnosis related to the 2-FAL.

1.4

1.2 Oil
changes

0.8
[2FAL], ppm

6 years

0.6

0.4

0.2

0
1/1/94 2/1/96 2/1/98 3/1/00 3/1/02 4/1/04 4/1/06

11B 12B

Figure 13: Effect of oil change on 2-FAL concentrations.

Maintenance procedures, which do not effectively dry the solid insulation such as oil processing
(degassing) or oil reclaiming normally lead to a temporary removal of the furanic compounds
(such as 2-FAL) from the oil [32]. After some time (e.g. 3 to 6 months) equilibrium can take
place between the furanic compounds absorbed in the paper and those dissolved in the oil. In
some cases, however, longer time is needed (see Figure 14) [33].

1,2

1,0
oil

0,8
[2-FAL], mg/kg

0,6

0,4

0,2

0,0
30 32 34 36 38 40
oil replacement oil
Years in service
degassing

Figure 14: Depletion and re-equilibrium of 2-FAL in oil after maintenance procedures, e.g. oil
replacement or oil degassing.

Page 24 of 70
Maintenance procedures, which are designed to dry the solid insulation, such as Low Frequency
Heating(LHF)-drying, hot air drying or stationary bypass drying, lead to a lower amount of
furanic compounds (2-FAL) in the oil (through changed distribution coefficients between paper
and oil) which remain constant over years (see Figure 15) [34].

3,50

3,00 150 MVA


2,50
2-Furfural [ppm]

2,00
38 MVA
1,50 Oil reclamation
1,00
0,50

0,00 1999

2000
1995

1997

2001

2002

2003
Figure 15: A 150 MVA (blue) & a 38 MVA (red) transformer subjected to off-site hot air dry
out and oil reclamation. Combined dry out and oil reclamation produced long term
reduction in 2-FAL content.

It should also be highlighted that as this recovery is highly dependent on the transformer itself
(technical characteristics and operating conditions), we cannot extrapolate the rate of recovery of
the concentration from one transformer to another. Three different cases of recovery of 2-FAL
concentration after the oil replacement in different transformers [35] are presented in Figure 16 to
illustrate that.

Page 25 of 70
40 MVA, 220kV, Open type
OLTC, Shell Diala D oil, 1970
Oil replacement: 1997

20 MVA, 60kV, Open type


OLTC, Nynas Nytro 11 EN, 1976
Oil replacement: 2003

80 MVA, 150kV, Open type


OLTC, Nynas Nytro 11 EN D oil, 1959
Oil replacement: 1997

Figure 16: Evolution of 2-FAL after oil replacements

Page 26 of 70
5 DIAGNOSIS METHODS

5.1 CORRELATION WITH DEGREE OF POLYMERIZATION

As was mentioned earlier, the degree of polymerization (DP) is a measure of the average
molecular weight of the cellulose molecules (which typically makes up 75 to 85% of the
composition of insulating paper). It is the most important parameter for assessing the ageing
status of the cellulose insulation of electrical equipment and ultimately the only way to estimate
the residual life of the insulation.

In the past, CIGRE WG 15.01 (TF 03) [7] had shown the correlation between the DP and the
furanic compounds evolution obtained in laboratory ageing tests. This correlation is shown in
Figure 17.
g/g (ppb) in oil

Figure 17: 2-FAL and other related furanic compounds in oil vs DP: collected laboratory data.

The correlation between the papers loss of mechanical properties, which are intrinsically linked
to a decrease in its degree of polymerization, and changes in some furanic compounds (mainly 2-
furfural), has been verified (logarithmic relation was established by Shroff and Stannett [36] in
1985.

In 1991, Xue Chendong [37] proposed another logarithmic correlation between the degree of
polymerization and the 2-FAL concentration.

Log (2-FAL in mg/l) = 1,51-0,0035DP

Page 27 of 70
In 1996, Hill, Darveniza and Saha [38] attempted to establish another model for generating
furanic compounds based exclusively on the splitting of glycosidic chains, which is directly
related to a decrease in the degree of polymerization. In 1999, based on research conducted
by CIGRE WG 15.01 (TF 03) [7], De Pablo [39] developed a new degradation model that
takes into account the fact that paper rarely decomposes evenly (e.g. effect of hot spots,
thermal gradient). Later validated by Serena [40], the model suggests that each time three
cellulose chains are broken; one 2-furfural (2-FAL) molecule is generated.

Minimum degree of polymerization (DPmin.) = 800/ [(0,186 x 2-FAL) + 1]

The establishment of these relationships is important for interpreting the furanic compounds but it
is clear that the use of any of these models and the resulting diagnosis will depend on our
knowledge of the amount of paper involved. Moreover, the non-negligible effect of the water and
acidity of the oil on the formation kinetics of these compounds will never allow the authors to
establish a foolproof model. The kinetics could also probably vary if the equipment involved is
operated in a nitrogen atmosphere or is insulated with thermally upgraded paper [41] Shkolnik et
al. [42] also noted this variation and proposed a new model in 1999 exclusively for these papers,
which can be expressed as follows:

Log 2-FAL= 2,88 x 10-3 DP + 4,17

The main aim of investigating insulation material samples taken from aged power transformers is
to establish the correlation between DP and the content of 2-FAL in the oil for operating
transformers. In some cases the CO2/CO quotient and particularly its trend can be used to
evaluate the degree of degradation of the insulation paper [43]. However it is not obvious that the
performance of in-service transformers can be based on information from these models.

Moreover, work done by Mulej et al. [20] showed why many workers find much lower
concentrations of 2-FAL in transformer oil samples as compared to results from laboratory ageing
tests for the same DP. According to these authors, the type and quality of the oil influences the
quantity, stability and solubility in oil in addition to the speed of ageing and the production of
furanic compounds. Lutke et al. [22] even concluded that due to these limitations, it will not be
possible to predict a transformers residual life solely based on the content of furanic compounds.
This issue, which has been debated at great length in the past, will be covered in section 5.5.

The correlation between DP of the paper and 2-FAL in oil seems to be possible within specific
transformer families with similar service conditions. Established correlations in different
countries and different populations may differ significantly, showing that there is no single
calculation formula universally applicable [19], [43], [44], [45], [46], [47].

5.2 PRESENCE OF THE OTHER RELATED FURANIC COMPOUNDS

Although 2-FAL seems to be the most popular thermal ageing marker, the other related furanic
compounds mentioned before can also be detected in operating transformers.

Some experiences were reported [48] observing changes in furanic compounds profiles over time
in the presence of known conditions. According to these authors this may lead to a new and

Page 28 of 70
reliable way to use the other related furanic compounds. The proposed diagnosis model is
presented in Table 5.

Table 5: Possible causes of specific furanic compound presence.

Compound Diagnosis proposed

5-HMF Oxidation
2-FOL High moisture
2-FAL General overheating or normal Ageing
2-ACF Rare, causes not fully defined
5-MEF High temperatures

5.3 CRITICAL THRESHOLDS

Although the analysis of furanic compounds combined with DGA analysis presents a certain level
of reliability, it is still difficult to use furanic compounds limits or threshold values or ratios for
diagnostic purposes due to a lack of knowledge of how these compounds are formed and how
they evolve over time.

The relative concentration of the different furanic compounds is always about the same, i.e. 2-
FAL is always the main compound, followed by 5-HMF then 5-MEF, and finally 2-ACF. Only
some authors were able to note major changes in 2-FOL which may be caused by degradation at
higher temperatures [48]. This was not systematically noted in thermal degradation studies.
However, based on a CIGRE WG 15.01 (TF 03) study [7] where the furanic compounds absorbed
on the paper had also been analyzed, 5-HMF was the compound primarily found. However, it
appeared to be retained to a far greater extent than 2-FAL on the paper, thereby making it less
accessible in oil.

In general, most of the authors agree that there is a single critical threshold for equipment of
about a few hundred ppb for 2-furfural. According to an U.S. consulting firm, for a representative
population of U.S. transformers, 90% have less than 0,1 mg/kg of 2-furfural. For European
transformers, the 90% value calculated based on a statistical survey including more than 5000
transformers was about 5 mg/kg (only around 60% have less then 0,1 mg/kg of 2-furfural)
which is very different [7][10].

It should be noted that when European statistics are compared with North American figures, the
levels of furanic compounds that are generated are greater in Europe. This fact can be related to
the introduction of thermally upgraded paper in North America since the 60s. Paper can be
thermally upgraded by impregnating it with dicyandiamide (dicy) or by other processes such as
the Insuldur method. In this respect, it has been suggested [50] that these molecules, which are
added to the paper to increase its thermal resistance, could lead to a degradation of the furanic
compound by-products over time. Another possible hypothesis is that these compounds are not
formed as readily in upgraded paper. It is interesting to mention that in some transformers
mixtures of upgraded and non-upgraded paper are used, and in almost all transformers the
pressboards are non-upgraded even if all the paper is upgraded.

It has been observed by many users that 2-FAL levels in oil are completely different for inhibited
oils (like the majority of USA power transformer have and also part of Europe, as for example the
Slovenian power transformers [20]) and for uninhibited oils, which have been used for the
majority of European power transformers. In fact, it seems that for inhibited oils, the

Page 29 of 70
concentration of 2-FAL is much lower than for uninhibited oils, if we compare transformers of
the same age. Inhibitor may influence the formation of 2-FAL or may have an effect on its
migration from the paper to the oil. This phenomenon it still not really well understood by the
scientific community.

Even though some critical thresholds based on laboratory experiments have been proposed [51],
the use of these classifications is still very difficult in the field. The summary table in Annex 3 in
fact proposes a large number of approaches. As these statistical studies have not been conducted
with the same parameters, it is very difficult, even impossible, to compare them or to draw any
general conclusions.

This is why Working Group D1.01 (TF 13) incorporated into its activities a vast field-data
collection campaign in order to conduct a statistical study that was better structured and more
homogenous. The results of this study are presented in Chapter 6.

5.4 RATIOS OF DIFFERENT FURANIC COMPOUNDS

Numerous authors have attempted to establish ratios between the various furanic compounds in
order to determine the default temperature involved and to help determine the equipments
expected lifespan. In 1988, Burton [52] attempted to determine the fault temperature between
140C and 350C by using a 2-FAL/5-MEF ratio. In 1990, a CIGRE working group [53]
proposed a 2-FAL/5-HMF ratio which would allow the fault temperature to be determined
between 60 and 130C. In 1992, Grant [54] proposed a 2-FAL/2-FOL ratio, while in Japan,
Nanba et al. [55] established a relation between 2-FAL and 5-HMF that appeared to be highly
linear. In 2001, a study by Soares et al. [56] showed that the 5-HMF/2-FAL ratio doubles
between 80 and 120C.

As these interesting results have still not been applied to real transformers, this approach does not
show promise as an effective diagnosis tool.

5.5 RESIDUAL LIFE ESTIMATION

The residual life estimation of transformers has been studied and discussed for decades. The life
of transformer has been assumed to be given by the life of the solid insulation, in particular [57],
[58]. The main reason for this has been obviously that once the solid insulation has lost its ability
to accomplish its mechanic function there is a possibility that the transformer could fail thereby
failing to reach its technical end-of-life. The concept of end-of-life is controversial in nature. For
example, a transformer which no longer fits into a power system due to its inability to serve a fast
growing load might have reached its strategic end-of-life. Another transformer which has
severe maintenance issues, high losses and so forth might be replaced and hence deemed to have
reached its economical end-of-life. For the purpose of this report, though, and considering the
use of furanic compounds for the estimation of solid insulation ageing (related to DP values), let
us accept the concept of life of the transformer as being the life of the solid insulation, and let us
understand how dramatically this has changed over decades. Life of insulation has generally been
simulated in laboratory experiments, using paper-oil sealed tubes in oven. Mechanical properties
of insulating paper have been used for decades to characterize ageing, the three main reasons for
this being: its ease of measurement, good variation with ageing and correlation to mechanical
supportability of the windings. Figure 18 shows a very aged transformer active part as compared
to a new winding structure. Figures 19 and 20 show laboratory results for the tensile index

Page 30 of 70
(strength) of Kraft paper, aged under different thermal stresses.

Figure 18: Aged insulation compared to a new transformer winding.

Figure 19: Change in tensile strength of paper, aged in air and oil at 120 to 160 C [57].

Figure 20: Correlation between tensile strength and DP [27].

Page 31 of 70
It is worthwhile mentioning here that the state of the solid insulation determines the life
expectance of a transformer. The tensile strength of paper determines mechanical strength of
paper that, in case of severe degradation, may lead to an insulation failure, particularly in a
situation of short circuit where the windings are mechanically stressed. The degree of
polymerisation (DP) of paper directly correlates with the paper tensile strength. As showed in
Figure 21, up to 500 to 600 units DP is the tensile strength constant, afterwards it decreases.

In 1920 Montsinger tested varnished tape insulation in a series of test tube experiments. He
heated the tapes and measured their mechanical tensile strength. The end-of-life was determined
when tensile strength reduced to about 50%. The industry has used this value for many years,
despite the fact that Montsinger himself acknowledged that his experiments were not accurate for
high temperatures.

There is of course some question whether laboratory ageing tests, made on isolated
strips of paper in sealed tubes can be applied directly in estimating the life of
insulation in a transformer. Montsinger, 1944.

Montsinger noted that the rate of deterioration of mechanical properties doubled for each 5C to
10C increase in temperature. The doubling factor was found to be about 6C in the temperature
range from 100C to 110C and 8C for temperatures above 120C. However, people tend to
remember the doubling factor as constant and the IEC loading guide [60] uses 6C.

In 1947 Dakins findings indicated that the deterioration of transformer insulation followed a
modified form of Arrhenius' chemical reaction rate theory. This stated that the rate of change of a
measured property can be expressed in the form of a reaction rate constant K. Mathematically, the
rate constant can be expressed by

B
273
K A e

where A and B are empirical constants and is temperature in C.

An insulation life curve was then developed to relate the insulation's degradation to a given
operating temperature. The industry took Dakin's work and Montsinger's 50% residual tensile
strength as the end-of-life point of the paper and arrived at loss-of-life percentages based on time
at various temperatures. The IEEE loading guide [61] defined the time span of 65000 h
(7,42 years) for the insulation end-of-life at a temperature of 110C; the criteria being 50% tensile
strength retention. The industry later recognized that those values were mistaken by a factor of 2
to 3:
Both the results of functional tests and service experience suggest that a normal life
of 15 to 20 years at a winding hot spot temperature of 110C is a reasonable
expectation for both distribution and power transformers with well-dried and oxygen-
free insulation systems.

It is good to mention here that the most important difference between the IEC and the IEEE
loading guides is that IEC proposed two different reference temperatures; one for typical Kraft
paper insulation (98C) and one for upgraded paper insulation (110C same as IEEE guide).

There are many reasons for the difficulty experienced in establishing a clear and universally
applied end-of-life criteria. Firstly it must take into account the fact that real transformers do not

Page 32 of 70
operate under a continuous hot-spot temperature, as hypothetically assumed by the Dakin &
Arrhenius ageing model. It is necessary to take into account load oscillation, ambient temperature
variation, different levels of cooling (ONAN/ONAF/OFAF)2 applied on the same transformer,
etc.

Also, under dynamic loading, any transformer will react to load or ambient temperature changes
according to the oil and winding time constants, meaning that any eventual change will have a
delayed response by the thermal components of the transformer (the copper respond faster, then
the copper-paper system, then the oil, etc.). These changes may also impose dynamic temperature
gradients on the windings (mainly the longitudinal temperature gradient) which, in turn, will
interact differently with the cooling environment (oil, radiators, etc.). This dynamic behaviour is
very complex and indicates that a uniform and constant hot-spot temperature is a technical
chimera. There is also the recurrent concept of end-of-life to contend with. How realistic is it
taking the reduction of a given property of the solid insulation to prescribe the functionality of the
transformer? The literature has many examples to where very aged insulation structure has not
impeded regular transformer operation. This suggests therefore that aged insulation only
increases the risk of transformer failure. End-of-life criteria should therefore be associated with a
higher risk of transformer failure.

Firstly, it is clear from the Table 6 that transformer experts cannot agree on one universal
criterion which clearly represents transformer technical end-of-life. Depending on the base used
to judge end-of-life, the same transformer might be in acceptable operating condition or nearing
its end-of-life. The second aspect is the importance of the cumulative effect. Notice that the table
sums up a given number of operating-hours at a given average temperature of the hot-spot which
again brings the importance of load profile into the discussion. In order to use Table 6 one must
estimate the actual transformer equivalent operating hot-spot temperature over a given number of
operating-years.

Table 6: Normal insulation life of a well-dried, oxygen-free 65C average winding


temperature rise insulation system at the reference temperature of 110C [61].

The issues discussed above are some of the main factors to be taken into account in trying to
establish a correlation between transformer ageing insulation and technical parameters such as the
amount of furanic compounds found in the oil over the ageing period. Hence, thermal aspects and
the amount of ageing acceleration factors in contact with the cellulose during the ageing period

2
ONAN: oil natural and air natural circulation ONAF: oil natural and air forced circulation OFAF: Oil forced and air forced
circulation.

Page 33 of 70
such as oxygen and moisture are the most significant elements to be considered when explaining
the expected life of a transformer.

There have been many attempts to explain the correlation between the levels of 2-FAL found in a
given transformer oil and the expected DP of the paper. Although these attempts have helped in
understanding the complex phenomena taking place inside a real transformer while in operation
they have been unable to precisely indicate the actual ageing or life expectancy of the paper.

Some investigations have been carried out under strict laboratory conditions [1] while others have
been based on real transformer data [37] and others have used mixture of laboratory and site tests
(see Figure 21) [62].

4
log furaldehyde (ug/g per g of paper)

1 120 C
140 C
160 C
0

-1
1200 1000 800 600 400 200 0
Degree of Polymerisation

Figure 21: Residual life estimation lab experiment, Chendong equation.

To provide a few additional example models, the case given in Figure 22 is provided. These are
among the most commonly used or discussed models to correlate 2-FAL to DP. Notice, from
Figure 23, that some of the models give similar answers to the correlation at some specific 2-FAL
levels but none will agree with the others at most 2-FAL levels. On the contrary, notice that the 2-
FAL scale is logarithmic to facilitate the distinction between the models. Notice also that at some
2-FAL levels the difference between some of them can reach up to 100% (maximum model
deviation at about 8 ppm of 2-FAL). It is good to mention that these models are strictly related to
traditional Kraft paper.

Page 34 of 70
1000

900

Degree of Polymerization, DP
800
700

600

500

400 Burton
300 Vuarchex
Chendong
200
DePablo
100
0.01 0.1 1 10
2FAL, mg/Kg paper

Figure 22: Example attempts of model correlation cases available in the technical literature.

Figure 22 provides an estimate of the relative deviation of all models illustrated in Figure 23,
giving the 90% confidence interval of all DP for a given 2-FAL level, from the selected models in
this example (Figure 24).

100
90
80
Deviation, %

70
60
50
40
30
20
10
0
0.01 0.1 1 10
2FAL, mg/Kg paper

Figure 23: Maximum deviation between the correlation models (see Figure 22).

1000
900
DP, models average value

800
700
600
500
400
300
200
100
0.01 0.1 1 10
2FAL, mg/Kg paper

Figure 24: Estimate of Models 90% Confidence Interval for DP (see Fig. 23).

Page 35 of 70
The large error can be observed when comparing different models. Transformers, which are from
different countries, experienced different climatic conditions, different operating characteristics
and load profiles, different manufactures, have different types of insulating paper, different
maintenance criteria, different breathing systems, levels of oxygen and moisture, history of oil
treatment and some transformers have on-load tap changer (OLTC) and some do not. Some
transformers have even had the oil completely replaced (oil reclamation process) at some point in
time.

The correlation of DP with 2-FAL from transformers that have been examined after scrapping
does not necessarily agree with results found from laboratory experiments. Figure 25 shows the
correlation of 2-FAL with data from scrapped transformers, with the predicted Chendong
equation as comparison [19]. Uniformly aged transformers/reactors are unusual and it is only for
such cases where the Chendong equation can be used to estimate the DP from the 2-FAL with
any real accuracy.

10

9
DPs
500 - 220
(mg/kg) in oil

6
[2FAL]

5
2-FAL

DP
4
200
3

0
1 2 3 4 5 6 7 8 9 10

1,000 / DP
Highest [2FAL] Chendong

Figure 25: 2-FAL detected in the oil (mg/kg) versus degradation factor from scrapped
transformers.

The rate of increase in 2-FAL, in certain cases is more useful at determining ageing activity than
calculating residual lifetime from 2-FAL levels, as there are too many variables affecting the
2-FAL level in real transformers. Figure 26 shows four different transformers that had rising 2-
FAL levels [19].

Page 36 of 70
2

1.8

1.6

1.4

1.2
[2FAL], ppm

0.8

0.6

0.4

0.2

0
3/3/91 3/3/93 4/3/95 4/3/97 5/3/99 5/3/01

T4260 T6669 T5434 T5533

Figure 26: 2-FAL rates (ppm or mg/kg) of increase in 400 and 275 kV transformers.

More important than a precise correlation between 2-FAL and DP is the understanding of the
physical phenomena involved in the 2-FAL formation. The on-going effort is aimed at providing
the user with some meaningful statistical reference so that one can make adequate decisions
based on relative values of furanic compounds.

And for an increasing number of experts, the determination of the residual life of a piece of
equipment, through the monitoring of furanic compounds, becomes unrealistic. Furanic
compounds certainly give information about the papers degradation, but this is likely the extent
of it. Years of experience reveal that their progression will also be dependent on test conditions or
on a transformers operating conditions. A literature review of the statistical populations for the
study of 2-FAL in transformers is presented in Annex 3.

Page 37 of 70
6 STATISTICAL STUDY

6.1 UNIVERSAL VALUES: ARE THEY POSSIBLE?

The Working Group was able to collect approximately 30000 records. These data came from
twelve different sources across North and South America, Europe, and Asia. Through cooperation
with IEC TC 10 WG 33 we were able to add another 12000 data point from their database which
was primarily from Europe. Because of a lack of detailed information about the transformers
from which the data come from, it was difficult to perform any significant analysis of the furanic
compounds levels (principally 2-FAL as shown in Figure 27).

Figure 27: 2-FAL (mg/kg in the oil): 30K data points from 12 countries.

To address this, the Working Group D1.01 (TF 13) tried to develop a data input form which
included those data fields which will enable the analysis of 2-FAL levels according to specific
parameters including:

Type of transformer.
Type of oil.
Type of oil preservation system.
Type of cooling system.

These are the primary factors to be taken into account in attempting to establish a correlation
between transformer ageing and technical parameters, such as the amount of 2-FAL found in the
oil over the ageing period. Additionally there are many other factors that may have a major
influence on the ageing of the cellulose and must be taken into account when identifying factors
such as the 2-FAL production rate or the 2-FAL levels found in the 30000 samples forming part
of the WG D1.01 (TF 13) database shown in Figure 27. These include:

Thermal aspects of transformer operation.


Ageing acceleration factors in contact with the cellulose during the ageing period (e.g.
oxygen and moisture).
Design characteristics.
Oil processing.
Type of paper employed to insulate the transformers.

Page 38 of 70
6.2 TYPICALS VALUES

In order to assess possibilities for establishing norms for furanic compounds levels in
transformers, WG D1.01 (TF 13) utilized a similar approach to that used by the IEC standard
60599 in determining its recommended levels for Dissolved Gases Analysis (DGA). The outcome
of this approach is to establish typical levels for a piece of equipment. If the 2-FAL level is within
these typical levels then the probability that the condition of the insulation system is acceptable
will be high. If the 2-FAL levels are above these typical levels then the probability that the
insulation system is approaching its end-of-life will be high as well. Ideally these typical levels
should be based on actual values of 2-FAL which have been correlated with the condition of the
transformer insulation.

The WG D1.01 (TF 13) addressed the identification of a single typical value for 2-FAL levels for
all transformers. The data suggests that there are distinct typical levels for 2-FAL for different
transformer types. Transformer size and voltage rating can have an impact on the oil to paper
ratio which can influence the amount of 2-FAL produced. Transformer loading will also have an
impact on the generation of furanic compounds. Generator Step-Up transformers, which are
typically loaded close to nameplate throughout the day while have different typical levels to
distribution substation transformers which are frequently loaded at 50% of nameplate and will see
daily load cycles.

An important criterion when evaluating 2-FAL data is whether the data came from transformers
with known problems or if the data was from routine sampling of transformers. All of these
factors will have an influence on the calculated typical values.

Because of the above, it is now obvious that typical values should be calculated for each
transformer type as has been done by IEC TC10 WG 33 who used the following transformer
types in their statistical evaluation:

Generator Step Up (GSU).


Transmission (Trans).
Large Distribution (L Dist).
Small Distribution (S Dist).

Unfortunately, at the time of development of this brochure, the task force did not have sufficient
data to determine typical levels based on actual condition of the insulation and type of equipment
as IEC TC 10 WG 33 was able to do. However, most of collected reported data in this brochure
are coming from companies where mainly suspected transformers were tested for furanic
compounds which can be strongly misleading for the approach.

The WG D1.01 (TF 13) decided then to perform a statistical evaluation of each set of data that
was provided. In this study we calculated the 90%, 95% and 98% typical concentration level for
2-FAL for each data source as well as for the entire database. Each level represents the
percentage of 2-FAL that would be equal to or less than that particular level. For each sample
considered, the cumulative number of 2-FAL analyses where the 2-FAL concentration is below a
given value is calculated, and then plotted as a function of 2-FAL concentration. For example the
90% norm curve illustrates the typical levels of 2-FAL for 90% of the transformer population.
The results, including the results of the IEC TC 10 WG 33 database presented above, are given in
Figure 28. Each laboratory data set is identified by a different colour. The typical concentration of
2-FAL for the entire database of participating laboratories, excluding the IEC TC 10 WG 33 data,
is given in Table 7.

Page 39 of 70
2-FAL Percentiles

7000

6000
B
C
5000 D
2-FAL Concentration (ug/kg)

E
F
G
4000 H
J
K
3000 L
M
N
GSU
2000
Trans
L Dist
S Dist
1000

0
90 95 98
Percentile Values

Figure 28: Typical concentrations of 2-FALin ug/kg (ppb) in oil from various laboratories.
(B: USA, C: Canada, D: German, E: Thailand, F: Belgium, G: Poland, H: Portugal, J: Brazil, K: England,
L and N: Italy, M: Austria, GSU,Trans, L Dist, S Dist: IEC TC 10 WG 33)

Table 7: Typical concentration of 2-FAL for the entire data base (excluding the IEC TC10 WG33
data)

Percentile 2-FAL 2-FAL


(%) (g/kg) or (ppb) (mg/kg) or (ppm)
90 840 0,84
95 1657 1,66
98 3080 3,08

It is very important to stress that the values in Table 7 , which are typical percentile levels for the
entire database, should NOT be used to establish norms for 2-FAL in transformers because of the
variation in the total population of results. This variation can be explained by:

The difference between North America and Europe (probably related to the more
prominent use of thermally upgraded paper in North America which has a lower
concentration of 2-FAL observed with that paper.
Data coming from normal operating transformers versus those with known problems.
Transformer size and type included in the database.
Loading practice of the transformer owners included the database.

Another use of the analysis of furanic compounds in the transformer oil is to determine whether
an active problem is causing the 2-FAL generation at an abnormal rate. Typical faults causing
abnormal generation rates include cooling system failure for example. The task force calculated
the generation rate of 2-FAL from the database in those cases where several records from the

Page 40 of 70
same transformer were available. The 90, 95 and 98th percentile typical generation rates were then
calculated. These levels are shown in Figure 29.
400

350

300
B
C
D
250
E
2-FAL(ug/kg)/year

F
2-FAL/year

G
200
H
J
K
150
L
M
N
100

50

0
90 95 98
Percentile

Figure 29: Typical generation rates of 2-FAL (ug/kg)/year in oil. (B: USA, C: Canada, D: Germany,
E: Thailand, F: Belgium, G: Poland, H: Portugal, J: Brazil, K: England, L and N: Italy, M: Austria)

6.3 CONCLUSIONS FROM STATISTICAL STUDY

The ideal methodology of determining appropriate limits for 2-FAL content in transformers is to
analyze data from decommissioned or failed transformers and compare its polymerization degree
profile (as is well known now that different samples of paper taken in the transformer will present
different values of DP which will be influenced by temperature gradient, default localisation, etc.)
to furanic compounds levels in the oil. As this protocol takes time and is very expensive, it is not
commonly used by utilities. Unfortunately, sufficient records of this data are not available.

However, there is a significant amount of data available on furanic compounds (without relation
to DP results) in samples taken from transformers in service. The task force was able to collect
over 30.000 data sets from twelve different sources. This database was supplemented by over
12.000 data sets which had been collected by IEC TC 10 WG 33. This data was analyzed utilizing
the methodology currently used to establish norms for DGA, which is to establish percentile
values of 90%. The 95th and 98th percentiles and the 2-FAL generation rate percentiles were also
calculated.

The large variation in the percentile values for each datasets indicates that a universal value for
establishing normal levels for 2-FAL should not be utilized. Because the source of the data varies
in regards to transformer type, loading practice, presence of thermally upgraded paper and oil
inhibitor, ambient temperature, and test practice, there is a significant difference in the percentile
levels. The task force recommends that any database be first subdivided into subsets which
represent similar equipment in regards to 2-FAL generation and materials (upgraded or not,
breathing or not, etc.). Subsets can then be analyzed to establish normal and abnormal levels by
percentiles.

Page 41 of 70
7 DIAGNOSTIC EXPERIENCE

7.1 SURVEY

7.1.1 Introduction
The production of 2-FAL and its relationship to the age of paper is well-known in the laboratory
environment but how does this relate to the behaviour of 2-FAL in real operating transformers?
To answer this question WG D1.01 (TF 13) prepared a questionnaire that was directed to
potential users.

7.1.2 Methodology
In total, the questionnaire was completed by more than 27 participants around the world. More
than half of the respondents were electricity utilities. The remainings came from laboratories or
research centers while a small percentage came from transformer manufacturers. Figure 30
summarizes the type of respondents who completed the questionnaire. As some participants have
demanded confidentiality, no laboratory or company name is presented.

Electrical utilities

Manufacturers

33%
Reseach centers and
laboratories
52%

15%

Figure 30: Type of respondents who completed the questionnaire.

WG D1.01 (TF 13) identified five (5) areas of interest that were primary focus of the data
gathered in the questionnaire. These areas were:
General issues.
Analytical approaches.
Statistics.
Diagnosis.
Benefits.

Page 42 of 70
7.1.3 Results

General issues
As it is well known now that the evolution of furanic compounds is dependent on the type of
insulation, questions were developed to determine whether users were aware of the nature of oil
and the paper used in their equipment. Results from the questionnaire are presented in Figure 31
and identified that:

30% of respondents were unaware of the nature of the oil or the paper.
Uninhibited oils appear to be used slightly more frequently than inhibited oils.
37% of respondents recognize the existence of thermally upgraded paper in some of their
equipment.
Although the year of introduction of thermally upgraded paper was difficult to determine
(mid 60s), some respondents advised that this type of paper had been used since the
beginning of the 1980s.

Type of insulating oil Type of paper

Unhibited Kraft only


Inhibited Thermally upgraded
30% Both
33% N/A
30% 30%
N/A

15% 37%
26%

Figure 31: Insulation type used in respondents equipments (oil and paper types).

Analytical approach
Questions also focused on identifying the percentage of respondents who use the furanic
compounds analysis as a diagnostic tool.

Results from the questionnaire are presented in Figure 33 and identified that:
Only 4% of the 27 participants indicated that they do not use the method
Almost half of the respondents advised that they use the 5 derivatives
Approximately a third use 2-FAL only
11% cited that they do not use the 2-FOL (which has had considerable stability problems
in laboratory studies)

Page 43 of 70
4%
11% All furanic derivatives

2-FAL only

All but not 2-FOL


52% No use
33%

.
Figure 32: Furanic compounds utilisation preconized by the respondents.

Some respondents were interviewed to gather additional information on the interpretive methods
that were used. The results of these interviews are illustrated in Figure 33.

4% 4%
Absolute value of 2FAL

2FAL rates

Total of 5

Presence of other furanic


59% derivatives
96%

Figure 33: Principal analytical approaches used by the respondents with furanic compounds
results.

It is interesting to note that only a few of the respondents advised that they utilise the derivatives
other than 2-FAL.

Statistics
The questionnaire identified that the most widely used approach is the use of a threshold value.
The majority of respondents (85%) use a threshold of 2-FAL only (see Figure 34). The origin of
the threshold values in most cases (81%) is from personal databases and not values from literature
or laboratory experience (see Figure 35).

This situation was consistent with the findings of this WG. As the presence of the 2-FAL can be
influenced by many parameters, the construction of data banks that take in account the different
parameters found in the companys population of transformers (ex: device type, type of
insulation, temperature operation, age of the equipment, loading, etc.) is the best approach to
adopt.

Page 44 of 70
4%

11%
2-FAL
All
No
l

85%

Figure 34: Type of threshold used by the respondents with furanic compound results.

7%
4% Data banks

Literature only

Lab experiments

19% No answer

81%

Figure 35: Origin of the thresholds values used by the respondents for 2-FAL diagnosis.

Page 45 of 70
Diagnostics
Results from the questionnaire also identified how the respondents think that 2-FAL is related to
equipment (Figure 36).

Age
Load
19%
Type of equipment
11
11% Type of cooling
67 Nothing in particular
% No data
22%

30%
41%

* Note: Percentages are calculated individually for each selection

Figure 36: Relation between the 2-FAL and transformers in service.

An important issue was to determine how the methods are actually applied by respondents (see
Figure 37 for results). It is interesting to note that only 15 percent of surveyed users use the
furanic analysis on a routine basis. The method is primarily used (70%) for condition assessment
or in alarm condition (56%) when it is used as a complementarily method to Dissolved Gas
Analysis (DGA).

A small number of respondents (11%) use the method to estimate the degree of polymerization
(DP) of the cellulosic insulation. It is important to note here that even though the relation between
DP and the evolution of 2-FAL is well known in laboratory tests, this relationship is less
applicable in the field. The difficulty is that DP profiles throughout the insulating system of real
transformers can be quite steep (depending on the type of cooling and design), whereas the
concentration of furanic compounds detected in the oil is a mean value.

A significant number of respondents (26%) indicated that they use furanic derivatives as
diagnostic tool during factory heat run testing. It is assumed that this information is collected with
other diagnostic information such as DGA. It should be noted however that the short time
involved in factory heat run testing is usually not enough to develop representative amounts of 2-
FAL.

Page 46 of 70
Condition assessment

15% 4% Alarm diagnosis test


Factory Heat run test
11%
Preventive
15% 70% maintenance
Insulation thermal fault
DP estimation

15% Routine base


No answer

26%

56%

* Note: Percentages are calculated individually for each field of application

Figure 37: Field of application for furanic compounds as a diagnosis tool.

Benefits
For the majority of respondents (44%), furanic compounds analysis (particularly 2-FAL) is
particularly useful when planning interventions (relocation, refurbishment, prioritization on
scrapping activity updates etc.). This analysis is also well recognized (33% of respondents) for
the detection of thermal faults involving the paper insulation. It is evident from the most answers
here, that DGA is a very important additional criterion, especially when the ratio CO2/CO > 10.
Oil values, e.g. humidity can also be essential. The detailed results are presented in Figure 38.

Planning action

Detection of thermal fault in the


7% paper
4% Load management

Condition assessment
15%
44% Data bank construction

No data
7%

33%

Figure 38: Expected benefits when using 2-FAL analysis.

7.1.4 Conclusion
The survey enabled the Working Group to understand better how furanic compounds (primarily
2-FAL) are used in field applications. It was noted that although most of the surveyed users

Page 47 of 70
integrate their results in data banks, only a very few use them to their full capacity. Indeed very
few laboratories perform this analysis on a routine base.

It was evident that the use of furanic analysis is appreciated by all respondents in specific and
sometimes very different situations. Comparison of replies on diagnosis (Figure 37) and added
benefits (Figure 38) identified that the potential and the limitation of furanic compound analysis
as a diagnostic tools may not be perceived in the same way by all users. However it is clear that
in association with other diagnostic methods, 2-FAL analysis provides extremely useful
information that can be used in the asset condition decision process.

7.2 CASE STUDIES

In many cases the furanic compounds analysis can be very helpful to detect the degradation of
cellulose materials in transformers and together with DGA and other oil tests results can prevent
costly damages. Eight (8) examples are given below to demonstrate the effectiveness of this
method in different field situations.

7.2.1 Case 1: Local hot spot wrapped by cellulose insulation


Generator transformer 181 MVA; 16/220 kV; 18 years in service
The first insulating oil investigation showed a normal DGA pattern but with a rather unusual
CO2/CO - ratio of about 30 (see Table 8). Furanic compounds analysis yielded a total content of
furanic compounds of 0,55 ppm (mg/kg). As the measurement one year later showed no
substantial changes in the gas or furanic compounds the situation was not estimated to be
dangerous. However, examination of the analysis performed two years later exhibited
accelerating degradation (see Figure 39). The decomposition gases (particularly the contents of
ethylene and the carbon oxides) and the furanic compounds had increased considerably. The
transformer was taken out of service and the inspection showed highly deteriorated cellulose
insulation as the result of poorly designed electric connections between the low voltage bushings
and the low voltage windings.

Thermally overstressed insulation Paper deterioration by overheating

Figure 39: Case 1: Insulation degradation.

Page 48 of 70
Table 8: Case 1: Laboratory results (DGA and Furanic compounds).

Date of test 2001 2002 2004


H2 15 48
CH4 18 55
C2H6 10 47
C2H4 41 170
Similar
Composition of gases [ppm] C2H2 0 0
as in
(l/l) CO 1000 3600
2001
CO2 30000 103000
CO2/
30,0 28,61
CO
TCG 1084 3920
2-FAL 0,47 1,81
5-HMF 0,01 0
Similar
Furanic compounds [ppm] 2-FOL 0,03 0,05
as in
([mg/kg]) 2ACF 0,01 0,02
2001
5-MEF 0,03 0,03
Total 0,55 1,91

7.2.2 Case 2: General ageing of cellulose insulation caused by insufficient cooling


Single-phase-excitation transformer of 1000 MVA generator; 134,3 MVA; 27/1,4 kV; 18 years in
service
The poor condition of the cellulose insulation was identified after a Buchholz alarm which was
caused by an electric fault not in direct contact with cellulose material.
The problem was traced to the location of the transformer which had been placed in a closed
room with insufficient cooling. Average oil temperature was about 60C. It is expected that these
conditions caused high levels of CO and CO2 in the oil (940 and 17000 ppm respectively) and
deterioration of dielectric-chemical properties revealed one year before the failure occurred as
presented in Table 9.

Table 9: Case 2: Laboratory results (DGA and Oil tests).

Date of test One year before


Date of test One year before failure
failure
H2 23 Visual inspection clear
CH4 13 Colour 4
C2H6 Neutralization
10 0,05
Dielectric and number[mgKOH/g]
Composition C2H4 7 chemical Water content[ppm] 13
of gases C2H2 parameters Interfacial tension
[ppm] <1 20
[mN/m]
(l/l) CO 940 tg (90C) [%] 0,26
CO2 17000 Breakdown voltage[kV] 76
CO2/
18,1
CO
TCG 994

Page 49 of 70
During internal inspection of the transformer 11 different paper samples were taken and their
degrees of polymerisation (DP) were evaluated in accordance with IEC 60450. The following
results as presented in Table 10.

Table 10: Case 2: Laboratory results (DP and Furanic compounds).

No Sampling place Result


1 Connection winding-bushing (faulty coil), outer layer 230
2 Connection winding-bushing (faulty coil), middle layer 180
3 Connection winding-bushing (faulty coil), inner layer 140
DP of
4 Connection winding-bushing (faultless coil), outer layer 390
paper 5 Connection winding-bushing (faultless coil), middle layer 310
sampled 6 Connection winding-bushing (faultless coil), inner layer 250
from Connection winding-bushing (faulty coil), outer layer, part with
7 360
different black deposits
places 8 HV-winding, outer layer 360
9 HV-winding, inner layer 320
10 Paper below bakelite tube (faulty coil) 200
11 Paper below bakelite tube (faultless coil) 260
Average value 270 80
2-FAL 3,44
5-HMF 0,03
2-FOL 0,48
Furanic compounds [ppm] ([mg/kg])
2-ACF 0,02
5-MEF 0,11
Total 4,1

Using de Pablos 5% equation3 of CIGRE WG 15.01.03 [7]


1850
DP
2,3 Furanic compounds
(ppm)
the average DP value of 290 was estimated, which is in a good agreement with measured data. As
a consequence of these investigations the complete set of the 3 single phase transformers has been
replaced by new instruments to prevent unexpected failure.

7.2.3 Case 3: As complimentary test to DGA in paper overheating


Transformer 75 MVA; 121/10,5 kV; year of manufacture 1982; year of repair 1998; uninhibited
oil; Kraft paper; installed in power station.
Less then two years after repair of the transformer DGA revealed high levels of CO and CO2,
with small amounts of hydrogen and other hydrocarbons. The composition of gases indicated
thermal overheating of paper insulation. Results of subsequent DGA analysis revealed mainly
significant increase of carbon oxides (Figure 40). Furanic compounds analysis identified a
gradual increase of 2-FAL and of the development of other furanic compounds (2-MEF, 5-HMF
and 2-ACF) over time (Figure 41). Analysis indicated progressive process of paper degradation
and suggested that the fault was diagnosed correctly on the base of DGA. Six years later, when

3
This means that 5% of the total cellulose material is involved in the degradation process, and that the remaining 95% of the cellulose
material is absorbing the furanic compounds

Page 50 of 70
CO and CO2 levels reached respectively 780 ppm and 13250 ppm (Table 11), the user decided to
carry out internal inspection of the transformer which revealed deformations of edge rings made
of soft insulation on the phase B (see Figure 42). The deformed edge rings caused significant
difficulties in the oil flow resulting in an excessive rise of winding temperature and overheating
of the paper insulation. Sampled paper showed noticeable brittleness of some fragments.
Measured degree of polymerization of 498 corresponded to more then 50% ageing of paper
insulation, which was evaluated as excessive for a transformer of this age with its loading history.
As replacement of the transformer was planned within two years with a new unit, the repair was
limited to mounting of additional insulating washers to improve oil transfer of the oil ducts and
an improvement in heat transfer.

Table 11: Case 3: Laboratory results (DGA and Furanic compounds).

Composition of Time of operation after repair [years]


gases [ppm]
0,1 0,8 1,4 1,8 2,4 3,8 5,3 6,2 7,3 7,6
H2 0 0 5 3 5 6 14 15 13 14
CH4 2 2 5 47 7 999 9
C2H6 0 0 1 12 2 334 4
C2H4 11 13 28 26 49 56 86 94 103 106
C2H2 0 0 0 00 0 000 0
C3H8 0 0 1 00 0 011 1
C3H6 0 0 1 11 1 223 4
CO 296 185 639 381 770 764 780 765 713 780
CO2 2857 2598 6165 5027 9234 9185 10900 11270 12469 13250
CO2/ CO 9,7 15,7 9,7 13,2 12,0 12,0 14,0 14,7 17,5 17,0
TCG 309 200 680 416 834 836 894 889 846 918
2-FAL 0,35 0,62 0,77 0,77 0,82
5-HMF 0 0 0,02 0,02 0,02
Furanic compounds 2-FOL 0 0 0 0 0
[ppm]
([mg/kg]) 2-ACF 0 0 0 0,01 0,02
5-MEF 0,02 0,03 0,03 0,03 0,04
Total 0,37 0,65 0,82 0,83 0,90

14000

12000

10000
gas concentration [ppm]

8000

6000

4000

2000

0
0 1 2 3 4 5 6 7 8
time of operation [years]

CO CO2 CnHm

Figure 40: Case 3: Fault gases evolution.

Page 51 of 70
0,9

0,8

0,7

0,6

0,5
[ppm]
0,4

0,3

0,2

0,1

0
3 3,5 4 4,5 5 5,5 6 6,5 7 7,5 8
Time of operation [years]

2FAL 5HMF 2FOL 2ACF 5MEF

Figure 41: Case 3: Furanic compounds evolution.

Figure 42: Case 3: Insulation damage in transformer 75 MVA.

7.2.4 Case 4a & b: Different 2-FAL results in two transformers with similar internal faults
assessed by DGA
Case 4a. Regulating transformer; 20/20/10 MVA; 15,65/6,3/6,3 kV; year of manufacture: 1972;
year of internal inspection and repair: 2005; uninhibited oil; Kraft paper; installed in power
station.
After 28 years of operation the concentration of gases in the transformer was low and carbon
dioxide level did not exceed 3300 ppm. A planned internal inspection was performed connected
with changing of the oil and the cooling system. Only one year of operation after the planned
work a high level of carbon dioxide (15380 ppm) and other combustible gases characteristic for
internal low-temperature overheating were identified. Diagnosis was accelerated decomposition
of transformer insulation, including also paper. Analysis of furanic compounds revealed
2,5 mg/kg 2-FAL and 0,05 mg/kg 5-HMF, which confirmed that significant degradation of the
paper insulation had already occurred. Two consecutive measurements revealed further excessive

Page 52 of 70
CO2 increase but lower increase of other gases, particularly saturated hydrocarbons. The ratio
CO2/CO gradually rose. Simultaneously 2-FAL reached the level of 5,49 mg/kg with total
furanic compounds rate of increase of 0,13 mg/kg/month. In addition to 2-FAL and 5-HMF there
appeared to be also other furanic compounds (2-ACF and 5-MEF) in measurable amounts (see
Table 12). DGA results and furanic compounds analysis in this case agreed indicating accelerated
thermal decomposition of the paper insulation caused either by defect within windings or
ineffective cooling (in spite of changed cooling system). Unfortunately, as DPv is an intrusive
measurement, no samples were taken when the transformer was opened and repaired.

Table 12: Case 4a: Laboratory results (DGA and Furanic compounds).

Time of operation [years]


Composition
of gases Before repair After repair
[ppm] 23 27 28 0,1 1 2 2,5
H2 24 53 56 2 320 323 330
CH4 27 19 49 1 229 246 250
C2H6 17 11 18 0 277 329 330
C2H4 16 25 55 1 45 108 161
C2H2 16 17 11 0 0 0 0
C3H8 66 7 0 71 76 84
C3H6 8 13 22 0 15 20 23
CO 174 404 409 25 345 362 369
CO2 1158 2309 3272 187 15380 20840 25290
CO2/ CO 6,7 5,7 8,0 7,5 44,6 57,6 68,5
TCG 288 548 627 29 1252 1464 1547
2-FAL 2,5 4,72 5,49
5-HMF 0,05 0,13 0,15
Furanic
compounds 2-FOL 0 0 0
[ppm] 2-ACF 0 0,01 0,02
([mg/kg])
5-MEF 0 0,06 0,06
Total 2,55 4,92 5,72

Case 4b. GSU transformer 240 MVA; year of manufacture 1971; after modernization in 1996
rated power increased to 263 MVA; 250/15,75 kV; uninhibited oil; Kraft paper; installed in
power station.
The amount of carbon dioxide increased continuously after the transformer had been uprated.
Currently the level of CO2 is above 21000 mg/kg (Table 13). The rate of increase of combustible
gases has also risen. Routine DGA analysis indicates, as in the previous case, significant ageing
of paper insulation in the transformer suggesting a presence of low temperature overheating of
medium activity.
It is expected that one of the causes of excessive thermal ageing of insulation (that is suspected) is
malfunctioning of the cooling system, after uprating of the transformer. Furanic compounds
analyses performed several times show low 2-FAL level, rising from only 0,09 mg/kg to
0,13 mg/kg over three years (see Table 13). In this case furanic compound measurements do not
confirm high degradation of cellulose nor its accelerated character. This equipment is still under
surveillance.

Table 13: Case 4b: Laboratory results (DGA and Furanic compounds).

Page 53 of 70
Time of operation after
10,0 10,4 10,8 11,0 12,2 12,7 13,0
modernization [years]
H2 151 135 166 182 181 198 233
CH4 104 97 97 114 115 126 137
C2H6 130 110 118 128 127 130 141
C2H4 54 46 62 73 73 84 92
Composition C2H2 28 23 20 28 30 39 50
of gases C3H8 91 79 75 79 80 83 82
[ppm] C3H6 21 23 22 24 23 22 24
CO 281 267 306 318 325 368 383
CO2 1840 12718 15180 18870 19313 20177 21324
CO2/ CO 6,5 47,6 49,6 59,3 59,4 54,8 55,7
TCG 860 780 866 946 954 1050 1142
2FAL 0,09 0,12 0,12 0,12 0,12 0,12 0,12
Furanic 5HMF 000 0 0 0
compounds 2FOL 000 0 0 0
[ppm] 2ACF 000 0 0 0
([mg/kg])
5MEF 000 0 0 0
Total 0,09 0,12 0,12 0,12 0,12 0,13 0,13

7.2.5 Case 5: Restoring of 2-FAL after oil changing; accelerated process of cellulose
degradation mutually confirmed by DGA and 2-FAL results
Transformer 150 MVA; 125/15,75 kV; year of manufacture: 1998; uninhibited Nytro 10GBN*
changed into NytroTaurus**; Kraft paper; installed in power station.
*Nytro 10 GBN _ uninhibited oil, which was proved to be one of most corrosive ones, currently not manufactured
**Nytro Taurus _ not corrosive, uninhibited oil of new generation

After several years of transformer operation a significant increase in carbon oxides and
combustible gases characteristic for low temperature overheating was found. Consecutive DGA
confirmed slow development of the fault. Furanic compound analysis revealed the presence of
1,44 mg/kg 2-FAL and two other derivatives, 5-HMF (0,05 mg/kg) and 5-MEF (0,02 mg/kg)
suggesting significant degradation of the cellulose in the transformer for a transformer of such a
short operation duration. Oil parameters deteriorated (particularly dissipation factor tan,
resistivity, neutralization value and interfacial tension), until sludge appeared, indicating a high
level of aging of the insulating oil. Analysis of corrosive sulfur (acc. to IEC 62535 Ed.1) revealed
that the oil was potentially corrosive. Based on these results it was decided to change the oil.
After the oil change, DGA carried out two times during six months following the oil change
revealed a dynamic increase of carbon dioxide, the level of which is now about 16000 ppm (see
Figure 43 and Table 14). Measurement of furanic compounds revealed in a very short time a
result similar to that obtained before the oil change. It is expected that in this case the relatively
high level of 2-FAL (1,26 mg/kg) following the oil change is a result of not only the release of
the furanic compounds from the paper, (on the basis of achieving equilibrium), but comes from
generation of additional products of paper decomposition with high rate of formation. It indicates
an accelerated process of cellulose degradation which should further be monitored.

Page 54 of 70
1,6 18000
2FAL 1.44
1,4 16000
2FAL 1.26
14000
1,2
CO2
12000
1
10000
[ppm]

[ppm]
0,8
8000
0,6
oil change 6000

0,4
4000

0,2 2000

0 0
15-11-2006 26-2-2007 23-5-2007 10-8-2007 22-11-2007 22-2-2008
date

Figure 43: Case 5: Evolution of 2-FAL and CO2.

Table 14: Case 5: Laboratory results (DGA and Furanic compounds).

10.08.07
Date of sampling 15.11.2006 26.02.2007 23.05.2007 22.11.2007 22.02.2008
Oil change
H2 45 51 53 0 10 14
CH4 98 112 130 0 14 16
C2H6 173 188 293 1 18 24
C2H4 22 22 40 0 12 18
Composition C2H2 2 2 202 3
of gases C3H8 48 52 74 2 5 6
[ppm] C3H6 14 14 20 0 5 5
CO 429 430 440 3 319 425
CO2 6098 6325 7223 49 8867 16197
CO2/ CO 14,2 14,7 16,4 16,3 27,8 38,1
TCG 831 871 1052 6 385 511
2-FAL 1,44 1,26
Furanic 5-HMF 0,05 0,05
compounds 2-FOL 0 0
[ppm] 2-ACF 0 0
([mg/kg])
5-MEF 0,02 0,02
Total 1,51 1,33

7.2.6 Case 6: Restoring of 2-FAL after repair


Six rectifiers; 48 MVA; 120 kV; 45 tons of inhibited oil Nytro 10X; open breathing; no OLTC;
year of manufacture: 2002; year of repair 2003 or 2004.
Six rectifiers, identified by numbers 201-206, were put into service in 2002, with some of them
suddenly failing in 2003 and 2004 without warning from DGA (gas levels were very low for all
gases). The Buchholz alarm suggested a fault which involved paper. Internal inspections
confirmed the diagnosis of overheating in oil and paper. All rectifiers were repaired. Several
DGAs carried out after repair revealed increasing CO2 levels and CO2/CO ratios (see Figure 44
and Figure 45).

Page 55 of 70
CO2content
CO2/CO
no
no v

0
1000
2000
3000
4000
5000
6000
7000
8000
9000

0
5
10
15
20
25
30
35
40
ja 0 3
ja 03 n
n m v0
m v 0 ar 4
ar 4 s
s m 04
m 04 ai

ai
ju 04
ju 0 4 i
i se l04
se l04 pt
pt no 04
and 2004
n o 0 4 v
v ja 0 4
ja 0 4 n
n m v0

After repair in
m v 0 ar 5

2003 and 2004


ar 5 s
After repair in 2003

s m 05
m 05 ai
ai
ju 05
ju 0 5 i
i se l05
se l05 pt
pt
no 05
n o 0 5 v
v ja 0 5
ja 0 5 n
n m v06
m v06 ar
ar s
s m 06
m 06
201

ai

have remained very low (see Figure 46).


ai

201
ju 0 6 ju 06
i il
se l06 se 06
pt pt
no 06
n o 0 6 v
202

202
ja 0 6 ja 0 6
n nv
m v 0 m 0
ar 7
ar 7
s s
m 07
Month

m 07

Figure 45: Case 6: CO2/CO versus sampling date.


ai
203

203
ai
ju 0 7 ju 07
i il
se 07

Month
se l07 pt
pt
Figure 44: Case 6: CO2-content versus sampling date.

Page 56 of 70
no 07
n o 0 7
v
204 v
ja 0 7
204

ja 0 7 nv
n m 0
m v 0 ar 8
ar 8
s s
m 08 m 08
ai

205

ai
ju 08
205

ju 0 8 il
i se 08
se l08 pt
pt
no 08
n o 0 8 v
206

v ja 0 8
ja 0 8
n n
206

m v 0 m v0
ar 9
ar 9
s s
m 09 m 09
ai
ai
ju 0 9 ju 09
i il
se l09 se 09
pt pt
n o 0 9 no 09
v v
ja 0 9
ja 0 9
n n
m v 1 m v10
ar
ar 0
s s
m 10 m 10
ai
ai
ju 1 0 ju 10
il il
10 10

(2 and 3 years after). This example shows, that in spite of elevated CO2 levels, results of 2-FAL
Furanic compounds in the oil were also analysed: once before and only two times after repair
201 202 203 204 205 206

0,06

0,05

Afterrepairin2003
Valuesbeforerepair and2004
0,04
2-FAL

0,03

0,02

0,01

0
2003 2004 2005 2006 2007 2008 2009
Year

Figure 46: Case 6: 2-FAL content (mg/kg) versus sampling date.

7.2.7 Case 7: Restoring of 2-FAL after oil reclaiming; mutually confirming CO2, CO2/CO
and 2-FAL results
Four different transformers; 10/13 MVA; 60/63 kV; 5-7 tons of oil; open breathing.
Three built in 1961 with Kraft paper, one built in 1970 with thermally upgraded paper (Insuldur).
Four transformers, identified by numbers T1-T4, were reclaimed in 2002. In comparison with
results obtained before the reclamation, the values of CO2 and CO2/CO following reclaiming
were found to be decreasing. The transformer with upgraded paper shows lower values than other
ones (Figure 47 and Figure 48).

25000
T1
T2
T3
T4
20000

15000
CO2contentinul/l

10000

5000

0
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

YEAR

Figure 47: Case 7: CO2 content versus sampling date.

Page 57 of 70
30,0
T1
T2
T3

25,0 T4

Reclaimedin2002forall
20,0

CO2/CO

15,0
Newreference

10,0

Thermalupgradedpaper

5,0

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

YEAR

Figure 48: Case 7: CO2/CO versus sampling date.

Additionally furanic compounds analysed after reclaiming have drastically decreased in relation
to reference values obtained before (there is big difference between 2-FAL level in 1999 and
2006). Lowest values are for thermally upgraded paper. There is the same tendency in changes of
carbon dioxide, CO2/CO ratio and 2-FAL (Figure 49).

1,6

T1
1,4 T2

T3

1,2 T4
Reclaimedin2002forall

1
2FALmg/kg

0,8

0,6
Newreference

0,4

0,2

Thermalupgraded
0
1999 2000 2001 2002 2003 2004 2005 2006 2007

YEAR

Figure 49: Case 7: 2-FAL content versus sampling date.

Page 58 of 70
8 CONCLUSION

8.1 CONCLUSIONS

The WG D1.01 (TF 13) had the task of evaluating the behaviour of the furanic compounds and
their potential as a diagnosis tool. The work performed by the WG presented in this report has
produced the following clear statements:

The measurement of furanic compounds in the insulating oil of transformers is well


controlled in the majority of the labs involved in this study. Recommendations will be
presented to IEC TC 10 in this regard.
Only 2-FAL is used in common practice in asset management decisions.
The use of other furanic compounds also detected in oil (2-ACF, 5-MEF, 2-FOL,
5-HMF), even though it is not doing unanimity, has shown interesting results which may
lead to new diagnosis information.
Generation of 2-FAL and other furanic compounds is influenced by the ambient
conditions (humidity, temperature, oxygen etc.).
Generation of 2-FAL and other furanic compounds is also dependent on the quantity and
type of paper involved. It is dependant on whether it is general ageing or localised ageing.
Concentrations of 2-FAL and other furanic compounds measured in oil will be influenced
by their ability to remain in the paper after their formation or to dissolve in the oil.
Concentrations of 2-FAL and other furanic compounds are in equilibrium between the oil
and paper, but the partition between the two is highly influenced by temperature,
moisture, oxygen, oil acidity, type of paper and also the design of the electrical
equipment.
Concentrations of 2-FAL and other furanic compounds measured in oil seem to be also
influenced by the presence of inhibitor in oil.
It is difficult to correlate 2-FAL with ageing (DP) of the paper insulation in electrical
plant such as transformers. Correlation in laboratory aged experiments on the other hand
is very good.
Absolute values of 2-FAL are used as a general indication of ageing, whereas the rate of
increase of 2-FAL being produced can be more helpful in specific cases at determining
the extent of ageing.
2-FAL threshold values cannot be universally applied as a diagnostic tool as there are too
many factors influencing the evolution of 2-FAL and also the concentration in the oil.
2-FAL threshold values can be determined on a group or family of equipment (plant),
where enough information is available from scrapped units. Many utilities use a data bank
of furanic compounds or only 2-FAL data with success.
Working Group D1.01 (TF 13) recommends that any database be first subdivided into
subsets which represent similar equipment in regards to 2-FAL generation. The subsets
can then be analyzed to establish normal and abnormal levels by percentiles.

Page 59 of 70
8.2 RECOMMENDATIONS

2-FAL analysis is a valuable condition monitoring tool for determining the ageing of solid
insulation. It is more useful in combination with other oil diagnostic methods to determine
insulation condition. This analysis should be done at least each time thermal degradation of the
paper insulation is suspected. However, it is clear that setting up a database will also require 2-
FAL measurements on equipment not showing any faults. In this respect, a routine analytical
campaign (that could be incorporated into the oil quality control campaigns already in place)
could be considered.

The analysis of the five other furanic compounds could be stored on an experimental basis. As
additional data become available, the few diagnostic models proposed for these other compounds
(i.e. specificities, ratios) could be more easily explored.

Further work is needed to improve the understanding of 2-FAL as a diagnostic tool, such as
Improvement of the statistical analysis (data mining) and general
acknowledgment
Collaboration with IEC.
Collaboration with IEEE.
Influence of inhibitors, passivators and other additives.
Impact on diagnosis for new insulating fluids (e.g. ester base liquids, bio-fluids).
Complementarity with other recent compounds of interest which may also
provide information on the condition of the paper insulation (methanol,
ammonium, acetone, etc.).

Page 60 of 70
ANNEX 1: SUMMARY OF METHODS USED TO DETECT FURANIC COMPOUNDS IN OIL

Method Detection limit Advantages Disadvantage s References


(g/kg)
100-200 Fast and low in cost Semi-quantitative Dominelli et al. (6) (1993)
Wet test
Field use potential
Thin-layer chromatography 500 Low in cost; sequential analysis Semi-quantitative Cataldi et al. (9)(1990)

Fast and low in cost Includes an extraction procedure Ortel et al. (10) (1992)
Colorimetry
Pahlavanpour et al. (11)
5-100
(Complexing with aniline (1995)
acetate)
Gas chromatography Quantitative; detects other furanic Oil interference in most cases requires that the
compounds sample be properly purified
Solid-phase extraction (SPE) 100-200 Azizian et al. (12) 1989)
and flame ionization detection
(FID) Aralkelyan (13) (1993)

Headspace extraction 300-1000 Combined dissolved-gas analysis High detection limit Gilbert et al. (14) (1994)
And universal detector (screening test)

High-performance liquid
chromatography
Ultraviolet visible 1-30 Sensitive; detects other furanic Purification by liquid-liquid or solid-phase Burton (1) (1984)
detector (with or without a compounds and phenols extraction required (12)
diode array system) Azizian et al. (1989)
Potential interference of oxidized oil (15)
Grant (1992)
Dominelli et al. (16) (1991)
FairHolme (17) (1991)
Direct injection 5 Does not require any purification Lessard et al (1992)
by extraction

Coupled with mass nd Useful for evaluating Complex and costly Korh et al. (18,19) (1998)
spectrometry experimental models

Page 61 of 70
ANNEX 2: ROUND ROBIN TEST ADDITIONAL RESULTS
Table 2-FAL
2-FAL LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB
(mg/kg) No. 1 No. 2 No. 3 No. 5 No. 6 No. 7 No. 8 No. 9 No.10 No. 12 No. 13 No. 14 No. 11 No. 11 *

A 0,27 0 ,28 0,24 0,30 0,29 0,26 0,30 0,29 0,22 0,27 0,28 0 ,3 0,12 0 ,24

B 0,03 <0 ,05 0,08 0,03 0,04 0,05 0,03 0,06 0,04 0,05 <0,002 <0,05 0,01 0 ,03

C 1,59 1 ,54 1,45 1,56 1.60 1,45 1,49 1,60 1,55 1,55 1,52 1 ,65 1,65 1 ,51

Table 5-HMF
5-HMF LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB
(mg/kg) No. 1 No. 2 No. 3 No. 5 No. 6 No.7 No. 8 No. 9 No. 10 No. 12 No. 13 No. 14 No. 11 No. 11 *

A 0,24 0 ,25 00 ,30 0,28 0,19 0,26 0,9 *** 0,24 0,25 0 ,27 0,12 0,23

B < 0,01 < 0,05 0 < 0,01 < 0,002 < 0,001 < 0,002 < 0,03 *** 0,02 <0,002 < 0,05 < 0,01 0,01

C < 0,01 < 0,05 0 *** 0,004 < 0,001 < 0,002 < 0,05 *** 0,07 0.01 < 0,05 < 0,01 0,03
*** This product is not analysed by this laboratory
Table 2-ACF
2-ACF LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB
(mg/kg) No. 1 No. 2 No. 3 No. 5 No. 6 No. 7 No. 8 No. 9 No. 10 No. 12 No. 13 No. 14 No. 11 No. 11 *

A 0,30 0 ,27 0 ,42 0,30 0,29 0,29 0,28 0,28 0,21 0,32 0,29 0 ,29 0,16 0,30

B < 0,01 < 0,05 0,12 < 0,01 < 0,002 0,01 < 0,002 < 0,03 < 0,02 0,02 < 0,004 < 0,05 < 0,01 < 0,01

C < 0,01 < 0,05 0,22 < 0,01 0,01 0,04 < 0,002 < 0,03 < 0,02 0,05 < 0,004 < 0,05 0,07 < 0,01

Table 2-FOL
2-FOL LAB LAB LAB LAB LAB LAB LAB Lab LAB LAB LAB LAB LAB LAB
(mg/kg) No. 1 No. 2 No. 3 No. 5 No. 6 No. 7 No. 8 No. 9 No. 10 No. 12 No. 13 No. 14 No. 11 No. 11 *
A 0,20 0 ,20 00 ,30 0,23 0,13 *** 0,27 0,14 0,20 0,28 0 ,26 0,12 0,18

B < 0,01 < 0,05 0 < 0,01 < 0,002 0,02 *** < 0,05 < 0,02 < 0,02 < 0,01 < 0,05 < 0,01 0,02

C < 0,01 < 0,05 0 < 0,01 < 0,002 0,03 *** < 0,05 0,02 < 0,02 < 0,01 < 0,05 < 0,01 0,02
*** This product is not analysed by this laboratory
Table 5-MEF
5-MEF LAB LAB No. LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB LAB
(mg/kg) No. 1 2 No. 3 No. 5 No. 6 No. 7 No. 8 No. 9 No. 10 No. 12 No. 13 No. 14 No. 11 No. 11 *

A 0,26 0 ,26 0,58 0 ,30 0,26 0,24 0,27 0,27 0,25 0,24 0,26 0 ,28 0,13 0,26

B < 0,01 < 0,05 0,37 < 0,01 < 0,002 0,001 <0,002 < 0,03 < 0,02 0,024 <0,004 <0,05 < 0,01 < 0,01

C 0,05 < 0,05 0,37 < 0,01 0,07 0,04 <0,002 < 0,03 0,08 0,040 0,06 <0 ,05 < 0,01 0,05

Page 62 of 70
ANNEX 3: LITERATURE REVIEW OF THE STATISTICAL POPULATIONS RETAINED FOR THE STUDY OF 2-FAL IN TRANSFORMERS
Document prepared by R. Gilbert, J. Jalbert and M.-C. Lessard
Institut de recherche dHydro-Qubec (IREQ), June 2005
Authors Year Country Number Nature of statistics Other information
of units (Type of equipment, of insulation,
Oil/antioxidant, Open/closed system
Cooling mode, etc)

Burton, Carballeira, 1988 United Kingdom 100 (United Field measurements to determine a logarithmic relationship between 2-FAL concentrations (mg
Duval, Fuller, Graham, and elsewhere Kingdom) of 2-FAL per kg of paper) and the DPV.
de Pablo, Samat and CIGR Task Force
Spicar 15.01.03
Domun 1990 United Kingdom 500 Determination of normal 2-FAL concentrations in an old healthy transformer (2-FAL < 2 mg/L)
Griffin 1991 United States Data on normal and abnormal concentrations obtained by Griffin in the literature (references:
(Doble Eng. Domun 1990 (2-FAL < 2 ppm healthy transformer), Grant 1990 (2-FAL < 5 ppb
Company) transformer under normal operation), Azizian 1989 (various furanic compounds 40-170 ppb
units withdrawn from service) and Burton 1988 (2-FAL 1-10 ppm old units with inadequate
cooling))
Dominelli 1991 Canada (Powertech 350 User guides for 2-FAL levels with DPV equivalencies (0-0.1 ppm representing a DPV between Generator transformers (64%), shunt
Labs Inc., BC- 1200 and 700 healthy transformer, 0.1-1.0 ppm representing a DPV between 700 and 450 reactors (9%), current transformers (25%),
Hydro) moderate deterioration, 1-10 ppm representing a DPV between 450 and 250 extensive autotransformers, etc.
deterioration and > 10 ppm representing a DPV < 250 end-of-life criterion) (information
taken from Powertech Labs Web site)
Allan, Jones and Sharp 1991 Australia (QEC) 200 Changes in normal 2-FAL levels over time (0-40 years) (also for CO and CO2) for a relatively Power transformers
homogenous family of equipment (addresses lifespan)
Chendong 1991 China (EPRI) 77 + 345 Statistics on 77 step-up transformers in service with which a relationship over time could be 100-500-kV generator transformers (77) +
established for 2-FAL levels, based on which three graphical areas could be identified substation transformers (212)
corresponding to different types of aging: slow, normal and abnormal. This relationship was then
tested on 345 transformers of all types.
Andersson and 1992 Finland (Imatran 57 Determination of average 2-FAL levels and values corresponding to 90% of concentrations Power transformers rated 400 kV (18),
Heinonen Voima Oy (IVO)) measured by type of equipment and by voltage level. The authors provide a table on the statistical 220 kV (8), 110 kV (20), and 400-kV
distribution of the furanic compounds found in all the transformers (56% of units with 0-0.1 mg/L current transformers (6) and 10-kV shunt
of 2-FAL; 70% with 0-0.3 mg/L of 2-FAL, 89% with 0-1.0 mg/L of 2-FAL and 95% with 0-3.0 reactors (5) //Column transformers with
mg/L of 2-FAL). open conservator and cooling
Hiironniemi, 1992 Finland (ABB ~100 Power transformers (transmission and
Nordman, Elovaara, Strmberg Power step-up transformers (GSU) > 2 MVA)
Ojanen and Andersson Oy and IVO)
Ortel 1992 United States 518 Statistical distribution of 2-FAL concentration levels based on the type of equipment and voltage Power transformers > 2 MVA: 13/69-kV
(Potomac Electric levels (0-0.1 mg/L, 0.101-1 mg/L, 1.001-5 mg/L, 5.001-10 mg/L, above 10.001 mg/L) on distribution transformers (242), 115/230-
Power Company equipment originating from the same electrical service. Readers note: interesting since the kV transmission transformers (122), 4-kV
System (PEPCO)) units all come from the same electrical service in the U.S. (may serve as a basis for comparing distribution transformers (83), auxiliary
Hydro-Qubecs sealed and unsealed equipment) transformers (45) and step-up
transformers (GSU) (26)

Page 63 of 70
Authors Year Country Number Nature of statistics Other information
of units (Type of equipment, of insulation,
Oil/antioxidant, Open/closed system
Cooling mode, etc)
Myers and Sans 1992 United States (S.D. 315 Determination of 2-FAL transition concentration levels between acceptable, questionable and
Myers) unacceptable areas (units with total furanic compound content < 100 ppb would be acceptable,
those with more than 100 ppb would be questionable, and the paper of those with more than 250
ppb would be severely damaged)
Allan 1993 Australia (QEC) 40 Determination of normal 2-FAL levels based on a homogenous family of equipment (addresses
lifespan)
dePablo, Andersson, 1993 Spain, Finland, 5005 Statistical distribution of 2-FAL concentration levels by participating laboratory and breakdown Systems with conservator that is opened
Knab, Pahlavanpour, Switzerland, UK, of data by power rating, voltage level and age (data from five European countries). Determination or closed by a diaphragm //Oil without
Randoux, Serena and Belgium and Italy of 95% level (within a range of 1 to 5 ppm for all labs except one) and 99% level (within a range oxidant
Tumuatti of 1 to 5 ppm for labs A and B, 5 to 10 ppm for labs C, D and F, and up to 10 ppm for lab E) for
the concentrations measured. As a result, 2-FAL is shown to be the main product and includes 5-
CIGR Task Force
HMF and 5-MEF, and the lower concentrations are found in transformers with high voltage and
15.01.03
power ratings (54% of the 5005 units with 2-FAL < 0.1 ppm; 80.9% with 2-FAL < or = 0.50
ppm; 89.2% with 2-FAL < or = 1 ppm; 98.5% with 2-FAL < or = 5 ppm and 99.6% with 2-FAL
< or = 5 ppm).
Griffin, Lewand and 1993 United States 104 Statistical distribution of 2-FAL concentration levels (0-10 ppb, 11-50 ppb, 51-100 ppb and 101- Power transformers (> 69 kV, > or = 10
Finnan (Doble Eng. 500 ppb) and determination of values corresponding to 90% (87 ppb) and 95% (105 ppb) of the MVA, no recent degassing and no recent
Company) concentration levels measured. Readers note: interesting since larger in scope than Ortels oil treatment) representative of power
1992 study but still limited to power transformers in the U.S. transformers in the U.S.// Population
representative of units in the U.S. use
of thermostabilized paper
Ali, Eley, Emsley, 1996 United Kingdom 200 The normal 2-FAL content varies between 100 and 1000 ppb with an upper limit of the normality
Heywood and Xaio (National Grid range at around 1 to 2 ppm (all cases with 5 ppm and over should be a source of concern)
Company and Readers note: interesting as the population is limited to the transformers from a single country,
University of the United Kingdom
Surrey)
dePablo and 1997 Spain, Finland, 5005 Statistical distribution of 2-FAL concentration levels by participating laboratory and breakdown Oil without antioxidant
Pahlavanpour Switzerland of data by power rating, voltage level and age (data from five European countries). Determination
(Andersson, Camino, United Kingdom, of 95% and 99% levels for the concentrations measured (see description for dePablo et al, 1993).
Carballeira, de Pablo, Belgium, Italy
etc.)

CIGR Task Force


15.01.03
Sans, Bilgin and 1998 United States and 12231 Statistical distribution on the sum of all compounds with user guides for the values with DPV
Kelley elsewhere (S.D. correspondence (< 100 ppb 75 percentile, usually healthy equipment with DPV between 1200
Myers Inc.) and 444; between 101 and 250 ppb 87 percentile, equipment health questionable with DPV
between 443 and 333; between 251 and 1000 ppb 96.5 percentile, equipment in poor
condition with DPV between 332 and 237; between 1001 and 2500 ppb 98.5 percentile,
equipment with reduced reliability with DPV between 236 and 217; > 2500 ppb > 98.5
percentile, equipment with DPV < 217 where the reliability is so low that the transformer must be
rewound or basically replaced) (addresses lifespan).

Page 64 of 70
Authors Year Country Number Nature of statistics Other information
of units (Type of equipment, of insulation,
Oil/antioxidant, Open/closed system
Cooling mode, etc)

Bilgin, Shkolnik and 1999 United States and 18763 Furanic compound content is lower than that reported by other authors, which makes the authors
Kelly elsewhere (S.D. believe that dicyandiamide would have an effect on 2-FAL levels in the oil. Dicyandiamide is
Myers Inc.) used as an additive in the insulating paper used in transformers in the U.S. (thermostabilized
paper). This type of impregnation would not be used elsewhere than in the U.S. and Canada
(reported by Griffin, Lance, Lewand, Finnan and Barry, 1993). Extensive statistics on 18,763
units reveal 2-FAL concentrations of 0-100 ppb for 77% of the units (14,373) (healthy
transformers), between 0 and 250 ppb for 87.2% of the units (16,358), between 0 and 1000 ppb
for 89.9% of the units (17,983) and between 0 and 2500 ppb for 98.5% of the units (18,485), for
an average value of 211 ppb for all the units.
Soares, Ricardo, 2001 Brazil and United These authors add in relation to the maximum concentration levels proposed by Ali, E ley,
Heatley and Rodrigues Kingdom Emsley, He ywood and X iao in 1996 that concentration levels greater than 10 ppm in a
Departamento de transformer are indicative of overheating, which will inevitably lead to electrical failure.
Quimica e
Inorganica,
Universitade
Federal do Ceana,
Brasil and
Department of
Chemistry,
University of
Manchester,
Manchester, UK
Shang, Yang, Guo and 2001 China (Xian 77 Based on the statistical measurements of 2-FAL levels over time in 77 transformers, they Step-up transformers
Yan Jiaotong established graphical areas corresponding to different types of aging: slow, normal and abnormal.
University) Readers note: strangely similar to what Chendong presented in 1991!
Mulej, Varl, Konan 2003 Slovenia (Milan 300 Statistical distribution of 2-FAL concentrations based on age group. Determination of levels for Transformers rated 110, 220 and 400 kV
and Gradnik Vidmar Electric different percentiles (0-5 years 90% of units with levels between 0 and 0.05 ppm; 6-10 years //High-quality oil with antioxidant//
Power Research 90% of units with levels between 0 and 0.05 ppm; 11-15 years 84% of units with levels Sealed equipment
Institute) between 0 and 0.05 ppm; 16-20 countries 83% of units with levels between 0 and 0.05 ppm;
21-30 years 84% of units with levels between 0 and 0.1 ppm; and > 30 years, 88% of units
with levels between 0 and 1 ppm) and comparison with the results of five European countries
(CIGR) and the U.S. The comparison shows that their 2-FAL concentrations for all the
transformer age groups are still substantially lower than those of other European countries. The
authors note that it is well known that most European transformers are filled with uninhibited oil
with low oxidation resistance, while in Sovakia, the equipment has traditionally been filled with
Technol. high-quality oils and that over the last 25 years these oils in most cases contained an
oxidant.
Dong, Zhang and Yan 2004 China (Xian 699 (137 + 562) The authors report that the literature proposes maximum 2-FAL levels in equipment: < 0.1 mg/L Step-up transformers (137) and substation
Jiaotong (ppm) acceptable; > 0.1 mg/L questionable; and > 0.25 mg/L unacceptable. The authors transformers (562) (< or = 110 kV, 200
University) provide graphs for the statistical distribution of the 2-FAL log based on voltage levels. kV and > or = 330 kV)

Page 65 of 70
ANNEX 4: CIGRE WG D1.01 (TF 13) TRANSFORMER OIL LAB SURVEY

Participant: _____________________________
Contact:________________________________

1-Does your company include analysis of furanic compounds? YES or NO


If yes : Which one? 2-FAL 2-FOL 5-HMF 5-MEF 2-ACF

2-What use do you make of the furanic data?

Preventive diagnostics test (Please indicate the frequency) _____

Alarm diagnostics test _____


When DGA suspected a hot spot (IEC, Duval. Rogers, etc..)____
Others (ex: CO2/CO ratio are below 3 and above 10) ____

Factory heat run test _____

Polymerization degree evaluation _____

Condition assessment _____

3-Can you report specif ic benefit you have obt ain from the used of furan ic
compounds analysis?

Replacement of equipment ____

Load management ____

Detection of thermal fault including insulating paper ____

Help for planning refurbishment, relocation, other ____

4- Any other information which you fell will be benefit for the CIGRE into the use
of furanic analysis?

5-Typical Data

Do you have typical data for furanic compounds?


Which compounds ? 2-FAL 2-FOL 2-HMF 5-MEF 2-ACF Values:

Page 66 of 70
Origin of those values ?

Data bank study____


Literature ____
Own experience ____

Are you following

Absolute data____
Trending ____
Other derivative compound following ____

Are the typical data related to some characteristics of the equipment ?


General ____
Age ____
Loading ____
Type of equipment ____
Type of cooling ____
Other : ____________________________________

Information about the transformers population

Type of oil filling: inhibited _____ non inhibited _____


Type of paper : Kraft paper_____ thermostabilised ______
Date of introduction of thermostabilised paper

Page 67 of 70
REFERENCES

[1] P.J. Burton et al., Dveloppements rcents au CEGB pour lamlioration de la prvision et de
la surveillance des performances des transformateurs, CIGRE Paper No. 12-09, 1984.
[2] J. Vergne, tude du vieillissement physico-chimique du complexe papier-huile dans les
transformateurs lectriques, Doctoral thesis, Institut National Polytechnique de Toulouse,
1992.
[3] J. Scheirs, G. Camino, M. Avidano and W. Tumiatti, Study of the mechanism of thermal
degradation of cellulosic paper insulation in electrical transformers, Die Angewandte
Makromolekulare Chemie, 259, pp. 19-24, 1998.
[4] A.M. Emsley, The kinetics and mechanism of degradation of cellulosic insulation in power
transformers, Polymers Degradation and Stability, Vol 44, pp. 343-349, 1994.
[5] J. Unsworth and F. Mitchell, Degradation of electrical insulating paper monitored with high-
performance liquid chromatography, IEEE Trans. on Electrical Insulation, Vol. 25, No. 4,
pp. 737-46, 1990.
[6] N. Dominelli et al., Recent developments in paper degradation products analysis, Minutes of
the 60th Annual International Conference of Doble Clients, Sect. 10.41, 1993.
[7] De Pablo and B. Pahlavanpour, Furanic compounds analysis: a tool for predictive maintenance
of oil filled electrical equipment, Electra, CIGRE, Group 15.01.03, No. 175, 1997.
[8] L. Lundgaard, W. Hansen, D. Linhjell & T. Painter, (2002) Ageing of Oil Impregnated paper in
power transformers, IEEE Trans Pow.Del., Vol 19, No.1, 2004, pp. 230-238.
[9] D. Allan, Recent advances in the analysis and interpretation of aged insulation from operating
power transformers, Proceedings of the 5th International Conference on Properties and
Applications of Dielectric Materials, Korea, pp. 202-205, 1997.
[10] D.S. Myers and J.R. Sans, Field study of furan formation in transformer fluids as an indicator
of damage to paper insulation, Minutes of the 59th Annual International Conference of Doble
Clients, Sec. 10-6, 1992.
[11] P.J. Griffin, L. R. Lewand, A practical guide for evaluating the condition of cellulosic
insulation in transformers, Minutes of the 62th Annual International Conference of Doble
Clients, Sect, 5-61,1995 B. Pahlavanpour, M. Augusta G. Martins, A. De Pablo, Experimental
Investigation in to the thermal ageing of Kraft paper and mineral insulating oil, IEEE, ISEI,
pp. 341-345 , 2002.
[12] B. Pahlavanpour, M. Augusta G. Martins, A. de Pablo, Experimental Investigation in to the
thermal ageing of Kraft paper and mineral insulating oil, IEEE, ISEI, pp. 341-345 , 2002. B.
Pahlavanpour, M. Eklund, M. Augusta G. Martins, Insulating paper aging and furfural
formation, EIC, 2003.
[13] B. Pahlavanpour, M. Eklund, M. Augusta G. Martins, Insulating paper aging and furfural
formation, EIC, 2003.
[14] I. Hhlein and J. Kachler, Aging of cellulose at transformer Service Temperatures. Part 2.
Influence of Water content and Temperature on Degree of Polymerization and Formation of
Furanic Compounds in Free-Breathing Systems, IEEE Elect. Insul. Mag., Vol. 21, No. 5,
(Sept./Oct.), pp. 20-23, 2005.
[15] M.C. Lessard, M. Koncan-Gradnik, I. Hhlein, B. Finlay, Furan Analysis as a Transformer
Life Management Tool, CIGRE Cession Paris, 2006, A2, PS1, Question 1.2.
[16] T. Gradnik, M. Koncan-Gradnik, Cooling system optimization and expected lifetime of large
power transformers, Proceedings of the 2006 JASME/WSEAS International Conference of
Energy & Environment, Chalkida, Greece, May 6-10, 2006, pp. 194-201, 2006.
[17] M.C. Lessard in Paper Ageing, Report of TF D1.01.10, Convenor Lars Lundgaard, CIGRE
Brochure, 2006.
[18] J. Kachler, I. Hhlein, Ageing of Cellulose at Transformer Service Temperatures, Part 1.

Page 68 of 70
Influence of Type of Oil and Air on the Degree of Polymerisation of Pressboard, Dissolved
Gases and Furanic Compounds in Oil, IEEE Electrical Insulation Magazine, Vol. 21,
March/April 2005, pp. 15-21.
[19] J.A. Lapworth, R. J. Heywood, P.N. Jarman, C. Myers, Transformer Insulation: Towards more
reliable residual life assessments, CIGRE A2-D1 Colloquium in Brugge, 2007.
[20] M. Mulej, A. Varl, M.Koncan-Gradnik, Up-to-date experience on furans for transformer
diagnostics , XIIIth International Symposium on High Voltage Engineering, pp. 1-6,
Netherlands, 2003.
[21] A. de Pablo, Furfuraldehyde Evolution in Passivated Oils, Document for IEC TC 10, WG 33,
Turin 20.06.2006.
[22] H. Ltke, I. Hhlein, , A. J. Kachler, Transformer Ageing Research on Furanic Compounds
Dissolved in Insulating Oil, CIGRE paper 15-302, Paris CIGRE session, 2002.
[23] M. C. Lessard, L. Van Nifterik, M. Masse, J. F. Penneau, R. Grob, Physicochemical
characterization of the thermal ageing of insulating paper in power transformers, Conference
record of the 1996 IEEE International symposium on electrical insulation, Canada, Quebec,
June 16-19, 1996.
[24] L. Lewand, Practical Experience Gained from Furanic Compound Analysis, Doble
Engineering Company, 2006.
[25] L. Lundgaard, W. Hansen, D. Linhjell & T. Painter, (2002) Ageing of Oil Impregnated paper in
power transformers, IEEE Trans Power, Del., Vol 19, No.1, pp. 230-238, 2004.
[26] M. C. Lessard, Contribution to CIGRE WG D1.01.13, 2008.
[27] A.M. Emsley, R.J. Heywood, M. Ali, X. Xiao, Degradation of Cellulosic Insulation in Power
Transformers, Part 3, Effects of oxygen and water on ageing in oil, IEE, Prosci. Meas.
Technol., Vol. 147, 3, pp. 115-119, 2000.
[28] P.J. Griffin et al., Measurement of Cellulosic Insulation Degradation, Annual International
Conference of Doble Clients, Sec. 10-3, 1993.
[29] A.M. Emsley, X. Xiao, R.J. Heywood, M. Ali, Degradation of Cellulosic Insulation in Power
Transformer; Part 2, Formation of furan products in insulating oil, IEE, Prosci. Meas.
Technol., 147, 3, pp. 110-114, 2000.
[30] J. Scheirs, G. Camino, M. Avidano and W. Tumiatti, Origin of furanic compounds in thermal
degradation of cellulosic insulating paper, Journal of Applied Polymer Science, Vol. 69,
pp. 2541-2547, 1998.
[31] L. Lewand, Doble Engineering, Practical experience gained from furanic compounds
analysis, 2006 Doble Engineering Company 73rd Annual International Doble Client
Conference, USA, 2006.
[32] T. Buchacz, Contribution to CIGRE WG D1.01.13, 2010.
[33] M.A. G. Martins, Contribution to CIGRE WG D1.01.13, 2010.
[34] A. Petersen, A2PS2Q12, Proceedings of CIGRE Session 2004.
[35] M.A. G. Martins, Rate of 2FAL recovery, after oil replacement by new oil, in power
transformers of the Portuguese Electrical System Labelec, Insulating Materials Department,
2010.
[36] D.H. Shroff and A.W. Stannett, A review of paper aging in power transformers, IEEE
Proceedings, Vol. 132, Pt. C, No. 6, pp. 312-319, 1985.
[37] X. Chendong, Monitoring paper insulation aging by measuring furfural contents in oil, EPRI,
7th International Symposium on high Voltage Engineering, pp.139-142, 1991.
[38] D. Hill, T. Le, M. Darvenia , T. Saha, A study of degradation of cellulosic insulation in a
power transformer, Polym. Degrad. Stab., 48, pp. 79-87, 1995.
[39] A. de Pablo, Furfural and ageing: How are they related, IEE, 5, 1999.
[40] E. Serena, Degradation of transformer insulation, CIGRE WG 15.01.05, Turin, 1997.
[41] R.M. Morais, M. Carballeira, J.C. Noualhaguet, Furfural analysis for assessing degradation of
thermally upgraded papers in transformer insulation, IEEE Transactions on Dielectrics and
Electrical Insulation, Vol. 6, No. 2, April 1999.

Page 69 of 70
[42] A. B. Shkolnik, K. Bilgin, Creating a preliminary model for estimating degree of
polymerization of thermally upgraded paper based on furan concentrations in transformer oil,
66th Annual International Conference of Doble Clients, Section 5-8, 1999.
[43] I. Hhlein. A. J. Kachler, U. Thie, M. Stach, Th. Leibfried, Diagnostic and Maintenance as
Prerequisites for Transformer Reliability, CIGRE SC A2 Colloquium Transformer reliability
and Transients, June 20-24, Moscow, 2005.
[44] J. Rimez, P. Leemans, S. Eeckhoudt, Impact of furan analysis on asset management of high
voltage transformers in the Belgian transmission network, CIGRE A2-D1 Colloquium in
Brugge, 2007.
[45] H. P. Gasser, Ch. Krause, T. Prevost, R. Schmid, Assessing the condition of power transformer
insulation by post mortem investigation on apparatus to be scrapped, CIGRE A2-D1
Colloquium in Brugge, 2007.
[46] M.A. G. Martins, Envelhecimento trmico do papel isolante de transformadores. Investigao
experimental. Modelos de degradao Cincia e Tecnologia dos Materiais. Vol.22, n 1/2
Jan/Jun 2010, pp. 77- 86.
[47] M.A. G. Martins, M. Fialho, J. Martins, M. Soares, M. Cristina, R.Castro Lopes, H. M.R.
Campelo - Power Transformer End-Of-Life Assessment - Pracana Case Study, Advanced
Research Workshop on Transformer, Santiago de Compostela , Spain. 3-6 October 2010.
[48] R. D. Stebbins, D. S. Myers, and A. B. Shkolnik, Furanic Compounds in Dielectric Liquid
Samples: Review and Update of Diagnostic Interpretation and Estimation of Insulation Age,
Proceedings of the 7th International Conference on Properties and Applications of Dielectric
Materials, pp. 921-926, 1-5, June 2003.
[49] D.H. Grant, Furfuraldehyde-in-oil: How to be sure your transformer has not been weakened by
heat, Minutes of the 57th Annual International Conference of Doble Clients, Insulating Fluids,
1990.
[50] K. Bilgin, A.B. Shkolnik, J. Kelly, D. S. Myers The effect of dicyandiamide in insulation on 2-
furfuraldehyde concentration in oil, H2B, Techcon, pp. 207-216, 1999.
[51] A.M. Emsley, R.J. Heywood, M. Ali, X. Xiao, Degradation of Cellulosic Insulation in Power
Transformers, Part 4, IEE, Prosci. Meas. Technol., 147, 285, 2000.
[52] P.J. Burton et al., Applications of liquid chromatography to the analysis of electrical insulating
material, CIGRE, Section 15-08, 1988.
[53] M. Carballeira, HPLC contribution to transformer survey during service or heat run tests,
Electra, CIGRE, No. 133, pp. 44-50, December 1990.
[54] D. Grant, A study of furanic compounds generated in transformers during heat run, Minutes
of the 59th Annual International Conference of Doble Clients, Insulating Fluids Section 10-4,
1992.
[55] S. Naban et al., Adsorption phenomena of furfural in insulation paper (fundamental test for life
diagnosis of transformer), T. IEE Japan, 112-A, No. 2, pp. 139-144, 1992.
[56] S. Soares, N. Ricardo, F. Heatly and E. Rodrigues, Low temperature thermal degradation of
cellulosic insulating paper in air and transformer oil, Poly. Int. 50, pp. 303-308, 2001.
[57] L. Lundgard et al, TF D1.01.10, Ageing of cellulose in mineral oil insulated transformers,
Brochure 323, October 2007.
[58] J. Declercq et al, CIGRE WG A2.24, Thermal performance of transformers, CIGRE
Technical Brochure 393, October 2009.
[59] A.M. Emsley, G.C. Steven, Review of chemical indicators of degradation of cellulosic
electrical paper insulation in oil-filed transformers, IEE, Pro-Sci. Meas. Technol., 141,
pp. 324-334, 1994.
[60] IEC Loading Guide for oil-immersed power transformers IEC 600-76-7, 2005.
[61] IEEE Guide for loading Mineral-oil-Immersed Power Transformers, C57.91-1995.
[62] L. Cheim and C. Dupont, A new transformer ageing model and its correlation to 2-FAL
CIGRE International Transformers Colloquium, Merida, 2003.

Page 70 of 70