Stp998-Eb 2198

Copyright by ASTM Int'l (all rights reserved); Thu May 6 14:02:13 EDT 2021
Downloaded/printed by
(UFRN) Universidade Federal do Rio Grande do Norte ((UFRN) Universidade Federal do Rio Grande do Norte) pursuant to License Agreement. No further reproductions auth
STP 998
Electrical Insulating Oils
Herbert G. Erdman, editor
ASTM
1916 Race Street
• Philadelphia, PA 19103
ASTM Publication Code Number (PCN): 04-998000-21
ISBN: 0-8031-1179-7
Copyright © by AMERICAN SOCIETY FOR TESTING AND MATERIALS 1988
NOTE
The Society is not responsible, as a body,
for the statements and opinions
advanced in this publication.
Peer Review Policy
Each paper published in this volume was evaluated by three peer reviewers. The authors
addressed all of the reviewers' comments to the satisfaction of both the technical editor(s) and
the ASTM Committee on Publications.
The quality of the papers in this publication reflects not only the obvious efforts of the authors
and the technical editor(s), but also the work of these peer reviewers. The ASTM Committee on
Publications acknowledges with appreciation their dedication and contribution of time and ef-
fort on behalf of ASTM.
Printed in Ann Aibor, Ml

Septeniber 1988
Second Printing. Ann Arbor. Ml
May 1992
Foreword
This publication, Electrical Insulating Oils, contains papers presented at the symposium of
the same name held in Bal Harbour, Florida on 19-20 Oct. 1987. The symposium was spon-
sored by ASTM Committee D-27 on Electrical Insulating Liquids and Gases. Herbert G. Erd-
man, PSE&G Research Corp., Maplewood, NJ, presided as symposium chairman and was edi-
tor of this publication.
Contents
Overview
SECTION I—REFINING AND SPECIFICATION LIMITS
Mineral Insulating Oil Manufacture and Safekeeping—T. G. LIPSCOMB, II 5

Discussion 24
PCBs: The Beginning, the Muddle, the End?—H. C. MANGER 25
SECTION II—SIGNIFICANCE OF APPLICATION
Insulating Oil Qualification and Acceptance Tests from a User's Perspective—

D. L. JOHNSON 37
Discussion 46
Significance of Transformer Oil Properties—H. A. PEARCE 47

Discussion 54
SECTION III—ANALYSIS OF OIL
Newly Developed Analytical Techniques for Characterization of Insulating Oils—

J.-p. CRINE 59
Discussion 80
SECTION IV—DISSOLVED GAS IN OIL
Transformer Fault Gas Analysis and Interpretation—A User's Perspective—L. J. SAVIO 83
Criteria for the Interpretation of Data for Dissolved Gases in Oil from Transformers
(A Review)—p. i. GRIFFIN 89
Discussion 106
A Review of the Operating Principles and Practice of Dissolved Gas Analysis—

T. J. HAUPERT AND F. JAKOB 108
SECTION V—ELECTROSTATIC BUILDUP IN TRANSFORMER OIL
An Engineering Model for Streaming Electrification in Power Transformers—

J. F . R O A C H A N D I. B. TEMPLETON 119
Static Electrification in Power Transformers—o. w. CROFTS 136
Discussion 151
STP998-EB/Sep. 1988
Overview
This symposium was organized to update information on the important role of petroleum-
based oil as an insulating and heat transfer agent in electrical apparatus.
The properties of these insulating oils are of vital importance to the service life of electrical
equipment and have become increasingly important as operating voltages have increased to 500
kV and above and internal equipment spacings have decreased. The papers in this symposium
have been directed towards the changing requirements of the insulating media as demanded by
the changing design of the equipment, and they are of particular value to three main groups: the
refiners of the oil; the manufacturers of the electrical equipment; and the end users (principally
the electric utilities).
The papers in this book address the factors involved in the manufacture and use of the prod-
uct that will meet these stringent requirements.
From the refiners' standpoint, the papers discuss how the crude oil is selected and what must
be done to produce an end product with high dielectric strength, heat transfer capability, low-
temperature pour point, resistance to oxidation, and long service life. These factors determine
the specifications of the product.
From the users' standpoint, the papers discuss the significance of these specifications and test
methods to determine if the specifications have been met, as well as tests to check the condition
of oil after use in the equipment. Included is the very important diagnostic test for electric
equipment in service, the analysis of dissolved gases in the oil.
Of special interest is section 5, which is devoted to a relatively new problem caused by electro-
static charges built up in the oil due to forced cooling in large transformers.
This volume is of value to all who refine or use insulating oils because it addresses the latest
technologies involved.
The first section of this book, "Refining and Specification Limits," addresses methods used
in refining crude oil to produce a product which is suitable for both electric insulation and heat
transfer. Included is the problem of handling and shipping the product once produced. Con-
tamination can occur from tank cars and drums that are not completely devoid of foreign mate-
rials or not properly sealed. The paper by Manger, which discusses cross contamination with
polychlorinated biphenyls (PCB), is included to describe a specific problem. This paper updates
the local and federal regulations dealing with PCB contamination.
The second section, "Significance of Application," addresses the reasons behind the various
specifications for insulating oils and the significance of these specifications to the users.
The third section, "Analysis of Oil," discusses the methods of analysis to determine if these
specifications have been met. There are two conditions to be considered: (1) analyzing new oil as
shipped; and, perhaps more important, (2) the analysis of oil after certain periods of time in the
electrical equipment to determine deterioration and predict further usefulness of the oil.
The fourth section, "Dissolved Gas in Oil," discusses the analysis of gases dissolved in oil and
the significance of various gases to the condition of the electrical equipment involved. This is a
very useful diagnostic test, particularly to monitor the performance of large high-voltage
transformers.
The fifth section, "Electrostatic Buildup in Transformer Oil," perhaps one of the most pro-
vocative, has to do with a recent problem that has arisen in large transformers where the pump-
ing of the oil is required to limit the temperature rise in the transformers. Pumping a liquid with
highby dielectric
Copyright ASTM Int'l (all strength and
rights reserved); Thu very low quantities
May 6 14:02:13 EDT 2021 of moisture over various materials in the pump
1
Copyright'^^ 1988 b y A S T M International www.astm.org
2 INSULATING OILS
and in the insulating material, in many cases, has caused an electrostatic charge buildup in the
oil and has resulted in electrical failure in several large transformers throughout the country.
The papers in this section present the latest information on this phenomena. There is much
more to be learned of this complex process, and much work is being done throughout the coun-
try to try and solve this problem.
The papers in this book should provide the reader with the latest technologies in selecting and
handling this very important insulating media for electric equipment. This is a very significant
area for equipment designers. Large capacity size and weight of electrical equipment is of vital
importance, resulting in smaller internal spacings requiring the best dielectric material
available.
ASTM Committee D27 on Electrical Insulating Liquids and Gases is continually working to
update this material, improve test procedures, and investigate newly refined products that ap-
pear on the market. Refiners are continuously producing new insulating oils from petroleum
crude oil. In addition, manufacturers are continually producing synthetic insulating fluids.
Committee D27 on Electrical Insulating Liquids and Gases addresses these new products as
they come on the market.
Herbert G. Erdman,
PSE & G Research Corp.; Maplewood, NJ
07040; symposium chairman and editor
Section I—Refining and Specification Limits
Thomas G. Lipscomb, IP
Mineral Insulating Oil Manufacture and

Safekeeping
REFERENCE: Lipscomb, T. G. II, "Mineral Insulating Oil Manufacture and Safekeeping,"

Electrical Insulating Oils, STP 998, H. G. Erdman, Ed., American Society for Testing and Mate-
rials, Philadelphia, 1988, pp. 5-24.
ABSTRACT: This paper presents a general overview of the methods used to refine conventional
mineral insulating oils. The discussion will outline the individual refining processes and the effect
each may have on the composition of the oil. Included will be a brief discussion of the Occupa-
tional Safety and Health Administration's Hazard Communication Standard, 29 CFR 1910.1200
as it can apply to mineral insulating oil and some suggestions for handling and storage of the oil to
reduce contamination and degradation.
KEY WORDS: transformer oil, mineral insulating oil, refining, naphthenic transformer oil, par-
affinic transformer oil, storage, handling, cancer hazard warning, OSHA hazard communication
standard
In planning a seminar on the subject of insulating oil of petroleum origin, it was recognized
that many people involved in using "mineral insulating oils" or "transformer oils" have proba-
bly never known how they are made. Only those persons involved in their manufacture or in
research on the subject have direct knowledge.
The author was asked to provide information of a nonproprietary nature on the general sub-
ject to serve as a basis for understanding some of the properties of transformer oils that will be
discussed in subsequent papers.
Discussion
Refining
Mineral transformer oil today is a compromise between what is needed by the equipment,
what is desired by the customer, and what can be achieved without overwhelming complexity in
the refining process. This compromise is well exemplified by ASTM Specification for Mineral
Insulating Oil Used in Electrical Apparatus (D 3487). The need to meet this specification in the
most efficient manner is the aim of the refiner.
The manufacture of transformer oil begins with a choice of crude oil type: naphthenic, paraf-
finic, or mixed base. Convention in the United States has been to use naphthenic crudes, since
they generally contain very much lower concentrations of n-paraffins (wax), and the resultant
product has a low pour point without dewaxing being required. This convention at one time
applied everywhere, but necessity has driven some regions of the world to use mixed-base or
paraffinic crudes as sources. All useable lube crudes can be manipulated by suitable combina-
tions of processes to result in a useable transformer oil. The degree of difficulty in refining and
the product yield will vary with the crude.
There is also another little-appreciated benefit of using naphthenic oils; their viscosities
change much faster with temperature than those of paraffinic-derived oils. Normally, this might
Copyright'Technical
by ASTM Int'ladvisor, ExxonThu
(all rights reserved); Company, U.S.A.,
May 6 14:02:13 EDT 2021Houston, TX 77252.
5
6 INSULATING OILS
be considered a detriment, but not in this case. As the temperature rises in a transformer, the
oil becomes less viscous and the heat transfer rate is improved. For oils of equal viscosity at
40°C, the heat transfer coefficient can be as much as 8 to 11% greater at 100°C for an oil of 30
VI (viscosity index) versus an oil of 100 VI.
Since the world supply of good "low cold test" naphthenic crudes has declined, we have seen
paraffinic crudes being used as sources, particularly in South America, Mexico, and Europe.
One such oil was commercially introduced in the United States. At the time this paper was
written, it was withdrawn from the market.
The definition of a naphthenic crude by the U.S. Bureau of Mines contains no mention of the
pour point of the distillate nor of the n-paraffin content. This classification system is shown in
Fig. 1. By this system it is possible to have nine different crude types, depending upon the
various combinations of light key and heavy key classifications.
ASTM Committee D27 was more pragmatic and chose to use the definitions in ASTM Defini-
tions of Terms Relating to Electrical Insulating Liquids and Gases (D 2864) as shown in Fig. 2.
Although not technically correct, they represent the main concerns about pour points, low-
temperature viscosities, and the means needed to achieve the desired properties.
Refining processes are used to eliminate undesirable compounds and to retain those that are
desirable. The elimination procedure may be by simple removal or by conversion to more desir-
able compounds. Both types of elimination procedures are represented by refining processes
commonly in use today.
Crude Distillation
The first step in producing a transformer oil is distillation of the crude to produce a suitable
distillate feedstock. This is usually done on a pipe still. The distillate can be taken as a side
stream product from either the atmospheric tower or the vacuum tower. Or a very broad stream
can be taken, processed partially, and then redistilled to the desired narrow boiling range. Fig-
ure 3 shows a schematic flow diagram of a typical pipe still operation.
In this process, "light ends" are removed in a prefractionator tower (distillation column) not
shown in the diagram, and "middle distillate" type products are removed in the atmospheric
tower. The "reduced crude" from the bottom of the atmospheric tower then goes to the final, or
vacuum, tower. This distillation column is run under vacuum to keep the boiling point of the
individual fractions well below the cracking temperature, that is, the temperature at which ther-
mal decomposition of the oil molecules would begin. Cracking of hydrocarbons usually becomes
rapid enough to detect at about 370°C (700°F).
Steam is usually added to the bottom of the vacuum tower to further lower hydrocarbon par-
tial pressure and reduce the temperature required for distillation still more. The transformer oil
distillate, being relatively low boiling, is removed near the top of the tower. As mentioned ear-
lier, the transformer oil distillate could also have been removed as a lower side stream on the
atmospheric tower.
A question sometimes asked is, "What portion of crude can be made into transformer oil?"
Gravity of Key Fraction, "API

Light Key Fraction Heavy Key Fraction
( 4 8 2 - 5 2 7 ' 'F) (740-790°F)
Paraffinic >40 >30
Intermediate 33-40 20-30
Naphthenic <33 <20
FIG. 1—Crude oil classifications (U.S. Bureau of Mines}.
LIPSCOMB ON MINERAL INSULATING OIL 7
Naphthenic Oil
A term applied to mineral insulating
oil derived from special crudes having
very low, naturally occuring n-paraffin
(wax) contents. Such an oil has a
low natural pour point and does not need to
be dewaxed nor does it usually require the
use of a pour depressant.
Paraffinic Oil
A term applied to mineral insulating
oil derived from crudes having substantial
contents of naturally occurring n-parafflns
(wax). Such an oil must be dewaxed
and may need the addition of a pour
depressant in order to exhibit a low pour
point.
FIG. 2—Definitions of naphthenic oil and paraffinic oil.
For most useable crudes, the distillate yield can be from 3 to 10% of the whole crude. There are
two limiting factors. On the front end, the flash point limits the amount of light molecules that
can be present. The second limitation is the maximum viscosity. As increasing amounts of
heavier molecules are included, the viscosity of the whole fraction increases until the specified
limit is reached.
There are other practical considerations. The molecules boiling above and below the desired
fraction are valuable for other uses. Pipe stills are not perfect; there is overlap between frac-
tions. If the maximum amount of transformer distillate is taken, then a shortage of the next
lower and/or higher boiling fractions may develop.
In looking at some current typical transformer oils versus ASTM D 3487, none seems to be at
the minimum flash point, and none is at the maximum viscosity, so compromise has won again.
Table 1 shows the boiling range of a distillate from a naphthenic crude and one from a paraf-
finic crude [by ASTM Test Method for Boiling Range Distribution of Petroleum Fractions by
Gas Chromatography (D 2887)]. Both distillates were used to make finished transformer oil.
Atmospheric Vacuum
Distillates
r^K^
ates
i(Transformer
Transformer Oil
(
^
. l a - I C o on Cru
rude
100°!.
Crude •
Oil
i^
Atmospheric Vacuum
Residuum Residuum
Atmospheric Atmospheric Vacuum Vacuum

Furnace Distillation Furnace Distillation
Tower Tower
FIG.do3—Distillation
(UFRN) Universidade Federal do Rio Grande to make
Norte ((UFRN) Universidade transformer
Federal do Rio Grandeoil base pursuant
do Norte) stock.to License Agreement. No further reproductions auth
8 INSULATING OILS
TABLE 1—Comparison of naphthenic and paraffinic distillates

boiling ranges.
Naphthenic Paraffinic
Initial boiling point, °C 283 239

5% distilled, °C 271 306
Final boiling point, °C 429 531
The data for the naphthenic distillate can be translated into an average molecular weight of
255, or an average molecule with 18 carbon atoms (carbon number = 18), if all of the com-
pounds were paraffins. Similarly, the molecular weight of the paraffinic distillate would be
about 300.
Not all of the compounds present in naphthenic distillates are paraffins, far from it. There is
a great variety of compounds present, and, in general, the same types of compounds are present
in all crudes. The difference in crudes can be found in the relative amounts of the various com-
pound types and in the degree and kind of substitution on the ring structures.
Molecular Components of the Distillate

Before examining the next processing steps, we need to illustrate the various molecule types
present in the distillate. Refining is essentially the effort to remove or preserve these com-
pounds.
Paraffins, normal-, iso-, and cyclo-, are shown in Figs. 4 and 5. Remember, we said that the
average carbon number was 18, yet the structures shown do not have 18 carbons. We are illus-
trating the compound types present, not the actual compounds themselves in order to simplify
the drawings.
A naphthenic crude-derived distillate will have very low n-paraffin content. There will be
some present, but the values will be so low as to be almost unmeasurable. Not all naphthenic
H H H /H H H
I I I I I I I
H-C-C- C- C -C-C-H
I I i l l I I it-nonar»e
H H H |H
Normal-Paraffin
H
H-C-H
H-C-H
H H , H H H
I I ' i l l 3 methyl, 4 ethyl hexane
H - C - C - C - C - C - C - I-
I I 1 , 1 1 (iso - nonane)
H H H ' H H
H-C-H
I
H
iso - Paraffin
Downloaded/printed by FIG. 4—Paraffins.
H H
H
H-C C-H
H2
H ,C H
H2 H,
^ ^ hexane decalln
1 - Ring Naphthene 2 - Ring Naphthene

(cyclo-Paraffin)
FIG. S—Naphthenes.
crudes are wax free. There are some which meet the U.S. Bureau of Mines definition, but which
contain enough wax to make them unsuitable for "low cold test" products.
The naphthenic distillate will then contain considerable amounts of iso-paraffins and cyclo-
paraffins. The iso-paraffins have very low crystallization temperatures, as do the cyclo-paraf-
fins, provided that they do not have very long n-paraffin side chain substitutions. All these
cyclo-paraffins, also called naphthenes (hence the name naphthenic crude), will have some sub-
stitution of side chains, probably several short or branched chains.
We want to keep the iso-paraffins and cyclo-paraffins and, if necessary, reject n-paraffins.
Olefins are shown in Fig. 6. The double bond makes them highly reactive to form sludges,
acids, aldehydes, etc. There are several ways to eliminate them, which will be discussed later.
Aromatics are a major constituent of the distillate. Figure 7 does not show any side chain
substitution, but at this boiling range all of the aromatics present will have side chains. The
degree and kind of substitution strongly affects the viscosity/temperature properties of these
aromatics. Note also mixed molecules, naphtheno-aromatics, where part is aromatic and part is
naphthenic. Indane is an example.
Aromatics are both good and bad. We need to keep some but not all. In general, as the
aromatic content is decreased, the gassing tendency of the oil increases, and the impulse
H H H H H
I t I I I
-H n-Hexene
H-C- C = C- C - C - C -
H H H H H
H
H-c-H
H H 1 H
H-C- C = C- C - C - H -hexene
H H H H
V^
H-C C-H,
cyclo-hexene
II 1
H-C C-H,
1
Hj
(UFRN) Universidade Federal do Rio Grande do Norte ((UFRN) Universidade -Olefins
FIG. 6-Federal do Rio Grande do Norte) pursuant to License Agreement. No further reproductions auth
10 INSULATING OILS
H
I
"v^x^^^
Indane Fluorene
FIG. 7—Aromatics.
strength goes up. Also, some aromatics seem to be mild oxidation inhibitors. The refiner must
then try to remove some aromatics, leave some aromatics, be selective about the process, and
compromise between the desire for good antigassing properties, good impulse strength, and
good oxidation resistance.
Polar compounds, and there are many types, are shown beginning in Fig. 8. Heterocyclic
aromatics containing nitrogen or oxygen are generally undesirable. They are not stable as to
oxidation, color, etc. Thiophenes may be an exception. Certain thiophenes are known to be
moderate oxidation inhibitors.
Polar oxygenates shown in Fig. 9 are all undesirable. If they are not already, they soon will
degrade further to become acids. As acids they contribute to corrosion, low interfacial tension,
and poor dissipation factor.
Sulfur compounds, except for thiophene, are shown in Fig. 10. Some of these may be mild
antioxidants, but they also react with copper and have objectionable odors. These compounds
must be removed.
Benzofuran
H
Benzothiophene Pyridine
Downloaded/printed by FIG. 8—Polars, heterocyclics.
H
1
H H H H H H H H
1 1 1 1 I I I I
H-C-C-C-C-OH H -C-C-O-C-C-H
H'-^^^^^H I I I I
1
H H H H H H H H H
phenol n-butyl alcohol diethyl ether
H O H H H H H H H H H O
1 II 1 1 1 1 1 1 I I I / /
H-C-C-C-C-H H-C-C-C-C=0 H-C-C-C-C
1 1 1
' ' ' ÔH
H H H H H H H H H
lethyl ethyl ketone butyradehyde butyric acid
FIG. 9—Polars. oxygenates.
Actually, there are two more ever-present materials we must consider: water and particulates.
These must be removed because they are as deleterious to the performance of the finished oil as
any of the molecules previously mentioned.
Refining the Distillate

In this section we will describe each of the major refining processes. In a later section we will
compare the physical and chemical effects of the processes on each molecule type present in the
distillate. In examining the refining processes, it is well to keep in mind that it is rare to find a
single process that will do all. Combinations of processes are the rule. We will list some of the
more popular combinations, but not by any means all combinations.
Acid Treating—The first refining process is acid refining or acid treating. Many years ago it
was the "one" accepted method, but in reality it was a dual process, for it was combined with
clay treating. Figure 11 shows a block diagram of the process.
In this process, the transformer oil distillate is initially contacted with 90 to 99% sulfuric
acid. A very light treat, 2 to 4 volume % acid based on the distillate, is used. The contact time of
acid and distillate varies between about 0.1 and 4 s. Following mixing, the acid sludge that is
formed from the chemical reaction of acid and oil is separated from the oil by centrifuging. At
this point some harmful sulfonic acids and nitrogen bases are removed as sludge from the oil.
The oil from the centrifuging operation contains oil-soluble acidic compounds resulting from
the reaction with sulfuric acid. Upon neutralization with sodium carbonate or caustic soda and
H
H-S-H H-C-S-H
H
Hydrogen Sulfide Methyl Mercaptan
H H H H H H
H-C-C-S-C-C-H H-C-S-S-C-H
H H H H H H
Diethyl Sulfide Dimethyl Disulfide
FIG. 10—Sulfur compounds.
12 INSULATING OILS
Ac d Treating Clay
Caustic
Alcohol & Water
90-99%
Sulfuric Acid Alcohol Clay
Water
'^ " t "

Treated
Oil
V "
Spent
t
steam Spent
"
Acid Caustic Clay
Sludge
Mixer Centrifug B Mixer - stripper Perc

Se t t l e r
FIG. 11—Acid treating/day treating.
extraction of the oil with alcohol, these acidic compounds are removed as acid salts. The neu-
tralized oil is then stripped with steam to remove alcohol and water.
Processing of the oil to this stage may be described as a pretreatment operation, since the oil
does not meet the electrical insulating oil criteria as outlined in ASTM D 3487. A finishing step,
percolation through a bed of fine clay, is undertaken to produce an acid treated insulating oil of
acceptable quality. The clay percolation step is necessary not only to remove compounds not
reacted out by the acid treating, but also to remove residual traces of sodium sulfonates created
by the neutralization of the sulfonic acids. The finished electrical insulating oil is neutral in
acidity, noncorrosive, and has good electrical properties. More detail on the changes in chemi-
cal/physical properties that take place will be given later in this paper.
Hydrotreating—The next processes that will be described are hydrotreating and hydrogena-
tion. The block flow diagrams for both processes are shown in Fig. 12. The two processes differ
in severity of treatment of the oil. In hydrotreating (a relatively mild process), the aim is to
remove unwanted sulfur, nitrogen, oxygen, and olefinic compounds without materially remov-
ing the aromatics present and to improve the odor, color, and color stability. For the purposes
of this discussion, "hydrogenation" will mean a process that is so severe that a significant por-
tion of the aromatics present is removed (converted) also.
In hydrotreating, hydrogen gas and the transformer oil distillate are chemically reacted over a
fixed catalyst bed at elevated temperatures and pressures. In the hydrotreater, harmful material
Hydrogen
Gas
n j 1112 5
/ NH,, etc.
Vacuum
A t X
\
Treated
Oil
Steam
Furnace Reactor Separator Stripper Drier

FIG. 12—Hydrotreatiitg/hydrogenation.
(such as sulfur compounds, naphthenic acids, and nitrogen compounds) react with hydrogen.
The products of these reactions are: (1) stable hydrocarbon compounds similar to naturally
occurring hydrocarbon compounds; and (2) gases such as hydrogen sulfide, ammonia, and wa-
ter. The conditions of pressure and temperature used in the hydrotreating process are not suffi-
cient to cause cracking of the oil. The highest temperatures experienced by insulating oil occur
in the pipe still during distillation.
After the hydrotreating step, the oil goes to a steam stripper where the gaseous reaction prod-
ucts (such as hydrogen sulfide, ammonia, and water) are removed. The oil can be dried at this
stage by using a subsequent vacuum stripper.
The hydrotreated oil by itself is not usually a satisfactory product. The process is too mild to
remove all deleterious compounds. Some other process such as clay percolation or solvent ex-
traction must also be used to arrive at a suitable finished oil. Both combinations have been used
by the industry.
Hydrogenation can differ from hydrotreatment in temperature and pressure, but usually also
employs a different catalyst, one that is more reactive for the hydrogenation of aromatics to
cyclo-paraffins.
There are commercially available a wide variety of catalysts used in hydrotreating and/or
hydrogenation. The science of catalysis is now well enough advanced to permit the user to select
a catalyst that will accentuate a specific reaction, such as desulfurization, dearomatization,
denitrogenation, etc.
Hydrogenation can, by itself, serve to perform the functions of both solvent extraction (aro-
matics reduction) and hydrotreating (polar compound reduction), but it is so severe as to re-
move most naturally occurring antioxidants at the same time it removes oxidation promoters
and unstable compounds. In this case, synthetic antioxidants must be used with the oil for the
product to meet the oxidation requirements of ASTM D 3487.
Solvent Extraction—The solvent extraction block flow diagram is shown in Fig. 13. In this
process the oil feed is contacted countercurrently in a treating tower or centrifugal extractor
using an immiscible (or partly miscible) solvent that is more dense than the oil and is selective
Extraction
Solvent
Wate r T
or Vacuum
Raffinate
•A-
•J p ^
L,ri
I Treated
9"
steam ^ T ^ l I
Feed orNg ^
Stock
Extract
-A- / \ Vacuum
Extract
•
Steam
or N2
Extraction Solvent Driers

Furnaces Strippers
Tower
Copyright by ASTM Int'l (all rights Recovery
reserved); Thu May 6 14:02:13 EDT 2021
FIG. 13—Solvent extraction.
14 INSULATING OILS
for aromatics. The extracted oil (now called raffinate) containing some dissolved solvent goes to
a stripping system where the solvent is recovered and reused.
The bulk of the solvent from the treater tower containing some oil (now known as extract)
goes to a separate stripping system where the solvent is removed from the extract for reuse. The
extract produced has a high content of aromatics.
Solvents that have been used to treat transformer oil are: phenol, furfural, N-methyl-2-pyrro-
lidone, and liquid sulfur dioxide. All are selective for removal of aromatics, but the degree of
selectivity will differ somewhat among the solvents and can even be changed by various means.
It should be noted that none of these solvents is perfectly selective for aromatics. Desirable
compounds are partly soluble, and so some are removed in the extract. Likewise, some of unde-
sirable compounds are left behind. In general, all of the above solvents tend to remove a greater
proportion of the more highly condensed ring aromatics first, the less condensed ring systems
next, etc. Solvent extraction by itself is not a sufficient process. It must be used in combination
with other processes.
Solvent Dewaxing—Dewaxing is not required if the crude source used is naphthenic (low cold
test), but is required for oils made from paraffinic crudes or waxy naphthenic crudes.
Figure 14 is a block diagram of a solvent dewaxing unit. Waxy oil is mixed with a suitable
antisolvent for wax (n-paraffins) and chilled to a low temperature to cause the n-paraffins to
precipitate. The wax is then filtered out, usually on a rotary filter. The solvent is recovered from
the dewaxed oil and wax for reuse.
Typical solvents that can be used In this process are liquid propane, methyl ethyl ketone, and
methyl iso-butyl ketone. The ketones are especially good antisolvents for wax but typically must
be used with another cosolvent to prevent rejection of desired molecules.
When using solvent dewaxing, the pour point of the dewaxed oil is usually limited to — 18°C
(0°F) to — 23°C ( —10°F) as a practical matter. It is theoretically possible to achieve a lower
pour point, but most refinery equipment is not designed for very low temperature operation.
Obviously, if a pour point of —40°C (—40°F) or lower is desired, either a further processing
step is required, such as hydrodewaxing, or a pour depressant must be added to the oil. Pour
depressants are in use in Europe but are not used in the United States.
Catalytic Dewaxing—A process capable of very low pour points is that of hydrodewaxing or
catalytic dewaxing. Figure 15 gives a block diagram of this process. The equipment used is very
similar to that used in hydrotreating. Only the catalyst needs to be different.
Dewaxed
Oil
Stock
Refrigerant Wax
Steam
Chillers Solvent Rotary Stripper Drier

Recovery Fitter
Downloaded/printed by FIG. 14—Solvent dewaxing.
^Hydrogen
'Gas
C, Ha, etc.
Feed Vacuum
Stock
Steam
Furnace Reactor Separator Stripper Drier
FIG. 15—Catalytic dewaxing
The dewaxing catalyst (for example, mordenite) can best be visualized as a hollow sphere
with holes in the sides leading into the interior cavity. The hole size is such that a straight chain
hydrocarbon (n-paraffin) can protrude through the hole into the interior, but a branched-chain
(iso-paraffin) or cyclic structure cannot.
The waxy oil feed to the unit is mixed with hydrogen gas at elevated temperature and pressure
and allowed to flow over the catalyst just as in hydrotreating. As a normal paraffin molecule
protrudes into a side hole of the catalyst it is "cracked" off, typically in units of three carbons.
The olefins formed immediately react with the hydrogen present to make saturated hydrocar-
bons. Instead of solid wax as a by-product, the wax is converted to light hydrocarbons, chiefly
propane, by this process.
This process is capable of producing dewaxed oils with pour points of —45°C (—50°F). No
pour depressants are required.
Physical and Chemical Effects of Refining

Having discussed the molecular compound types present and the refining processes com-
monly in use, let us now see how the processes affect the molecular mixtures and, in turn, how
that may affect the properties of the finished transformer oil.
Figure 16 shows that n-paraffins are removed only by the dewaxing processes.
Figure 17 shows that iso-paraffins and naphthenes are not affected by any process, except
that small amounts may be solubilized into the extract phase in solvent extraction.
Figure 18 shows that olefins are removed by acid treating, while hydrotreating/hydrogena-
tion converts them to saturated hydrocarbons which remain in the oil.
Acid Treating
Clay Treating
IHydrotreating No Effect
Hydrogenation
Solvent Extraction
Solvent Dewaxing
• Solvent
CH 3 (CH 2 ) n CH 3 *- Removal By Filtering
Hydro-Dewaxing
CH3(CH2)nCH3+ H j - ^ CsHe.Etc.
FIG. 16—Hydrocarbon reactions: n-paraffins.
16 INSULATING OILS
Acid Treating
Clay Treating
Hydrotreating
Hydrogenation | No Effect
Solvent Extraction*
Solvent Dewaxing
Hydro-Dewaxing
'Small Amounts May Be Solubilized Into The Extract Phase.
FIG. 17—Hydrocarbon reactions: iso-paraffins, naphthenes.
Figures 19 and 20 show the slight removal of highly condensed aromatics by acid treating
under the mild conditions used for transformer oil. Hydrotreating converts some of them to
partially saturated hydrocarbons. Depending upon the severity of the process, hydrogenation
can fully saturate the aromatics, converting them to naphthenes. Solvent extraction will remove
substantially all of the highly condensed aromatics. Clay treating, solvent dewaxing, and hydro-
dewaxing have no effect. It is desirable to leave a substantial amount of aromatics (of the right
kind) in the oil to balance the impulse strength, gassing tendency, and oxidation resistance of
the fmished oil. This is achieved by adjusting the severity of the processes that affect aromatics
content.
Figures 21 and 22 illustrate the reactions with benzothiophenes. Acid-treating can remove
them; hydrotreating may convert a limited amount, and hydrogenation can totally convert them
if severe enough. Benzothiophene, in solvent extraction, behaves very much like any other single
ring aromatic, so some is removed with the extract. Clay treating, solvent dewaxing, and hydro-
dewaxing have no apparent affect on benzothiophene.
The effects on polar nitrogen compounds (Figs. 23 and 24) can be illustrated with quinoline
as an example. Acid treating will remove it, as will the subsequent clay treating. Hydrotreating
will convert a small amount, and hydrogenation can, if desired, convert all of it to a saturated
hydrocarbon. In solvent extraction, quinoline behaves like an aromatic, so it is partitioned be-
tween the raffinate and extract phases, with most going to the extract phase. The dewaxing
processes would not affect the quinoline content.
A d d Treating
^ 1^ Removed With
R-C=CH2 + Ho SO4 — > • R-C-OSO3 H S'"**9* ° ' ' "
^ ^ * riu Neutralization
^"3 step
(Aikyl Hydrogen
Sulfate)
Clay Treating - Limited Adsorption
Hydrotreating/Hydrogenation
R-CsCHg + H j - • R - C H j - C H a Hvdrocarb<
Hydrocarbon
Solvent Extraction L
Solvent Dewaxing 1 No Effect
Hydro-Dewaxing '
(UFRN) Universidade Federal do Rio Grande do Norte FIG. 18—Hydrocarbon
((UFRN) Universidade Federalreactions:
do Rio Grandeolefins.
do Norte) pursuant to License Agreement. No further reproductions auth
Acid Treating
Removed With
+H2O Sludge Or In
+H2SO4
Neutralization
-SO3H
Step
* At Treating Conditions (Sulfonic Acid)
Clay Treating - No Effect
Hydrotreating
R Slight <p
Partially Saturated
Aromatic
4> At Treating Conditions

FIG. 19—Hydrocarbon reactions: aromatics.
Hydrogenation
Substantial" R Full To Partially
• I Saturated
+ 7H.
Cat. Hydrocarbon
** At Treating Conditions
Solvent Extraction
Substantial" Higiily Condensed-Ring
> Compounds Substantially
+ Solvent
Removed With The Extract
Phase
" At Treating Conditions
Solvent Dewaxing V
Hydro-Dewaxmg f
FIG. 20—Hydrocarbon reactions: aromatics (continued).
Acid Treating
Ramovcd With
00=S+"2S04 ->OUl=S03H+"20r.?:.rc
Nautralization
Step
Clay Treating - No Effect
Hydrotreating
Desulfurized Hydrocarbon
*At Treating Conditions
(UFRN) Universidade Federal do Rio Grande FIG. 21—Hydrocarbon
do Norte reactions:
((UFRN) Universidade Federal do Riobenzothiophenes.
Grande do Norte) pursuant to License Agreement. No further reproductions auth
18 INSUUTING OILS
Hydrogenation
•At Treating Conditions 9*f"""f"!*u'^

Saturated Hydrocarbon
Solvent Extraction
rSff^v, ^_R ^ Partial Removal Into

^__JL j J _ H + Solvent ^ Extract Phase
Solvent D e w a x i n g ^
H y d r o - D e w a x i n g f "** ^ " • * = *
FIG. 22—Hydrocarbon reactions: benzothiophenes (continued).
Acid Treating
Removed In Sludge
N Step
Clay Treating
Clay
09 >• Adsorbed On Clay
Hydrotreating
Slight* ^.
—^—>. r;=^-CH2-CH2-CH3
Cat. W + N Hst
* At Treating Conditions De-Nitrogenated Hydrocarbon
FIG. 23—Hydrocarbon reactions: polar nitrogen compounds.
Hydrogenation
0 0 * 7 "2 "EiT^ Q-CH2-CH2-CH3+NH3t

De-Nitrogenated,
Saturated Hydrocarbon
Solvent Extraction
r ' ^ ^ l r ^ + Solvent >• M o d e r a t e T o Substantial

^=!s--*'^l^"** Removal In Extract
Solvent D e w a x i n g V . , _ . . .
„ . _ . " 1 No Effect
Hydro-Dewaxing f
Downloaded/printed by FIG. 24—Hydrocarbon reactions: polar nitrogen compounds.
Acid Treating
(1)2RSH + H 2 S O 4 — • R j S j + 2H2O + SO2 t
|2)R2S2 + H2SO4—^2 RS'SOjH —»- Removed With
Sludge Or In
Sulfonic Acid Neutralization Step
R-SH +2H2 —• RH3 + H2St
Cat. Desulfurized Hydrocarbon
R2S2 +5H2 —• 2RH3 + 2H2St
Cat. Desulfurized Hydrocarbon
Clay Treating I
Solvent Extraction k _ ^ NQ Removal
Solvent Dewaxing \
Hydro-Dewaxing |
FIG. 25—Hydrocarbon reactions: polar sulfur-mercaptans and disulfides.
Polar sulfur compounds, illustrated by mercaptans and disulfides in Fig. 25, are removed in
acid treating and converted easily by hydrotreating and hydrogenation. The other processes do
not seem to affect these compound types.
Polar oxygen compounds, illustrated by naphthenic acids in Fig. 26, are removed in acid
treating and converted easily by hydrotreating and hydrogenation. Clay treating also is effective
in removing naphthenic acids, hence its prevalent use in reclaiming used transformer oil. The
other processes do not affect naphthenic acids.
Water and particulates are removed by deliberate drying and filtering, but can also be re-
moved incidentally in some of the processes. Unfortunately, clay beds, catalyst beds, etc. make
excellent filters. Over a long period of time, this may produce flow problems, such as pressure
drop, in such operations.
Now that the refining processes and their chemical effects have been examined in some detail,
let us turn to a different way of listing the effects that corresponds to our normal perception of a
transformer oil. Table 2 illustrates the effects on the bulk oil properties that are of major inter-
est. Most of the effects are self-evident from the previous discussions, but a few need comments.
Acid Treating
Removed In
00
Clay Treating
^°°" • Neutralization
Step
00 — COOH • Adsorbed On Clay
QQ_COOH +3H2 - ^ Q Q -CH3.2H,.
Solvent Extraction
Solvent Dewaxing
Hydro-Dewaxing } ^ No Effect
FIG. 26—Hydrocarbon reactions: polar oxygen compounds-naphthenic acid.
20 INSULATING OILS
&
Q
(«
a 2
o o. ' ca
Cu i
Q
X
«
» ' OQ « « Q
4)
Q
J.
e •§
[»«</} : ca Q Q >-< oa oa ; BO ^
^
"3 * •
i/i
w
c
s
•a
X>.
c •1 mOQW JCOQQ^^CQ -m^
Si*
a
6
t-t
> j
•o a mm I |«QQ-icocQ -cp^
o
X>>••-
w
>. •5
ctf V a 00
u s
•o ta
'3 n BO ;BOQQ»OaBO ;fiO^
<
s
p
11
>>
••J
a
3
C
siiit*
liiillllii
2-5'
£0 z
u
H
o
Hydrogenation as considered in this paper will, by itself, make the oxidation stability worse.
Combined with the use of added oxidation inhibitors, it makes the oxidation stability much
better.
The flash point is largely unaffected by any of the processes except distillation. The degree of
stripping in some of the processes can marginally affect the value.
The gassing tendency is made poorer by any process that removes aromatics. Since none of
the refining processes discussed increase aromatics, there is no way described herein to improve
this property. The need to preserve a good gassing tendency must be balanced against the desire
for a good impulse strength.
As mentioned earlier, it is possible to make a satisfactory transformer oil by many different
process sequences. Some of the various combinations that have been reported are:
1. Acid treating, clay treating.
2. Hydrotreating, clay treating.
3. Hydrogenation.
4. Solvent extraction, hydrotreating.
5. Solvent extraction, hydrotreating, clay treating.
6. Solvent extraction, hydrotreating, dewaxing.
7. Hydrotreating, solvent extraction, dewaxing, clay treating.
These are probably not the only refining sequences used, and there will be others used in the
future as the refiners cope with changing crude oils and new technology.
Table 3 shows a limited comparison between transformer oils made from differing crude
sources and types using different processing. This table illustrates the fact that it should be
possible to make a satisfactory conventional transformer oil from any good lube crude though
paraffinic crudes require at least one additional processing step.
OSHA Regulation Effects

Adding complexity in today's environment in the United States is the Occupational Safety
and Health Administration's (OSHA) Hazard Communication Standard, 29 CFR 1910.1200,
which became effective 25 Nov. 1985. This regulation was subsequently modified in a Notice of
Interpretation published in the Federal Register on 20 Dec. 1985 at 50 FR 51852. Under the
requirements of this standard, some mineral transformer oils are classified as potentially car-
cinogenic based only on processing itself.
A full discussion of the OSHA standard and the Notice of Interpretation would require a
separate presentation, but the results can be summarized as follows:
1. Oils requiring a cancer hazard warning:
a. Oils made by acid treating.
b. Oils made by mild hydrotreating.
c. Oils made by mild solvent refining.
2. Oils exempt from a cancer hazard warning:
a. Oils made by severe solvent refining.
b. Oils made by severe hydrotreating.
c. Oils made by sequential processing of mild hydrotreating and mild solvent refining.
OSHA has defined mild hydrotreating, but has not yet defined either mild or severe solvent
refining.
Under the OSHA standard, oils made by acid treating plus clay treating and oils made by
mild hydrotreating
Downloaded/printed by plus clay treating now require a cancer hazard warning. There has been a
(UFRN) Universidade
shift away Federal
from dotheseRio Grande
twodoprocesses
Norte ((UFRN) by
Universidade Federal do Rio Grande do Norte) pursuant to License Agreement. No further reproductions auth
the industry.
22 INSULATING OILS
cQ
60 . *< •^ ** !S o •« f^
B, '5 H »°-|r~-oZZ°° ZZ Z d " - +
u
cD
S I - Z ' t d ' is
u
(/5
•IS
z ao UJ
o ^ ^" - V ^ ^
- 5, lOl oQo
lo " ; o 1^ <N o
I/)
00 CM
.1/) >o °o
. (N (N S <^
00 00
O <N
3S I O '^ i d d
(1. </)
(N H
P S 5; ^ ^ o c^ 2; 00 o r^ _; H ^ r~- I ".'
r- (N ° o '^ ' o I
•T:?
in
Z •^ - o ? :?,^
Tl- O
I
aI
W
a §§3a?5zog?
< f2Sod"^+ z^
3< o
gi H
I ji U
U 00 M
00 :~; c
c a. - •o ac •5
.. U » ô s
cu 5 X ilsl 1"= <yii4
^ 00
i:
•T3
- E •5
fco2 3 Si j : .
TT £ >
CS
II
liii
H lO W5 2 S ^
U 111 u ^—
u ao.S ^ 3
H
QQ U
DD O
OS Qi: ofi
U ( J BH
II II "II1JI << <
For more detailed information on this subject the reader is referred to the OSHA standard
published in the Federal Register on 25 Nov. 1983 at 48 FR 53280 and to the mandatory refer-
ence of most significance, lARC (International Agency for Research on Cancer) Monograph
Volume 33 (Pages 90, 150-151).
Handling and Storage

Only a brief discussion of handling and storage is included here. This is a subject worthy of a
separate discussion also.
Transformer oil is a delicate product that must be treated with extreme care and dedication to
prevent contamination. Water and particulates are the most common contaminants and the
easiest to remove. There are other materials, such as motor oils that are extremely difficult to
remove and must particularly be avoided.
Transformer oil received in drums, pails, etc. is subject to water contamination during stor-
age. The bung closures and seals are not perfect and may over a period of time begin to
"breathe". The mechanism is simple. The container becomes heated during the day and cools
at night. The resultant expansion and contraction can be enough to "pump" moist air in and
out of a drum sitting upright on a slab. This is accentuated, if the drum is stored outside.
It is preferred that drums be stored indoors and in racks laying on their sides with the two
bungs horizontally opposed so as to keep the closures under the surface of the oil. This will
greatly minimize chances of "breathing". Drums stored outdoors should never be stored up-
right. Water will collect on the head and can be pulled into the drum if "breathing" occurs.
Finally, keep the drum supply "fresh". Do not expect drums two and three years old to con-
tain dry oil. They will not.
Transformer oil transported in bulk is subject to other sources of contamination. It should be
transported in tank trucks or tank cars that are dedicated to that service alone. It is also desir-
able to have the conveyances equipped with dessicant breathers or nitrogen blanketing to re-
duce moisture contamination. Dessicant breathers may not be needed on tank trucks, if the
delivery time is relatively short.
Tank trucks or tank cars that have been in service handling compounded oils (motor oils,
turbine oils, etc.) or any vegetable or animal oil should not be used to haul electrical oils. The
dissipation factor of the electrical oil will be adversely affected by these products. Conveyances
that have been in these services are almost impossible to clean adequately for electrical oil use.
It is strongly recommended that any system for receiving transformer oil in bulk do so with
totally dedicated hoses, lines, pumps, and tanks. It is also advisable to have the storage tank
equipped with dried air or dried nitrogen blanketing and to provide a system to redry and filter
the oil.
Drying is best accomplished with a vacuum dehydration unit or by means of dried gas (air or
nitrogen) sparging in the tank. If sparging is used, a continuous purge of the gas space above
the oil with dried gas will be required to prevent recondensation of the moisture on the tank
roof.
It may seem strange to have a refiner caution the receiver of bulk product by saying the fol-
lowing: "Check the quality of the transformer oil, before it is unloaded." Nevertheless, there is
justification to always check the quality of the oil upon receipt. It will save time, work, and
anxiety in the long run.
This discussion of transformer oil handling has been brief. It, too, is a subject worthy of a
separate discussion.
Conclusion
This paper has presented an overview of the manufacture of mineral insulating oils of petro-
leum
(UFRN) originFederal
Universidade anddorelated
Rio Grandetopics. No startling
do Norte ((UFRN) new
Universidade concepts
Federal have
do Rio Grande been
do Norte) presented,
pursuant but rather
to License Agreement. it is
No further reproductions auth
24 INSULATING OILS
intended that this discussion will improve the reader's understanding and appreciation of some
of the complexities of transformer oil manufacture and handling.
Acknowledgment
The author wishes to express his indebtedness to the pioneering work of J. R. Lawley and
Homer Jennings in comparing hydrotreating and acid treating as processes for refining mineral
transformer oil.
DISCUSSION
C. L. Sobral Vieira^ (written discussion)—According to ASTM D 3487, Type I oil can have
up to 0.08% of DBPC. Is the permission given to the producer of the oil to add artificial inhibi-
tor at this level a help to them to produce an oil that meets the specification (good oil) or is it a
way to improve the life of an already good oil (an oil that would meet the specification even
without the DBPC)?
T. Lipscomb (author's closure)—Both situations probably exist today. Remember that
OSHA has said that a cancer hazard warning is required if the oil is made by mild hydrotreating
plus clay treating or by acid treating plus clay treating. Some refiners have switched to severe
hydrotreating to avoid a cancer hazard warning. Severe hydrotreating usually destroys enough
of the natural inhibitors that addition of a synthetic inhibitor is required to pass the oxidation
tests. Other methods that are severe but more selective may not require synthetic inhibitors. An
example of this is solvent extraction plus hydrotreating. In this case the refiner may elect to add
synthetic inhibitor to enhance the quality of an oil that already meets the oxidation tests.
'CEPEL, P.O. Box 2754, Rio de Janeiro, Brazil.
Howard C. Manger^
PCBs: The Beginning, the JVIuddie,

the End?
REFERENCE: Manger, H. C , "PCBs: The Beginning, the Muddle, the End?" Electrical Insu-
lating Oils. STP 998, H. G. Erdman, Ed., American Society for Testing and Materials, Philadel-
phia, 1988, pp. 25-34.
ABSTRACT: Over the years, there have been many changes in the federal regulations regarding
PCBs. From the beginning, utility personnel were in a dilemma as to how to deal with the
problem.
First, we had trouble realizing that the askarel we were using in our transformers was more than
50% polychlorinated biphenyl, and even if it was, could it possibly be a toxic or a hazard to any-
one's health? After all, we and those before us had been up to our elbows in the stuff for years. It
may have burned our eyes a little, but as far as we know it did us no physical harm. We also found
it was many things in our everyday life, at work and at home.
Soon those of use close to the situation realized that the utility industry, in particular, was in
deep trouble. Congress passed the Toxic Substance Control Act in 1976 and singled our PCB to be
controlled. Three years later EPA published the first of a number of rules that have changed the
way utilities do business. The industry has been forced to spend millions of dollars and countless
man-hours dealing with regulations that we questioned from the start and, at best, learned to live
with. In addition, the rules change so often it is truly difficult to keep up with them.
This paper tries to take us from the start, thru the confusing muddle of regulations, to where we
are today.
KEY WORDS: askarel, Aroclor, Environmental Protection Agency, federal regulations, insulat-
ing fluids, polychlorinated biphenyls (PCBs), PCB, Toxic Substance Control Act
It was obvious in the early 1970s that Congress was serious about enacting legislation regard-
ing control of toxic substances in the environment and in commerce. Polychlorinated biphenyls
apparently were not one of the major concerns early in the process.
However, during the Senate debate on S.3149, March 1976, Senator Nelson introduced an
amendment on polychlorinated biphenyls. His remarks were primarily on the control of PCB
manufacturing and the environmental and health effects from PCB environmental
contamination.
In the House of Representatives in August 1976, Representatives Dingell and Gude intro-
duced an amendment on polychlorinated biphenyls. Their remarks were primarily on environ-
mental levels and exposure. Other remarks by members were addressed to the question of sub-
stitutes for commercial PCBs and to why PCBs should be singled out and not controlled by other
sections of the Act.
The Toxic Substance Control Act (Public Law 94-469) was enacted on 11 October 1976. Sec-
tion 6(e) of the act required that polychlorinated biphenyls specifically be regulated. It is the
only substance that was singled out.
In 1976, the cumulative U.S. production of PCBs since 1919 was estimated at 1.4 billion
pounds. Of the 1.25 billion pounds used in the United States, about 760 million pounds were in
service; 290 million pounds are in dumps and landfills; 55 million pounds had been destroyed;
and 150 million pounds were believed currently located in soil, water, air, and sediment.
At that time, there were continuous discussions within EPA on the definition of "Polychlori-
Copyright'General supervisor,
by ASTM Int'l Baltimore
(all rights reserved); Thu MayGas and Electric,
6 14:02:13 EDT 2021 Baltimore, MD 21203.
25
26 INSULATING OILS
nated Biphenyls" as a chemical and on the scope of coverage which should be included by the
definition. Section 6(e) only mentions "Polychlorinated Biphenyls" and does not use any other
commonly used acronym, trade name, or chemical name; nor is there offered a definition.
It was clear to EPA from the wording of the Act and from the debates on the amendments
that Congress meant for the Agency to control manufacturing, processing, distribution, use,
and disposal of polychlorinated biphenyls, and the environmental contamination from all
sources. Therefore, the marking and disposal regulations were to apply not only to newly manu-
factured PCBs, but also to those PCBs in use in electrical and other industrial equipment and
those PCBs which occur as a significant impurity in other chemicals.
Commercial PCB
Commercial polychlorinated biphenyls or PCBs are prepared with anhydrous chlorine react-
ing with biphenyl in the presence of ferric chloride. This crude PCB mixture is distilled to obtain
the finished product.
The biphenyl (diphenyl, phenylbenzene) ring is numbered as shown in Fig. 1.
Chlorine can be placed at any or all of the sites. There are 209 possible chlorobiphenyl iso-
mers, but only about 100 different isomers have been found in commercial PCBs.
The synthesis of chlorinated biphenyl was reported as long ago as 1881, and successful com-
mercial production of PCBs was commenced in this country in 1929 by the Monsanto Co.,
which has been the sole U.S. producer of Aroclors (Monsanto trade name).
The molecular composition of six typical Aroclors is shown in Table 1. Monsanto adopted a
four-digit designation for its Aroclors; the last two digits indicate the approximate chlorine
content by percentage weight; the first two digits indicate the type of material: biphenyl,
triphenyl, or a mixture of the two.
Physical Properties
The three most important physical properties of PCBs are: low vapor pressures, low solubil-
ity, and high dielectric constants. They are miscible with most organic solvents.
The chemical properties that make PCBs desirable industrial materials are their excellent
thermal stability, their strong resistance to both acid and basic hydrolysis, and their general
inertness. They are quite resistant to oxidation.
The higher chlorinated biphenyls are nonflammable and have extremely low volatilities. Un-
fortunately, some of the same characteristics, such as stability and nondegradability, make
them highly persistent in the environment.
Uses
Their chemical stability, low volatility, high dielectric constant, and compatibility with other
chlorinated hydrocarbons have resulted in many and varied industrial applications for the
PCBs.
3 2 2 3
Downloaded/printed by FIG. 1—Biphenyl ring.
MANGER ON PCBs 27
TABLE 1—Molecular composition of typical Aroclor.
Aroclor
1221 1016 1242 1248 1254 1260
Biphenyl 11 T T T T
Monochloro 51 1 1 T T
Dichloro 32 20 16 2 05
Trichloro 4 57 49 20 1 T
Tetrachloro 2 21 25 40 21 T
Pentachloro T 1 8 32 48 12
Helachloro T 1 4 23 28
Heptachloro T 2 6 41
Octachloro T T 18
Nonachloro 1
NOTE: T = trace.
Major uses of PCBs were as dielectric liquids in capacitors (99% PCB) and in transformers
(60 to 99% PCBs), as dye carriers in carbonless copy paper (3V2% PCB in the paper), and as
plasticizers in various paints and plastics (1 to 10% PCB).
Uses of PCB also included formulation into ballasts for fluorescent fixtures; impregnation for
insulation of electric cables and electronic capacitors; and plasticizers of vinyl chloride polymer
films.
Because of their thermal stability and fire resistance, PCBs also found application in high-
pressure hydraulic fluids, specialized lubricants and gaskets, heat transfer agents, and machine
tool cutting oils. Miscellaneous uses include: formation into some epoxy paints; protective coat-
ing for wood, metal, and concrete; adhesives; and in carbonless copy paper.
Askaiels
Dielectric and cooling fluids used in electric equipment that contain significant amounts of
PCBs and blends of chlorinated benzenes are generally referred to as askarels.
Manufacturers either modified the Aroclor themselves or had Monsanto prepare special for-
mulations for them. These "askarel" have been marketed and used by the manufacturers in
their products under various trade names. The more common names are listed in Table 2.
Toxic Substance Control Act (TSCA) (Start of the Muddle)

Congress passed TSCA in 1976 and appointed EPA the watchdog. Since then, we have been
in a continuing muddle as how to deal with PCBs. One must realize that, with the swipe of a
pen. Congress legislated a fluid used with no obvious ill effects into a hazardous substance.
However, the Toxic Substance Control Act allowed EPA to write regulations so there could be
three ways for continued use of a PCB:
1. Totally enclosed.
2. Use authorization.
3. Exemption.
Totally Enclosed
The
Copyright main
by ASTM Int'lbenefit of having
(all rights reserved); a use
Thu May considered
6 14:02:13 EDT 2021 totally enclosed is that it may fall outside of the
Toxic Substance Control Act if it can be shown that it offers insignificant exposure to human
28 INSULATING OILS
TABLE 2—Various trade names for

askarel.
Name Manufacturer
COMMON DOMESTIC TRADE NAMES

Aroclor Monsanto
Aroclor B Mallory
Asbestol American Corp.
Askarel Hevi-Duty Corp.
Askarel Ferranti-Packard Ltd.
Askarel Universal Mfg. Co.
Chlorextol Allis-Chalmers
Clorinol Spraque Electric
Clorphen JARD Company, Inc.
Diaclor Sangamo Electric
Dykanol Cornell Dubilier
Elemex McGraw Edison
Eucar Electric Utilities Corp.
Hyv Aerovox
Inerteen Westinghouse Electric
No-Flamol Wagner Electric
Pyranol General Electric
Saf-T-Kuhl Kuhlman Electric
COMMON FOREIGN TRADE NAMES

Clophen Bayer (Germany)
Fenclor Caffaro (Italy)
Kennechlor Kenneclor (Japan)
Phenoclor Prodelec (France)
DK Caffaro (Italy)
Pyralene Prodelec (France)
Solvol USSR
health and the environment. Unfortunately, insignificant exposure, by EPA definition, is no

exposure.
Use Authorization
However, EPA can authorize a use, not considered totally enclosed, if it can be shown that
such use does not offer unreasonable risk. The Toxic Substance Control Act may allow this
authorized use for extended periods of time. To reiterate: Subject to EPA approval, a use con-
sidered totally enclosed does not have to be regulated; however, if the risk is reasonable, a use
can be authorized.
Exemption
Exemptions are applied for by the user, who must show no unreasonable risk, indicate effort
for improvement, and reapply for the exemption every year.
Regulation
With these ground rules, the first rule was published on 31 May 1979. EPA looked at the use
activities
Copyright by ASTMofInt'lPCB and
(all rights deemed
reserved); the6 14:02:13
Thu May use of EDT
capacitors
2021 and transformers to be totally enclosed. EPA
felt that this equipment posed an insignificant exposure
(UFRN) Universidade Federal do Rio Grande do Norte ((UFRN) Universidade Federal to do
do Rio Grande the environment.
Norte) pursuant to LicenseEPA also
Agreement. No autho-
further reproductions auth
MANGER ON PCBs 29
rized a number of uses, some of which were railroad transformers, small capacitors, and carbon
paper. There were a few exemptions sought under that 31 May 1979 final rule.
The rule established regulatory cutoff of 50 ppm for manufacturing, processing, distribution
in commerce, or use of PCB.
It covered:
1. Marking.
2. Recordkeeping.
3. Restrictions on repairs.
4. Disposal (basically, incineration was the recommended method).
5. Storage.
6. Inspection.
This May 1979 rule took effect on 2 July 1979.
Rule Challenged
As new and disruptive as the regulation was, electrical equipment owners were learning to live
with it. However, the Environment Defense Fund (EDF) challenged the 31 May 1979 rule in the
U.S. Court of Appeals in Washington D.C. on three grounds:
1. EDF felt that there was no record for EPA to establish a 50-ppm cutoff.
2. That EPA had no data to back up the claim that capacitors and transformers were totally
enclosed.
3. The authorizations that were allowed were improper.
The court ruled that transformers and capacitors were not totally enclosed, and that EPA
lacked substantial evidence to support a regulatory cutoff of 50 ppm for manufacture, process-
ing, distribution in commerce, or use of PCB.
Had the court's decision gone into effect, making any use of PCB illegal, EPA said it would
have caused a major economic impact. I think we would have had to turn the lights out.
However, EPA, EDF, and certain industry and electric utility intervenors involved in the case
filed a joint motion seeking a stay of the court's mandate until further rulemaking could be
completed. The court granted this request.
Additional Regulation (as a Result of the Court Decision)

On 22 April 1982, EPA issued a proposed rule governing the use and servicing of electrical
equipment containing PCBs. The final rule appeared in the Federal Register of 25 August
1982. This final rule was issued as a result of the court's decision and dealt only with electrical
equipment. It authorized certain uses of PCB in electrical equipment and set 50 ppm as a cutoff
for electrical equipment.
This Rule:
1. Prohibited the use of PCB transformers and PCB electromagnets posing a risk to food and
feed facilities after 1 Oct. 1985.
2. Expanded the definition of electrical equipment posing an exposure risk to food or feed
facilities.
3. Authorized the use of all other PCB transformers for the remainder of their useful lives,
with inspections. (Take note: it "authorizes," where May 1979 considered transformers totally
enclosed.)
4. Set requirements for inspection frequency and retention and availability of records.
Copyright
5.by Authorized
ASTM Int'l (all rights
thereserved);
use of Thuall
MayPCB-containing
6 14:02:13 EDT 2021 mineral oil-fill electrical equipment.
6. Presumed circuit breakers, reclosers,
(UFRN) Universidade Federal do Rio Grande do Norte ((UFRN) and
Universidade cables
Federal do Rio to contain
Grande do Norte)less than
pursuant SO-ppm
to License PCBs.
Agreement. No further reproductions auth
30 INSULATING OILS
7. Authorized the use of large PCB capacitors located in electrical substations and indoor
installations.
8. Prohibited the use of all other large PCB capacitors after 1 Oct. 1988. (Here we see a time
period tied to authorization.)
9. Allowed storage of large PCB capacitors and PCB-contaminated equipment outside of
qualified storage facilities.
10. Defined "disposal" as including leaks and spills, but went on to say: spills, leaks, or
uncontrolled discharges resulting from the authorized use (or storage) of electrical equipment
shall not constitute a disposal violation provided adequate cleanup measures are initiated
within 48 h after notice of the discharge.
11. Required records of inspection and history to be kept for three years after disposing of
PCB transformers.
Nonelectrical Rules
The 25 Aug. 1982 rule authorized a 50-ppm cut only for electrical equipment. What about
the other items and systems that have PCB in them (even a molecule of PCB)? There are many
nonelectrical items that fall into this category.
EPA has stated that any time you have a carbon, chlorine, and some heat, the possibility
exists that you can create a polychlorinated biphenyl. It follows that any manufacturing process
that uses these parameters is suspect of the incidental generation of PCBs. Industrial represen-
tatives convinced EPA that there are some manufacturing systems in which PCB may be cre-
ated, but where it is not released into the environment. EPA issued Phase One of a two-part rule
on 21 Oct. 1982 to address the 50-ppm regulatory cutoff in "closed and controlled" waste man-
ufacturing processes.
The second phase of regulation dealt with processes involving "uncontrolled" PCBs gener-
ated in other than "closed" and "controlled waste" processes.
These non-Aroclor, inadvertently generated PCBs were the principal subject of a 10 July 1984
rule. The exclusions announced in this 1984 rule expanded upon and superseded the exclusions
for closed and controlled waste manufacturing processes. The generic exclusion for inadver-
tently generated PCBs applied to manufacturing processes which qualified as "excluded manu-
facturing processes." These excluded processes were defined in terms of established limits for
PCB releases in products, air emissions, water effluents, and wastes.
Transformer Fires and the Regulation

At the time these rules were written, information available to the EPA indicated that fires
involving electrical transformers were rare, isolated incidents. However, several transformer
fires in buildings brought into question EPA's earlier assumptions.
Binghamton, NY
The first and most notable event involved an 18-story state office building, the tallest and
most prominent landmark in Binghamton, a city of some 60,000 people in southeastern New
York. On 5 Feb. 1981, an intense fire occurred in the electrical switching equipment in the
basement mechanical room of the building. The switching equipment was located adjacent to
two transformers filled with askarel (65% polychlorinated biphenyls and 35% chlorinated ben-
zenes). The building and transformers were owned by the State of New York.
The heat from some allegedly burning automobile fan belts stored next to the transformers
cracked a porcelain bushing on one of the transformers and about 180 gal of askarel sprayed
ontobythe
Copyright ASTM floor and
Int'l (all rightsinto theThu
reserved); hotMayswitch gear.
6 14:02:13 The fire produced a fine soot that spread up an open
EDT 2021
vertical chase which carried the exhaust throughout
(UFRN) Universidade Federal do Rio Grande do Norte ((UFRN) Universidade every
Federal do Rio floor
Grande of pursuant
do Norte) the building.
to License Agreement. No further reproductions auth
MANGER ON PCBs 31
Immediately following the fire, and before the extent and nature of the contamination be-
came known, more than 300 people reentered the building to help with the cleanup or to retrieve
documents and personal effects. The cleanup effort was begun. However, three weeks after the
fire, the extent of the contamination was discovered.
Analysis of soot samples recovered from the building indicated high concentrations of PCBs
and the presence of polychlorinated dibenzodioxins at about 3 ppm and dibenzofurans at con-
centrations of around 300 ppm.
The building was closed and secured and cleanup was started. Every movable item was
scrapped. Walls and partitions were ripped out and thousands of barrels of material were
hauled to the Niagara Falls landfills, including six acres of floor tiles. The State had hoped to
reopen the building in April 1985. It has not opened yet (January 1988) and the cost of cleanup,
so far, is approaching twice the original cost of the building.
San Francisco
About two years later, on 15 May 1983, a fire occurred in a transformer vault located under a
sidewalk adjacent to One Market Plaza, an office building in downtown San Francisco. The
transformers involved in Binghamton were inside the building and were the property of the
owner. This askarel transformer was outside in a sidewalk vault and was owned by Pacific Gas
& Electric Co. The smoke from this incident was drawn into the building thru an outside intake
vent adjacent to the sidewalk grating. The lower floors of this building were contaminated with
PCB and furans. Although the upper floors were reoccupied after a short time, it took months
before the lower floors were cleaned to a level low enough for reopening.
Tulsa
A similar sidewalk-type transformer vault incident occurred about a year later in Tulsa,
Oklahoma. The high side switch of a network transformer overheated and exploded, atomizing
between 20 and 30 gal of PCB onto a city street and into an adjacent basement of the Beacon
building. A cleanup was commenced by Public Service of Oklahoma, who owned this
transformer.
Subsequent sampling for PCB indicated that the cleanup was not successful, and a profes-
sional cleanup company with experience at the San Francisco fire was hired.
For the protection of their workers, this contractor tested for the presence of furans and diox-
ins. These tests showed the presence of furans in surprisingly large quantities. The basement
was again evacuated some four months after the accident, and a complete cleanup was
commenced.
The cleanup of the basement was in its very last stages when a second identical explosion
occurred in the high side switch of the replacement transformer. This time there was consider-
able contamination on the outside of the building and on the outside of an adjacent building
containing a hot dog restaurant and bar, but, because of the work performed in the vault, no
PCBs penetrated the basement/vault wall of the Beacon building.
However, before a cleanup of the vault could commence, a 12-in. rain deluged Tulsa and
flooded the vault. This water breached the basement wall of the Beacon building with PCB-
contaminated water, requiring another involved and costly cleanup.
Transformer Fire Rule

Because of these and other occurrences, the Agency issued an advance notice of proposed
rulemaking (ANPR) on 23 March 1984, a proposed rule on 11 Oct. 1984 to gather data on the
specific
(UFRN) risks
Universidade posed
Federal do Rio by fires,
Grande and
do Norte a final
((UFRN) rule for
Universidade electrical
Federal transformers
do Rio Grande onLicense
do Norte) pursuant to 17 July 1985.
Agreement. No further reproductions auth
32 INSULATING OILS
Briefly, EPA imposed the following amendments to the August 1982 Rule. This rule required
immediate notification to the National Response Center of a PCB transformer fire-related inci-
dent and prohibited the further installation of PCB transformers in or near commercial build-
ings. A number of requirements were to be completed by 1 Dec. 1985:
1. The registration of all PCB transformers with fire departments or fire brigades with pri-
mary response function.
2. The registration of all PCB transformers owned by utilities located in or near buildings,
with the owners of the building.
3. The marking of the exterior of all PCB transformer locations (excluding grates and man-
hole covers).
4. The removal of combustible materials stored within 5 m of a PCB transformer enclosure.
The regulation also set another date as a milestone: 1 Oct. 1990. It prohibited the continued
use of higher secondary voltage network PCB transformers (equal to or greater than 480 V) in or
near commercial buildings beyond 1 Oct. 1990. This final rule also required by 1 Oct. 1990 the
installation of enhanced electrical protection on lower secondary voltage network PCB trans-
formers (less than 480 V) and on all radial PCB transformers used in or near commercial
buildings.
Challenge to the Fire Rule

The 17 July 1985 rule, which became known as the "fire rule," was challenged by Mississippi
Power Co. in the U.S. Court of Appeals for the 5th Circuit in Louisiana. In the settlement of
that appeal EPA agreed to undertake good faith efforts to issue and cause to be published in the
Federal Register by 31 Dec. 1986, a document clarifying EPA's position concerning the inter-
pretation of certain provisions of the rules. (The Document is a series of questions and answers.
EPA made the deadline. They were published on the last day, 31 Dec. 1986.)
In addition, some important salient amendments to the PCB Rule were proposed. The re-
striction in the "fire rule," that would not allow a PCB transformer to be installed in a location
in or near a commercial building has been changed to allow such in (1) emergency situations
(EPA states they would not want to "deny electric service to the entity or entities served by such
transformer") and (2) "in situations where the transformer has been retrofilled and is placed in
service in order to qualify to be reclassified."
Although the requirement that a network PCB higher voltage transformer, that is, with sec-
ondary greater than or equal to 480 V in or near commercial buildings, must be reclassified to
less than 500 ppm or be replaced by 1 Oct. 1990 has not changed, the settlement amendment
states that radial PCB transformers in or near commercial buildings must be equipped with
electrical protection by 1 Oct. 1990, which can deenergize the transformer within several hun-
dreds of a second. (The 17 July 1985 rule called for several tenths of a second.)
PCB network lower voltage transformers (less than 480 V) have been further reclassified into
those located in sidewalk vaults and those not located in sidewalk vaults. Those located in side-
walk vaults must be removed from service by 1 Oct. 1993. The ruling is not specific, but I would
think reclassification is an option.
PCB network lower voltage transformers not located in sidewalk vaults may continue in ser-
vice after 1993 if electrical protection is supplied to deenergizing the transformer within several
tenths of a second. This must be done by 1 Oct. 1990. In addition, if a transformer has not been
so protected by 1 Oct. 1990, the transformer must be registered with EPA.
These changes to the "fire rule" should be published in the spring of 1988.
MANGER ON PCBs 33
Spill Cleanup
A very important item has finally come to fruition. The administrator of EPA signed a PCB
spill policy on 20 March 1987. It was published in the Federal Register as 761.20 on 2 April
1987.
There have been efforts to come to some sort of an agreement about spill cleanup with EPA
since 1984. A number of organizations were involved, including the Utility Solid Waste Activi-
ties Group, EEI, NEMA, the Chemical Manufacturers' Association, the Environmental De-
fense Fund, and the Natural Resource Defense Fund. Early on, environmental factions and
EPA were talking about cleaning up almost every molecule of spilled PCB. Further, for some
time, EPA has considered a spill or a leak as improper or unauthorized disposal, and, as such,
they could require a cleanup to background levels. It is a credit to all the organizations that this
policy was the result of a consensus agreement.
There are a few salient points concerning this spill cleanup policy.
It is very significant that no cleanup requirements exist for leaks or spills of fluids containing
less than 50 ppm PCB, that is, leaks or spills from non-PCB transformers are not regulated by
this policy.
For any spill or leak of fluids containing more than 500 ppm (a PCB fluid), EPA requires not
only cleanup of the fluid, contaminated soil, and contaminated surfaces, but also analytical
sampling of the soil and surfaces to assure specific low concentrations of PCB following comple-
tion of the cleanup. These levels are as low as 10 ppm, depending on where the spill is, and are
as low as 10 ^g per 100 cm^, again depending on where you might have spilled this fluid.
For spills and leaks of less than 270 gal of fluid between 50 and 500 ppm PCB, EPA requires
only cleanup of visible traces of fluids and double rinsing of surfaces, but does not require
analytical sampling to confirm the adequacy of the cleanup. These cleanups must be completed
within 48 h of the time you were aware of the spill. However, if some of this fluid should enter a
residence, the cleanup levels could be as low as 10 fig per 100 cm^. Any spill or leak of material
containing more than 10 lb of PCB must be reported to EPA.
One of the more important things that came out of the consensus and this policy is that
penalties for improper disposal can be avoided if there is a good faith compliance with the
cleanup requirements. Basically, this is saying that if you make a good faith effort to clean up
the spill, and it is realized later that you didn't fully comply with the regulations, you can be
asked to clean up to a lower level, but you won't be fined.
Health Effects
PCBs can certainly be classified as an environmental contaminant. They are stable, do not
biodegrade easily, and tend to bioaccumulate in organisms. As they move up in the food chain,
they may increase in concentration. Laboratory studies on fish and animals have shown PCBs to
cause chronic (long-term) toxic effects in some species even at low concentrations; however,
PCBs are not acutely toxic or lethal. Numerous health effect studies have been made by some of
the most prestigious health consultants in the country. In summary, they conclude:
1. PCBs have not been found to be implicated as human carcinogens.

2. PCBs have not been shown to affect adversely human reproduction or cause birth defects.
3. PCBs do not affect enzyme production in humans in a manner that leads to adverse health
effects.
4. The only proven adverse human health effect resulting from PCBs is chloracne or skin
irritation, bya reversible health effect resulting in no long-term adverse health consequences.
Downloaded/printed
34 INSULATING OILS
The End ??????

Available evidence today indicates that some level of concern about PCBs is well founded.
PCBs are persistent, ubiquitous, and bioconcentrate. Subsequent studies, however, show that
adverse health effects have not resulted from past levels of exposure to PCBs. Future effects are
even more unlikely because recent studies of PCB levels in the environment, in the food chain,
and, indeed, in human adipose tissue are definitely trending downward.
It should then follow that additional PCB regulations are unnecessary, and it is time to move
on to other issues. While this may be logical and reasonable based on past experience, it is
doubtful that this is the end of the ever-expanding PCB regulations.
Bibliography
Environmental Research, Vol. 5, No. 5, September 1972.
Wood, D., National Conference On PCBs, 19-20 Nov. 1975.
Druley, R. M. and Ordway, G. L., The Toxic Substances Control Act, The Bureau of National Affairs,
Inc., Washington, DC, 1977.
Draft Preamble to Proposed PCB Marking and Disposal Rule, Part 761, 4 Feb. 1977.
Federal Register, 31 May 1979, Vol. 44, No. 106, Supplementary Information and Rule.
EPA's Final PCB Ban Rule, Over 100 Questions & Answers to Help You Meet These Requirements, June
1980.
Manger, H. C , "Status of EPA Rules and Regulations," Doble Conference Minutes, 1981.
"PCB Regulations Status Report," Doble Conference Minutes, 1982.
Federal Register, 25 Aug. 1982, Vol. 47, No. 165, Supplementary Information and Amendments.
Craddock, J. H., "The History and Toxicology of PCBs," PCB Seminar, Baltimore, MD, March 1983.
EPA's Final PCB Ban Rule, Over 100 Questions & Answers to Help You Meet These Requirements,
Revision 3, August 1983.
Manger, H. C, "Progress Report on Status of EPA's Latest Rules," Doble Conference Minutes, 1983.
"Update on PCB Rules and Regulations," Doble Conference Minutes, 1984.
Federal Register, 11 Oct. 1984, Vol. 49, No. 198, Supplementary Information and Amendments.
Manger, H. C, "Reclassification of PCB and PCB Contaminated Transformers," Doble Conference
Minutes, 1985.
"Fires Involving PCB Transformers—Further Regulations," Doble Conference Minutes, 1986.
Federal Register, 17 July 1985, Vol. 50, No. 137, Supplementary Information and Amendments.
Code of Federal Regulations 40 Part 700 to End, revised as of 1 July 1985.
Kump, R. K., "EPA Update: Fire Rules," Doble Conference Minutes, 1986.
Association of Edison Illuminating Companies, Report of Committee on Power Distribution 1985-1986,
PCB Transformer Regulation Implementation.
Manger, H. C, "Status of EPA Rules and Regulations," Doble Conference Minutes, 1987.
Federal Register, 2 April 1987, Vol. 47, No. 250, Supplementary and Amendments.
Section II—Significance of Application
Dennis L. Johnson^
Insulating Oil Qualification and

Acceptance Tests from a User's
Perspective
REFERENCE: Johnson, D. L., "Insulating Oil Qualification and Acceptance Tests from a User's
Perspective," Electrical Insulating Oils, STP 998, H. G. Erdman, Ed., American Society for
Testing and Materials, Philadelphia, 1988, pp. 37-45.
ABSTRACT: To be acceptable, insulating oil must have several specific electrical, chemical, and
physical attributes. Tests that demonstrate that the oil does indeed possess the desired character-
istics are required. The tests specified by one utility are outlined in the paper with limits listed.
Also described in the paper are insulating oil uses on a utility system. An attempt is made to relate
oil use to each required test. The relative significance of each test is addressed.
KEY WORDS: insulating oil, technical specifications, acceptance tests, qualification tests
My occupational involvement with insulating oil is through my employment as a manager in a

large federal utility, the Bonneville Power Administration (BPA). EPA is the largest of five
power marketing agencies in the United States Department of Energy. The BPA service area
covers Oregon, Washington, Idaho, Western Montana, and corners of California, Utah, and
Wyoming. We have electrical equipment located near the Pacific Ocean, in the high deserts of
Eastern Oregon and Washington, and in the arctic-like environs of the Northern Rockies and
Grand Tetons. Our equipment faces all weather extremes.
BPA transmits bulk power to its 100 plus customers in the Northwest United States. The
customers are the local public and private utilities and electroprocess industries. There are
more than 14 000 miles of transmission lines connecting some 400 substations. Installed in
those substations are approximately 1100 large, high-voltage power transformers and reactors
and approximately 1600 oil-insulated, high-voltage power circuit breakers. The facilities also
include numerous other related oil-filled equipment items such as station service and instru-
ment transformers, and voltage regulators. We estimate that altogether we have at least
6 000 000 gal of insulating oil in energized equipment. Of course, it goes without saying that the
integrity of each gallon, both as insulant and as a media for heat transfer, is of considerable
significance to the reliable transmission of electrical energy to the distribution or end-use cus-
tomer.
Prociuement Practice
For our utility, oil performance is especially important because our equipment specifications
encourage manufacturers to minimize the amount of oil needed. Further, we specify reduced
basic insulation levels (BIL) in equipment. Thus, tremendous reliance is placed on the oil to
withstand the rigors of thermal and electrical stress during service life. Our practice is to buy the
oil we use separately from major equipment. This practice gives us better control and knowl-
edge of what liquid goes into our electrical equipment. Equipment suppliers are prenotified of
Copyright'Manager,
by ASTM Int'l Substation Maintenance,
(all rights reserved); Bonneville
Thu May 6 14:02:13 EDT 2021 Power Administration, Vancouver, WA 98666.
37
38 INSULATING OILS
the oil we intend to use, and, with their consent, our new-product warranty is therefore pre-
served.
We prepare equipment for service ourselves. We have excellent oil-handling apparatus and
our crews have developed substantial expertise through experience. In all cases, our equipment
preparation practices either match or exceed manufacturers' requirements. Through past prac-
tice and thorough documentation of equipment preparation activity, a reputation has been es-
tablished that withstands manufacturers' scrutiny in the event of trouble during the warranty
period.
By being large and by virtue of being an agency of the federal government, it seems our pro-
curements are uniquely complicated. The procurement process is a function of federal procure-
ment regulations intended to promote competition and result in high value for the dollars spent.
Whether or not this goal is attained is a topic for other symposiums. The purchase specifications
for insulating oil that I am familiar with include a thick layer of "boiler plate," which also is
best left for other discussions. What we do want to focus on is the examination given to insulat-
ing oils available on the market and what we learn about quality of individual oils from that
examination.
Technical Specifications
The specification requirements for insulating oil used in electrical equipment at BPA are that
the oil shall be in accordance with Type II of ASTM Specification for Mineral Insulating Oil
Used in Electrical Apparatus (D 3487). Tests, including submittal of test reports and data,
assure that the oil being furnished does meet the sampling and testing requirements of D 3487.
Oil suppliers are also requested to provide information about:
1. Supplementary design as listed in Appendices XI and X2 of D 3487.

2. The generic refining process used.
3. The type of crude.
4. The type of oxidation inhibitor added.
5. Polychorinated biphenyl (PCB) concentration, equipment used for PCB tests, and accu-
racy and minimum detection range of PCB test equipment. ASTM Method for Analysis of
Polychlorinated Biphenyls in Mineral Insulating Oils by Gas Chromatography (D 4059) is to be
followed.
The concentration of PCBs in the oil shall not exceed 2 ppm when tested according to D 4059.
Qualification Testing
Vendors of insulating oil are required to demonstrate that the product they offer is of suitable
quality. To qualify to supply oil to the government, a vendor must furnish representative sam-
ples taken in accordance with ASTM Method for Sampling Electrical Insulating Liquids (D
923) for a rigorous testing regimen. Samples for qualification testing are taken after final refin-
ing processes are complete but before loading for transportation. Once a vendor becomes quali-
fied, that testing process is not repeated unless a significant product change occurs.
Acceptance Testing
Qualified vendors must provide further samples for acceptance testing for individual deliv-
eries. Acceptance testing involves only tests needed to assure that batches of known products
retain the expected characteristics.
The specifications require that samples of oil for acceptance testing be drawn by the vendor
from his storage tanks. We draw samples from our dedicated oil transport trucks after they are
loaded.
(UFRN) (Use
Universidade of doour
Federal Rio own
Grandeoil tank((UFRN)
do Norte trucks has eliminated
Universidade frequent
Federal do Rio Grande disconcerting
do Norte) pursuant to Licenseinstances
Agreement. Noof con-
JOHNSON ON QUALIFICATION TESTS 39
tamination that occur during shipping.) Both samples are brought to BPA's Materials Labora-
tory in Vancouver, Washington for acceptance testing. The supplier's samples are the "referee"
samples. On occasion we have taken additional samples from the transport vehicle at our site if
we suspected the oil had become contaminated during transit.
The oil samples provided are examined in detail. Individual tests referenced in D 3487 are
listed in Table 1 with a brief comment on each following. After this review of the D 3487 tests,
the paper will be concluded with remarks about additional oil characteristics of importance that
should be investigated.
TABLE 1—Qualification & acceptance tests and limits.
Qualifi- Accep-
ASTM Method Limit cation tance
ELECTRICAL TESTS
Dissipation Factor 0.05%, maxat25°C X X
(Power Factor) (D 924) 0.3%, max at 100°C X
Dielectric Breakdown (D 877) (disks) 30-kV, min X X
Dielectric Breakdown (D 1816) 28-kV, min at 0.040-in. gap
(spheres) 56-kV, min at 0.080-in. gap X X
Gassing Tendency (D 2300) + 1 5 fiL/min, max (Procedure A) X
+ 30 (iL/min, max (Procedure B) X
Impulse Breakdown (D 33(X)) 145 kV, min at 25°C X
CHEMICAL TESTS
Water (D 1533) 35 ppm, max X X
Acidity (D 974) 0.03 mg KOH/g, max X X
Corrosive Sulfur (D 1275) Noncorrosive X
DBPC Content (D 2668) 0.3% by mass, max X
DBPC Content (D 1473) 0.3% by mass, max X
Oxidation Stability (D 2112) 195 minutes, min X X
Oxidation Stability (D 2440) [72 h; 0.1% sludge, max ]
[o.3 mg KOH/g acid, max]
PHYSICAL TESTS
IFT (D 971) 40 dynes/cm at 25°C, min X
IFT (D 2285) 40 dynes/cm at 25°C, min X
Color (D 1500) 0.5, max X X
Visual Exam (D 1524) Bright and clear X X
Pour Point (D 97) - 4 0 ° C , max X
Flash/Fire Point (D 92) 145°C,min X
Aniline Point (D 611) 63-84°C, max X
Viscosity (D 88) p . O cSt at 100°C, max^
12.0 cSt at 40'='C, max X
Viscosity (D 445) 176.0 cSt at 0°C, max j
Specific Gravity (D 1298) 0.91 at 15°C, max X
DESIGN TESTS
Coefficient of Thermal Expansion (D 1903)
Thermal Conductivity (D 2717)
Specific Heat (D 2766)
ADDITIONAL TESTS
Analysis of Gases Dissolved in Insulating Oil by Gas Chromatography (D 3612)
Static Charging Tendency
Analysis of PCBs in Insulating Liquids by Gas Chromatography (D 4059)
Dielectric
Copyright by ASTMStrength of reserved);
Int'l (all rights Oil in ThuMotion
May 6 14:02:13 EDT 2021
Specific Resistance
Downloaded/printed by (Resistivity) of Electrical Insulating Liquids (D 1169)
40 INSULATING OILS
Electrical Tests
ASTM Test Method for A-C Loss Characteristics and Relative Permittivity
(Dielectric Constant) of Electrical Insulating Liquids (D 924)
Limit: 0.05% max at 25°C, 0.3% max at 100°C
Commonly called the power factor test, this is a valuable screening test. If new oil has a power
factor greater than 0.05%, it may include water, oxidation products, or other polar contami-
nants or combinations of these. Oil testing above 0.05% is not acceptable and should be looked
at very closely. The 25°C test is always performed for acceptance, whereas the 100°C test is
made for qualification of new oils and not normally repeated unless problems are suspected.
ASTM Test Method for Dielectric Breakdown Voltage of Insulating Liquids Using
Disk Electrodes (D 877)
Limit: 30-kV, minimum
This test determines if an oil has adequate insulating strength and is used as a vendor qualifi-
cation test. It is also routinely run as an acceptance test on samples of new oil as received. Low
breakdown voltages usually mean that free moisture and conducting particles are present in the
oil. Filtration will normally remove contaminants that cause low dielectric strength. This test
will not detect any moisture in solution, which could amount to as much as 45 to 50 ppm for new
oil at 25°C.
ASTM Test Met hod for Dielectric Breakdown Voltages of Insulating Oils of Petroleum
Origin Using VDE Electrodes (D 1816)
Limit: 28-kV, minimum at 0.040-in. gap
56-kV, minimum at 0.080-in. gap
This sensitive dielectric strength test is used as both an acceptance and as a qualification test,
but since the method is highly sensitive to moisture, new oil has to be first treated to dehydrate it
and to remove gases. The VDE dielectric test is also routinely used to test oil as received. Failure
to meet the minimum strength limit due to moisture or gas content will not cause rejection
because the oil will be dried and degassed before it is used in energized equipment. For new oils
before processing, typical results range from 40 to 60 kV with a 0.080-in. gap. (We use the
0.080-in. gap exclusively.)
ASTM Test Method for Gassing of Insulating Oils Under Electrical Stress and Ionization
(Modified Pirelli Method) (D 2300)
Limit: + 1 5 /xL/min, max (Procedure A)
+ 3 0 /iL/min, max (Procedure B)
New equipment designs that reduce materials and costs may result in greater electrical stress
on the liquid insulation, causing gases to be evolved or absorbed. Oils with slightly negative
gassing tendencies are preferred. However, price and availability may override decisions to
avoid oils with high gassing tendencies. Gassing tendency is an important qualification test, but
we do not perform it. When needed, the services of an external laboratory would be obtained.
TestbyMethod
Copyright ASTM Int'l for Dielectric
(all rights Breakdown
reserved); Thu Voltage
May 6 14:02:13 EDT 2021 of Insulating Oils of Petroleum Origin
Under Impulse
Downloaded/printed by Conditions (D 3300)
Limit: 145-kV, minimum, at 25°C.
This is a crucial qualification test. Impulse strength tests demonstrate the ability of an insu-
lating oil to withstand the voltage stress caused by transients from lightning or switching. The
test can be done in the BPA Laboratory if needed. Our primary concern is with impulse strength
of total transformer systems, however. The ability to withstand transient voltages is very depen-
dent on geometry, spacing, and polarity.
Chemical Teste
ASTM Test Method for Water in Insulating Liquids (Karl Fischer Method) (D 1533)
Limit: 35 ppm, max
This test is used for both vendor qualification and routine acceptance. If sample handling has
been sloppy, high readings not indicative of the true moisture level in the oil can result. To avoid
misleading results, sampling procedures require: (1) use of clean, dry nonpermeable-to-air con-
tainers; (2) precautions to preclude ingress of moisture (which in the rainy Pacific Northwest
can be challenging); and (3) speed in shipment of samples to the laboratory. Temperature/
pressure cycling can draw water into the sample. Low moisture content in mineral insulating oil
is needed to minimize metal corrosion, maximize life of cellulosic insulation, and achieve ade-
quate electrical strength and low dielectric loss characteristics. Water in oil levels greater than
10 ppm are considered to be high and cause for concern. Levels below 10 ppm can be achieved
and maintained without extreme efforts.
ASTM Test Method for Neutralization Number by Color-Indicator Titration (D 974)

Limit: 0.03 mg KOH/g max
This is one of the tests we rely on to determine when to reclaim service-aged oil. We also apply
it for both vendor qualification and delivery acceptance purposes. It is important to us to always
start with a low acid content oil (less than 0.01 mg KOH/g) to help attain full amortized life
expectancy from installed equipment and to protect the investment in capitalized facilities. In
our opinion, the acceptable limit should be reduced from 0.03 mg KOH/g max to 0.025 mg
max KOH/g for new oil.
ASTM Test Method for Corrosive Sulfur in Electrical Insulating Oils (D 1275)
Limit: Noncorrosive
This is still a necessary test method even though corrosive sulfur is not a problem with most
modern oils. The apparatus for the test is in our laboratory and the test is run for qualification
purposes. Free sulfur or sulfur compounds in oil can result in corrosion of copper in transform-
ers and may help form acids.
ASTM Test Method for 2,6-Ditertiary-Butyl Para-Cresol (DBPC) and 2,6-Ditertiary-Butyl

Phenol (DBP) in Electrical Insulating Oil by Infrared Absorption (D 2668)
Limit: 0.3% by mass, max
This is a good semiquantitative test used for vendor qualification. Inhibitor content in oil is
important to slowing down oxidation aging rates and maintaining long service life of transform-
ers.Universidade
(UFRN) Our laboratory
Federal do Riobelieves the ((UFRN)
Grande do Norte alternative gasFederal
Universidade chromatograph method
do Rio Grande do Norte) now
pursuant being
to License developed
Agreement. byreproductions auth
No further
ASTM for measuring inhibitor content will provide more reliable quantitation.
42 INSULATING OILS
ASTM Test Method for 2,6 Ditertiary-Butyl Para-Cresol (DBPC) in Electrical Insulating Oils
(Intent to Withdraw) (D 1473)
Limit: 0.3% by mass, max
Method D 1473 is our preferred method for measurement of inhibitor content because the
infrared absorption method (D 2668) is sensitive to mixtures of 2,6 ditertiary-butyl phenol and
2,6 ditertiary-butyl para-cresol.
ASTM Test Method for Oxidation Stability of Inhibited Mineral Insulating Oil by
Rotating Bomb (D 2112)
Limit: 195 minutes, minimum
New oils inhibited with DBPC are tested according to D 2112 for qualification. Oils with
oxidation stability are essential to long service life of equipment. Oils that do not accept inhibi-
tors are less desirable. The test also gives a clue about the consistency of production processes at
various refineries, which makes it useful for acceptance purposes as well.
ASTM Test Method for Oxidation Stability of Mineral Insulating Oil (D 2440)
Limit: 72 h; 0.1% sludge max, 0.3 mg KOH/g acid max
This oxidation stability test is a useful qualification test; however, D 2440 is not currently in
use at BPA. The method assesses the resistance of an oil to oxidation under accelerated aging
conditions. The Doble Power Factor Valued Oxidation (PFVO) test gives information adequate
to assess oxidation stability. The PFVO test characterizes the power factor of an oil specimen
under accelerated aging conditions, measuring oil quality up to the point sludging starts and
revealing the sludge-free life of the specimen.
Physical Tests
ASTM Test Method for Interfacial Tension of Oil Against Water by the Ring Method (D 971)
Limit: 40 dynes/cm at 25°C, minimum
This test, which uses a platinum ring, is used by our laboratory for qualification of vendors.
An alternate method [ASTM Test Method for Interfacial Tension (IFT) of Electrical Insulating
Oils of Petroleum Origin Against Water by the Drop Weight Method (D 2285)], which is more
practical for field application as a diagnostic aid, is used by the laboratory for acceptance pur-
poses. The platinum ring method is, of course, used to officially reject any oil batches. IFT
values less than 40 dynes/cm indicate that unacceptable polar contamination is present, mak-
ing this test a useful screening device for new oil exposed during transport to soaps, acids,
paints, varnishes, or solvents. Oils with low IFT initially may result*in accelerated aging of oil
and solid insulation in transformers. The limit of acceptability could probably be raised to 45
dynes/cm.
ASTM Test Method for ASTM Color of Petroleum Products (ASTM Color Scale) (D 1500)
Limit: 0.5 max
Because of this test, used for both acceptance and qualification, some suppliers are elimi-
nated.
(UFRN) Slightly
Universidade Federal colored
do Rio Grandenew oil ((UFRN)
do Norte does not indicate
Universidade poor
Federal do Rioproduct quality.
Grande do Norte) pursuantThis testAgreement.
to License is useful, how-
ever, in detecting deterioration during service (generally over long time periods). New oils that
are bright and clear are desired. Clear oil (low color numbers) allows visual inspection of inter-
nal equipment components and is some assurance that the oil is probably in serviceable condi-
tion.
ASTM Method for Visual Examination of Used Electrical Insulating Oils of Petroleum
Origin in the Field (D 1524)
Limit: Bright and clear
Section 7.2, the Tyndall Beam examination, is very useful for identifying visible contami-
nants present in new oil. A narrow beam of light projected upward through a sample will reflect
from particles of metal, carbon, paper, or other material. Cloudiness indicates that water or
sludge are present. Valuable information may be obtained by filtering and identifying the resi-
due. If oil is anything other than sparkling clear, examine the sample for sediment and look at
the dielectric strength, neutralization number, and IFTtest data. Oxidation products can cause
cloudiness, but will also cause a high neutralization number and low interfacial tension.
ASTM Test Methods for Pour Point of Petroleum Oils (D 97)

Limit: - 4 0 ° C , max
Many sites within the Northwest United States can experience wintertime ambients below
—40°C. The extremely cold sites are located in the Montana Rockies, Southern Idaho, and the
Grand Tetons. To my knowledge, the influence of extreme cold on today's naphthenic-based
insulating oil has not yet caused an equipment failure. Our practice is to run the pour point test
for qualification of vendors, but not for routine acceptance purposes. We are not aware of the
pour point of an oil being affected by normal aging in a transformer or other electrical equip-
ment unless the oil becomes contaminated with compounds containing n-paraffins or waxy-like
substances.
ASTM Test Method for Flash and Fire Points by Cleveland Open Cup (D 92)
Limit: 145°C, minimum, flash (fire point limit not given in D 3487)
Because of the possibility of a fire related to failure of electrical equipment, it is important to

have knowledge of the flash and fire points of oil. We use the method as a qualification test, but
not for acceptance purposes. The flash point test provides adequate information on the flamma-
ble properties of oil, and the fire point, a higher temperature, is not usually required. Note 1 in
D 92 refers to D 93 [ASTM Test Methods for Flash Point by Pensky-Martens Closed Tester]. By
virtue of Note 1, we consider D 93 as binding on suppliers as part of D 92. In other words, we
expect new oil to be free of flammable gases (such as methane) that may escape detection by the
open cup method.
ASTM Test Method for Aniline Point and Mixed Aniline Point of Petroleum Products and
Hydrocarbon Solvents (D 611)
Limit: 63-84°C, max
This is a qualification test. It is not performed at BPA, however. A contract laboratory would
do it for us
Downloaded/printed by when needed. Typically, reports from the supplier and information on the type of
crude
(UFRN) are accepted.
Universidade The do
Federal do Rio Grande aniline point
Norte ((UFRN) estimates
Universidade Federalthe total
do Rio Grandearomatic content
do Norte) pursuant and
to License indicates
Agreement. the
44 INSULATING OILS
solvency of the oil for materials that are in contact with the oil. It may relate to impulse and
gassing characteristics of oil.
ASTM Test Method for Saybolt Viscosity (D 88)

Limit: 3.0 cSt at 100°C, 12.0 cSt at 40°C, 76.0 cSt at 0°C, max
BPA does not perform test D 88 (see D 445 below). Viscosity of oil influences heat transfer
rates and consequently temperature rise of apparatus. Viscosity of oil also influences the speed
of moving parts in circuit breakers or tap changers, for example. Oil viscosity controls oil treat-
ment rates as well. High viscosity oils are less desirable, especially in cold climates.
ASTM Test Method for Kinematic Viscosity of Transparent and Opaque Liquids
(and the Calculation of Dynamic Viscosity) (D 445)
Limit: 3.0 cSt at 100°C, 12.0 cSt at 40°C, 76.0 at 0°C, max
Viscosity measurements are made for qualification purposes. D 445, measuring flow through
an orifice, is used as a referee method.
ASTM Test Method for Density, Relative Density (Specific Gravity), or API Gravity of Crude
Petroleum and Liquid Petroleum Products by Hydrometer Method (D 1298)
Limit: 0.91 max, at 15°C
Specific gravity measurements are made for acceptance. Specific gravity is important where
there is a concern about water in oil freezing and rising to float on top of the oil. A lower limit for
specific gravity would eliminate this potential problem. The BPA Laboratory uses a PAAR digi-
tal density meter, based on a different measuring principle for routine testing. Normally, den-
sity is checked at room temperature. The PAAR meter results compare favorably with D 1298
results.
Design Tests
ASTM Test Method for Coefficient of Thermal Expansion of Electrical Insulating

Liquids of Petroleum Origin, and Askarels (D 1903)
This is considered as a design test with little importance to transformer and circuit breaker
applications. The test is used for oil-filled cable installations at BPA.
ASTM Test Method for Thermal Conductivity of Liquids (D 2717)

D 2717 test results are useful in engineering calculations relating to the manner in which a
given system reacts to thermal stress. The thermal conductivity test measures temperature
change across a liquid sample for a known amount of energy input. The test is not practical for
acceptance or qualification; it is a design test.
ASTM Test Method for Specific Heat of Liquids and Solids (D 2766)
This standard method is also basically a design test, measuring capacity of oil to absorb ther-
Copyright
malbyenergy
ASTM Int'l(amount
(all rights reserved); Thu May input
of energy 6 14:02:13to
EDT 2021 a temperature change). The D 2766 test is not
cause
essential
(UFRN) forFederal
Universidade qualification
do Rio Grande door acceptance
Norte purposes.
((UFRN) Universidade Federal do Rio Grande do Norte) pursuant to License Agreement. No further reproductions auth
Additional Tests Not in ASTM D 3487
ASTM Method for Analysis of Gases Dissolved in Electrical Insulating Oil by

Gas Chromatography (D 3612)
Refining method by-products can include hydrogen and methane. To find out what is being
purchased, a dissolved gas analysis test is an appropriate acceptance test. Dissolved gas is re-
moved from oil during normal treatment (drying, dehydrating, and degassing) before being put
in transformers. But, the test can be very revealing of unique manufacturing circumstances. On
large transformers received without oil, it is of interest to make sure any residual oil left after
factory tests is free from combustibles.
Static Charging Tendency

This utility is actively following research and development in this area and is interested in
seeing a standard method for evaluating the ability of liquid insulations to generate and trans-
port entrained charge.
ASTM Method for Analysis of Polychlorinated Biphenyls in Insulating Liquids by

Gas Chromatography (D 4059)
Limit: 2 ppm, max (lower limit of detection)
Movement to add D 4059 to D 3487 is concurred with. It is important to underscore the

legislated requirement restricting commerce of PCB-containing electrical equipment. BPA's
current specification requires oil with 2 ppm or less PCB as tested by D 4059 anyway.
Dielectric Strength of Oil in Motion

No standard test method is available for evaluation of the dielectric strength of insulating
liquid moving in restricted channels using dissimilar wall materials. Many transformers have to
force oil flow with pumps to achieve cooling. Static electrification research reveals that the di-
electric strength of oil flowing in regions of electrical stress is a function of flow rate, field
strength, temperature, moisture content, and particulate contamination.
ASTM Test Method for Specific Resistance (Resistivity) of Electrical Insulating Liquids
(D 1169)
It may be appropriate to add D 1169 to D 3487. The resistivity of oil depends on many factors,
including temperature, moisture, and contaminating particles. Some European transformer
manufacturers have specified oil requirements to include this or a similar test. For many years,
resistivity has been used by cable and capacitor manufacturers as an insulating fluid qualifica-
tion and acceptance test. Comparing different oils to be used in forced cooled transformers for
this characteristic would be beneficial. Test Method D 1169 has special relevance for oil to be
used in HVDC equipment.
Bibliography
BPA Invitation for Bids No. DE-FB79-86BP60194, "Insulating Oil," Bonneville Power Administration,
Vancouver, WA.
BPAbySubstation
Copyright ASTM Int'l (allMaintenance Standard
rights reserved); Thu No.EDT
May 6 14:02:13 6476-000-2,
2021 "Insulating Oil," Bonneville Power Administra-
tion, Vancouver,
Downloaded/printed by WA.
Doble
(UFRN) Transformer
Universidade Federal doOil Purchase
Rio Grande Specification
do Norte (TOPS),
((UFRN) Universidade Federalrevised, Doble
do Rio Grande Engineering
do Norte) Co., Agreement.
pursuant to License Watertown, MA.
46 INSULATING OILS
1986 Annual Book of ASTM Standards. Vol. 10.03, ASTM, Philadelphia, PA.
Reference Book on Insulating Liquids & Gases, Section III: Insulating Oil Test Guide. Doble Engineering
Co., Watertown, MA.
DISCUSSION
C. L. S. Vieira^ (written discussion)—I understand from your presentation that you use
ASTM D 3847 (all tests) for the qualification of a new supplier of transformer oil and only
repeat some of the tests as a criteria for acceptance. My question is divided into two parts:
1. Are the tests in ASTM D 3487 enough to evaluate a new oil (for example, a European
paraffinic oil)?
2. If no, what additional tests do you recommend to evaluate a new product?
D. L. Johnson (author's closure)—Evaluation of a new oil would include all of the D 3487
tests plus any other standard tests considered important by the oil user. For example, as noted
in the paper, Standard Test Methods D 1169 and D 3612 could provide results useful in com-
paring new products to familiar products. The application for the new oil should be considered.
ASTM Test Method for Compatibility of Construction Material with Electrical Insulating Oil of
Petroleum Origin (D 3455) is suggested to verify that the oil and other material in contact do not
change the electrical, physical, or chemical properties of each other.
Henry A. Pearce^
Significance of Transformer Oil

Properties
REFERENCE: Pearce, H. A., "Significance of Transformer Oil Properties," Electrical Insulat-

ing Oils, STP 998. H. G. Erdman, Ed., American Society for Testing and Materials, Philadel-
phia, 1988, pp. 47-54.
ABSTRACT: Transformer oil is a refined mineral oil obtained from the fractional distillation of
crude petroleum. It is then treated to remove impurities and to obtain the most desirable proper-
ties to make it suitable as an insulating and cooling liquid.
The most important property for an insulating liquid is the high dielectric strength and for a
cooling liquid the low viscosity. Other properties are important to show the absence of impurities
and to show the degree of stability for long life. Properties indicating purity are also important in
the consideration of the ability of the oil to be compatible with the other materials used in trans-
former construction.
The primary enemies of the transformer oil are oxidation, contamination, and excessive tem-
perature. Processing and transformer construction are designed to minimize the effect of these
enemies, and tests are conducted to evaluate these considerations.
ASTM Specification for Mineral Insulating Oil Used in Electrical Apparatus (D 3487) has been
designed to cover the properties of transformer oil for use as an insulating and cooling medium.
Quoting its scope: "This specification is intended to define a mineral insulating oil that is func-
tionally interchangeable and miscible with existing oils, and is compatible with existing apparatus
and with appropriate field maintenance, and will satisfactorily maintain its functional characteris-
tics in its application in electrical equipment."
Much effort is expended by manufacturers to obtain and maintain quality oil for use in their
equipment.
KEY WORDS; transformer oil, insulating oil, cooling liquid, dielectric liquid
Transformer oil is truly a unique material for doing the job for which it is famous, that is,
insulating and cooling transformers. Our friend George Westinghouse first built a transformer
100 years ago using oil, copper, cardboard, and steel. And after 100 years of outstanding mate-
rials developments we are today building most of our transformers using oil, copper, cardboard,
and steel. So, while there have been many changes in crudes, processing, additives, and extrac-
tions, the basic insulating oil has a very long history of reliable performance. Actually we are
talking about insulating oil rather than transformer oil because it is used in other electrical
apparatus such as circuit breakers, capacitors, bushings, etc., as well as transformers.
From the viewpoint of the electrical equipment manufacturer, insulating oil is a refined min-
eral oil obtained from the fractional distillation of crude petroleum. It is free from moisture,
inorganic acid, alkali, sulfur, asphalt, tar, vegetable oils, or animal oils. It offers the following
advantages:
1. High dielectric strength.
2. Low viscosity.
3. Freedom from inorganic acids, alkali, and corrosive sulfur.
4. Good resistance to emulsification.
5. Freedom from sludging under normal operation.
6. Rapid settling of arc products.
Copyright'Manager,
by ASTM Int'l Insulating Materials,
(all rights reserved); Thu May 6Westinghouse
14:02:13 EDT 2021Electric Corp., Sharon, PA 16146.
47
48 INSULATING OILS
7. Low pour point.

8. High flash point.
Insulating oil has few enemies, and the manufacturers and users make a concerted effort to
protect it from its enemies.
1. Oxidation. Oxidation is the most common cause of oil deterioration. The transformer
manufacturer puts forth a significant effort to make sure that the tank or case is well sealed
from the atmosphere. Careful drying and vacuum processing is conducted to remove air and
moisture prior to sealing of the tank after which it is filled with dry air or nitrogen to minimize
the exposure to oxygen.
2. Contamination. Moisture is the chief among potential oil contaminants. Its presence can
provide a source of reactive products with the oil in the presence of heat. It also tends to lower
the dielectric properties of the insulating oil. The same precautions that are exercised to protect
the oil from oxidation are used to protect it from moisture.
3. Excessive temperature. Excessive heat is an enemy of oil. It will cause decomposition of
the oil itself and/or it will increase the rate of oil oxidation. The best way to protect against
excessive heat is to avoid overload of the transformer.
4. Corona discharges. Sparking and local overheating can also break down the oil molecule
producing gases and water, which can lead to the formation of acids and sludge.
The properties of insulating oil are very important to the manufacturer of electrical appa-
ratus. Therefore, much time and effort are spent testing the oil and following its properties at
the various stages in the design, manufacture, and life of a transformer or other piece of electri-
cal equipment.
Each manufacturer has their own specification for the desired properties of insulating oil.
Historically, there were several properties that all manufacturers agreed on and there were
some, such as oxidation stability, that involved the use of widely divergent test procedures. In
recent years the manufacturers of the electrical apparatus, the producers of insulating oil, and
the users of the apparatus have worked together to create one specification in which everyone
could agree. This is ASTM Specification for Mineral Insulating Oil Used in Electrical Appa-
ratus (3487). Almost all manufacturers and users have now patterned their requirements after
D 3487.
The functional property requirements of mineral insulating oil as listed in D 3487, as well as
some typical limiting values, are reproduced in Table 1.
All of the listed properties are important to the equipment manufacturer. Some have higher
levels of importance than others. Also, the properties have significance for different areas of
consideration. Some properties are used for design calculations, some are used to indicate the
uniformity of the oil from the producers, and some are used to reveal normal or abnormal oper-
ation of the apparatus in service.
Significance of Test
Dielectric Breakdown ASTM D 877 AND ASTM D 1816, 60 Cycle

The dielectric breakdown voltage is a very important measurement of the electrical stress
which an insulating oil can withstand without failure. This is an important design parameter
and is used by the design engineer to calculate permissible clearances. Therefore, the manufac-
turers must observe that the value for the oil being put into the apparatus is maintained at the
desired level.
This property is measured by applying a voltage between two electrodes under prescribed
conditions
Copyright under
by ASTM Int'l
the
(all rights liquid.
reserved); There
Thu May are EDT
6 14:02:13 two2021
ASTM procedures: One is ASTM Test Method for
Dielectric
(UFRN) UniversidadeBreakdown Voltage
Federal do Rio Grande do Norteof Insulating
((UFRN) Liquids
Universidade Federal doUsing Disk
Rio Grande Electrodes
do Norte) (D 877),
pursuant to License which
Agreement. spec-
PEARCE ON TRANSFORMER OIL PROPERTIES 49
TABLE l—ASTM D 3487.
The following table lists the physical properties for insulating oil and the appropriate ASTM test numbers
for each test procedure.
Specification Typical
Property Value Value ASTM No.
Color 0.5 max 0.5 D 1500

Dielectric breakdown at 60 Hz
0.100" gap 30 kV min 35 kV D877
0.040" gap 28 kV min 30 kV D1816
0.080' gap 56 kV min 60 kV D1816
Neutralization number 0.03 mg max 0.01 mg D974
Free or corrosive sulfur Noncorrosive Noncorrosive D1275
Flash point 145°C (293°F) min 150° D92
Pour point -40°C(-40°F)max -55°C D97
Viscosity
max. cST at 100°C 3 s max 3s D88
max. cST at 40°C 12 s max 10 s D88
max. cST at 0°C 76 s max 70s
Moisture content 35 ppm max 20 ppm D1533
Specific gravity at 60°F 0.910 max 0.890 D1298
Inorganic chlorides or sulfates None None D878
Interfacial tension 40 dynes min 45 dynes D971
Power Factor, 60 Hz
25°C (77°F) 0.05% max 0.01% D924
100°C (212°F) 0.30% max 0.10% D924
Oxidation stability
Sludge after 72 h 0.15% 0.1% D2440
Sludge after 164 h 0.30% 0.2% D2440
Total acid after 72 h 0.5 mg of KOH 0.2 mg of KOH D2440
Total acid after 164 h 0.6 mg of KOH 0.3 mg of KOH D2440
Aniline point 63 to 83°C 75°C D611
Dielectric breakdown, impulse,
needle-negative 145 kV 150 kV D3300
ifies a test cup equipped with 1-in.-diameter vertical disk electrodes spaced 0.100 in. apart and
is reported as the standard test value. It has the longest history of the two procedures and is used
for acceptance of oil from the producer. The second procedure is ASTM Test Method for Di-
electric Breakdown Voltage of Insulating Oils of Petroleum Origin Using VDE Electrodes (D
1816), which specifies a test cup equipped with spherical electrodes spaced either 0.040 or 0.080
in. apart. This cup includes a stirrer and thus is sensitive to small amounts of contaminants. It
is designed for evaluation of new degassed and dehydrated oil and therefore is a good measuring
stick of the processing of the oil and the apparatus in the manufacturing location. Figure 1 is a
Baur Dielectric Tester setup for D 877 test. Figure 2 is the same tester setup for D 1816.
Dielectric Breakdown, Impulse, ASTM D 3300

This test, ASTM Test Method for Dielectric Breakdown Voltage of Insulating Oils of Petro-
leum Orign Under Impulse Conditions (D 3300), covers the determination of dielectric break-
down voltage of insulating oil under impulse conditions. The electrodes consist of a brass or
steel sphere and a steel point. Electrode configuration may be specified sphere to sphere and
point
(UFRN) to sphere.
Universidade Federal doThis information
Rio Grande is important
do Norte ((UFRN) to the
Universidade Federal designers
do Rio of equipment
Grande do Norte) asAgreement.
pursuant to License it indicates the
50 INSULATING OILS
FIG. 1—Baur dielectric tester setup for D 877 test.
ability of the oil to withstand transient voltage stresses from such causes as nearby lightning
strokes or high-voltage switching operations.
Dissipation Factor (or Power Factor), ASTM D 924

The power factor is the ratio of the power dissipated in the oil in watts to the product of the
effective by
Downloaded/printed voltage and current in volt-amperes, when tested with a sinusoidal field under pre-
FIG. 2—Baur dielectric tester setup for D 1816 test.
scribed conditions. This is a good measuring stick for oil quality. A high power factor value is an
indication of the presence of contaminants or deterioration products. (See ASTM Test Method
for A-C Loss Characteristics and Relative Permittivity (Dielectric Constant) of Electrical Insu-
lating Liquids (D 924).
Moisture Content, D 1533

The presence of free water may be observed by visual examination in the form of separated
droplets or as a cloud dispersed throughout the oil. Water in solution is normally determined by
chemical means and is measured in parts per million. Water is important because increased
quantity normally decreases the dielectric strength of the oil. The moisture content is important
to the manufacturer because it is an indication of the quality of the processing and drying that
has been performed on the transformer. It is important in the field operation because increasing
values can indicate a breathing tank or the deterioration of cellulose insulation. [See ASTM
Test Method for Water in Insulating Liquids (Karl Fischer Method) (D 1533).]
Neutralization Number, ASTM D 974

The neutralization number is the number of milligrams of potassium hydroxide required to
neutralize
Copyright the(allacid
by ASTM Int'l rights in 1 g of
reserved); Thuoil.
May It measures
6 14:02:13 the acid content of the oil. This is most important in
EDT 2021
Downloaded/printed
indicatingbychemical change or deterioration of the oil in a unit in service. This normally indi-
52 INSULATING OILS
cates abnormal operation. Increased acidity may indicate the need to reclaim or replace the oil.
[See ASTM Test Method for Neutralization Number by Color Indicator Titration (D 974).]
Interfacial Tension, ASTM D 971

The interfacial tension between the insulating oil and water is a measure of the molecular
attractive force between the unlike molecules and is expressed in d3mes per centimeter. The test
provides a means of detecting soluble contaminants and products of deterioration. It is used by
the equipment manufacturer as an indication of the level of contamination of new oil or oil after
it has been put into a unit. Figure 3 shows an interfacial tension test being done. [See Test
Method for Interfacial Tension of Oil Against Water by the Ring Method (D 971).]
Color. ASTM D 1500

The primary significance of color is to observe darkening in short periods of time, which may
indicate either contamination or deterioration of the oil. The color is determined by means of
transmitted light and is expressed by a numerical value based on comparison with a series of
color standards. [See ASTM Test Method for ASTM Color of Petroleum Products (ASTM
Color Scale).]
Downloaded/printed by FIG. 3—Interfacial tension test being run.
Viscosity, ASTM D 88
Viscosity of insulating oil is one of the significant parameters in determining the heat transfer
capability of the fluid. This information, of course, is very important to the equipment designer.
The viscosity is the resistance of oil and its resistance to uniformly continuous flow without
turbulence inertia and other forces. It is usually measured by timing the flow of a given quantity
of oil under controlled conditions. A viscosity increase in an operating unit could indicate dete-
rioration of the oil and formation of sludge. [See ASTM Method for Saybolt Viscosity (D 88).]
FIG. 4—HIACparticle analyzer.
54 INSULATING OILS
Other Specification Requirements

Other specification requirements are important to characterize a particular oil, but because
they are not likely to vary significantly they are not run on a frequent basis by the manufacturer.
These include: pour point, ASTM Test Methods for Pour Point of Petroleum Oils (D 97), which
is the temperature at which insulating oil will just flow under prescribed conditions. This is
important in cold climates to determine start-up conditions. The flash and fire points, ASTM
Test Method for Flash and Fire Points by Cleveland Open Cup (D 92), determines the potential
fire hazard involved. Specific gravity, ASTM Test Method for Density [Relative Density (Spe-
cific Gravity), or API Gravity of Crude Petroleum and Liquid Petroleum Products by Hydrome-
ter Method (D 1298)] is another consideration for heat transfer calculations and also what ma-
terials will sink or float in the liquid. Inorganic chlorides or sulfates, ASTM Test Method for
Inorganic Chlorides and Sulfates in Insulating Oils (D 878), indicate the degree of treatment of
the oil. The oxidation stability, ASTM Test Method for Oxidation Stability of Mineral Insulat-
ing Oil (D 2440) or ASTM Test Method for Oxidation Stability of Inhibited Mineral Insulating
Oil by Rotating Bomb (D 2112), is an indication of the ability of the natural inhibitors or the
added inhibitors to resist the oxidation of the oil.
ASTM standard D 3487 has been designed to cover the properties of transformer oil for use as
an insulating and cooling medium. It is presently well structured for this purpose. Quoting its
scope: "This specification is intended to define a mineral insulating oil that is functionally inter-
changeable and miscible with existing oils, and is compatible with existing apparatus and with
appropriate field maintenance, and will satisfactorily maintain its functional characteristics in
its application in electrical equipment."
A somewhat recent observation that is being used to evaluate oils in units in service is the
particle count and the related atomic absorption. The particle count provides the information
or the size and number of particles in the oil, and the atomic absorption is a metal-in-oil analy-
sis that tells us what metals are present and their concentrations. Conductive particles are
harmful to the dielectric strength of the oil. The generation of metallic particles is usually re-
lated to wear and fault problems. Pump bearing wear or arcing and other damage can add
metallic particles. This evaluation is increasing in importance in determining what is taking
place inside of a transformer. Figure 4 shows a HIAC particle analyzer.
We have not mentioned the determination of PCB content of insulating oil. While this has
nothing to do with the functional operations, it is very important to all of us in determining how
we may handle, store, transport, or dispose of the oil.
The quality of insulating oil is most important in the operation of electrical apparatus. There-
fore, much time and effort is put forth by the producer, the manufacturer, and the user to
assure that the desired properties are maintained.
DISCUSSION
C. L. S. Vieira^ (written discussion)—During the presentation of the paper, the test of gas-
sing tendency was not mentioned as one important test to control the oil in the manufacturer
point of view.
I would like discussed the real need of this property and the changes, if any, that the manu-
facturers are taking into account to design the new EHV and UHV transformers due to the
introduction of this new characteristic.
'CEPEL, P.O. Box 2754, Rio de Janeiro, Brasil.
DISCUSSION ON TRANSFORMER OIL PROPERTIES 55
Normally, today, the users are looking for gas absorbent oils (negative in ASTM D 2300) but
as we saw in the paper by D. L. lohnson, the Bonneville Power Administration accepts oil with
values of +15 /iL/min (Proc. A) or +30 /iL/min (Proc. B) for all transformers. What is really
the actual practice of the manufacturers and/or the utilities concerning this property?
H. Pearce (author's closure)—Vfe as a manufacturer do not test the gassing tendency [ASTM
Test Method for Gassing of Insulating Oils Under Electrical Stress and Ionization (Modified
Pirelli Method) (D 2300)] on a regular basis. There are many differences of opinion as to what is
the most desirable range of values. We have tested each of our suppliers' oils but we do not have
this requirement as part of our purchasing specification.
Section III—Analysis of Oil
Jean-Pierre Crine^
Newly Developed Analytical Techniques

for Characterization of Insulating Oils
REFERENCE: Crine, J.-P., "Newly Developed Analytical Techniques for Characterization of In-
sulating Oils," Electrical Insulating Oils, STP 998, H. G. Erdman, Ed., American Society for
Testing and Materials, Philadelphia, 1988, pp. 59-80.
ABSTRACT: A review of newly developed or improved analytical techniques for characterization

of insulating oils is made. The emphasis is put on recent techniques, procedures, and applications
developed at IREQ in five major areas of oil characterization: thermal aging, electrical properties,
particles and metals in oil, gases in oil, and low-temperature applications. It is shown that among
other things, the antioxidant content, as measured by HPLC, can be a simple and reliable way to
estimate oil oxidation. The influence of particles and metals in oil on a-c breakdown strength are
also stressed, and an ultrasonic agitation procedure is proposed to rehomogenize oil samples be-
fore electric breakdown measurements. Finally, new trends and future developments, including
static electrification and scale model studies, are pointed out.
KEY WORDS: oil characterization, analytical techniques, thermal aging, inhibitor, electric
strength, particles, metal in oil, homogenization, gases in oil, paraffins
Electrical engineers dealing regularly with dielectric liquids (mainly oils) need various analyt-
ical techniques and methods to evaluate the overall physical, chemical, and electrical properties
of these fluids. Over the years a wide variety of tests [see ASTM Specification for Mineral Insu-
lating Oil Used in Electrical Apparatus (D 3487)] have been devised, especially to determine the
basic functional property requirements of the liquid (see Table 1). With the development of new
dielectric liquids (silicone oil, synthetic fluids, etc.), several of the tests listed in ASTM D 3487
have become obsolete or inadequate. The properties and operation conditions of these new flu-
ids are often very different from those of the conventional mineral oils. This has lead to the
development of new or improved analytical techniques for the characterization of insulating
liquids.
Dielectric oils are prone to thermal degradation and gas generation. The thermal resistance
of any material is generally determined from aging tests relying on accelerated aging conditions
[as in ASTM Test Method for Oxidation Stability of Mineral Insulating Oil (D 2440)]. Al-
though these accelerated aging conditions are not representative of the real aging conditions in
service, they nevertheless allow comparisons to be made between different materials. There is a
need for better (that is more realistic) aging tests, but before these tests be implemented it is
necessary to select the analytical techniques that best characterize aging.
Dielectric oils being mostly subproducts of the petroleum industry, their characterization is
made more from the chemical than from the electrical point of view (see for example the proper-
ties listed in Table 1). With the use of new liquids and the reduction in physical size of modern
high-voltage apparatus, the electrical industry has become aware of the necessity to better de-
fine its selection criteria for insulating liquids. Their specific behavior under high voltage has
also received more attention, and specifications are now more stringent for parameters that do
not appear in Table 1, such as gassing, partial discharges, particle concentration, etc. One
Copyright 'Senior research

by ASTM Int'l (all rightsscientist, Institut
reserved); Thu de recherche
May 6 14:02:13 EDT 2021 d'Hydro-Quebec, Varennes, Quebec, Canada JOL 2PO.
59
Copyright® 1988 b y A S T M International www.astm.org
60 INSULATING OILS
TABLE 1—Functional property requirements for oils

(ASTM D 3487).
Property ASTM Test Method
CHEMICAL
Oxidation stability D 2440 or D 2112
Oxidation inhibitor content D 1473 or D 2668
Corrosive sulfur D1275
Water content D 1315 or D 1533
Neutralization number (acidity) D974
PHYSICAL
Aniline point D611
Color D1500
Interfacial tension D971
Pour point D92
Specific gravity D1298
Viscosity D 445 or D 88
Visual examination D1524
ELECTRICAL
Dielectric breakdown, at 60 Hz D 877 or D 1816
Dielectric breakdown, impulse D3300
Power factor D924
objective of this paper is to present new or improved analytical techniques for a better character-
ization of insulating oils. This presentation is arranged along five major themes:
1. Thermal aging characterization.
2. Electrical properties characterization.
3. Particles and metals in oils.
4. Gases in oils.
5. Characterization techniques for low-temperature applications.
The third item is just beginning to receive the attention it deserves by the industry and engi-
neers, and the last one addresses to a specific application.
Another objective of this paper is to review some of the analytical techniques and methods
developed at IREQ during the last 15 years. As the research arm of a utility (Hydro-Quebec), a
lot of efforts have been devoted in our laboratory to the development of more reliable methods
for the evaluation of transformer oil aging. Thus, the emphasis here is put on mineral oil char-
acterization, although the techniques discussed could also be used with other insulating liquids.
Thermal Aging Characterization

New oils can be characterized by their interfacial tension, acidity number, density, and vari-
ous other parameters (see Table 1). However, as informative as these values can be, they cannot
predict the oil lifetime in service, especially for reclaimed oils. A certain amount of new infor-
mation, such as the molecular weight distribution, nature of aromatics, etc., can be gained by
high pressure liquid chromatography (HPLC) analysis [1]. This technique can also be used to
evaluate some specific transformations (polymerization, cracking, and unsaturation) during se-
vere electrical and thermal aging [/]. One of the most important application of HPLC with
dielectric liquids is to estimate the antioxidant content of inhibited oils. It is known that adding
di-t-butyl
Copyright para
by ASTM Int'l cresol
(all rights (DBPC)
reserved); Thu Mayto new oils
6 14:02:13 will prolong considerably their oxidation induction
EDT 2021
time [2]. The
Downloaded/printed by estimation of DBPC content in new oil can be done by infrared (IR) spectroscopy
CRINE ON CHARACTERIZATION OF OILS 61
[ASTM Test Method for 2,6-Ditertiary-Butyl Para-Cresol and 2,6-Ditertiary-Butyl Phenol in

Electrical Insulating Oil by Infrared Absorption (D 2668)]. However, the technique is not very
sensitive, and oxidation by-products absorb IR light at the same wavelengths, which means that
it is extremely difficult to evaluate the antioxidant content as a function of aging time. A HPLC
procedure developed at IREQ allows the detection of DBPC to content as low as 50 ppm (see
Fig. 1) [3]. In fact, this method has also been used with success to determine the antioxidant
content in lubricating oils. The possibility of developing a commercial instrument based on this
technique is currently examined.
Another technique developed at IREQ to evaluate DBPC content in mineral oil is differential
impulsional cyclovoltametry [4]. This technique is less precise than HPLC for DBPC content
lower than = 200 ppm and is also more time consuming.
Knowledge of the antioxidant content (if any) in mineral oil is very important because it was
shown that its depletion corresponds to a drop of interfacial tension value, a rapid increase of
acidity number, and sludge formation [5,6]. This critical point, called the oxidation induction
time, depends very much on the chemical structure of the oil and on the antioxidant amount.
The fact that the oxidation induction time varies very much with the oil nature implies that
considering only the results of measurements performed after 72 and 164 h of accelerated aging
at 110°C, as suggested by ASTM D 2440, could possibly be misleading. The 72 and 164 h-
values were originally selected because it was thought that an inhibited mineral oil will not be
oxidized before 72 h and will be completely oxidized after 164 h. However, some inhibited oils
(especially the new ones, as shown in Fig. 2) are oxidized before 72 h, which makes the compari-
960
INFRARED (ASTM test)
Q DIFF. IMPULSIONAL VOLTAMMETRY
24 36 48 60 84
Oxidation lime (h)
FIG. 1—Variations of DBPC content in paraffinic oil as a function of oxidation time (ASTM D 2440) as
measured by three different techniques: note that IR spectroscopy does not allow the detection of contents
smaller
(UFRN) thanFederal
Universidade = 120 do Rioppm.
Grande do Norte ((UFRN) Universidade Federal do Rio Grande do Norte) pursuant to License Agreement. No further reproductions auth
62 INSULATING OILS
50| T
o VOLTESSO 35 (before 81)

• VOLTESSO 35 (1981)
n VOLTESSO 35 (1984)
24 48 72 96 120 144 168
OXIDATION TIME (ASTM 2440) (hours)
0.40 T
0.35
o VOLTESSO 35 (before 1981)
. VOLTESSO 35 (1981)
0.30
• VOLTESSO 35 (1984)
:o.25
-0.20
§0.15 INDUCTION
<
PERIOD
go.to
c
0.05
0,
0 24 48
OXIDATION TIME (hours)
FIG. 2—Interfacial tension and soluble acidity of samples of a naphthenic oil manufactured on different
years as a function
Downloaded/printed by of oxidation time (ASTM D 2440). The abrupt variations correspond to sludge forma-
tion.
son between the results at 72 and 164 h worthless. It is therefore suggested to oxidize several oil
samples at various aging times, as shown in Figs. 1 and 2, rather than oxidizing two samples for
72 and 164 h. Thus, one can distinguish three phases in the oxidation process (Fig. 2): (1) an
induction period which ends when there is no more antioxidant left in the oil; (2) the beginning
of oxidation where acidity increases and sludge and peroxides are formed; and finally (3) a
saturation period where acidity has reached a maximum value. Note that it is not necessary to
measure all these parameters (acidity, peroxides, sludge, antioxidant) to determine the onset of
oxidation. The measurement of antioxidant content by HPLC is sufficient to give precise and
reliable estimates.
Another method to determine the oxidation onset is to measure the acidity number [ASTM
Test Method for Neutralization Number by Color Indicator Titration (D 974)], and the begin-
ning of oxidation corresponds to a rapid increase of soluble acidity (Fig. 2). Note that in aged oil
it is difficult to evaluate the neutralization point by visual observation because of sludge forma-
tion, particles, and oil darkening. We have improved the sensitivity of the ASTM test method by
determining photometrically the neutralization point with a Sybron-Brinkman Digital Probe
Colorimeter Model PC800 equipped with a 660-nm filter.
The onset of oxidation corresponds also to the formation of polar products which are the
precursors of acid groups. These polar products get unnoticed by standard techniques such as
IR spectroscopy, interfacial tension, and acidity number. However, these products are ex-
tremely detrimental to the chemical and thermal resistance of the oil (more details in Refs 5-8).
The polar products can be separated from the aromatic constituents of the oil and their content
estimated in relative units by reverse-phase HPLC (Fig. 3) [ 7]. Elution of oxidized oil on Fuller's
earth removes acidity products, sludge, and particles. However, it does not remove these polar
products, and their presence in reclaimed oils (in which 3000 ppm of DBPC was added) explain
the poorer thermal resistance of these oils [6-8]. In fact, it was possible to demonstrate that the
best time for oil reclaimation corresponds to antioxidant depletion [ 6-8]. This conclusion would
have not been made possible without the use of HPLC for the determination of antioxidant and
polar products content.
Finally, an alternative method of evaluating oil thermal stability is to measure the oxidation
induction time (OIT) by differential scanning calorimetry (DSC). The OIT is the time where the
heat of reaction at a given high temperature ( > 150°C) increases rapidly. The technique is rou-
tinely used for the characterization of lubricating oils [ 9], and it could also be used to character-
ize insulating oils. The measurement should be made under a high-pressure (500 psi or more) of
O2 in order to reduce the OIT.
Electrical Characterization of Oils

It is well known that it is difficult to characterize dielectric liquids (and especially aged ones)
from only their electrical properties. Nevertheless, electrical measurements such as dielectric
losses, d-c conductivity, inception voltage for partial discharges, and dielectric strength are of-
ten the ultimate criterion for selection of new oils by electrical engineers. We have recently
shown that the tan 8 and d-c conductivity of mineral oils aged under accelerated conditions
depend essentially on three parameters: (1) the dissolved copper content; (2) the peroxides con-
tent; (3) the soluble acidity content [5]. Empirical equations fitting the experimental results
obtained at 22°C with naphthenic oil (see Figs. 4 and 5) were [5]:
a = (1.5 X 1 0 - " X Ccu) + (10-'2 X CSA) + (1.5 X lO-'^ Cp) in (12 cm)"'
tan 6(60 Hz) = (10-^ X Ccu) + (2 X 10"^ CSA) + dO"^ Cp)
Copyright
where by ASTM
Ccu,Int'lCSA
(all rights
andreserved);
Cp are Thuthe
May copper,
6 14:02:13 EDT 2021 acidity, and peroxides content of the oil, respec-
soluble
tively.
(UFRN) These
Universidade equations
Federal suggest
do Rio Grande that some
do Norte ((UFRN) electrical
Universidade properties
Federal do couldpursuant
Rio Grande do Norte) be predicted from No
to License Agreement. thefurther
deg- reproductions auth
64 INSUUTING OILS
Elution, m ^
FIG. 3—Polar products detected by reverse-phase HPLC in transformer oil aged under various condi-
tions. The UV-response scale is the same for all samples.
radation products content or vice Versa. However, in-service mineral oils are used under a wide
variety of conditions and in the presence of impurities, paper, particles, etc. The above equa-
tions
Copyright are therefore
by ASTM less useful,
Int'l (all rights reserved); Thu May although the
6 14:02:13 EDT 2021electrical properties of oils aged in service depend
very muchby on acidity and on dissolved copper (as shown later).
Downloaded/printed
icr^2-
E
o
60 80
t{h)
FIG. A—Variation of d-c conductivity (22°C) as a function of oxidation time (ASTM D 2440). Exp
mental results: solid line: calculation (see text): open circles.
Another dielectric test commonly used in the electrical industry is the so-called "PFVO Doble
test," where the dielectric losses of an oil sample heated in air at 95°C are recorded as a function
of time [10]. Again, it is extremely difficult to correlate the results thus obtained with real in-
service conditions. However, it can be used to compare different new or aged oils.
A major difficulty in the comparison of tan 6 or conductivity values of different oils (or differ-
ent test methods) is the influence of the applied electric field. At low and moderate a-c field
intensities, tan 6 is nearly constant, but it increases rapidly at high field intensity. Although
ASTM D 924 specifies that measurements should be made at voltage stress larger than 2 kV/
cm, most results are obtained with a few applied volts per cm and only few are obtained with
fields of several kV/cm. Aged oils with their conductive impurities are much more sensitive to
field intensity than new oils. In our opinion, the measurements of electrical properties should
always be made at, at least, three field intensities ranging from a few volts per cm up to field
values similar to those encountered in service. That would certainly reduce the large scatter of
values often reported by various laboratories (and therefore various experimental conditions).
CopyrightEquipment
by ASTM Int'l (alland
rights test methods
reserved); Thu May 6for electrical
14:02:13 EDT 2021 properties measurements have not been greatly
modified over the last 20 years. One promising and recent development is a complex permittiv-
66 INSULATING OILS
10-2 P""T—1—r-
X O rP
o
c 10- NAPHTHENIC OIL
co
icrt0 ±_J 1 \ I I I I I I I L_i.

20 40 60 80 100 120 140
t(h)
FIG. 5— Variation of tan 5 (60 Hz. 22°C) as a function of oxidation time (ASTM D 2440). Same symbols
as in Fig. 4.
ity measurements system called Microdielectrometry and manufactured by Micromet Instru-

ments, Inc. [//]. The original feature of the instrument is that it uses coplanar interdigital
(comb-like) electrodes deposited on a silicon chip which is acting as a sensor. The small size of
the sensor and the fact that electrodes are coplanar mean that small liquid samples are re-
quired. The sensor could also be used in locations that would have been otherwise unaccessible
forbyconventional
Copyright parallel
ASTM Int'l (all rights disk
reserved); Thu Mayelectrodes.
6 14:02:13 EDT Finally,
2021 since the floating electrode on the sensor is a
charge integrating
Downloaded/printed by device, there is no lower limit of frequency. This system seems well suited for
very low-frequency measurements (down to = 10~^ — 10"'' Hz) where ionic currents and inter-
facial phenomena can then be estimated with more accuracy. Note that such low-frequency
measurements can now be relatively easily achieved with the recent development and use of
frequency response analyzers | / 2 ] .
The detection of partial discharges and the significance of their inception voltage are still a
matter of debate [ 13,14]. In fact, the major problem is that there is yet no standard experimen-
tal procedure and no agreement on what should be detected: Are the apparition of bubbles,
light 115], or sound emission [ 16] the characteristics of partial discharges? Another experimen-
tal difficulty is the nature of electrodes: Metallic or metal covered with paper or any insulating
material? [ 14] Much remains to be done in the basic understanding of the phenomena underly-
ing partial discharges and much more before a rational and reliable interpretation of the results
can be done. However, great progress has been made in their detection, mainly due to the exten-
sive use of computers [17],
Electric breakdown strength is one of the most important parameter characterizing an insu-
lating liquid. However, it is very difficult to correlate its value with a given lifetime in service.
This is essentially due to the fact that breakdown is associated with many parameters (water,
gases in oil, sampling conditions, particles, etc.). The following section is devoted to particle
characterization and their influence on electric breakdown measurements.
Particles and Metals in Oils

It is well known that solid particles in suspension in insulating liquids affect their dielectric
strength depending on their concentration, size, and nature [18-20]. However, up to recently
only limited attention has been paid to this problem. This was essentially due to the lack of a
simple and fast analytical technique. There is presently no specific ASTM test method devoted
to particle counting in insulating liquids, although ASTM Method for Microscopial Sizing and
Counting Particles from Aerospace Fluids on Membrane Filters (F 312) could be used. Particle
counting with the advent of automatic particle counters has now become a simple, fast, and
inexpensive technique. However, it is plagued with experimental difficulties that can make its
results absolutely worthless. The major problem is particle deposition at the bottom at the sam-
pling vessel. It was shown that particle deposition in transformer oil is relatively rapid, and in
only 1 h all particles larger than 50 nm are deposited [20]. Smaller ones take more time, but, as
shown in Fig. 6, particle deposition is very important in only one day, which results in a signifi-
cant improvement of the a-c breakdown strength of the oil. Note that in agreement with ASTM
Test Method for Dielectric Breakdown Voltage of Insulating Oils of Petroleum Origin Using
VDE Electrodes (D 1816), the oil container was inverted gently prior to the breakdown test; it
was shown elsewhere that this procedure does not restore the original particle distribution [22].
This means that oil samples stored several days before measurements (as is often the case) are
not representative of the oil situation at the sampling time. This also means that the oil shown in
Fig. 6 had a breakdown voltage (Vj) below the "acceptable" 50-kV value (for 2 mm) just after
sampling. However, one day later its Vj, was over the critical 50-kV value and it had an even
more artificially high V/, value after seven days at rest. In other words, the breakdown voltage
value, which is one of the most important routine measurements performed by most electrical
utilities, may give a completely misleading indication of the real condition of the oil in the trans-
former if it is performed several days after sampling. It should be noted that the results shown in
Fig. 6 correspond to a particularly dirty transformer oil. Most oil samples taken in transformers
in service have particle concentrations of the order of some thousands per 10 c^.
It is therefore absolutely necessary to rehomogenize oil samples before measurements such as
particle count, metal content, dielectric strength, etc. At IREQ, we have selected ultrasonic
agitation
Copyright by ASTMbecause
Int'l (all rightsitreserved);
is easyThutoMay
use6 14:02:13
and requires
EDT 2021 low cost equipment. Bubble formation has been
occasionally observed during ultrasonic agitation but they disappear very rapidly, and it was
68 INSULATING OILS
I I 1111
10'
After sampling
'Vb=47.6kV
Ul
7 days at rest
10- Vb=63.2kV
1 day at rest —
10' J I I I I I II I
10" 10
Size ()un)
FIG. 6—Particle distributions in aged transformer oil as a function of resting time. The breakdown volt-
age (ASTM D 1816) Vb is also indicated.
verified experimentally that this does not affect particle count and dielectric strength measure-
ments. The oil sample can absorb some water during ultrasonic agitation, but for a relatively
dry sample (that is, < 15 ppm of water) the water content increase is less than 2 to 3 ppm (as
long as the oil is not directly exposed to air). Interestingly, an International Electrotechnical
Commission (lEC) worlcing group is currently developing a procedure of oil homogenization
also based on ultrasonic agitation [21].
Detailed experimental conditions are given in Ref 22; to summarize, the recommended ultra-
sonic homogenization is as follows:
1. Fill a small ultrasonic bath (150 W) with 1 L of water.
2. Agitate one oil bottle (1 L or less) for 25 min (do not forget to unscrew the bottle's cap to
avoid overpressure induced by liquid expansion).
3. Perform the measurement (particle counting, electric breakdown test, metals content,
etc.) immediately afterward.
Particle deposition is not only important for particle counts and dielectric strength measure-
ments but also affects other measurements such as the metal content of the oil. It is well known
that metals, especially copper and iron, are dissolved during oil aging [2,5,20], Figure 7 shows
the copper
Downloaded/printed by content of some transformer oils measured after various deposition times; it is obvi-
CRIME ON CHARACTERIZATION OF OILS 69
MONTHS AT REST
FIG. 7—Deposition as a function of resting time of copper in samples of oil taken in various transformers
in service.
ous that significant differences are observed and again it could wrongly be concluded that, in
some cases, the oil is still acceptable when, in fact, it is aging very rapidly. Thus, homogeniza-
tion is requested before metals in oil analysis.
In addition to the dissolved copper induced by aging, transformer oil may also contain some
copper (and other metal) particles in suspension related to the windings or contact problems
[23]. Since most suspended particles can be removed by filtering, important information re-
garding transformer aging can be gained by separating the two kind of metallic contents. Table
2 shows some typical results obtained with two apparatus where the copper and lead contents
are sensitive to filtering whereas the iron and zinc contents are not; the last two metals are
therefore essentially dissolved in oil as submicron particles.
The very low metal contents reported in Table 2 affect significantly several electrical proper-
ties of oils, as discussed in the section "Electrical Characterization of Oils." They are also ex-
tremely difficult to estimate, which explains, in fact, why there are only very few references on
that aspect [20.23]. To the best of our knowledge the best analytical technique to measure such
low metal contents is atomic absorption spectroscopy (AAS) performed with a graphite furnace.
Without a graphite furnace, ordinary flamme AAS or Ion-Coupled Plasma AAS are not sensi-
tive enough (minimum of = 1 ppm for most elements) to detect 10 to 100 ppb of impurities in
oil. Another analytical technique that has proven to be reliable for copper content measure-
ments is neutron activation analysis (NAA) [5,20]. This technique can reliably detect very low
contents
Copyright by ASTMofInt'lcopper
(all rights (less than
reserved); 5 ppb),
Thu May butEDTit 2021
6 14:02:13 is not very sensitive to iron (more than 1000 ppm are
Downloaded/printed
required) by
and cannot detect lead. By any means, detection of copper by NAA (see Fig. 8) is
70 INSULATING OILS
TABLE 2—Influence offilteringon some metal-in-oil contents.
Copper, Iron, Lead, Zinc,

Oil sample ppb ppb ppb ppb
No. 1
Before filtering 110 3.7 3.7 <2
After filtering 81 3.7 1.9 <2
No, 2
Before filtering 105 2.3 8.8 <2
After filtering 69 2.0 3.0 <2
much simpler and more precise than the photometric analysis recommended by ASTM Test
Method for Copper in Electrical Insulating Oils by Photometric Analysis (D 2608). Whatever
the technique used, sample preparation [see ASTM Test Method for Copper in Electrical Insu-
lating Oil by Atomic Absorption Spectrophotometry (D 3635)] and container cleanliness are
crucial steps in obtaining reliable results.
As useful as they are, particle counting and metals in oil analysis cannot give a description of
the shape and nature of all particles. In transformers, for example, most particles are long
cellulose fibers (whose real length is rarely measured correctly by particle counters) and small
( < 2 /im) carbon particles. They can be identified by microscopic observation after deposition
on a membrane, as suggested in ASTM F 312. However, this is a long and tedious procedure
which may be of limited interest for routine measurements. In addition, some particles get
through the membrane, and it is not an easy job to distinguish a cellulose fiber on an acetate
membrane [24]. These difficulties may be overcome by using ferrography, a technique origi-
nally developed to study wear and abrasion particles, where all particles are deposited and per-
manently fixed on a glass plate under the action of a magnet (see Fig. 9).
Magnetic particles are aligned by the magnetic field and can be easily detected (see Fig. 10)
under microscopic observation. The approximate nature of other particles can also be deter-
mined by an appropriate choice of various light filters (Fig. 11) [24]. A rapid microscopic obser-
vation of the glass plate gives interesting information on the shape and nature of the particles.
In addition, particles being permanently fixed on the glass plate can be precisely identified at
any time by a technique such as electron microprobe. Thus, ferrography combined to particle
counting can give a qualitative estimation of the shape, kind, and nature of particles in an
insulating liquid. Note that ferrography glass plates could eventually be analyzed by an image
analysis system for automated particle size distribution measurements.
Gases in Oils
When a dielectric liquid is subjected to electric arcs, localized hot spots, or partial dis-
charges, it generates various gases whose chemical composition depends on the nature of the
fault. The analysis of dissolved gases has become a very useful method of diagnosis of incipient
faults in transformers [25.26]. The lEC is currently reviewing its standards of gas analysis in oil,
especially with regards to contents below the ppm range.
A method has been developed at IREQ which addresses this problem. It consists in the prepa-
ration of standard oil samples containing volumetrically determined amounts of dissolved gases
[27]. This method has been successfully used by Hydro-Quebec to control the quality of the
analyses performed by various laboratories bidding for services. It has also been used by lEC to
evaluate the repeatability and reproductibility of various gas extraction and chromatographic
techniques
Copyright used
by ASTM Int'l worldwide.
(all rights This
reserved); Thu May has allowed
6 14:02:13 EDT 2021 us to show that ASTM Method for Analysis of
Downloaded/printed
Gases by
Dissolved in Electrical Insulating Oil by Gas Chromatography (D 3612) is a very good
402r:
P-l \ T -I—I—r •T—I—r-^
NAPHTRENIC OIL
10^
E
a.
o.
l - i o o -/V
c_>
10-^ o—o—-o-
-IO-21 I I I I I \ I I I I I
20 40 60 80 100 120 140
t(h)
FIG. 8—Dissolved copper content (as measured by neutron activation analysis) in transformer oil as a
function of oxidation time (ASTM D 2440). Filled symbols: unfiltered: open symbols: filtered oil. The arrow
corresponds to sludge formation.
method as compared to those used in Europe and Japan. Note that this applies only to the
original method of ASTM D 3612 and not to the modified versions. Special diagnosis methods
are used at Hydro-Quebec for the interpretation of gas-in-oil results [26,28], which have been,
in several instances, more reliable than standard codes [29].
One of the most dangerous gases generated by partial discharges, arcs, or hot spots is hydro-
gen. In order to be able to follow continuously the time-evolution of hydrogen content in HV
transformers, a detector was developed at IREQ [30]. Basically, it is an inverted fuel cell, and
the presence of hydrogen induces a current proportional to the gas content. Commercial instru-
ments
Copyright called
by ASTM Hydran
Int'l (all 101 Thu
rights reserved); andMay201 (Fig.EDT
6 14:02:13 12)2021
have subsequently been marketed by Syprotec Inc.
[31]. To the best of our knowledge this hydrogen detector has proven satisfactory to users [32].
72 INSULATING OILS
"^
FIG. 10—Photograph of particles in aged oil deposited on a ferrogram. Note the magnetic particles
aligned parallel to the magnetic field. Other metallic particles and fibers are also visible. Magnification:
XIOO.
It should be noted that the hydrogen analysis should be complemented by a more complete gas
analysis to determine the origin of the fault.
For transformer applications, the tendency is to develop a simple and inexpensive gas moni-
tor that could be installed in or near the apparatus. Westinghouse Electric Corp. is currently
developing an in-situ gases detector to monitor power transformers [33]. A multigas in-situ
monitor is also under study at IREQ [34], and one Japanese team has recently presented results
obtained with a new and inexpensive gases detector [35].
Characterization Techniques for havi Temperature Applications

The dwindling reserves of mineral naphthenic oil have stirred a continuous interest for other
types of mineral liquids, especially for the so-called paraffinic oils. Paraffinic and naphthenic
oils have essentially the same electrical and chemical properties [36]. However, oils with larger
paraffinic content have significantly larger viscosities at low temperatures. It was shown that a
paraffinic oil made from Canadian crude was completely solid at —23°C [37]. The wax forma-
tionbywas
Copyright ASTMalso observed
Int'l (all in oils
rights reserved); of 6various
Thu May origins,
14:02:13 EDT 2021 and it was determined that it occurs every time
that the mineral
Downloaded/printed by oil contains more than 4% of linear paraffins [38]. This solidification problem
74 INSULATING OILS
FIG. 11—Photograph of the ferrogram of another transformer oil sample taken in a transformer in ser-
vice. Longfibers and carbon particles are the major contaminants. Note the wide variety ofparticle size and
shape. Magnification: XIOO.
of paraffinic transformer oil could be disastrous in a cold weather country. It is therefore impor-
tant to evaluate the paraffins content in mineral oils. This can simply be done by measuring the
cloud point or the pour point as a function of paraffins content [38]; these are not very precise
methods. We obtained much more precisely the paraffins content of mineral oils by differential
scanning calorimetry (DSC) (Fig. 13) or by X-ray diffraction measurements [38]. The latter
method is less precise for low wax content but it has the advantage of characterizing the nature
of paraffins (linear or amorphous).
A complete characterization of low-temperature liquids should, of course, include viscosity
measurements. This can be done with commercial viscosimeters such as the Brookfield or by the
so-called Stoke method [37],
It is possible to improve, that is, to reduce, the viscosity of oils at low temperatures by adding
a flow-improver.
Copyright The
by ASTM Int'l (all rights amount
reserved); of6flow-improver
Thu May 14:02:13 EDT 2021 in unaged or aged oil can be precisely evaluated
by gel permeation
Downloaded/printed by chromatography (GPC) [39].
FIG. 12—Photograph of an Hydran 201R hydrogen monitor installed on a transformer.
New Trends and Futme Developments
Static Electrification
Static electrification in transformer oils has recently gained a renewed interest due to some
unexpected and yet unexplained transformer faults [40]. When an insulating fluid flows over a
conducting component charge, separation occurs and the liquid gets charged. If the voltage
thus generated exceeds the breakdown strength of the liquid, discharges will occur. The separa-
tion of charges at the solid/liquid interface is affected by several parameters which are not yet
all known. In fact, the basic phenomenon of static electrification is poorly understood and is
now the subject of several studies all around the world [40-43]. There is currently a need for a
more detailed standard experimental procedure than the present ASTM test method D 2679
[discontinued—to be replaced by D 4470]. A meaningful definition for the "electrostatic charg-
ing tendency" of a liquid is also needed.
Metal Deactivators
It has been known for a long time that adding a metal deactivator to insulating liquids, espe-
cially transformer oils, leads to a better thermal resistance [44,45]. It was recently shown that it
may also reduce the electrostatic charging tendency of oil [42]. Research has to be done to
determine whether metal deactivators should be really added to transformer and capacitor oils
to prolong
Copyright their
by ASTM Int'l lifetime.
(all rights reserved); An analytical
Thu May procedure
6 14:02:13 EDT 2021 to detect its content and variation with time
has also to
Downloaded/printed bybe devised.
76 INSULATING OILS
-60 -50 -40 -30

Temperature ("C)
-10 -1 1 1 1 1 1 r
20-
i
o
V)
o
2 4 6 10
Linear paraffins (%)
FIG. 13—DSC evaluation of the paraffins content in transformer oils, (a) Typical DSC curves showing
the onset of crystallization ofnaphthenic oil (A), paraffinic oil (B), and partially dewaxed oil (C). (b) Influ-
ence of the amount of linear paraffins on the onset of crystallization and on the cloud point of various oils.
Ionic Impurities in Liquids

Although it is agreed that ionic impurities are the major charge carriers in most insulating
liquids, there are still few experimental measurements of the ionic contents of these liquids.
With the advent of ion chromatographs, one may expect that more data would be soon avail-
able. Again, a standard experimental procedure is required since the ionic content could vary
greatly with the experimental conditions.
Scale Model Studies

CopyrightDielectric liquids
by ASTM Int'l (all are generally
rights reserved); used in
Thu May 6 14:02:13 EDTpower
2021 equipment such as transformers, capacitors, or
cables where
Downloaded/printed by they are in contact with various insulating and conducting materials. On the long
High voltage Guard

electrode Stress
copper cone
conductor
r=04mni
Electrical insulotion
5 layers of 76/im
Kraft paper
a) Schematic representation of the scale models used by Ref. 46
H.V. N2
! f
f
to bridge — Lead weight
1
Mz: #-Phenolic lid
• Teflon tube
-Sample
i:; r (up to 7)
-Glass vessel
Copper
connector -
Brass plate-
^
T>T~
u Teflon spacer
•Teflon leg
b) Experimental set-up used to test the above scale models.

FIG. 14—Scale models developed and tested at IREQ to simulate cable and transformer aging, (a) Sche-
matic representation of the scale models used by Ref 46. (b) Experimental setup.
f^ INSULATING OILS
TABLE 3—Some new analytical techniques developed or used at IREQ.
Complement
Property or Phenomenon or Improve
Technique Characterized ASTM Test
HPLC Antioxidant content (onset of

oxidation) D 2668, D 1473, D 2446
Reverse-phase HPLC Polar groups D 2440, D 974, D 1902
Neutron Activation Analysis Dissolved copper content D3635
Atomic Absorption (with graphite
furnace) Impurity contents D3635
Automatic particle counting Number of particles F312, D1816
Ferrography Nature and shape of particles F 312, D 1816
Ultrasonic agitation Sample homogenization D 3613, D 1816, D 3635
Differential Scanning Calorimetry Onset of oxidation (OIT), paraffins
content D2766
GPC Flow improver content
Hydran monitor Detector of hydrogen D3612
term, these materials tend to interact with the liquids either by dissolution, swelling, electro-
chemical reactions, etc. This means that a complete characterization should take into account
the liquid itself and also other parameters or phenomena associated with its environment.
One approach in this direction is to use scale models made with various metals, paper, var-
nish, etc. (Fig. 14). At IREQ we have developed a scale model that could simulate the materials
and conditions in transformers or in oil-filled cables [46]. Scale model aging studies are very
useful because they provide data obtained midway between service aging conditions in a full
scale apparatus and accelerated laboratory testing, as in ASTM D 2440.
Identification of Oil Constituents by GC/MS

Identification of the various components of mineral oils by gas chromatography/mass spec-
trometry (GC/MS) is not a new topic. Although some work has started more than 30 years ago,
it is still far from being complete, largely because mineral oils are natural and very complex
compounds. Development of new experimental procedures and analytical techniques are re-
quired before significant progress can be made.
Conclusions
The advantages and limits of a wide variety of newly developed analytical techniques (see
Table 3) for insulating liquids have been discussed. New trends and future developments have
also been pointed out. One may expect that the current implementation of computer-controlled
instruments will eventually lead to significant modifications of laboratory procedures and anal-
yses.
Acknowledgments
It is a pleasure to thank my colleagues C. Lamarre, M. Duval, C. Vincent, D. Couderc,
G. Belanger, and R. Malewski for their assistance in the preparation of this review.
References
(/] Duval,
Downloaded/printed by M. and Lamarre, C , "The Characterization of Electrical Insulating Oils by HPLC,'
IEEE
Transactions
(UFRN) Universidade Federal doon RioElectrical Insulation,
Grande do Norte Vol. 12,
((UFRN) Universidade 1977,
Federal p. Grande
do Rio 340. do Norte) pursuant to License Agreement. No further reproductions auth
[2] Melchiore, J. J. and Mills, 1. W., "The Role of Copper During the Oxidation of Transformer Oils,"
Journal of the Electrochemical Society, Vol. 112, 1965, p. 390.
[3] Lamarre, C , Duval, M., and Gauthier, J., "Dosage par chromatographie liquide haute performance
du DBPC dans les huiles de transformateur neuves ou usagees," Journal of Chromatography, Vol.
213, 1981, p. 481.
[4] Belanger, G., Lamarre, C , andCrine, J. P., "Antioxidant Analysis in Transformer Oils," Minutes of
48th Annual International Conference Doble Clients, Section 10-601, 1981, Doble, Inc., Watertown,
MA.
[5] Lamarre, C , Crine, J. P., and Duval, M., "Influence of Oxidation on the Electrical Properties of
Inhibited Naphtenic and Paraffinic Transformer Oils," IEEE Transactions on Electrical Insulation,
Vol. 22, 1987, p. 57.
[6] Lamarre, C , Crine, J. P., andSt-Onge, H., "Antioxidant Functionality in Liquid and Solid Dielectric
Materials," CIGRE Symposium on New Materials, Vienna, May 1987.
17] Duval, M., Lamarre, C , and Giguere, Y., "Reverse-Phase HPLC Analysis of Polar Oxidation Prod-
ucts in Transformer Oils," Journal of Chromatography, Vol. 284, 1984, p. 273.
[ 8] "Study of the Thermal Resistance of Reclaimed Transformer Oils," Report No. 203 T 444, Canadian
Electrical Association, Montreal, Canada, 1986.
[91 Walker, J. A. and Tsang, W., "Characterization of Lubricating Oils by Differential Scanning Calo-
rimetry," SAE Technical Paper Series No 801383, Society of Automotive Engineers, Warrendale, PA,
1980.
[10\ Doble Engineering Co., 85 Walnut St., Watertown, Mass. 02172 (see also Ref 45).
\Il] Micromet Instruments Inc., 21 Erie St., Cambridge, MA 02139.
112] See among others J. Pugh, 1984 IEEE Conference Dielectric Materials & Applications, London, No.
239, 1984, p. 247.
[13\ "Engineering Dielectrics Vol. \," m Engineering Dielectrics, Vol. 1, Corona Measurement and Inter-
pretation, ASTMSTP669, R. Bartnikas and J. McMahon, Eds., 1979.
[14] Fallou, B., Perret, J., Samat, J., and Vuarchex, P., "Evolution des criteres de selection des liquides
isolants," CIGRE Conference, Paper 15-10, 1986.
[IS] Pfeiffer, W., "Fast Measurements Techniques for Research in Dielectrics," IEEE Transactions on
Electrical Insulation, Vol. 21, 1986, p. 763.
[16] Harrold, R. T., "Acoustic Theory Applied to the Physics of Electrical Breakdown in Dielectrics,"
IEEE Transactions on Electrical Insulation, Vol. 21, 1986, p. 781.
[17] James, R. E., Trick, F. E., Phung, B. T., and White, P. A., "Interpretation of Partial Discharge
Quantities as Measured at the Terminals of HV Power Transformers," IEEE Transactions of Electri-
cal Insulation, Vol. 21, 1986, p. 629.
[18] Oomen, T. V. and Petrie, E. M., "Particle Contamination Levels in Oil. Filled Large Power Trans-
formers," IEEE Power Engineering Society Summer Meeting 1982, IEEE, New York, NY.
[19] Miners, K., "Particles and Moisture Effect on Dielectric Strength of Transformer Oils Using VDE
Electrodes," IEEE Transactions PAS. Vol. 101, No. 3, 1982.
[20] Vincent, C. and Crine, J. P., "Evaluation of In-Service Transformer Oil Condition from Various Mea-
surements," Conference Record 1986 IEEE International Symposium of Electrical Insulators, IEEE,
New York, NY, 1986, p. 310.
[21] lEC SCIOA—Special Working Group on Particle Sizing and Counting, 1985.
[22] Vincent, C , Crine, J. P., and Olivier, R. G., "Ultrasonic Homogenization of Particles in Transformer
Oil," Doble Client Conference, April 1987, Doble, Inc., Watertown, MA.
[23] Skog, I. E., de Giorgio, J. B., Jakob, F., andHaupert, T. J., "Location of Incipient Faults by Metal in
Oil Analysis," Doble Client Conference, 1980, Doble, Inc., Watertown, MA.
[24] Olivier, R. G., Vincent, C , and Crine, J. P., "Analysis of Particles in Transformer Oil by Ferrog-
raphy," Doble Clients Conference, Sec. 10-401, 1986, Doble, Inc., Watertown, MA.
[25] Baker, A. E., Griffin, P. J., and Locke, C , "An Update on Fault Gas Analysis. A Review," Doble
Clients Conference, Section 10-302, 1986, Doble, Inc., Watertown, MA.
[26] Duval, M., "Fault Gases Formed in Oil-Filled Breathing EHV Power Transformers. The Interpreta-
tion of Gas Analysis Data," IEEE Power Engineering Society Summer Meeting, Paper C74 476-8,
IEEE, New York, NY, 1974.
[27] Duval, M. and Giguere, Y., "Preparation of Standard Samples of Dissolved Gases in Insulating
Oils," Doble Clients Conference, Section lOC-01, 1984, Doble, Inc., Watertown, MA.
[28] Dind, J. E. and Regis, J., "How Hydro-Quebec Diagnoses Incipient Faults by Using Gas-in-Oil Anal-
ysis," Pulp and Paper Canada. Vol. 76, 1975, p. 61.
[291 See among others, Rogers, R. R., "IEEE and lEC Codes to Interpret Incipient Faults in Transform-
ers," IEEE Transactions on Electrical Insulation. Vol. 13, 1978, p. 349.
[30] Belanger, G. and Duval, M., "Monitor for Hydrogen Dissolved in Transformer Oil," IEEE Transac-
tions on
Downloaded/printed by Electrical Insulation. Vol. 12, 1977, p. 334.
[31]
(UFRN) Syprotec,
Universidade Inc.,
Federal 88Grande
do Rio Hymus Road,
do Norte Pointe-Claire,
((UFRN) Qc,doCanada
Universidade Federal Rio GrandeH9R 1E4.
80 INSULATING OILS
[32] Sparks, R. D. and Turney, J. H., "Tennessee Valley Authority's Application of the Hydran 201 on-
line Hydrogen Monitor to 500 kV Transformers," Doble Clients Conference, Sec. 6-401, 1986, Doble,
Inc., Watertown, MA.
[33] Nilsson, S. L., "Development of New Techniques for Operation and Maintenance of High Voltage
Substations and Substation Equipment," Doble Clients Conference, Sec. 2-301, 1986, Doble, Inc.,
Watertown, MA.
[34] Malewski, R., Douville, J., and Belanger, G., "Insulation Diagnostic System for HV Power Trans-
formers in Service," 1986 CIGRE Conference, Paper 12-01, 1986.
[35] Tsutokia, H., Sugawara, K., Mori, E., and Yamaguchi, H., "New Apparatus for Detecting Trans-
former Faults," IEEE Transactions on Electrical Insulation, Vol. 21, 1986, p. 221.
[36] Duval, M., Cauchon, D., Lamothe, S., and Giguere, Y., "Paraffinic Transformer Oils for Use at Low
Temperatures," IEEE Transactions Electrical Insulation, Vol. 18, 1983, p. 586.
[37] Langhame, Y., Castonguay, J., Crine, i. P., Duval, M., and St-Onge, H., "Physical Behavior of
Paraffinic Oils at Low Temperatures," IEEE Transactions on Electrical Insulation, Vol. 20, 1985, p.
629.
[38] Crine, J. P., Duval, M., and St-Onge, H., "Determination of the Nature and the Content of Paraffins
in Mineral Transformer Oils," IEEE Transactions on Electrical Insulation, Vol. 20, 1985, p. 419.
[39] Duval, M., Lamothe, S., Cauchon, D., Lamarre, C, and Giguere, Y., "Determination of Flow-Im-
prover Additives in New and Aged Insulating Oils by Gel Permeation Chromatography," Journal of
Chromatography, Vol. 244, 1982, p. 169.
[40] Griffin, P. J., "Update on Static Electrification in Transformer Oils," Doble Clients Conference, Sec.
6-1201, 1986.
[41] Crofts, D., "The Static Electrification Phenomena in Power Transformers," Record of Conference
Electrical Insulation Dielectric Phenomena, 1986.
[42] Yasuda, M., Goto, K., Okubo, H., Ishii, T., Mori, E., and Masunaga, M., "Suppression of Static
Electrification of Insulating Oil for Large Power Transformers," IEEE Power Engineering Society
Winter Meeting, Paper 82WM 197-2, 1982, IEEE, New York, NY.
[43] Brzostek, E. and Kedzia, J., "Static Electrification in Aged Transformer Oil," IEEE Transactions on
Electrical Insulating, Vol. 21, 1986, p. 609.
[44] Hughes, F. and Haydock, P. T., "The Industrial Use of Passivated Transformer Oil," Journal of the
Institute of Petroleum, Vol. 50, 1964, p. 239.
[45] Duval, M. and Crine, I. P., "Dielectric Behavior and Stabilization of Insulating Oils in EHV Current
Transformers," IEEE Transactions on Electrical Insulation, Vol. 20, 1985, p. 437.
[46] "Life Testing of Catalytically Dewaxed Paraffinic Transformer Oil," Report No. 000 T 453, Canadian
Electrical Association, Montreal, Canada, 1985.
DISCUSSION
C. L. S. Vieira' {written discussion)—The study of thermal aging presented was made with
oils that had at least 800 ppm of DBPC. Considering that some oils do not have DBPC added,
what will be the evaluation of the decomposition products? In this case the suggested control of
remaining DBPC content during the life of the oil when in service will not be useful. What
additional control should be added?
J. P. Crine (author's closure)—The most sensitive method to evaluate thermal aging of an
insulating oil (with or without antioxidant) is to measure its polar groups content by HPLC (see
Refs 7 and 8). This technique requires relatively sophisticated equipment and a skilled opera-
tor. A less expensive, though much less sensitive, method is to measure the interfacial tension; a
value lower than 25 to 30 mN/m generally indicates significant thermal degradation of the oil
sample.
Finally, note that the so-called noninhibited oils may contain up to 800 ppm of DBPC. There-
fore, it may be worth verifying whether some antioxidant is present in the oil before performing
other measurements.
'CEPEL, Federal
(UFRN) Universidade P.O. doBox 2754,doRio
Rio Grande de((UFRN)
Norte Janeiro, Brazil.Federal do Rio Grande do Norte) pursuant to License Agreement. No further reproductions auth
Universidade
Section IV—Dissolved Gas in Oil
Leo J. Savio^
Transformer Fault Gas Analysis and

Interpretation—A User's Perspective
REFERENCE: Savio, L. J., "Transfomier Fault Gas Analysis and Interpretation—A User's Per-
spective," Electrical Insulating Oils, STP 998, H. G. Erdman, Ed., American Society for Testing
and Materials, Philadelphia, 1988, pp. 83-88.
ABSTRACT: All transformers generate gases during their normal operation as a result of aging
and oxidation exclusive of abnormalities. From an operational point of view, it is important to,
first, detect and recognize the existance of a problem by observing or detecting deviations from
gases considered normal, second, to evaluate the impact on the operation of the transformer (serv-
iceability), and, lastly, to take appropriate action, such as removal from service or increased moni-
toring frequency. This paper will address these considerations from a utility's perspective using
IEEE C57.104 as a guide.
KEY WORDS: transformer, combustible gas, analysis, oil, gases, diagnosis, dissolved gas, IEEE
C57.104
Transformers are an important and normally a very reliable component of a power system.
Over the years various techniques have evolved to monitor the serviceability of the power trans-
former, such as testing of mineral oil, its insulating and cooling fluid, and various electrical
tests of its paper and oil insulation system. One of the most important indicators, in this writer's
opinion, is testing the transformer's mineral oil for dissolved combustible gases. This is analo-
gous to blood testing of humans to detect various biological abnormalities.
The significance and use of dissolved gas as a diagnostic tool to determine the serviceability of
a transformer will be discussed in this paper. ANSI/IEEE C57-104 1978, "IEEE Guide for the
Detection of Generated Gases in Oil-Immersed Transformers and their Relation to the Service-
ability of the Equipment," hereafter referred to as C57.104, will be reviewed. C57.104 is under-
going a major revision in the area of interpretation and the decision process.
It is this aspect of C57.104 that will be addressed from the viewpoint of the user.
This paper is organized in the sequence proposed for the revised C57.104 and is the basis of
the decision process.
1. Detection.
2. Diagnosis.
3. Evaluation and action.
Detection of a Problem
As the mineral oil circulates through the transformer, it carrys with it solids and gases. Using
the human body and its blood circulation system as an analogy, a doctor can detect the early
warning signs of many abnormalities from tests of blood samples. It is likewise possible to detect
the early signs of abnormalities in a transformer by sampling its vital fluid (oil) and analyzing it
for combustible gas and other indicators.
There are several monitoring options available to the user which can provide an indication of
the presence of combustible gas. Some are a gas detector relay on conservator and membrane
Copyright'Consulting engineer.
by ASTM Int'l (all Consolidated
rights reserved); Edison
Thu May 6 14:02:13 EDTCo.
2021 of New York, New York, NY 10003.
83
Copyright® 1988 b y A S T M International www.astm.org
84 INSULATING OILS
sealed units, total combustible gas on gas-blanketed units, dissolved gas in oil using periodic
manual sampling, or continuous monitoring using installed hydrogen detection devices. These
may be used alone or in combination. The existing C57.104 indicates normal operating levels
for total combustible gas and gas in oil. In my company we manually sample transformer oil on
a routine schedule as indicated below for chromatographic analysis:
Generator step-up transformers Every two months

345-kV transformers Quarterly
138 and 69 kV Annually
We also monitor several units using fixed detection devices, which indicates hydrogen content
dissolved in the oil.
It is recognized that all transformers will produce combustible gases as a result of the normal
aging process. That being true, it should then be easy to determine abnormal transformer be-
havior by comparing it to the normal aging gassing quantities. That would be fine if one knew
what gases and levels are associated with a healthy transformer. Since this is still an art and not
a science, we must depend on our documented experience, which is small compared to that in
Europe.
In order to help determine norms for a healthy transformer, we must rely on published infor-
mation like that presented in C57.104 and other sources as indicated in Table 1. It is evident
from data in Table 1 that there is no agreement on acceptability levels.
Diagnosis of the Gas Data

Let us assume that we have determined that a transformer has abnormal gas content. What
next? At our company we resample the unit's oil to verify the gas content. If it is still high we
monitor the unit by sampling the oil every week or daily if the total gas content is high or if
acetylene is above 25 ppm. We observe the gassing rate, and if it is essentially stable we will only
monitor; however, if it is increasing more than 100 ppm per day we will take it out of service for
an internal inspection.
Before removing the unit from service, it would be helpful if one could diagnose the source of
the gassing, much like making a diagnosis of a human illness based on blood tests. There are
several available methods used to diagnose the source of gassing. Three of the most commonly
used methods are:
1. Domenberg ratio method.
2. Rogers ratio method.
3. Key gas method.
TABLE 1—Dissolved gas-in-oil acceptability limits from various sources.

Source H2 CO CR, CiYU C2H4 C2H2 TCG
Electra (1978) [2] 28.6 289 42.2 85.6 74.6 520

Hydro-Quebec [3\ 5(X) 900 35 20 85 2 1600
General Electric [4\
(3-year-old Trf) 200 200 50 20 100 20 595
General Electric \4]
(6-year-old Trf) 500 500 100 400 200 25 1725
Doble [2] 100 250 100 60 100 5 <700
IEEE [5]
Generator 240 580 160 115 190 11 1296
Transmission 100 350 120 65 30 35 700
NOTE: TCG = total combustible gas; trf = transformer.
SAVIO ON TRANSFORMER FAULT GAS ANALYSIS 85
TABLE 2—Transformer gassing history.
Sampling Dates
Gas 1/5/87, ppm 4/7/87, ppm 4/10/87, ppm
Hydrogen (Hj) ND 723 2 036

Methane (CH4) 48 1988 4 743
Carbon monoxide (CO) 14 28 28
Carbon dioxide (CO2) 412 496 421
Ethane (CjHt) 53 595 1303
Ethylene (CjR,) 5 3259 8 284
Acetylene (C2H2) ND 32 110
Total combined gas 120 6625 16 504

Rate/day (ppm/day) 3 293
NOTE: The transformer above is a 138-kV, 200-MVA Phase Angle Regulator. The volume of oil was
about 20 000 gal.
Domenberg Method
This method uses the ratio analysis of significant gases, that is, methane/hydrogen, acety-
lene/ethylene, acetylene/methane, and ethane/acetylene. Dornenberg requires, for accuracy,
that the significant gases exceed norms developed from the analysis gas samples from a large
population of transformers.
Rogers Method
This method also uses ratios of significant gases, except that Rogers uses three ratios, that is,
methane/hydrogen, acetylene/ethylene, and ethylene/ethane. Rogers does not indicate any
qualifying norms when using his method.
Key Gas Method

This method uses the percentages of various key gases to determine the cause, that is, in the
case of overheating oil the key gas is ethylene with smaller quantities of methane, ethane, and
hydrogen present. The key gas in the case of arcing is acetylene, and in the case of corona the
key gas would be hydrogen.
However, unlike the diagnosis of blood where rules have been established, the diagnosis of
gas in oil is not so clearly defined. It has been reported that the diagnosis of gases do not always
yield the same results when using different methods on the same sample [1].
The presence and quantity of gases are dependent on many variables, such as type, area, and
temperature of the fault; solubility and degree of saturation of various gases in oil; the type of oil
preservation system; the type and rate of oil circulation; the kinds of material in contact with the
fault; and, finally, variables associated with the sampling and measuring procedures them-
selves. Keeping in mind that this is an art and not a science, it is safe to use these methods as
indicators.
One of the obstacles to better accuracy is the lack of a large published data base in the United
States which relates gassing to actual conditions found. The Edison Electric Institute (EEI) is
beginning
Copyright to (all
by ASTM Int'l collect data Thu
rights reserved); onMay
gassing transformers
6 14:02:13 EDT 2021 where the cause of gassing has been deter-
mined. by
Downloaded/printed
86 INSULATING OILS
Evaluation and Action

For illustration purposes, let us assume that we have a transformer that has developed a
gassing history (see Table 2 for gas data). Using Fig. 1, a flowchart, I will address the evaluation
process being considered in the revision of C57.104.
1. Gas-in-oil monitoring indicates abnormal levels of hydrogen, methane, ethane, ethylene,
and acetylene based on chromatographic analysis.
2. Unit is resampled and the high gas levels are confirmed.
3. Gas level increasing.
4. Notified manufacturer.
5. Diagnosis made using the following methods with results as indicated in Fig. 2.
6. Continued to monitor unit and arranged for outage to inspect.
7. Inspection revealed that core laminations were touching the tank wall (shell form design).
8. Enter information in company and EEI data base.
Conclusion
The new C57.104 will be more informative and therefore more useful to the user. However, as
was pointed out, we need a broader data base of United States utility and industrial experience
with gassing transformers. This is necessary in order to improve the accuracy of diagnosis,
which is a key element in the decision-making process. No one wants to remove equipment from
I N P U T GAS L E V E L S
F R O M G A S , OIL MONITOR
O R G A S IN O I L
C O M P A R E TO S A F E LEVEL
OPERATING LEVELS OK —UNIT OK
R E F E R TO 6 . 2 . 2
DETERMINE
G A S S I N G RATES
R E F E R TO 6 . 2 . 2
E S T A B U S H OPEFiATlNC
DETERMINE INDIVIDUAL GAS .5 AD'/lSe MANUFACTURER
PROCEDURE * SAMPUNOfQ- CONCENTRATION IN GAS
INTERVALS S P A C E OR OIL
REFER TO 6 . 3 . 1 ic 0 . 3 . 2
IDENTIFY FAULT TYPE

THERMAL / ELECTRICAL
REFER TO 6 . 3 . 2 . 6 . 3 . 3 OR
6.3.4
il
DETERMINE THERMAL
3:
EXTERNAL SYSTEM
LOAD ANALYSIS
DEPENDENCE DETECTIO^^ EVALUATION
I.R. E T C .
a T A. T
PLAN
OUTAGE AND INTERNAL:£
INSPECTION
REFER TO 6 . 3 A N D M A N U F A C T U R E R
FIG. do1—Flowchart.
(UFRN) Universidade Federal Rio Grande do NorteReferences refer to
((UFRN) Universidade the do
Federal section of do
Rio Grande theNorte)
proposed
pursuant toC57.104.
License Agreement. No further reproductions auth
SAVIO ON TRANSFORMER FAULT GAS ANALYSIS 87
DORNEWBERB METHOD
CH4 C2H2 C2H6 C2H2

H2 C2H4 C2H2 CH4
Using the data from 4 / 7 / 8 7 i n Table 2:

1988 = 2 . 7 32 « 0.01 595 = 18.6 32 =0.02
723 3259 32 1988
Diagnosis: Hot spot - No cellulose
ROGERS METHOD
CH4 C2H6 C2H4 C2H2
H2 CH4 C2H6 C2H4
Using the data from 4/7/87 In Table 2:
1988 = 2 . 7 595 - 0.3 3259 = 5 . 5 32 = 0.01
723 1988 595 3259
Diagnosis: Circulating current core and tank.
KEY GAS METHOD

Using data from 4/7/87 In Table 2:
GAS PERCENT OF TOTAL
H2 11
CH4 30
C2H6 9.6
C2H4 49
C2H2 0.4
Diagnosis: Overheated oil
Condition found when inspected: Core steel touching tank wall.

All methods were in basic agreement.
FIG. 2—Example of diagnostic methods.
service without technical justification. I'm sure that data collection is not the whole answer, but
it is a step in the right direction and can eventually lead to expert systems that are dependable.
References
[I] Sobral, C. L. C , "Correlation Between Results of Dissolved Gas Analysis and Actual Transformer
Inspection," Doble Conference Paper, 1986, Sec. 6-501, The Doble Conference Minutes, Doble, Inc.,
Boston, MA.
Copyright
[2] by ASTM Int'l
Baker, A. (all
E.,rights reserved);
Griffin, P.Thu
J.,May
and6 14:02:13
Locke,EDTC 2021
, "An Update on Fault-Gas Analysis (A Review)," Doble
Conference Paper, 1986, Sec. 10-301, The Doble Conference Minutes, Doble, Inc., Boston, MA.
88 INSULATING OILS
[3] Dind, J. E. and Regis, J., "Preventative Electric Maintenance," Pulp and Paper Canada, Vol. 76, No.
9, September 1975, pp. 61-64.
[4] General Electric Bulletin GET-6552, "Power Transformer Dissolved Gas Analysis," General Electric,
Pittsfield, MA.
[5] ANSI/IEEE C57-107 1978, "IEEE Guide for the Detection and Determination of Generated Gases in
Oil-Immersed Transformers and their Relation to the Serviceability of the Equipment," under revision.
Paul J. Griffin^
Criteria for tlie Interpretation of Data for

Dissolved Gases in Oil from
Transformers (A Review)
REFERENCE: Griffin, P. J., "Criteria for the Interpretation of Data for Dissolved Gases in Oii
from Transformers (A Re\leyi)," Electrical Insulating Oils, STP 998, H. G. Erdman, Ed., Ameri-
can Society for Testing and Materials, Philadelphia, 1988, pp. 89-106.
ABSTRACT: The analysis of gases dissolved in oil has been used as a diagnostic tool for many
years to determine the condition of transformers. The criteria used in evaluating dissolved gas-in-
oil data are reviewed. The criteria are based on experience from failed transformers, transformers
with incipient faults, laboratory simulations, and statistical studies. The interpretation of results
is enhanced by including specific information on a particular transformer, including its history.
Comparison of a particular transformer of concern with a population of similar transformers pro-
vides additional information as regards "normal" gassing behavior. The interpretation of dis-
solved gas-in-oil data is not a simple matter and requires the integration of numerous criteria.
Diagnostic gases for silicone fluids are also discussed.
KEY WORDS: mineral oil, transformer oil, silicone fluid, dissolved gases, fault gases, combusti-
ble gases, gas solubility, key gases, ratios, overheating, corona, arcing
It has long been recognized that the degradation of transformer insulation results in the for-
mation of by-products, notably characteristic gases. In a sense, the generation of these gases,
quantified by routine methods, is rather fortuitous since the detection and measurement of
abnormal amounts provides a basis for appropriate measures to be undertaken to prevent un-
planned outages and/or catastrophic failures. Although the utility industry uses the dissolved
gas-in-oil test as a routine tool, there is no universally accepted means for interpreting results.
There have been many refinements of existing methods of interpretation; however, no single
"simple" method seems to provide a complete picture under all circumstances. A simple "lit-
mus paper" type of approach with black and white answers is not sufficient for resolving prob-
lems of this complexity. Although general rules can be used, as much specific information as
possible is important in providing the diagnosis on a particular transformer. This paper reviews
some of the important criteria in the interpretation of results of tests for dissolved gases in oil
from transformers.
Historical Information
Dissolved gas-in-oil data by themselves do not always provide sufficient information on which
to evaluate the integrity of a transformer system. The nameplate information and the history of
the transformer in terms of maintenance, loading practice, previous faults, etc. are an integral
part of the information required to make an evaluation. For example:
1. How old is the transformer?
2. Did a bushing fail at some point?
3. Did the transformer fail at some point?
Copyright'Laboratory
by ASTM Int'l (allmanager, Doble
rights reserved); Engineering
Thu May Co.,
6 14:02:13 EDT 2021Watertown, MA 02172.
89
90 INSULATING OILS
4. Is the unit heavily loaded or did the unit have to carry overload for a period of time?
5. Has the unit been repaired after a failure (residual gases)?
6. Have previous tests been performed to determine gas-in-oil results?
7. Has the percent total combustible gas (TCG) in the gas space risen suddenly?
8. Has the percent TCG in the oil risen suddenly?
9. Has the oil been degassed?
As one can see, the answers to these questions may very well temper the evaluation of a particu-
lar transformer.
Oil Pieservation Systems

It has been common practice since the 1950s to design transformers such that the oxygen (air)
concentration can be reduced and maintained at low levels to preserve the oil and cellulose
insulation. Prior to the advent of oil preservation systems, transformers were designed to freely
"breathe out" gases generated and to "breathe in" fresh air (oxygen). The oil preservation sys-
tems are of three basic types as seen in Fig. 1:
1. Gas blanketed (usually nitrogen).

2. Conservator (sealed with a bladder).
3. Conservator (opened to atmosphere).
The gas-blanketed transformer can be either a sealed unit or a pressurized system containing
a replenishing bottle. Overpressurization is prevented through the use of a pressure release
valve. The gas blanket type of system provides a cushion for expansion and contraction of the
oil. Gases partition between the gas blanket and oil in accordance to Henry's law, which essen-
tially says that the concentration of gas in the gas phase is directly proportional to its concentra-
tion in the liquid phase. The Ostwald coefficient, the volume of gas dissolved in a unit volume of
oil at equilibrium at a specified partial pressure and temperature, is used to calculate the con-
centration of gas in one phase when the concentration in the other phase is known. Table 1 gives
the Ostwald coefficients for an oil with a density of 0.880, a temperature of 25°C, and pressure
of 1 atm for nine characteristic gases [/].
When the concentration of gas in the oil is known, an estimate of the percent concentration of
each gas in the gas blanket is given by:
ppm gas in oil

Ostwald coefficient
X 100% = % each gas in gas space
., ppm gas in oil
Ostwald coefficient
The gases H2, N2, O2, and CO have relatively low solubilities in oil, with CH4 somewhat
higher. The gases CO2, C2H2, C2H4, and C2H6 have high solubilities in oil. An example of how
this affects the partitioning of gases between the two phases is given in Table 2.
The majority of the gas is typically nitrogen; however, for demonstration purposes an approx-
imate concentration of O2, as would be found in air, is included. As can be seen, the O2/N2 ratio
increases in the oil due to the higher solubility of O2 in oil. Of the combustible gases, H2 and CO
preferentially partition into the gas space and, therefore, are often the major components. Hy-
drocarbons, most importantly acetylene, tend to remain in the oil. These estimated values are
based on certain equilibrium conditions and therefore are only rough guidelines.
The solubility of gases in oil varies with temperature and pressure. H2, N2, CO, and O2 in-
crease
Copyright in solubility
by ASTM with temperature,
Int'l (all rights reserved); and2021
Thu May 6 14:02:13 EDT CO2, C2H2, C2H4, and C2H6 decrease in solubility
with temperature, while CH4 remains essentially unchanged
(UFRN) Universidade Federal do Rio Grande do Norte ((UFRN) Universidade Federal [2-4].
do Rio Grande do All the
Norte) pursuant gasesAgreement.
to License are proportion-
GRIFFIN ON TRANSFORMERS 91
O
H
u
w
z
o
o
z
u
a<
o
i
U
w
ooz
a
u
M
g
E-i
U
Ixj
(/)
CQ < U
O o
W
^
\\- , rf
^ N
Z
92 INSULATING OILS
TABLE 1—Ostwald solubility coefficients.
Ostwald
Gas Coefficient
O2 0.138
N2 0.0745
CO2 0.900
CO 0.102
H2 0.0429
CH4 0.337
C2H, 1.99
C2H4 1.35
C2H2 0.938
ally more soluble in oil with increasing pressure [3]. For example, doubling the pressure dou-
bles the gas concentration in the oil. The solubility of the gases also modestly increases as the
density of the oil decreases [1,3].
In conservator systems there is no gas blanket in the transformer main tank and the entire
transformer is filled with oil. An expansion chamber permits oil volume changes due to temper-
ature. The expansion chamber can be closed by a bladder type of arrangement which permits
changes in oil volume while restricting contact with the air, or can be open in contact with the
air. The closed system essentially results in complete solution of the gases into the oil except in
the case where gas bubbles are formed and collect in a special device, such as a Buchholz relay,
for such conditions. A Buchholz relay (positioned high on the transformer between the main
tank and conservator tank) collects gas that comes out of solution and allows a floating con-
tact(s) to move. If the float moves sufficiently it closes a contact and causes an alarm. A second
contact can initiate breaker tripping if sufficient gas is present. The open type conservator sys-
tem allows for a continuous exchange of gases at the air/oil interface of the expansion chamber;
however, access is somewhat restricted. In the case of the completely closed conservator system,
hydrogen, carbon monoxide, and to a lesser extent methane will be proportionally higher than
in the gas-blanketed-type system. In the open system these same gases will preferentially de-
crease in concentration as they are slowly lost to the atmosphere through the expansion cham-
ber.
TABLE 2—Estimation of gas composition in the gas space.
Estimated Estimated
Gas in Oil, ppm Gas in % Gas in
Gas ppm (Vol./Vol.) Gas Space' Gas Space
O2 28 400 206 000 20.6

N2 59 000 792 000 79.2
CO2 1 000 1 110 0.11
CO" 100 980 0.10
H2" 100 2 330 0.23
ca," 100 297 0.03
C2H," 100 50 0.01
Cjft," 100 74 0.01
C2H2" 100 107 0.01
Copyright by ASTM Int'l (all rights"Combustible

reserved); Thu Maygases.
6 14:02:13 EDT 2021
Downloaded/printed by 'Estimated value under equilibrium conditions at 25°C and 1 atm.
Homogeneity of Gases in Oil

The homogeneity of the gas in the oil is dependent upon factors such as rate of gas genera-
tion, access of the fault area to flowing oil, rate of oil mixing, presence of a gas blanket, and, to
a small degree, diffusion. In the vast majority of the cases, incipient fault gases are generated in
a specific area in the oil phase and then transferred throughout the system. If gas bubbles are
generated, the gases will immediately begin to partition between the gas bubble and the oil
according to their solubilities, with the more oil-soluble gases going in the oil. Hydro-Quebec
Institute of Research has developed a means for estimating the depth of a fault in oil by compar-
ing the proportions of H2, CH4, and C2H2 present in the gas in the Buchholz relay and in the oil
immediately following an alarm [5]. The further the distance the bubble travels in the oil, the
greater the differences in the relative composition of the gases in the relay and the oil. Some-
times an incipient fault can also be further located by comparing combustible gas levels for oil
samples taken at various levels in the transformer, that is, the greater concentration of gases is
nearest the source of the fault.
Gas Bubbles
The formation of gas bubbles in power transformer systems is a serious and complex prob-
lem. Gas bubbles can impair the dielectric integrity of the transformer system with severe conse-
quences. Three mechanisms have been postulated concerning their formation [6]. These are:
1. Supersaturation of the oil with gas.
2. Thermal decomposition of the cellulose insulation.
3. Vaporization of adsorbed moisture in the cellulose.
In a pressurized transformer which is rapidly cooled after being at a high temperature, there
is a large pressure drop due to the reduction in oil volume. The lower pressure and temperature
results in excess gas being present, and saturation or supersaturation of the oil with the blanket-
ing gas occurs. Consequently, gas bubbles may be released if initiated by mechanical vibration,
oil pump operation, or sufficient electrical stress [7-9]. The other known causes of bubble for-
mation, generation of CO and CO2 from the degradation of the cellulose insulation, and vapor-
ization of moisture, are subject to higher temperatures associated with transformer overloading
(140°C winding hot-spot temperature) [6,10,11]. It also appears that the rate of temperature
rise and the percentage of moisture in the paper insulation may be important factors [11].
Nitrogen and Oxygen

Nitrogen and oxygen are not typically considered diagnostic gases per se, but nevertheless are
useful to assess the presence of leaks, overpressurization, or changes in pressure and tempera-
ture on the transformer system.
Oxygen is one of the key promoters of oil and cellulose oxidation in transformer systems and,
therefore, is generally maintained at low concentrations. Increases in oxygen suggest a leak in
the system. However, a determination of high oxygen may also be the result of a leak during
sampling and testing of an oil and, therefore, should be confirmed. A rapid decrease in oxygen
with changes in oil properties and the generation of other gases may be indicative of overheat-
ing.
Changes in the nitrogen content can be due to changes in the temperature of the transformer
in gas-blanketed systems. As the insulating oil heats and cools, there is a corresponding in-
crease and decrease in the oil volume and compression and relaxation of the gas space. The
increase
Copyright by ASTMand decrease
Int'l (all in Thu
rights reserved); pressure result
May 6 14:02:13 EDTin
2021raising and lowering the dissolved N2 (and other
gases) content
Downloaded/printed by in the oil. Excessive N2 can lead to bubble formation, as previously discussed.
94 INSULATING OILS
Combustible Gases and Carbon Oxides

Combustible gases and carbon oxides are the characteristic diagnostic gases generated from
incipient fault conditions. As expected, the normal aging of transformers also results in the
generation of these gases. Excessive generation of these gases due to abnormal conditions or
accelerated deterioration of the insulating materials is determined by comparing individual test
results to empirical norms for a population of transformers, total combustible gases, key gases,
ratios, trends, and individual "fingerprints."
Norms
A means of establishing if a transformer with no measured previous dissolved gas history is
behaving normally is to compare it with the gassing characteristics exhibited by the majority of
similar transformers or a "normal population." As the transformer ages and gases are gener-
ated, the normal levels for 90% of a typical transformer population can be determined. From
these values, and based on experience, acceptable limits have been determined such as those
given in Table 3 [/2].
Considerable differences are apparent for what is considered a normal transformer with ac-
ceptable concentrations of gases. Some of the differences seen here may be due to differences in
the transformer populations studied; that is, gas blanketed versus conservator, generation ver-
sus transmission, age of transformer, etc. Although not listed here, H2 and C2H2 limits are
based on the main tank and load tap changer having separate compartments. If the oil is com-
mon to both compartments, then revised H2 and C2H2 values are required.
Norms are now being developed for factory tests on new power transformers. Typical results
from three sources for factory tests for dissolved gas in oil performed after the heat run are given
mJah\e A [13,14].
Heat run data are not always reported for the same time interval, and, therefore, comparison
of the different levels here is not possible; however, in general, results are typically low on a new
transformer.
By definition most transformers, when tested for gases dissolved in oil for the first time, will
be "normal" when compared to any of the acceptance limit values. When an abnormal situa-
tion is indicated, a testing schedule is devised with increased sampling frequency to determine a
trend. Only in very rare severe cases would an investigation be required based on a first-time
analysis.
TABLE 3—Dissolved gas acceptable limits—various sources.
Gas, ppm (Vol./Vol.)
H2 CO CH4 C2H6 C2H4 C2H2 CO2 TCG
Doble 100 250 100 60 100 5 610

IEEE"
Generator 140 580 160 115 190 11 1296
Transmission 100 350 120 65 30 35 700
Electra (CIGRE)» 28.6 289 42.2 85.6 74.6 3771 520
Manufacturer 200 500 100 100 150 15 1065
(250) (1000) (200) (200) (300) (35) 1985
"In the process of being revised.

'Corrected values 1978.
(UFRN) (Universidade
) = ValuesFederal6 do
toRio
7 Grande
years.do Norte ((UFRN) Universidade Federal do Rio Grande do Norte) pursuant to License Agreement. No further reproductions auth
TABLE 4—Factory dissolved gas-in-oil tests after heat

runs, typical values."
Gas 1 2 3"
H2 5 5 30
O2
N2
CO 10 75 75
CO2 100 400 400
CH4 5 5 5
C2H, 5 5 5
C2H1 5 5 5
C2H2 0 None None
detected
"Results reported in ppm (vol./vol.).

'Upper limits for 90% of units tested for 24-h heat run.
Total Combustible Gas (TCG) Limits

On a first-time sample, the severity of an incipient fault can be further evaluated by the TCG
present. Limits for TCG, modified since originally published in 1973 [15], are as follows:
Combustible Gas Limits, ppm

0-500
Unless individual gas-acceptance values are exceeded, TCG in this range indicates
the transformer is operating satisfactorily.
501-1500 Indicates some decomposition of the transformer insulation has occurred and
action should be taken to establish the trend to determine if an incipient fault is
likely to evolve as a problem.
1501-2500 A high level of decomposition of the transformer insulation has occurred and a
fault may be present. Immediate action should be taken to establish the trend to
determine if the fault is progressively becoming worse.
> 2500 Substantial decomposition may have occurred to the integrity of the transformer
insulation and therefore could result in a failure. The gassing rate and the cause of
gassing should be identified and appropriate corrective action should be taken.
Key Gases
Combined, the individual gas acceptance values and total combustible gas, while informa-
tive, do not provide sufficient information to assess a transformer's condition, even for a first-
time analysis. A further requirement is the identification of the type of fault. Characteristic
"key gases" have been used to identify particular fault types. Laboratory simulations, and com-
parison of results of dissolved gas-in-oil tests combined with observations from the teardown of
failed transformers, have permitted the development of a diagnostic scheme of the characteris-
tic gases generated from the thermal and electrical (corona and arcing) deterioration of electri-
cal insulation [16], Figure 2 lists the gases for the conditions of arcing, corona, overheating in
oil, and overheating in paper in the order of decreasing severity.
For each type of condition, the level of concern at each total combustible gas level would vary.
In the case of electrical problems, erratic and rapid deterioration of the insulation is more
likely, and (particularly in the case of arcing) either high or very low levels of gases can be
generated before failure occurs. In general, the overheating condition is considered the least
serious
Copyright because,
by ASTM in reserved);
Int'l (all rights most cases,
Thu Maydeterioration
6 14:02:13 EDT 2021of the insulation is slow and relatively high levels of
combustible gases are generated as by-products.
(UFRN) Universidade Federal do Rio Grande do Norte ((UFRN) This
Universidade Federal permits
do Rio ample
Grande do Norte) timeto License
pursuant to plan and carry
Agreement. out
96 INSULATING OILS
1. ARCING ARCING M OIL

too
Large amounts of hydrogen and acetylene are 90
produced, with minor quantities of methane 80
and ethylene. Carbon dioxide and carbon 70 60
60
monoxide may also be formed if the fault
so
involves cellulose. The oil may be carbonized. 40 30
Key gas—Acetylene 30
20
10
0
<0.01 1.6
Jd.
CO H, CH. C,H.
2. CORONA CORONA IN OH
100
Low-energy electrical discharges produce 90 86
80
hydrogen and methane, with small quantities
70
of ethane and ethylene. Comparable amounts 60
of carbon monoxide and dioxide may result 50
from discharges in cellulose. 40
Key gas—Hydrogen 30
70 13
10
0.2 0.5 0.2 0.1
0
CO H; CH, C2H, C2H4 C2H2
3. OVERHEATED OIL OVERHEAHO OIL

100
Decomposition products include ethylene and 80
80
methane, together with smaller quantities of
70
hydrogen and ethane. Traces of acetylene may 63
60
be formed if the fault is severe or involves so
electrical contacts. 40
Key gas—Ethylene 30
20 16 17
10
0
<0.01
CO H;
J_L
CH4 CjH, C2H4 CJHj
4. OVERHEATED CELLULOSE 100

OVERHEATED CEUUIOSE
92
Large quantities of carbon dioxide and carbon 80
80
monoxide are evolved from overheated 70
cellulose. Hydrocarbon gases, such as 60
methane and ethylene, will be formed if the so
fault involves an oil-impregnated structure. 40
Key gas—Carbon Monoxide 30
20
10 6.7
1.2<0.01<p.0l<0.01
0
CO H, CH, C2M, CjNf C2H2
FIG. 2—Key gases.

remedial action. It should be noted that the relative concentrations of the gases given in Fig. 2
are only guidelines and variances do occur. In Table 5 there are three examples showing the
difference in the concentration of characteristic gases for each condition which resulted in reme-
dial action by the owners of the transformers [17-19].
Transformer 1 involved a case where low energy arcing was occurring [IT]. Levels of combus-
tible gases were stable for about 13 months at the level shown in Table 5, except for hydrogen,
which showed a level of 150 ppm in one test. Test results on a sample seven months later again
showed an increased level of H2 (90 ppm) and an increasing level of C2H2 (78 ppm). The unit was
removed from service and investigated. A piece of sheet steel which was found in an insulating
phase barrier was apparently arcing to the transformer frame. Serious damage could have oc-
curred if this problem had not been identified and corrected.
Transformer 2 showed an increase in H2 in the oil from 1700 to 2300 ppm over a period of
seven months before it was removed from service and repaired for loose connections [ 18].
Transformer 3 had an overheating condition due to a recurring core ground problem which
was eventually corrected by burning it away [19].
Several refinements to the key gas method of identifying incipient faults can be seen for
arcing and overheating conditions. When arcing of low energy or of short duration occurs, there
will be less overheating of the oil and, hence, less C2H4, CH4, and C2H6 formed. Even though the
arc is of a sufficient hot-spot temperature to yield C2H2, cooler transition zones about the arc
will generate C2H4, CH4, and C2H6 if sufficient time or energy is available to effect a large por-
tion of the oil.
Overheating of oil can be further characterized based on thermodynamic considerations [20].
As temperature increases, the concentration of gases increases as follows:
Low Temperature ( s 120°C) CH4

C2H6
C2H4
High Temperature (>700°C) C2H2
At specific temperatures, one or two of these gases predominate while lesser quantities of the
other gases are formed. In a practical situation, two scenarios are likely to occur which obscure
this clear temperature discrimination. In one case, a thermal condition deteriorates over time.
TABLE 5—Transformers with incipient faults.
Transformer
Gas,
ppm (Vol./Vol.) 1 2 3
H2 0 1 700 540
O2 1 100 3 000 2 300
N2 79 000 110 000 87 000
ca, 9 43 1 300
CO 33 440 420
C2H, 7 6 160
CO2 510 8 400 2 000
C2H4 8 2 810
C2H2 9 0 2
TGC" 80 676 123 591 94 532
TCG* 57 2 191 3 232
"Total
Copyright by ASTM Int'l (all rights reserved); Thugas
May content.
6 14:02:13 EDT 2021
Downloaded/printed by *Total combustible gas.
98 INSULATING OILS
resulting in a changing spectrum of gases as the temperature increases. In the other case, a
large thermal gradient exists, giving rise to a wide spectrum of gases.
It should also be noted that other factors, such as high O2, can result in additional gases
being generated, that is, CO and CO2.
Table 6 shows four cases of transformers exhibiting evidence of overheating in oil by the
characteristic hydrocarbon gases (in decreasing order of severity).
In Case 1, the major gases are CH4, indicating low temperature overheating in oil, and C2H4
along with traces of C2H2, indicating very high temperature. This transformer had a core
ground problem which generated gases at a rapid rate, suggesting that high temperature over-
heating (hot spot) was occurring and a thermal gradient existed with a large area of more low-
to-moderate overheating also occurring.
In Case 2, the predominant gas is C2H6. This is unusual in that in most cases where C2H6 or
CH4 is the predominant gas (with C2H4 a small proportion) modest gassing rates are evident.
Gas generation due to heating follows Arrhenius-type kinetics [16] and, therefore, at low tem-
peratures the characteristic gases should be generated in relatively small quantities, although
varying with the size of the incipient fault area. In Case 2 there are a number of transformers
from this same manufacturer which show the same condition which appears to stabilize after an
initial "break-in" period. The gas content in this unit has not changed in three years. Cases 3
and 4 indicate lower temperature overheating (predominate hydrocarbon gas is CiHe and CH4,
respectively) and, as would be typical, lower total gaseous hydrocarbons.
Ratios
Several analytical schemes have been developed which employ and compare ratios of charac-
teristic gases generated under incipient fault conditions. The methods are used to determine the
type of incipient fault condition similar to the key gas method. The advantages to the ratio
methods are that they are quantitative, independent of transformer oil volume, and can be
computer programmed. The disadvantages are that they may not always yield an analysis or
may yield an incorrect one and, therefore, should only be used as an additional guideline to
other diagnostic methods such as key gas methods. In addition, ratio methods require mini-
mum levels of gases to be present before they are reliable and better precision on the part of the
analysts performing the dissolved gas-in-oil test. Several papers have discussed problems with
ratio methods [21-22].
TABLE 6—Transformers exhibiting overheating of oil.
Transformer
Gas,
ppm (Vol./Vol.) 1 2 3 4
H2 540 1 16 110
O2 2 300 7 200 3 700 3 100
N2 87 000 110 000 100 000 82 000
CH4 1 300 69 290 110
CO 420 400 240 140
CjH* 160 2 300 480 39
CO2 2000 6 800 4 400 1 500
C2H4 810 180 33 8
C2H2 2 0 0 0
TGC 94 532 126 950 109 159 87 007
TCG* 3 232 2 950 1 059 407
Downloaded/printed by "Total gas content.
(UFRN) Universidade Federal'Total combustible
do Rio Grande gas. Universidade Federal do Rio Grande do Norte) pursuant to License Agreement. No further reproductions auth
do Norte ((UFRN)
Two ratio methods most commonly used are those of Dornenburg [23] and Rogers [24-26].
In both methods the same gases are used, although some of the ratios are different, as can be
seen in Figure 3. In both cases involvement of solid insulation (CO, CO2) is handled separately.
These methods and ratio methods in general are to some extent dependent upon the distribu-
tion of the gases in the transformer system; that is, there are differences for conservator and gas-
blanketed units.
Both Dornenburg's and Rogers' ratio methods were devised from data from conservator-type
transformers with expansion chambers exposed to the atmosphere. This type of transformer
experiences a slow loss of the less soluble gases (due to restricted oil movement between the
main tank and the expansion chamber) at the oil/air interface. Oommen has determined cor-
rection factors for the ratios for conversion between sealed conservator and gas-blanketed trans-
formers based on the solubility of the gases in oil [3] (calculation of the rate of loss of gases from
the open expansion tank conservator is not a constant and, therefore, difficult to calculate). The
correction factor is usually only significant for the less soluble gases, that is, CH4/H2 ratio.
Dornenburg developed ratios for gases dissolved in oil and gases in the Buchholz relay with the
relation between the two based on the solubility of each gas in oil.
Although the Dornenburg's and Rogers' ratios yield similar interpretation of dissolved gas-
in-oil data, there is a difference in the complexity of the schemes. The Dornenburg method,
used when prescribed normal levels of gassing are exceeded, provides a simple scheme for dis-
tinguishing between overheating, partial discharge, and other types of discharge.
The Rogers' ratio method is a more comprehensive scheme which details temperature ranges
for overheating conditions (based on Halstead's research [20]) and some distinction of the se-
verity of incipient electrical fault conditions. A normal condition is also listed. Minimum gas
limits for using the Rogers method are very low (10 ppm H2, 2 ppm hydrocarbons [24]). It is this
author's opinion that the use of low minimum limits can result in erroneous conclusions and
that limits such as those used by Dornenburg are more appropriate.
Several examples may illustrate some of the advantages and disadvantages of using ratio
methods. If we examine the data given in Table 6 for the various overheating conditions, we
arrive at the following diagnosis, according to Rogers' code [24,26]:
Transformer 1: Core and tank circulating currents, overheated joints.
Transformer 2: Overheating 150 to 200°C.
ROGERS DORNENBURG
CH4 CH4
C2H6 C2H6
CH4 C2H2
C2H4 C2H2
C2H6 CH4
C2H2 C2H2
C2H4 C2H4
(UFRN) Universidade Federal do Rio Grande do Norte ((UFRN) FIG. 3—Ratio
Universidade methods.
Federal do Rio Grande do Norte) pursuant to License Agreement. No further reproductions auth
100 INSULATING OILS
Transformer 3: Overheating 150 to 200°C.

Transformer 4: Slight overheating—below 150°C.
As can be seen, this offers a refinement to the key gas method.
In contrast, Table 7 shows two cases published by Vieira [27] where both Rogers' and
Dornenburg's ratio methods did not provide a diagnosis, even though it is obvious a very serious
incipient fault condition existed.
Trend Analysis
When a possible incipient fault condition is identified by the first time analysis, the gassing
trend should be determined. Subsequent analysis provides information as to which gases are
currently being generated and the rate of generation of these gases. Even when subsequent
analyses are of a routine nature, they provide the baseline from which to judge future occur-
rences. In the examination of trends, key gases, total combustible gas, rate of gas generation,
and "fingerprints" (of normal trends) of particular types of transformers are considered.
As in the case of a first-time sample, key gases are used to determine the type of incipient
fault condition with subsequent samples, with the difference being that only those gases actively
being generated are used in the diagnosis of subsequent samples. If no significant change in gas
concentration is observed for a reasonable period of time, even though values exceed norms, or
ratios indicate that a problem exists, no problem may be indicated unless excessive insulation
has deteriorated (which usually results in more gassing). Ratio methods may be used to aid in
the analysis, but, depending upon the history of the transformer, they may not be able to distin-
guish between past and present conditions. For example, a transformer which has been de-
gassed after a fault will show an altered gas pattern as the less-soluble gases are preferentially
removed. Also, test results for a transformer with more than one type of fault may be difficult to
interpret using ratio methods. Again, the total combustible gas content (in oil) is used to judge
the relative deterioration of the insulation. It is possible that an apparent incipient fault condi-
tion will generate gases away from insulating structures and cause no harm to the transformer
insulation system. However, such a condition may also mask the effects of a more serious condi-
tion developing at the same time.
TABLE 7—Gas-in-oil test results.
Gas,
ppm(Vol./Vol.) Case 3 Case?
H: 650 76
O: 2 900 3 500
Nj 53 000 93 000
Cft, 81 6 000
CO 380 36
CaH<, 170 27 000
CO2 2 000 4 400
C2R, 51 120 000
C2H2 270 1 700
TGC 59 502 255 712
TCG* 1 602 154 812
NOTE: Actual Inspection. Case 3: Electrical

problem caused by poor shield contact. Case 7:
Metal particles, carbon located on core; burning
and erosion by electric current noted.
Copyright by ASTM Int'l (all rights reserved); Thu"Total gas content.
May 6 14:02:13 EDT 2021
Downloaded/printed by 'Total combustible gas.
The rate of gas generation which can be load-dependent provides information as to the sever-
ity of the condition and offers guidelines as to whether or not a unit should be removed from
service. Two means for assessing the gassing rate are to use the change in concentration of gas in
parts per million or to determine the actual amount of gas generated. General guidelines that
have been used for removing transformers from service are 100 ppm/day [14] and 0.1 ft^ (0.003
m^) gas per day [28,29]. Cubic feet can be converted to ppm using:
(gas in ft^) (7.48) (10*)

ppm =
gallons of oil
If results are in cubic metres they can be converted to cubic feet by multiplying cubic metres
times 35.3. Thus, in 10 000 gallons of oil, 0.1 ft^ (0.003 m'') gas per day is equivalent to 75 ppm/
day. When the nature of an incipient fault is known, the use of one of the limits may be pre-
ferred. It may be more appropriate to use the absolute rate of gas generation for situations
involving localized conditions (where dilution of the gas into different volumes of oil would alter
relative ppm values).
Conversely, the ppm limits may be more appropriate for use when nonlocalized conditions
are involved, such as general conductor overheating. A further refinement by McNuttetal. [//]
in determination of cellulose degradation takes into account surface area and temperature in-
volved. These guidelines for gassing rates are general "rules of thumb" and considerations such
as type of fault, previous history of the particular type of transformer, etc. must always be taken
into account.
Fingerprints
Transformers from the same manufacturer and of the same type exhibit what might be called
"normal trends," "characteristic trends," or "fingerprints" [12]. As a new transformer
"breaks in," some gas evolution is expected and may occur at an initially high rate which subse-
quently slows down or plateaus. Specific manufacturer's transformer designs may show levels
and types of gases which are characteristic and do not represent an incipient fault condition.
Tables 8A, 8B, and 8C show what may be considered "fingerprints" of three different trans-
former designs. In each case two examples with initial test values and those after two or three
years are given.
Table 8A gives two examples of typical values from a data base of 50 to 100 transformers
which can be considered as representative of the group.
It is obvious that one can single out the predominant "key" gases, hydrogen and carbon
monoxide, and by cursory inspection one might even conclude that a possible problem exists.
TABLE 8A—Transformer fingerprints.
Example 1 Example 2
Gas,
ppm (Vol./Vol.) I 3 I 3
H2 350 260 110 210

CH, 44 61 11 13
CO 670 650 520 630
CjH, 26 25 3 4
CO2 3000 1900 5000 3900
C2H4 9 5 8 10
C2H2
NOTE: I = initial.
102 INSULATING OILS
TABLE 8B—Transformer fingerprints.

Example 1 Example 2
H2 20 12 0 0
ca, 55 16 60 29
CO 2 9 0 0
C2H, 250 83 61 46
CO2 1900 1500 650 250
C2H4 500 620 200 150
C2H2 1 Trace 0 0
NOTE: I = initial.
TABLE 8C—Transformer fingerprints.

Example 1 Example 2
Gas,
H2 0 1 0 0
CH4 92 69 15 18
CO 370 400 33 57
C2H0 2300 2300 560 520
CO2 6000 6800 1800 2200
C2ll| 180 180 9 6
C2H2 0 0 0 0
NOTE: I = initial.
However, it should be noted that in each of the two data sets, the first row of values represents
the initial results obtained on the "new" transformer while the second row is the corresponding
set of gas values three years later. Consequently, one must reconsider since the gas values three
years later are nearly the same. It appears that: (1) there apparently is no problem because the
units are still in service; (2) the distribution of gases represents an equilibrium condition since
the values over three years are nearly constant; and (3) the predominance of hydrogen and car-
bon monoxide is a unique condition. As a matter of fact, these transformers are from the same
manufacturer and the gas data appear to be the normal attributes of a particular class of distri-
bution transformers.
In Table 8B, another two sets of dissolved gas results are presented in which the predominant
gases are ethane and ethylene. Again, as in the previous table, the first row shows the initial
values and the second row shows the respective gas values two years later.
Considering the predominant gases in this case, ethylene and ethane, along with a detectable
amount of acetylene appearing also in the first group, one might again entertain the thought of
a possible problem, at least in the first group.
Since the predominant (key) gases are relatively constant over a two-year period, it appears as
in the previous situation that the dissolved gas compositions represent an equilibrium condition
which constitutes a "norm" for these transformers. Actually, the equipment represented by this
gas data also is from a single source, but different from those depicted in Table 8A. The first
group with the greater ethylene content applies to generator step-up transformers and the sec-
ond group to substation units.
Table 8C provides another example where somewhat higher gas levels are apparent, but the
conclusions
(UFRN) are do
Universidade Federal the
Riosame.
Grande do Norte ((UFRN) Universidade Federal do Rio Grande do Norte) pursuant to License Agreement. No further reproductions auth
Transformer "fingerprints" are a means of establishing normal trends for transformers

based on examination of many transformers of the same manufacturer and design.
Transformer dissolved gas "fingerprints" may not always be as distinct or revealing, for con-
sideration must be given to the modifying influences of service conditions, for example, loads
and temperature, in addition to rating and venting. Fingerprinting obviously is enhanced when
service conditions are more uniform.
CO/CO,
Cellulose insulation yields mostly CO and CO2 when subjected to conditions which result in
its deterioration. Norms, generation rates, and the ratio of CO to CO2, or CO2 to CO, have been
used as an indicator of abnormal cellulose degradation. Normal values of CO and CO2 are given
in Table 3. McNutt et al. have developed a correlation between CO gassing rate and tempera-
ture and surface area involved [11], Rogers [24] determined the average rate of CO formation in
a population of generation transformers as:
ppm ( y / V) = 374 Logio 4 Y (where Y = age in years)
The use of CO/CO2 ratios suffers from the presence of "accidental" CO2 from air and from
variations in O2 concentration present in the transformer incipient fault area (high O2 favors
formation of CO2). Also, oil, if sufficiently heated in the presence of excess oxygen, will also
generate some CO and CO2. Experiments by Vitols and Fernandes [30] show that ratios of CO
to CO2 increase with increasing temperature and that, based on population studies, "normal
ratios" range from 0.07 to 0.30. Rogers [24] found that the average CO2/CO ratio on 1000
generator transformers was 7/1 (CO/CO2 = 0.14) with a standard deviation of 4. Extreme
ratios of CO/CO2, approaching unity or greater than 1/15, should be considered as unusual
behavior, with the unity ratio being of more serious concern as high temperatures/electrical
faults may be involved. For example, in a case given by Vieira [27], where severe arcing is
apparent (high C2H2—5390 ppm) and cellulose appears to be involved (high CO—1130 ppm),
the ratio of CO to CO2 was 1130/285 or 4 / 1 .
Water
Much effort goes into reducing the water content and maintaining low concentrations of wa-
ter in the transformer insulation due to its obvious deleterious effects. However, in some cases
where leaks are present or large operating temperature swings occur causing supersaturation
and formation of free water, water is converted in the presence of core steel and heat to H2 and
O2 [31]. The electrochemical breakdown of free water can also occur. Wet cellulose under co-
rona discharge condition will yield more H2 than dry cellulose [32]. It is important to monitor
water concentration to distinguish the cause of H2 formation which may be due to the presence
of free water or corona activity (dissolved water only).
Other Materials
Power transformers are mostly composed of insulating oil, cellulose paper and pressboard,
conductor, and core steel. Some other constituents used in lesser percentages, such as silicone
compounds, organic polymers, etc. may also be subject to degradation under severe thermal
and electrical discharge conditions. In most cases, these other materials under arcing or pyroly-
sis conditions
Copyright generate
by ASTM Int'l (all predominantly
rights reserved); Thu May 6 14:02:13H2
EDTand
2021 CO, although hydrocarbons, CO2, and water may
(UFRN) Universidade Federal do Rio Grande do Norte ((UFRN) Universidade Federalconditions
also be generated in some cases or under certain do Rio Grande do[33,34].
Norte) pursuant to License Agreement. No further reproductions auth
104 INSULATING OILS
Other Flaids
In special applications alternative fluids such as silicone fluids are used instead of mineral
oils. As these fluids are used, diagnostic schemes need to be developed from population studies,
laboratory simulations, and examination of failed transformers, as has been done for mineral
oils. Laboratory simulations on silicone fluids have revealed the diagnostic scheme given in Ta-
ble 9 [35].
As would be expected, silicone transformer fluid (polydimethyl siloxane) yields lesser quanti-
ties of C2S than mineral oils under the same conditions, although the total combustible gases
generated are of the same order.
It should be noted that air is more soluble in silicone fluids than mineral oils and, with high
oxygen present with overheating conditions not involving the conductor, the predominant gas
generated is CO with lesser quantities of CH4 and C2H2 found. It should still be possible in most
cases to distinguish between overheating in silicone versus overheating in paper because the
formation of CH4 and C2H4 occurs in significant quantities only in the case of overheating in
silicone. Also, in the case of a sustained arcing, silicone fluids generate a small percentage of
C2H2 relative to the other combustible gases and, therefore, the presence of any acetylene in a
silicone transformer should be considered as very serious. The percent of C2H2 in the total com-
bustible gas in mineral oil and silicone fluid under arcing conditions is about 50% and 3%,
respectively.
Little information is known about field experience with silicone transformers as they have not
been in service in large numbers in the United States until recent years. As service experience is
gained, norms can be developed for silicone transformers as has been done for mineral oil units.
Conclusions
Dissolved gas-in-oil analysis has been a useful tool for detecting problems in power trans-
formers which are not detectable by other means or in confirming problems indicated by other
tests. A comprehensive scheme is required for diagnostic purposes due to the many variables
which determine or affect the types and quantities of gases generated. (Prudent judgement is
also an integral part of the final analysis, particularly when incipient faults are indicated.) His-
torical and nameplate data and the type of transformer oil preservation system supply impor-
tant information used in diagnosing transformer health. Characteristic combustible gases gen-
erated by incipient faults are determined and used in interpretive schemes consisting of norms
(acceptance levels), total combustible gas levels, key gases, ratios, gassing rates, trends, and
"fingerprints."
Norms, total combustible gas, key gases, and ratios (for ancillary information) are used for
the analysis of first-time samples to establish the future sample frequency. Subsequent analysis
incorporates total gas, key gases, ratios, trends, gassing rates, and "fingerprints" to diagnose
transformer health.
The future diagnostic schemes for dissolved gas-in-oil analysis should include more definition
of norms for individual transformer manufacturers' designs ("fingerprints") based on large
TABLE 9—Diagnostic gases dissolved in silicone fluid.

Key Gases (Listed in Order of
Condition Highest to Lowest Concentration)
Overheating CH4(CH4, CO, C2H4, and H2 in presence of copper)
CO, CH4, C2H4 in presence of high O2
Corona Hj, C a ,
Arcing
Downloaded/printed by H2, CR,, CO, C2H2
d a t a bases. A comprehensive system such as the use of artificial intelligence (AI), which has
already started [36], may be used to diagnose simple cases ( 9 0 % ) . Further refinement of the AI
system should enable it to diagnose the more complex problems as has been done, for example,
in the medical field.
For computerized methods, a comprehensive system such as AI is needed to handle the com-
plexity of information necessary to diagnose incipient fault conditions and to supply the infor-
mation used in drawing a conclusion when requested.
References
[1] ASTM D 3612, Method for Analysis of Gases Dissolved in Electrical Insulating Oil by Gas Chroma-
tography, Annual Book of ASTM Standards, Vol. 10.03, Electrical Insulating Liquids and Gases:
Electrical Protective Equipment, ASTM, 1916 Race St., Philadelphia, PA, 1986.
[2] Baker, A. E., "Solubility of Gases in Transformer Oil," Minutes of the Forty-Sixth Annual Interna-
tional Conference of Doble Clients, 1979, See. 10-701, Doble Engineering Co., Watertown, MA.
[3] Oommen, T. V., "Adjustments to Gas-in-Oil Analysis Data Due to Gas Distribution Possibilities in
Power Transformers," IEEE Transactions, Vol. PAS 101, No. 6, June 1982, pp. 1716-1722.
[4] ASTM D 2779, Standard Method for Estimation of Solubility of Gases in Petroleum Liquids, Annual
Book of ASTM Standards, Vol. 5.02, Petroleum Products and Lubricants (II), ASTM, 1916 Race St.,
Philadelphia, PA, 1986.
[5] Duval, M. "FauH Gases Formed in Oil-Filled Breathing EHV Power Transformers—The Interpreta-
tion of Gas Analysis Data," IEEE Transactions, PAS C.74 476-8, 1974.
[6] McNutt, W. J. et al., "Mathematical Modelling of Bubble Evolution in Transformers," IEEE Trans-
actions, Vol. PAS 104, No. 2, Feb. 1985, pp. 477-87.
[ 7] Chadwick, A. T. et al., "Oil Preservation Systems; Factors Affecting Ionization in Large Transform-
ers," AIEE Transactions, April 1960, pp. 92-99.
[«] Blanchard, C. S., "Pownal Substation No. 1 Bank Oil Bubbling Problem," ECNE Fall Meeting, 1985,
Electric Council of New England, Bedford, MA.
[9] Kaufmann, G. H., "Gas Bubbles in Distribution Transformers," IEEE Transactions, Vol. PAS 96,
No. 5, Sept./Oct. 1977, pp. 1596-1601.
[10] Heinrichs, F. W., "Bubble Formation in Power Transformer Windings at Overload Temperatures,"
IEEE Transactions, Vol. PAS 98, No. 5, Sept./Oct. 1979, pp. 1576-81.
[//] McNutt, W. J. et al., "Short-Time Failure Mode Considerations Associated with Power Transformer
Overloading,"/£•£:£ Transactions, Vol. PAS 99, No. 3, May/June 1980, pp. 1186-97.
[12] Baker, A. E. et al., "An Update on Fault Gas Analysis (A Review)," Minutes of the Fifty-Third An-
nual International Conference of Doble Clients, 1986, Sec. 10-301, Doble Engineering, Co., Water-
town, MA.
[13] Doble Client Committee on Insulating Fluids, "Combustible Gas Content in the Oil of New Trans-
formers Prior to Initial Energization," Minutes of the Fiftieth Annual International Conference of
Doble Clients, 1983, Sec. 10-701, Doble Engineering Co., Watertown, MA.
[14] Oommen, T. V. et al., "Experience with Gas-in-Oil Analysis Made During Factory Tests on Large
Power Transformers," IEEE Transactions, Vol. PAS 101, No. 5, May 1982, pp. 1048-52.
[ 15] Pugh, D. R., "Combustible Gas Analysis," Minutes of the Fortieth Annual International Conference
of Doble Clients, 1973, Sec. 10-401, Doble Engineering Co., Watertown, MA.
[ 16] Pugh, D. R., "Advances in Fault Diagnosis by Combustible Gas Analysis," Minutes of the Forty-First
Annual International Conference of Doble Clients, 1974, Sec. 10-1201, Doble Engineering Co., Wa-
tertown, MA.
[ 17] Hays, T. W., "Investigation to Determine the Location of a Low-Energy, Audible Electrical Arcing in
a Power Transformer," Minutes of the Fifty-Third Annual International Conference of Doble Clients,
1986, Sec. 6-301, Doble Engineering Co., Watertown, MA.
[ 18] Oms, L. W., "Transformer Problems Detected or Confirmed by Use of Combustible Gas Analysis (A
Progress Report)," Minutes of the Forty-Eighth Annual International Conference of Doble Clients,
[19] Marquez, A., "Recurring Core Ground Found by Combustible-Gas Analysis," Minutes of the Forty-
Fourth Annual International Conference of Doble Clients, 1977, Sec. 6-301, Doble Engineering Co.,
Watertown, MA.
[20] Halstead, W. D., "A Thermodynamic Assessment of the Formation of Gaseous Hydrocarbons in
Faulty Transformers," Jouma/o/tAe/nstifuteo/Petfo/eum, Vol. 59, Sept. 1959, pp. 239-241.
[21] Manger, H. C , "Combustible Gas Ratios and Problems Detected," Minutes of the Forty-Fifth An-
nual International
Downloaded/printed by Conference of Doble Clients, 1978, Sec. 6-1101, Doble Engineering Co., Water-
town, MA.
106 INSULATING OILS
[22] Rickley, A. L. et al., "Analysis Techniques for Fault-Gas Analyses (A Progress Report)," Minutes of
the Forty-Fifth Annual International Conference of Doble Clients, 1978, Sec. 10-401, Doble Engineer-
ing Co., Watertown, MA.
[23] Dornenburg, E. and Strittmatter, W., "Monitoring Oil-Cooled Transformers by Gas Analysis,"
Brown Boveri Review, Vol. 5, No. 74, pp. 238-247.
[24] Rogers, R. R., "U.K. Experience in the Interpretation of Incipient Faults in Power Transformers by
Dissolved Gas-in-Oil Chromatographic Analysis (A Progress Report)," Minutes of the Forty-Second
Annual International Conference of Doble Clients, 1975, Sec. 10-201, Doble Engineering Co., Water-
town, MA.
[25] Rogers, R. R., "U.K. Experience in the Interpretation of Incipient Faults in Power Transformers by
Dissolved Gas-in-Oil Chromatographic Analysis (A Progress Report)," Minutes of the Forty-Fourth
Annual International Conference of Doble Clients, 1977, Sec. 10-50, Doble Engineering Co., Water-
town, MA.
[26] Rogers, R. R., "Concepts Used in the Development of the IEEE and lEC Codes for the Interpretation
of Incipient Faults in Power Transformers by Dissolved Gas-in-Oil Analysis," presented at IEEE 1978
Winter Power Meeting.
[27] Sobral Vieira, C. L. C , "Correlation Between Results of Dissolved Gas Analysis and Actual Trans-
former Inspection," Minutes of the Fifty-Third Annual International Conference of Doble Clients,
[28] Lyke, A. J. and Vitols, A. P., "Automated Monitoring of Dissolved Gas-in-Oil for Large Power Trans-
formers," Minutes of the Forty-Fourth Annual International Conference of Doble Clients, 1977, Sec.
10-601, Doble Engineering Co., Watertown, MA.
[29] MacDonald, J. D. and Vitols, A. P., "Gas-in-Oil Analysis as a Diagnostic Tool for Monitoring Power
Transformer Insulation Integrity," Minutes of the Forty-Seventh Annual International Conference of
Doble Clients, 1980, Sec. 6-901, Doble Engineering Co., Watertown, MA.
[30] Vitols, A. P. and Fernandes, R. A., "Incipient Fault Detection for EHV Transformers," IEEE Sum-
mer Power Meeting, Vancouver, BC, July 15-20, 1979, IEEE, New York.
[31] Sheppard, H. R., "The Mechanism of Gas Generation in Oil-Filled Transformers," Minutes of the
Thirtieth Annual International Conference of Doble Clients, 1963, Sec. 6-601, Doble Engineering
Co., Watertown, MA.
[32] Baker, A. E., "Gas Composition in Corona Discharge," Minutes of the Forty-Ninth Annual Interna-
tional Conference of Doble Clients, 1982, Sec. 10-701, Doble Engineering Co., Watertown, MA.
[33] Pugh, D. R. and Belanger, G., "Pyrolysis of Insulating Materials," Minutes of the Forty-Fourth An-
nual International Conference of Doble Clients, 1977, Sec. 10-901, Doble Engineering Co., Water-
town, MA.
[34] Hettwer, P. F., "Arc-Interruption and Gas-Evolution Characteristics of Common Polymeric Materi-
als," IEEE Transactions, Vol. PAS 101, No. 6, June 1982, pp. 1689-1696.
[35] Griffin, P. J., "Analysis for Combustible Gases in Transformer Silicone Fluids," Minutes of the Fifty-
Second Annual International Conference of Doble Clients, 1985, Sec. 10-701, Doble Engineering Co.,
Watertown, MA.
[36] Lowe, R. I., "Artificial Intelligence Techniques Applied to Transformer Oil Dissolved Gas Analysis,"
Minutes of the Fifty-Second Annual International Conference of Doble Clients, 1985, Sec. 10-601,
Doble Engineering Co., Watertown, MA.
DISCUSSION
C. Vieira^ (written discussionj—What is the policy in the United States in using dissolved
gas-in-oil tests after factory heat runs? Is there any consideration by ASTM to modify Method D
3612 based on this application?
It is important to remember that all the diagnostic criteria for dissolved gas-in-oil analysis
were developed to monitor the condition of in-service transformers. With factory tests the con-

^Musil, J. and Foschum, H., "Application of Dissolved Gas Analysis (DGA) During Factory Testing of
Power Transformers," Annual International Conference of Doble Clients, Sec. 6-801, 1980, Doble Engi-
neering
(UFRN) Co.,Federal
Universidade Watertown,
do Rio GrandeMA.do Norte ((UFRN) Universidade Federal do Rio Grande do Norte) pursuant to License Agreement. No further reproductions auth
DISCUSSION ON TRANSFORMERS 107
centration of each gas is very small, and sometimes a variation of 2 ppm corresponds to more
than a 100% increase.
P. Griffin {author's closure)—l\i the United States there are no standard limits for dissolved
gases in oil generated during heat runs performed in the factory. However, manufacturers have
developed guidelines based on their experience as shown in Table 4 and in information in Refs
13, 14, and in Footnote 2. Often these limits are divided into two categories. The first set of
limits indicates that a problem may exist and that more testing should be performed. A second
set of higher limits indicates that a definite problem exists and that a thorough investigation
should be performed. Some utilities in the United States specify that dissolved gas-in-oil tests
must be performed after the heat run. Limits may or may not be specified.
ASTM Method D 3612 limits of detection for gases dissolved in oil are:
Minimum Detection
Gas Limit, ppm
Hydrogen 5
Hydrocarbons 1
Carbon oxides 25
Atmospheric gases 50
Committee D-27, which is responsible for this test method, is currently examining the
method to evaluate what factors can be controlled to improve accuracy and precision. Although
the performance of this work was not instigated by concern for factory heat run tests, it certainly
would be most important in this case, where small numbers and changes are being evaluated.
The limits of detection of the test method are not presently being examined.
T. J. Haupert^ and Fredi Jakob^
A Review of the Operating Principles and

Practice of Dissolved Gas Analysis
REFERENCE: Haupert, T. J. and Jakob, F., "A Review of the Operating Principles and Practice
of Dissolved Gas Analysis," Electrical Insulating Oils. STP 998, American Society for Testing
and Materials, Philadelphia, 1988, pp. 108-115.
ABSTRACT: Dissolved gas analysis has gained worldwide acceptance as a method for the detec-
tion of incipient faults. When properly applied, the method provides a highly reliable and sensitive
method for the identification of fault type and for the assessment of the rate of fault development.
This paper investigates the origins of the individual fauh gases, factors that are responsible for
the extreme sensitivity of the method, and the factors that influence the observed gas concentra-
tions. The paper also examines the scope and limitations of the diagnostic techniques that have
been developed to interpret the data that the method provides. These techniques include such
concepts as key gases, ratios of gas concentrations, nomographs, and rates of gas generation.
KEY WORDS: dissolved gases, fault detection, incipient faults, gas chromatography, key gases,
gas concentration ratios, nomograph, gas generation rates
Fifteen years ago when Analytical Associates began its dissolved gas analysis (DGA) pro-
gram, it was not generally recognized that DGA could be a useful diagnostic and preventive
maintenance tool for the detection of faults in oil-filled electrical equipment. DGA has, in the
ensuing years, been widely accepted by the power industry, and it is currently recognized as a
significant diagnostic tool for the detection of faults in oil-filled transformers. The wide accep-
tance of DGA is based on its sensitivity, which is orders of magnitude greater than other gas
detection methods, and its reliability. As a result, faults can be detected and identified at the
earliest possible time, and the method provides an excellent means of monitoring incipient fault
development.
Fault gases are produced by degradation of the transformer oil and other insulating materials
such as cellulose. In the presence of an active fault, the rate of oil and cellulose degradation is
significantly increased, and the types of degradation products formed will vary with the nature
and severity of the fault. The composition of the gas mixture resulting from the decomposition is
energy dependent, and, since fault processes differ greatly in the energy they dissipate, different
mixtures of degradation products are produced for each tjrpe of fault process. By observing the
composition of the gases produced by the degradation of the insulating media, it is possible to
distinguish three basic fault processes which differ greatly in their energy characteristics. These
processes are arcing, corona or partial discharge, and pyrolysis or thermal decomposition.
Pyrolysis occurs whenever there is sufficient thermal energy to break chemical bonds. Pyrolysis
can occur in the absence of oxygen and produces different products than a combustion process.
Thermal decomposition is a similar process but usually implies lower temperatures and lower
reaction rates. The low-molecular-weight gases that are formed vary widely in terms of their
equilibrium solubilities in oil, from a low of 7% for hydrogen to a high of 400% on a volume/
volume basis for acetylene. But one must realize that even a solubility of 7% is very high in
terms of the concentrations of those gases that actually occur in a fault. Therefore, the oil is
rarely saturated with the fault gases. Coupled with the fact that the volume of oil is significantly
Copyright'Analytical
by ASTM Int'l Associates, Inc.,
(all rights reserved); ThuSacramento, CA2021
May 6 14:02:13 EDT 95826.
108
HAUPERT AND JAKOB ON DISSOLVED GAS ANALYSIS 109
greater than the volume of the gas blanket, if one is present, it is apparent that most of the fault
gases remain dissolved in the oil. Thus, it is far better to analyze an oil sample for the presence
of a fault gas rather than to look in the gas above the oil. The sensitivity of DGA is also better
than that of other methods of gas detection such as portable or fixed space gas monitors and
pressure-sensing devices. Using DGA, fault gases can be routinely detected at the part-per-
million (ppm) level. Thus, DGA is often successful in identifying transformer problems before
there is an indication of a problem with other test procedures.
Buchholtz and others were aware of the production of fault gases in transformers and tried as
early as 1928 to correlate the gases produced with a particular type of fault [/]. Historically, two
different approaches have been used to try to correlate fault gas composition and fault type.
One approach is empirical and is based on statistical evaluation of fault gas data from units that
have exhibited documented problems. The Central Electricity Generating Board of Great Brit-
ain (CEGB) has been a leader in this area [2]. They started extensive work on fault-gas correla-
tions as early as 1968. The nonempirical approaches have treated the production of fault gases
from both an experimental and thermodynamic approach. Experimental data have shown that
an increase in oil temperature led to an increase in the ratio of unsaturated to saturated hydro-
carbons, specifically, the ratios of acetylene to ethylene and ethylene to ethane. Halstead [3],
using a thermodynamic analysis, predicted that low-molecular-weight fault gases would be pro-
duced with increasing temperature in this sequence: methane, ethane, ethylene, and finally
acetylene.
The empirical and theoretical studies have led to similar conclusions. The consensus is that
certain key gases can be correlated with fault type and that the rate of gas production can be
correlated with fault severity. It is generally accepted that acetylene can only be produced at the
very high temperatures that occur in the presence of an arc. All other low molecular weight
hydrocarbons, when produced concurrently with hydrogen, indicate pyrolysis of the oil. Corona
or partial discharge is a relatively low temperature process and produces mainly hydrogen.
Dornenburg and Strittmatter [4] are generally credited with the development of the gas ratio
concept. They noted that fault gas ratios clustered according to fault type when plotted on a log-
log graph. Their plots are reproduced in Fig. 1. These investigators also reported gas levels that
were observed in transformers that have been operating free of faults for several years. These
normal gas values are tabulated in Table 1. Subsequent experience [5] has shown that levels
slightly higher than those recommended by Dornenburg and Strittmatter are applicable to
American transformers. The difference in normal levels between European and American
transformers can be ascribed to differences in design and operating parameters. The higher
levels recommended by the Bureau of Reclamation are tabulated in Table 2. R. R. Rogers, a
CEGB engineer, extended the ratio concept and published interpretive guides that facilitated
the interpretation of gas ratios [6].
The pioneering work of Dornenburg, Strittmatter, and Rogers led to the development of a
nomograph for the interpretation of DGA data. J. O. Church of the U.S. Department of the
Interior provided the leadership in this work, tested and evaluated the nomograph and coau-
thored a paper with us which describes the development of the nomograph [5]. Our objective in
developing the nomograph was to combine the concepts of fault gas threshold levels and the gas-
ratio method in a convenient manner to facilitate interpretation of DGA data. Fault gas concen-
trations are plotted in ppm on several vertical logarithmic scales as shown in Fig. 2. Lines are
then drawn to connect the concentration values on adjacent scales. Points higher on the scales
are given greater significance than lower points. The unshaded tick marks on each scale repre-
sent the threshold criteria of Dornenburg and Strittmatter. Our experience has shown that the
concentration limits can be raised as noted by the shaded symbols. The slope of the line con-
necting the two adjacent concentrations is thus directly related to the fault gas concentration
ratio. Figure 3 illustrates the method as applied to one of the gas pairs, methane and hydrogen.
For this gas
Downloaded/printed by pair, connection of any set of points representing equal concentrations of hydrogen
andUniversidade
(UFRN) methane generates
Federal do Rio Grande a dohorizontal
Norte ((UFRN) line. Thus,
Universidade a horizontal
Federal do Rio Grande doline
Norte)corresponds
pursuant to License to a methane
Agreement. No furtherto
reproductions auth
110 INSULATING OILS
\J\J-
lO-
l\J
CH. pyrolysis
1_
Hz 1
(methane) arcing
(hydrogen)
,1 •
1
corono
UM .7
.01 .1 I 10 100
(acetylene)
l-a (ethylene)
100
10
CK pyrolysis
(nf)ethone)
arcing
(hydrogen)
corona
.01
.01 10 100
(ethane)
l-b (acetylene) Cg Hg
(UFRN) Universidade Federal do FIG. 1—Domenburg
Rio Grande and
do Norte ((UFRN) Strittmatter's
Universidade Federal do plots offault
Rio Grande gaspursuant
do Norte) ratiotopairs.
100-
10
C H4 pyrotysis
" ^
(methane) arcing
(hydrogen)
corona
.01
.c)l .3 10 IC
(acetylene) ^2 '"'2
l-c (methane) CH.
FIG. 1—Continued.
TABLE 1—Dornenburg and Strittmatter suggest that

transformers which have been in service for a few years
may be regarded as operating without problems if gas
concentrations do not exceed the following
parts-per-miltion levels.
Hydrogen 200
Methane 50
Ethane 35
Ethylene 80
Acetylene 5
Carbon monoxide 500
Carbon dioxide 6000
TABLE 2—Threshold levels utilized by the Bureau of

Reclamation.
Hydrogen 500
Methane 125
Ethane 75
Ethylene 175
Acetylene 15
Carbon monoxide 750
Carbon dioxide 11 000
112 INSULATING OILS
DISSOLVED GAS-IN-OIL ANALYSIS

DIAGNOSTIC CHART
»i s- 3
l - 2
-
irf Iff
li t -
• -:
•»_:
-
•
•
7
•5
irf_
lO"-
lO"-
t -
•-
ri 7 -
6 -
S -
ti
c -
s -:
i
i
4— 4
J - 3
2 — - 2
'(% 10 •
10 .
« —
(i i
» -=<a
T —
6
5 —
4 —
ia I0'_
9 -=
10
5 -
M
li
10 _
9—
5 — «—
5 -^
4 —
S -_
»i 3 —
4 —
'0_
11 3 -
7
<
5
A D&H
\OiH H&D
« —
5 — 5 —
• \
(HYDROGEN) (ACETYLENE)
CA
(ETHANE) (HYDROGEN) (METHANE)
CH, CjHj CjH,
(ACETYLENE)(ETHYLENE)
ro 'A
FIG. 2—The Church nomograph. Open triangles indicate threshold levels for fault gases according to
Domenburg and Strittmatter. Filled triangles indicate threshold levels currently utilized by the Bureau of
Reclamation.
1000 • 1000
PYRaYSIEj.
m
103 100
t\::v;ÂRCiNo -i
10- -10
H2 CH4
FIG. 3—The interpretation of hydrogen to methane ratios. An illustration of how the Church nomograph
was developed.
hydrogen ratio of one. A ratio greater than one gives a line with a positive slope and indicates
pyrolysis as the fault mode. Negative slopes in the range of 0 to — 1, corresponding to a ratio of
hydrogen to methane in the range of 1 to 10, indicate arcing. Negative slopes greater than — 1,
corresponding to a hydrogen to methane ratio greater than 10, indicate corona or partial dis-
charge. Figure 3 illustrates these conditions. For example, if the oil contains 1000 ppm of hy-
drogen and 10 ppm of methane, the line has a negative slope (Line a), and the failure mode
indicated is corona. If the DGA revealed a hydrogen concentration of 200 ppm and 70 ppm of
methane (Line b) the connecting line would have a negative slope and would indicate arcing. If
the methane concentration exceeds the hydrogen concentration, a line with a positive slope
results. Such would be the case if the methane concentration was 200 ppm while the hydrogen
concentration was 20 ppm. This line (c) indicates pyrolysis as the failure mode. It is important
to note that in each of these examples at least one of the pair of gases was above the threshold
level.
Carbon monoxide and carbon dioxide are indicative of cellulose decomposition in a trans-
former. Domenburg and Strittmatter suggest threshold levels of 500 ppm for carbon monoxide
and 6000 ppm for carbon dioxide. The Bureau of Reclamation, based on their experience, has
raised these levels to 750 ppm of carbon monoxide and 10 000 ppm for carbon dioxide. Concen-
trations exceeding these values indicate that cellulose material is involved in the fault process. A
negative slope for the line connecting these two concentrations indicates a carbon monoxide to
carbon dioxide ratio greater than 0.1. This ratio is indicative of an accelerated solid insulation
decomposition rate which can be ascribed to higher temperatures associated with a fault.
Application of the nomograph can be illustrated by what we consider a classic case. The
results shown in Fig. 4 are from a mobile transformer that was being tested for the effects of
operating without the cooling system functioning. During shutdown an internal ground fault
occurred. DGA samples were fortunately taken before and immediately after the occurrence of
the fault. The before and after results are tabulated in Table 3. An immediate and significant
increase of all the fault gases is obvious. The results again illustrate the sensitivity and ability of
DGA to detect a problem at the earliest possible time, in this case minutes after the fault oc-
curred. It is also of interest to note how consistently the various gas ratios indicate that an arc
occurred
Copyright by ASTMinInt'lthis mobile
(all rights transformer.
reserved); The
Thu May 6 14:02:13 EDTadvantages
2021 of using the nomograph are summarized
in Table 4.
114 INSULATING OILS
DISSOLVED 6 A S - I N - 0 I L ANALYSIS
DIAGNOSTIC CHART
before
after
10 _,
li
'0*,
>-
t -
f -z
8 —
'to 'A
(HYDROGEN) (ACTTYLENE) (ETHANE) (HYDROGEN) (METWANE) (ACETYLENE)(ETHYLENE)
FIG. 4—A Church nomograph plot of the data in Table 3.
Although the theoretical basis of DGA is well established, several problems exist in the cur-
rent practice of the technique. These problems are related not only to the interpretation of DGA
results but also to the generation of the results themselves. The determination of dissolved gases
is usually a two-step laboratory process. In the first step, the dissolved gases are separated from
the transformer oil sample by extraction into an evacuated gas collection apparatus. The effi-
ciency
Copyright of this
by ASTM extraction
Int'l (all rights reserved);isThu
dependent
May 6 14:02:13upon the initial vacuum achieved in the system, the vol-
EDT 2021
ume of the evacuated system into which the gases
(UFRN) Universidade Federal do Rio Grande do Norte ((UFRN) Universidade diffuse,
Federal anddo the
do Rio Grande Norte)length ofLicense
pursuant to timeAgreement.
allowedNofor the
TABLE i—Before and after data for mobile

transformer (see Fig. 4) in parts per million
(ppm).
Gas Before After
Hydrogen 45 400
Acetylene 0 270
Ethane 0 43
Methane 10 200
Ethylene 8 230
TABLE 4—The use of the nomograph has been found to offer distinct advantages in analyzing data
obtained from DGA. These advantages can be easily summarized.
1. Data can be quickly and easily interpreted.

2. High gas levels are immediately apparent.
3. Relative changes in concentrations are easily seen.
4. The rate of change in gas evolution can be observed when subsequent tests are plotted on the same
nomograph.
extraction. The dissolved gases are then compressed back to atmospheric pressure. Some labo-
ratories compress the gas in the presence of the oil, which necessitates a correction to be applied
for the reabsorption of the extracted gas into the oil. The volume of the collected gas is mea-
sured and corrected to standard temperature and pressure (STP). An aliquot of the extracted
gas is then analyzed qualitatively and quantitatively by gas chromatography (GC). Here again
there is a wide range of procedures employed. Some laboratories use a single chromatographic
column, others use two with two detectors. Some employ a single sample injection, others inject
two aliquots of the sample. Thermal-conductivity detectors are used by some laboratories; oth-
ers use flame ionization detectors. Samples of standard gas mixtures are available so that the
analyses of the extracted gases can be correctly made despite differences in the analytical sys-
tems. However, standards of oil containing dissolved gases are not presently available. As a
result, differences in the extraction efficiencies of the gases among laboratories will affect the
calculated results of the concentration of the individual gases dissolved in the oil. If greater
interlaboratory consistency of results is to be achieved, two things are necessary. First, laborato-
ries must agree to use the ASTM Method for Analysis of Gases Dissolved in Electrical Insulat-
ing Oil by Gas Chromatography (D 3612) or a revised version of this standard method, and,
secondly, chemical standards of oil containing dissolved gases must be made available. Despite
differences that exist among laboratories in the extraction efficiencies that affect the reported
concentrations of dissolved gases, little or no problem exists if all analyses are performed using
the same analytical system. Also, the effect of differences in the extraction efficiencies will be
minimized if ratios of gas concentrations are used for diagnostic purposes, rather than the indi-
vidual gas concentrations.
References
[/] Buchholtz, M., Electrische Technichse Zeitschrift, Vol. 49, No. 34, 1928, pp. 239-241.
[2] Rogers, R. R., IEEE Transaction Electrical Insulations. Vol. EI-13, No. 5, October 1978.
[J] Halstead, W. H., Journal of the Institute of Petroleum. Vol. 59, No. 569, September 1973, pp. 239-
241.
[4] Dornenburg, E. and Strittmatter, W., Brown BoveriReview. Vol. 61, No. 5, 1974, pp. 238-247.
[5] Church, J. C , Haupert, T. J., and Jakob, V.,Electrical World. Vol. 201, No. 10, October 1987, pp. 40-
44. by
Downloaded/printed
[6]Universidade
(UFRN) Rogers, Federal
R. R.,do Doble
Rio GrandeClient Conference
do Norte 1977, Federal
((UFRN) Universidade Paperdo44
RioAIC 77,
Grande Doble
do Norte) Engineering,
pursuant Boston,NoMA.
to License Agreement. further reproductions auth
Section V—Electrostatic Buildup in
Transformer Oil
/. Franklin Roach^ and James B. Templeton^
An Engineering Model for Streaming

Electrification in Power Transformers
REFERENCE: Roach, J. F. and Templeton, J. B., "An Engineering Model for Streaming Electri-
fication in Power TtmHlonoKn," Electrical Insulating Oils, STP 998, H. G. Erdman, Ed., Amer-
ican Society for Testing and Materials, Philadelphia, 1988, pp. 119-135.
ABSTRACT; An engineering approach to assessing static electrification phenomena in power

transformers is presentedi A theoretical model is developed to estimate streaming currents and
space charge accumulation around the hydraulic oil flow loop in a three-phase transformer. The
streaming current model is applied to a 345-kV, 240-MVA three-phase auto transformer with
shell form winding design. The influence of space charge on the electric field and potential distri-
butions in the upper plenum is modelled by a one-dimensional theory. Significant electric stress
enhancement at coil edges on alternate half cycles of applied a-c stress is predicted for the 240-
MVA transformer. Potential failure modes in power transformers due to static electrification are
discussed.
KEY WORDS: transformers, power, electrification, streaming, current, shell form, core form,
charge, static, failure, oil, paper
In the last six years considerable attention has been directed at investigating streaming elec-
trification in power transformers. Power transformers are constructed with either shell form or
core form configurations and employ an insulation system comprised of cellulosic solid materi-
als with mineral oil as an insulating fluid and heat transfer medium.
As the MVA throughput of transformers increases, it becomes necessary to provide cooling
of the core and coil assembly by means of forced oil and forced air (FOA) heat transfer. The
FOA cooling is accomplished by flowing oil through the core and coil assembly, out of the trans-
former tank, through externally mounted oil to air heat exchangers, and back into the trans-
former tank in a closed loop. In a typical elevation cross section of a shell form power
transformer, for example, the oil is injected into a lower plenum via pumps and is directed
vertically through the core and coil assembly. The oil continues to flow vertically through a lead
support region above the core and coils referred to as the bridge or superstructure and then
enters the upper plenum, where the oil is drawn out of the tank and into the external heat
exchangers completing the loop at the pump.
The existence of an insulating fluid flowing past an insulating solid results in charge separa-
tion at the interface of the two materials. Physically, charges separate at the interface of the oil
and paper in any transformer regardless of construction type, that is, shell form or core form.
The importance of studying the streaming electrification phenomena is to enable one to predict
when the magnitude of streaming electrification becomes problematical. It is known that the
streaming electrification phenomena has been a major contributor to a number of failures of
large transformers in service. Therefore, it is necessary to understand the theory of streaming
electrification so that design criteria may be developed to avoid the problematic effects of
streaming electrification. The purpose of this paper is to explain some of the theory embodied in
'Senior research scientist, Research and Development Center, Westinghouse Electric Corp., Pittsburgh,
PA 15235.
^Manager, Shell Form Design, Power Equipment Division, Westinghouse Electric Corp., Pittsburgh, PA
15235.
119
120 INSULATING OILS
flow electrification in power transformers and to show a mathematical model which may be
used to predict the magnitude of the streaming currents and the impact they have on voltage
stresses within the transformer.
Flow Electrification
When a low conductivity liquid, such as transformer oil, flows past a solid surface, the liquid
acquires a space charge due to frictional charge separation at the liquid-solid interface. In gen-
eral, ionic species of one sign are preferentially absorbed by the solid, leaving a net charge of
opposite sign in the liquid. The space charge developed in the liquid is transported by the flow,
resulting in a streaming current or charging current being carried by the liquid. If the walls of
the flow system are insulated or floating, the flow electrification process also leads to an accu-
mulation of charge and the generation of high electrostatic surface potential at liquid-solid in-
terfaces. The phenomenon of charge generation by low-conductivity liquids flowing in pipes and
ducts is well known in the petroleum industry. A number of explosions and fires have been
attributed to discharges initiated in fuel tanks and handling equipment due to accumulation of
charge generated by flow electrification.
Electrostatic charge generation is also known to occur in forced oil cooled power transform-
ers. As the cooling oil flows through the insulation structure, charge separation takes place at
oil/paper interfaces. Charge accumulation results both in the oil and on insulating surfaces,
producing d-c fields which are superimposed on the a-c energization field. If static charge fields
are high enough, spontaneous partial discharges will occur in the transformer, even when un-
energized. It is also possible that impulse and switching surges of the right polarity in combina-
tion with the static charge potentials may produce localized stresses of sufficient magnitude to
initiate catastrophic failure of the insulation. Therefore, it is important to investigate and quan-
tify the phenomena of static charge generation in transformers.
The subject of flow electrification has been most extensively studied in the petroleum indus-
try. The much quoted work of Klinkenberg and Van der Minne [/] published in 1958 gives an
early account of electrostatics related to the transport of petroleum liquids in pipes and the
hazards of charge accumulation in storage tanks. In 1965 a study of charge separation in liquid
hydrocarbons was conducted by Shafer, Baker, and Benson [2] at the National Bureau of Stan-
dards. More recent work on static charge in petroleum liquid has been reported by Itoh [3].
The problem of static electrification in both shell form and core form power transformers was
recognized in the early 1970s. Notable papers on flow electrification in transformers have been
published by a number of Japanese authors, including Kan [4], Shimizu [5], Higaki [6],
Okubo [7], Tanaka [8], and Tamura [9]. A CIGRE paper by Takagi [10] on reliability im-
provement of 5(X)-kV power transformers also discusses the static charge problem. The static
electrification properties of oil have been studied by Oommen [11,12]. In an important recent
paper by Crofts [13], the static electrification phenomena in power transformers is reviewed in
detail. Crofts discusses the failures of several power transformers which were attributed to static
electrification phenomena.
Theoretical treatments of flow electrification for low conductivity fluids have been published
by Klinkenberg and Van der Minne [/], Gavis and Koszman [14-16], Schon [17], Abedian and
Sonin [18], and Tanaka, Yamada, and Yasojima [19]. The paper by Tanaka et al. addresses
the electrification theory specifically for transformer insulation and gives empirical streaming
current formulations for both laminar and turbulent flow regimes.
The primary factors which influence the flow electrification process in a forced oil cooled
transformer are: (1) the oil flow rate; (2) the oil bulk oil properties of temperature, electrical
conductivity, and viscosity; (3) surface conditions and structure of the solid insulating materials
of the oil ducts; and (4) moisture content of oil and paper. Other important factors are the
nature of the
Downloaded/printed by ionic content of the oil, which is influenced by the refining process; the time his-
tory
(UFRN) of oil as
Universidade it circulates
Federal do Rio Grande doinNorte
the((UFRN)
transformer; and do
Universidade Federal contamination particles.
Rio Grande do Norte) pursuant The
to License charging
Agreement. ten-
ROACH AND TEMPLETON ON POWER TRANSFORMERS 121
dency of a transformer oil is therefore a strong function of the ionic adsorption chemistry which
occurs at oil-paper interfaces. The net rate of adsorption of charges by paper surfaces will be
dependent upon oil deterioration products such as copper and aluminum ions, polar molecules,
etc., and upon physical and chemical changes of the insulation materials themselves. The
charge separation process is also dependent upon the application of voltage to the transformer
windings. In general it is found that a-c energization leads to enhanced charge separation.
In this paper a flow electrification model for a three-phase, shell form transformer is devel-
oped. The model is applied to a 345-kV, 240-MVA, three-phase auto transformer design. The
streaming currents and charge densities generated throughout the hydraulic oil flow loop are
calculated. The model predicts the dependence of streaming current on oil temperature and
velocity. The effect of space charge accumulation in the upper plenum region of the transformer
on the electric field under a-c energization is also modelled. Nontrivial enhancements in the
electric field at coil edges in the upper plenum are predicted. This result may help to explain the
failure mode of recent transformer failures that involved flashovers across large oil distances of
tens of inches at the top of conservator-equipped, shell form transformer windings [13].
Streaming Current Theory

The transformer model calculations to be performed in this paper assume that the exit
streaming current generated by oil flow electrification in insulating or metallic ducts is given by
/ = /»(1 - 7) + /o7 (1)
where y is the decay factor
y = e-ô'y (2)
In Eq 1, /„ is the current which enters the duct and /„ is the limiting streaming current which is
a function of duct geometry, oil velocity, oil temperature, and the adsorption rate of charge at
the duct surface. In Eq 2, Zg is the duct length, Vthe average oil velocity, and r is the character-
istic decay time for charges in the oil given by
T = eeo/a (3)
where e is the relative dielectric constant for oil and a is the oil electrical conductivity.
According to Eq 1, the exit streaming current approaches the limiting current/„ for very long
ducts, and any current /„ entering decays exponentially along the duct length.
The theoretical derivation of the limiting streaming current /„ is developed in the Appendix
and is defined by Eqs A15 and A18 as
/oo = fWV{ttoRT/ndZF)(b/bBr (4)
where the factor/has been introduced to account for enhanced charge separation due to surface
roughness and protrusions such as spacer blocks in the insulation construction. The parameters
in Eq 4 are:
/ = surface roughness factor, 2 for metal and 5 for insulation ducts,
W = width of rectangular ducts, W = icr for pipe of radius r,
V = average flow velocity,
e = dielectric constant, 2.26 for oil,
t„ = permittivity
Downloaded/printed by of free space, 8.85 X 10~ ^^ F/m,
R = gasFederal
(UFRN) Universidade constant, 8.314
do Rio Grande J/mole-°K,
do Norte ((UFRN) Universidade Federal do Rio Grande do Norte) pursuant to License Agreement. No further reproductions auth
122 INSULATING OILS
T — absolute temperature,
n — transference number of ions 0 < n < 1, assumed M = 1,
d = diffusion layer thickness at surface, Eq A16 in Appendix,
Z = unit charges per ion, assumed Z = 1,
F = Faraday number, 96 500 C/mole,
8 = Debye length, 8 = V D 7 where D — 10~' cm^/s is assumed,
8B = boundary layer subzone thickness, Eq A19 in Appendix, and
m = empirical parameter, m = 0.3 to 1.0.
Equation 4 calculates / „ at a fixed temperature, oil flow rate, and duct configuration. The
theory is, in general, valid in the turbulent flow regime and is expected to be less accurate for
laminar flow. The limiting streaming current given by Eq 4 can be adjusted to match experi-
mental data over a limited temperature range. Over a wider temperature range the temperature
dependence Ioo(T) is estimated by the empirical expression
looiT) = I^(ao/a)- (5)
where !„ is given by Eq 4, a„ is an adjustable conductivity, a is the conductivity at temperature

T, and a is an adjustable empirically determined parameter near unity. The explicit tempera-
ture dependence on streaming current predicted by Eq 5 may be written as
/=o(r) = I^(REF)(T/TMv„/vy^'"+^^'Hao/aY^*'"^'^ (6)
whereI„(REF) is calculated by Eq 4 at some set of reference values T„,v„, a^ of the temperature

dependent parameters T, v, and a where i>{T) and a(,T) are known functions of temperature,
and V is the kinematic viscosity.
In a number of flow electrification model tests performed by Westinghouse, Eq 5 was found
to correlate with test data for values of the adjustable parameters m = 0.3 and a = 1. With
these values for m and a, I„{T) is predicted by Eq 5 to vary as (j<-<'-887s ^ a~' '^ X T) or
approximately as T/va. The conductivity (f(T) is an increasing function of T while v^T) is a
decreasing function of T. The combined temperature dependence is found to result in a peak in
the streaming electrification for transformer oil at some critical temperature T*.
Shell Form Transformer Model for Flow Electrification Analysis

A schematic view of a three-phase, shell-form transformer is shown in Fig. 1. The trans-
former is divided into twelve sections, A through L, which comprise the hydraulic flow loop.
Due to the oil flow, charges are separated at the oil-paper interfaces in Sections C and D in the
pancake coils and insulating structure (leakage ducts) and at the oil-metal interfaces in the
cooling loop Sections H through K. The pumps L also contribute to charge separation. This
flow electrification process results in a net accumulation of charge in the bulk oil in the Sections
A, B and F, G below and above the coils where the oil velocity is low. The charge density distri-
bution developed in these plenum sections is important in understanding potential failure
modes. As charge accumulates in the bulk oil, charge of opposite sign accumulates on insula-
tion surfaces, such as pressboard and spacer blocks. The distribution of surface charge
throughout the insulation structure results in the development of high local electrostatic voltage
gradients which may be sufficient to trigger spontaneous surface discharges. In most trans-
former cases the bulk oil is charged positively and the insulation surfaces charged negatively.
The core is represented as Section E. It is assumed that the core plays no role in the electrifi-
cation process. However, the model does account for the fraction of the total oil flow volume
which
Copyright is lost
by ASTM torights
Int'l (all the reserved);
core. Thu May 6 14:02:13 EDT 2021
Under steady state oil flow and temperature conditions, the sum of streaming currents enter-
124 INSULATING OILS
ing each of the Sections A through L is equal to the sum of the streaming currents exiting in
each of these sections. By applying the streaming current analysis developed above to every unit
duct in each section of the transformer, the streaming current inventory and balance may be
determined. Equations 1 through 4 are used to compute the streaming currents exiting each
section by summing the contributions from individual ducts in a given section. The current
balance is achieved when
L L
E W) = £ I(j) (7)
that is, when the sum of entrance currents equals the sum of exit currents. The charge density at
the exit of the^'-th section is the ratio of the total exit streaming current and the volumetric flow
rate for that section, namely
P(J) = Hj)/QU) (8)
The exit streaming current l(j) is
N
Hj) = S Hij) (9)
where I{iJ) is the streaming current for the i-th unit duct in thej-th section and N is the total
number of unit ducts. I(i,j) is calculated by Eq 1 for each duct in the hydraulic oil flow loop.
Streaming Current Analysis for a 240-MVA Transformer Design

The transformer model in Fig. 1 was applied to a 345-kV, 240-MVA three-phase auto trans-
former design. The results of the static charge analysis are presented in Table 1. The calculated
entrance and exit streaming currents /„ and /, and the exit charge density p„ for each section A
through L of the transformer are given. The % flow, oil velocity, number of oil ducts, and the
Reynold's number for each section are also given. The calculation assumes a temperature of
30°C, a conductivity of 3 X 10"''' Q"' cm"', and a viscosity of 0.16 cm^/s. The total flow rate
with all eight of the transformer's 800 gallons per minute (GPM) pumps running is 14.3 ft^/s
(0.405 mVs).
In arriving at Table 1 the following assumptions are made. The entrance and exit Sections B
and F are assumed to play no role in the streaming current, that is, these sections neither gener-
ate nor relax charge. The total streaming current generated by the eight pumps is assumed to be
0.82 fiA or 102.5 nA/pump. Finally, the relaxation of charge in the plenum Sections A and G is
given by the factor 1/(1 + V/QT), where v is the total volume of the plenum, Q is the total
volumetric flow rate, and T the characteristic relaxation time.
Based upon the results in Table 1, several conclusions are reached for the 240-MVA design.
First, the heat exchanger Section I is a good sink, that is, relaxer for charge. Second, the major
generators of charge are the coil structure and leakage ducts, Sections C and D. Third, charge
densities as high as 6.7 ^C/m^ are generated above coil sections. Fourth, at the insulation exit
Section F, the total leakage current is 1.59 Â.
The results in Table 1 are for a fixed oil temperature and flow rate. The dependence of leak-
age current and charge density for the 240-MVA transformer at the coil structure exit. Region
C, as a function of oil velocity or number of oil pumps is presented in Figs. 2 and 3. For these
calculations
Copyright by ASTM Int'la(all
temperature ofMay
rights reserved); Thu 40°6 C, a conductivity
14:02:13 EDT 2021 of 3 X lO"'"*!]"' cm"' and viscosity of 0.12
cm^/s are assumed. The theory predicts a power
(UFRN) Universidade Federal do Rio Grande do Norte ((UFRN) Universidade dependence
Federal do Rio Grande do of leakage
Norte) pursuant tocurrent on oilNovelocity
License Agreement. further reproductions auth
TABLE 1—Summary of static charge analysis for 240-MVA transformer.
Effective
Region of Flow, Velocity, Number Reynolds
Transformer % ft/s" of Ducts Number /o,M /. nA p, iiC/m^
A. Lower plenum 100 0.10 1 0.90 0.61 1.50
B. Insulation entrance 95 0.12 3 0.61 0.61 1.50
C. Coil structure 80 2.00 384 494 0.48 1.18 3.64
D. Coil insulation 15 1.53 60 191 0.092 0.41 6.70
E. Core 5 0.031 0.031 1.52
F. Insulation exit 95 0.12 3 1.59 1.59 4.13
G. Upper plenum 100 0.07 1 1.62 0.13 0.32
H. Pipes to HE 100 15.14 4 36 000 0.13 0.15 0.36
I. Heat exchanger 100 0.20 6144 97 0.15 0.013 0.03
J. Pipes from HE 100 15.14 4 36 600 0.013 0.085 0.21
K. Pipes to pumps 100 9.10 8 26 400 0.085 0.080 0.20
L. Pumps 100 0.080 0.90 2.22
NOTE: HE = heat exchanger.

"1 ft/s = 0.3048 m/s.
Number of Pumps @ 800 GPM Each

4
10
•<
5 1 0 = 3 x 1 0 ^"jf^cm ^
0.1 j -
.5 8 10
V (tt/secl
FIG. 2—Leakage current at coil exit versus oil velocity.
126 INSULATING OILS
Number of Pumps @ 800 GPM Each

4
10
o = 3 x 10 ^''ff'cm ^
v = 0. IZcm /sec
0.1 _L _L _L
1 8 10
V (tt/sec)
FIG. 3—Charge density at coil exit versus oil velocity.
between 2.1375 and 2.75 for values of the adjustable parameter m = 0.3 and 1, respectively.
The calculations in Table 1 were generated for a value of m = 0.3. The charge density depen-
dence on oil velocity is shown in Fig. 3. The predicted charge density increases with oil velocity
to the 1.1375 power for m = 0.3 and to the 1.75 power for w = 1.
Temperature Dependence of Streaming Current
The calculations presented thus far assume a fixed oil temperature. Since the oil temperature
in operational transformers is variable, depending on electrical load and environmental condi-
tions, it is important to assess the dependence of the flow electrification process on temperature.
Equation 6 was used to predict the temperature behavior of the leakage current for the 240-
MVA transformer at the coil exit. Section C. For this analysis, the oil velocity in the coil ducts
was assumed to be 2 ft/s (61 cm/s) as in Table 1. The temperature dependence on oil conductiv-
ity is approximated by the expression
o = 0.30 X 10'^"° (10-'" fi-' cm"')
where t is °C. The oil viscosity as a function of t is approximated by

Downloaded/printed by V = 0.76 X lO-'-'ô (cmVs)
for 0 to 40° and
V = 0.30 X lO-'''<» (cmVs)
for 40 to 100°. The results are presented in Fig. 4. The two curves are for values of the parame-
ter m = 0.3 and m = 1 with the parameter a = 1 in both cases. The theory predicts a peak in
the coil exit streaming current between 40 and 50°C at 2 ft/s (61 cm/s) oil velocity, that is, all
eight pumps running.
Power Frequency Voltage Effect on Streaming Electrification

The transformer model results presented above are for an unenergized transformer. When
the transformer is energized the generation of streaming current increases. Large experimental
model tests performed by Westinghouse on shell form insulation structures have shown that
streaming current development in windings is increased 1.5 to 2.0 times with the application of
60-Hz voltage at practical applied stress levels. Since the effects of a-c voltage on streaming
current are not taken into account in the theory presented here, the streaming current and
charge density results presented in Table 1 for the 240-MVA transformer are expected to be
greater when the transformer is energized. The applied power frequency voltage will lead, there-
0 (10 ^ " Q ' c m ^
Downloaded/printed by FIG. 4—Leakage current at coil exit versus temperature.
128 INSULATING OILS
fore, to more accumulation of surface charge on insulation and increased space charge in the
lower and upper plenums. The true electrical stress distribution in the transformer is deter-
mined by the superposition of flow electrification fields on the fields generated by the power
frequency voltage and also by transient voltages impressed at terminals.
Theoretical treatments of the influence of a-c energization on streaming current have been
developed by Tanaka [19] and by Melcher [20] for oil-paper configurations. These theories
show that under a-c energization the oscillations of ions distributed in the oil phase near the oil-
paper interface due to the impressed field lead to enhanced streaming currents. At a fixed ap-
plied stress and oil temperature, the streaming current enhancement varies inversely with the
applied frequency.
Charge Density Distribution in Upper Plenum

As the streaming current emerges from the top of the coil structure, charge accumulates in
the bulk oil in the upper plenum. Under energized conditions this "cloud" of space charge
which develops in the plenum will distort the normal a-c electrostatic field and potential distri-
butions. In order to assess the impact of space charge on the applied a-c field and potential, the
profile of the charge in the plenum is required. Based upon a one-dimensional theoretical model
of the upper plenum as shown in Fig. 5, the charge density profile is found to fall off exponen-
tially from the coil exit value of Po to near zero at the tank wall. The a-c potential is assumed to
be V„ at the coil exit and zero at the tank wall. An approximation for the variation of the charge
density p from the tank cover to the coil edge is assumed to be
P = Po sinh(/3j:/Z,)/sinh(/3) (12)
where the parameter /3 = L/UT. The distance between the coil exit and the tank cover isJL, the
average oil velocity in the plenum is u, and T is the characteristic decay time in the oil. The
charge profile predicted by Eq 12 is shown in Fig. 6 for the 240-MVA transformer. In this
calculation u = 0.1 ft/s (3 cm/s), T = 20 s, i = 85 in. (2.2 m) and p„ = 4 fiC/xn^. As shown in
Fig. 6, the charge density decays gradually to zero at the wall under these assumptions. At the
bridge exit, that is, 25 in. (63.5 cm) above the coil, the charge density falls to 1.3 /^C/m^.
Tank Ground V = 0, p = 0
t 0^
Oil Flow - P(x), E(x), V(x)
I Bridge
V = Vo,p=Po
Phase Coil
(UFRN) Universidade Federal do Rio GrandeFIG. 5—One-dimensional
do Norte model
((UFRN) Universidade Federal do Rioof upper
Grande plenum.
Tank Wall 1.0
x/L .6
Bridge Exit
Coil Exit 0
p (MC/m I
FIG. 6—Charge density profile above coil.
Field and Potential Distributions in Upper Plenum

The effect of the space charge distribution shown in Fig. 6 on the electric fields and potential
V is found by solving the one-dimensional equations
dE
(13)
ax
and
E = -dV/dx (14)
with p given by Eq 12. The boundary conditions are V = 0 atx = 0 and V = Vo a t x — i . The
results are
y/V„ = 1 - x/L ± K„(p/po - x/L) (15)
and
E/EAV = 1 ± A:„ [1 - |8cosh(|8jc/Z,)/sinh(|3)] (16)
where E^y = V„/L. The parameter Kg is given by
(17)
The ratio E/EAV given by Eq 16 is plotted in Fig. 7 for the upper plenum region of the 240-
MVA
Copyright transformer.
by ASTM The ratio
Int'l (all rights reserved); V/V„
Thu May given
6 14:02:13 EDTby Eq 15 is plotted in Fig. 8. The following parameter
2021
values were used in these calculations: £ = 85 in. (2.2 m), u = 0.1 ft/s (3 cm/s), T = 20 s,
130 INSULATING OILS
Tank Wall Bridge Exit Coil

2 1 1
' 1 '
on Flow ~m 0.1 ft/sec [
Upper Plenum
1 /-
+ 1
1 y
Uniform Field
> 1
<
1 N
_ 1
1
1
1
0 1 1
1
, 1 1
.2 .4 .6 .8 1.0
x/L
FIG. 7—Effect of charge density on electric field above coil.
V„ = 345 V27V3lkV, andp„ = 6.7 ^C/m^, the maximum charge density at the coil exit, Region
D from Table 1.
As can be seen by Figs. 7 and 8, the influence of the space charge in this case is nontrivial.
Depending on the sign of the space charge relative to the sign of the a-c potential, the electric
stress at the coil edge relative to the average stress is alternately enhanced and depressed by
about ±65% at each half cycle of applied stress. Such high stress enhancements at high voltage
coil edges, if actually realized in practice, might help to explain some field failures where leads
and coil edges in the upper plenum were involved. It should be pointed out that the space charge
Bridge Exit Coil Exit
Downloaded/printed by FIG. 8—Effect of charge density on potential above coil.
in the upper plenum would also result in stress enhancement on leads crossing above the coil in
the bridge region. A more accurate analysis of the complex region of coil edges and leads in the
upper plenum is needed to fully assess potential failure mechanisms due to static charge
"clouds."
Conclusions
1. Charge separation and thus streaming currents occur in any oil/paper transformer insula-
tion system regardless of transformer construction, that is, shell form or core form.
2. Streaming electrification is a complex multivariable problem. Variables influencing
streaming electrification include oil velocity, oil temperature, oil chemistry, duct geometry,
solid surface roughness, oil moisture content, contaminants in the oil, and flow turbulence.
3. Streaming electrification in power transformers may enhance electrical stresses within the
structure to the point of initiating partial discharges.
4. An engineering model may be used to predict magnitudes of streaming currents and the
ensuing electrical stress enhancements.
5. Additional work for other transformer configurations is necessary for completeness and is
underway.
APPENDIX
Theory of Flow Electrification
Derivation of Streaming Current for Rectangular Duct

Consider the two-dimensional flow of a low conductivity fluid, such as transformer oil be-
tween two parallel surfaces (conducting or insulating) which forms a rectangular duct 2L X W
(see Fig. Al).
The equation describing the steady state charge distribution p(x,z) between the surfaces is
-DV^p + v-Vp + - = Q (Al)

T
where T — eeo/ff is the characteristic charge relaxation in the fluid with oil conductivity a, and D
is the diffusion constant.
m~
Downloaded/printed by FIG. Al—Two-dimensional flow electrification.
132 INSULATING OILS
In general Eq Al is difficult to solve for realistic fluid velocity profiles v. A useful solution is
obtained, however, assuming v = uz, that is a uniform velocity profile or plug flow where z is a
unit vector along the z-axis. The uniform velocity u is an effective velocity defined such that the
true streaming current in the fluid per unit distance in the î-direction is
I/W=\ pv-dx = u\ pdx (A2)
where p is the charge density given by the solution to Eq Al for v = uz. The parameter W is the
width of the rectangular duct 2L X W and it is assumed that W î> L.
Assume that the fluid enters the duct with a uniform charge density p„. Furthermore, assume
that the first derivative of the charge density p at the fluid-surface interface is a parameter a
which is independent of the coordinates x and z. These boundary conditions are
p(x,o) = PoV/u (A3)
and
dp/dx = ±a (A4)
x=±L
where V is the average flow velocity in the duct.

Following the treatments of flow electrification by Gavis and Koszman [14-16] and of elec-
trochemical processes by Glasstone, Laidler and Eyring [21], the parameter a is written as
ZFC
nd " (I + D/ndkAr' (A5)
where
Z = the number of unit charges per ion,
F = Faraday's number,
Co — the ionic concentration,
n = the transference number,
d = the diffusion layer thickness, and
kji — the rate of ionic adsorption at the oil-wall interface.
The +/— sign is chosen for positive/negative streaming current, that is, for negative/positive
ion adsorption at the fluid-surface interface. The rate of ionic adsorption of the surface kx ac-
cording to the theory of absolute reaction rates is
kT
kA=CA— e<^ - ^'^'"^ (A6)
h
where
CA = the concentration of adsorption sites on the surface,
AF = the free energy change due to external factors such as the application of an external
field, and
AF* = the free energy of activation for the adsorption process.
In arriving at Eq A5 the reverse reaction to ionic adsorption, that is, escape of ions from the
surface, is considered negligible.
Now given the boundary conditions Eqs A3 and A4, the solution to Eq Al for p can be shown
to be
a6cosh(x/6) / a6^\ / 2
sinh(L/6) \ L / \ UT
lab^ " (-1) / z ,

L cos(yirji;/X)expl Oj] (A7)
L 7=1 bj \ UT
where 6 = V D T is the Debye length. The coefficients a, and bj are
VI + Abj{b/urf - 1
a, = ; (A8)
Kb/urf
bj=\ + {jirb/LY (A9)
Now for b/uT « ; 1, which is the case in practical transformer situations, Eq A7 can be ap-
proximated by
ab cosh(jc/6)
smh(L/6)
The streaming current exiting from the duct at z = Z„ is found by inserting Eq AlO into Eq
A2 and integrating. The result is
/ = 2WuaDT{\ - g-Zo/-) + IWLVpê-ô'"^ (All)
Limiting Streaming Current

According to Eq A l l the limiting streaming current, that is, for Z„ -• 00, is predicted to be
ZFDCo
/ „ = IWuaDT = ±2WuT (1 + D/ndkA)~^ (A12)
nd
The bulk electrical conductivity a may be expressed in terms of the ionic concentration €„ as
a = 2ZFDC0 (ZF/RT) (A13)
If Eq A13 is inserted in Eq A12, then the limiting streaming current may be expressed as
'RT\ ( 1
/o. = ± W u t t , \ ^ 1 \—^](X^ D/ndkA)-' (A14)
When the electrification process is diffusion controlled D/ndkx « 1 and the limiting stream-
ing current reduces to
I^=±Wuee.{^)I^J (A15)
which is the form employed in modelling transformer electrification in this paper. In order to
compute / „ using Eq A15, the effective velocity u and the diffusion film thickness d must be
determined.
134 INSULATING OILS
The diffusion thickness is assumed to be the same as that for mass transport in the fluid
which for pipe flow may be approximated by
d = Ir/Nsu (A16)
where r is the pipe radius and Nû the Nusselt number. The Nusselt number can be approxi-
mated for high Schmidt number fluids by [15,16]
Ni^, = 0.0223 AfÂ^r (A 17)
where
Nsc — f/D is the Schmidt number and
Nue — 2rV/v is the pipe Reynolds number.
The effective velocity u depends upon the Debye length and boundary layer thickness. An
empirical estimate of this dependence is expressed as
u = V(b/bBr (A18)
where
6/5fl = the ratio of the Debye length to the boundary layer thickness, and
m = an adjustable parameter near unity.
It is assumed that bg can be approximated by the laminar subzone boundary layer which for
pipe flow is
bg = 50r/N]^,^ (A19)
For a rectangular duct where W S> i , the pipe radius r in Eqs A16 and A19 is replaced by i .
References
[1] Klinkenberg, A. and Van der Minne, J. L., Electrostatics in the Petroleum Industry, Elsevier Publish-
ing Co., Amsterdam, 1958.
[2] Shafer, M. R., Baker, D. W., and Benson, K. R., Journal of the National Bureau of Standards, Vol.
96C, No. 4, 1965, pp. 307-317.
[3] Itoh, M. et al., Transactions of the Institute Electronic Engineers of Japan, Vol. 93-A, No. 5, 1973,
pp. 175-183.
[4] Kan, H. et al., MitsubishiDenki Giho, Vol. 52, No. 12, 1978, pp. 915-919.
[5] Shimizu, S., IEEE Transactions on Power Apparatus and Systems, Vol. PAS-98, No. 4, 1979, pp.
1244-1250.
[6] Higaki, M. et al., IEEE Transactions on Power Apparatus and Systems, Vol. PAS-98, No. 4, 1979,
pp. 1275-1282.
[7] Okubo, H. et al., "Charging Tendency Measurement for Transformer Oil," Transaction A79051-4,
IEEE Winter Power Meeting, New York, NY, February 1979, Institute of Electrical and Electronics
Engineers, New York.
[8] Tanaka, T. et al., IEEE Transactions on Power Apparatus and Systems. Vol. PAS-99, No. 3, 1980.
[9\ Tamura, R. et al., IEEE Transactions on Power Apparatus and Systems, Vol. PAS-99, No. 1, 1980,
pp. 335-343.
[10] Takagi, T. et al., "Reliability Improvement of 500 kV Large Power Transformers," CIGRE Paper, 12-
02, 1978.
[//] Oommen, T. V. and Petrie, E. M., IEEE Transactions on Power Apparatus and Systems, PAS-103,
No. 7, 1984, pp. 1923-31.
[12] Oommen, T. V., "Static Electrification Properties of Transformer Oil," IEEE 1986 Annual Report,
Conference on Electrical Insulation and Dielectric Phenomena, 86CH2315-0, Institute of Electrical
and Electronics Engineers, New York, 1986, pp. 206-213.
[13]by Crofts,
Copyright ASTM Int'lD.
(all W., "The Static
rights reserved); Electrification
Thu May Phenomena in Power Transformers," IEEE 1986 Annual
6 14:02:13 EDT 2021
ROACH ANDTEMPLETON ON POWER TRANSFORMERS 135
Report, Conference on Electrical Insulation and Dielectric Phenomena, 86CH2315-0, Institute of

Electrical and Electronics Engineers, New York, 1986, pp. 222-236.
[14] Gavis, J. and Koszman, 1., Journal of Colloid Science. Vol. 16, 1961, pp. 375-391.
[15] Koszman, I. and Gavis, J., Chemical Engineering Science, Vol. 17, 1962, pp. 1013-1040.
1/6] Gavis, ] . , Chemical Engineering Science. Vol. 19, 1964, pp. 237-252.
[17] Schon, G., Elektrostatische Aufladungsvorgange und ihre Zundungsgefahren, Handbuch der
Raumexplosion, Verlag Chemie, 1965, pp. 302-366.
[18\ Abebian, B. and Sonin, A. k.. Journal of Fluid Mechanics. Vol. 120, 1982, pp. 199-217.
[19] Tanaka, T., Yamada, N., and Yasojima, "^^ Journal of Electrostatics, Vol. 17, 1985, pp. 215-234.
[20] Melcher, J., Lyon, D., and Zahn, M., "Flow Electrification in Transformer Oil/Cellulosic Systems,"
IEEE 1986 Annual Report, Conference on Electrical Insulation and Dielectric Phenomena,
86CH2315-0, Institute for Electrical and Electronic Engineers, New York, 1986, pp. 257-265.
[2/1 Glasstone, S., Laidler, K. ] . , andEyring, H., The Theory of Rate Processes. Chapter X, McGraw Hill,
NY, 1941.
Dan W. Crofts^
Static Electrification in Power

Transformers
REFERENCE: Crofts, D. W., "Static Electrification in Power Transformers," Electrical Insulat-

ing Oils, STP 998. H. G. Erdman, Ed., American Society for Testing and Materials, Philadel-
phia, 1988, pp. 136-151.
ABSTRACT: The subject of static electrification and related transformer failures has gained the
attention of the industry in the last decade. To some extent electrification activity occurs with
flowing insulating oil passing over insulation items within every transformer. The extent of the
electrification determines whether or not this activity is detrimental to the dielectric integrity of the
transformer.
Within the paper, failures are described—two of which occurred on the TU Electric system.
Several others have occurred in the United States and even more worldwide. Many of the actual
failures have not been well documented and statistical information is lacking in the literature.
Electrification activity is enhanced by many variables which are described in the text. Methods
of minimizing detrimental activity are discussed and the need for monitoring techniques is
stressed.
KEY WORDS: static electrification, electrostatic charging tendency, streaming electrification
Introduction
With a growing list of failures of large power transformers attributed to the static electrifica-
tion phenomena, many affiliated with the electric power industry are concerned with the search
to find an accurate understanding of the phenomena. Methods to monitor and minimize static
electrification activity are needed to prevent additional destructive events.
The paper briefly describes two failures that occurred on the TU Electric system and how
these failures relate to static electrification. In addition, an update will be given of the known
failures, which involves several manufacturers of transformers and several suppliers of insulat-
ing oil.
Also covered are transformer characteristics which enhance static electrification develop-
ment, including design, insulating fluid, insulation material, other construction material, con-
tamination, and operation of transformer ancillary systems.
Since the oil and the transformer externals are easily accessible, attention to better monitor-
ing methods are essential for risk assessment and management. The characteristics of the insu-
lating fluid can serve to provide an indicator of internal distress in a transformer, even though
the fluid itself may not be the sole contributor.
Failures
Japanese—In 1972, there were two (assumed from comments made at the 1986 Conference
on Electrical Insulation and Dielectrics Phenomena [22]) large power transformer failures while
undergoing tests at the factory in Japan. Both units had voltage ratings of 500 kV with power
ratings of 750 and 1000 MVA, respectively. One unit was of a shell-form design and one was of a
core-form design. Both failures were attributed to streaming electrification. There also is a high
probability of field failures not reported to the industry.
'Manager
Copyright of(all
by ASTM Int'l Transmission and
rights reserved); Thu Distribution
May 6 14:02:13 EDT Operations,
2021 TU Electric, Dallas, TX 75266-0268.
136
CROFTS ON POWER TRANSFORMERS 137
C/G/IE—Publications of the International Conference on Large High Voltage Electric Sys-

tems (CIGRE) have mentioned static electrification problems [2].
Consolidated Edison—In January 1979, an experimental compact d-c terminal constructed,
operated, and studied at the Astoria station on the Consolidated Edison Co. system encoun-
tered a static electrification problem in the converter valve cooling system. Cooling for the con-
verter valves is accomplished by circulating Refrigerant 113 (trichlorotrifluoroethane) through
fiberglass epoxy-reinforced tubes and insulating Tefzel (modified copolymer of ethylene and
polytetrafluoroethylene) tubes. Circulation of this refrigerant through these tubes resulted in
spark discharges that eventually punctured the tubing and resulted in coolant leaks.
In a solution to the problem, an antistatic additive (Mobil DCA-48) was added to the coolant.
To date, there has been no recurrence of the static electrification problem. This terminal has
seen only intermittent operation and has never been commissioned by Consolidated Edison
19,20,26].
Public Service Electric and Gas (PSE&G)—A unit manufactured for PSE&G experienced
internal flashovers from static electrification while undergoing tests at the manufacturer's fac-
tory. When the cooling pumps were run, discharges could be heard and were even seen by fac-
tory personnel. Other identical units at the factory did not display the same phenomena. The
only difference between the transformers was the oil in the troubled unit. The oil was from a
different refinery than that used in the other transformers.
TUElectric—On 12 Dec. 1983, TU Electric experienced a failure of a three-phase, 345/138-
kV, 450-MVA, shell-form, conservator-equipped autotransformer at Renner Substation. The
failure occurred approximately three months after carrying load for the first time, with no fore-
warning of distress [11-13,16].
The failure involved a flashover in the main tank from the "C" phase high-voltage winding
near the line end to a low-voltage lead going to the no-load tap changer near the neutral end (see
Fig. 1). The flashover traversed approximately 42 cm (16.5 in.) of oil space. The failure was
severe and resulted in an oil spill of approximately 79.5 m^ (21 000 gal).
FIG. 1—Top of coil showing failure location.
138 INSULATING OILS
The unit was returned to the manufacturer for teardown, inspection, and rebuilding. During
the latter part of March of 1984, manufacturer's personnel, in conjunction with TU Electric
representatives, began the layer-by-layer disassembly of the damaged winding phase packages.
We were searching for an underlying reason for the flashover of 42 cm (16.5 in.) of oil above the
"C" phase winding. Layer-by-layer disassembly of the unit disclosed many areas of potential
electrification failure. There was evidence of tracking apparently caused by static electrification
on coils, static plates, insulation pieces, and oil ducts throughout the transformer. This track-
ing followed oil circulation paths. There was also evidence of carbon deposits, but closer inspec-
tion revealed evidence of heavier deposits in areas of higher oil velocity (the carbon particles in
the oil apparently assume a positive charge and seek areas of negative potential).
In addition, low energy electrical punctures in the conductor insulation were found in a coil in
Phase "C." These could have eventually led to failure because the punctures were bridged by a
carbonized path in adjacent insulation pieces to other conductors in the same winding layer.
The activity was turn to turn. There was no evidence of a-c power follow current between the
conductors.
Figure 2 shows another type of tracking, which we have erroneously termed as "wormholes,"
found in insulation washers throughout the unit. These "wormholes" are subsurface paths pre-
sumably caused by static electrification (apparently negative charges seeking a path to ground
within the pressboard). Studies by the manufacturer have discounted moisture or contaminants
within the pressboard as causative. Nevertheless, notice that these electrification paths did fol-
low the oil flow. Throughout the transformer, some of these subsurface paths were carbonized
while others are not. Conversely to our manufacturer's opinion, the Japanese authors, during
the 1986 Conference on Electrical Insulation and Dielectrics Phenomena (CEIDP), offered the
theory that "wormholes" in the insulation are created when moisture within the pressboard
expands from the heat generated from static electrification leakage current at the surface. That,
in effect, causes bubbles to form within the insulation creating pressure with the obvious conse-
quences. The credence of this hypothesis remains to be determined.
Figure 3 shows evidence of static electrification on the bottom of "B" phase at the joint clo-
sure. This photograph was taken at the bottom end of the coil package. Also, the affected areas
are where oil exits from pumps into the lower plenum. Vorticity patterns and possible turbu-
lence are existent, yet additional velocity and vorticity changes occur as the oil enters the wind-
ings in the joint closure area. The disassembly of this phase and Phase "A" disclosed similar
evidence of static electrification throughout the phase packages [11-13,16],
Downloaded/printed by FIG. 2—Subsurface tracking or "wormholes.
FIG. 3—Lower section of winding showing electrification activity at the joint closure.
The particulars of the failure at Renner parallel many of the contributing factors discussed in
the IEEE transactions referenced. The operating temperature of our autotransformer was in the
intermediate range (18° to 48°C). The unit was lightly loaded with the cooling running in the
manual mode. This unit was equipped with eight pumps rated 4.476 kJ/s (6 hp) and 3.03 m^/s
(800 gal/min). The moisture content of this unit was 7 ppm.
The unit contained transformer oil displaying peculiar charging tendencies. Charging ten-
dency tests were made on the failed unit's insulating oil after the failure. As a result of the
failure, the oil samples were contaminated with carbon and other impurities. The charging
tendency of the oil prior to failure was estimated to be in the 600 to 700 /iC/m^ range from tests
made following failure.
On 24 May 1984, a second 450-MVA autotransformer, identical to the Renner unit (built on
the same shop order), failed in service at the TU Electric Anna Switching Station. An internal
inspection revealed damage similar to the Renner unit [16].
The failure again involved a flashover in the main tank from the "C" phase high-voltage
winding near the line end to the low-voltage lead going to the no-load tap changer near the
neutral end. The flashover again traversed approximately 42 cm (16.5 in.) of oil space at the top
of the winding. Another oil spill occurred as 79.5 m-' (21 000 gal) of flaming oil spilled from the
transformer. It was obvious that the transformer had been on fire and had self-extinguished.
The internal inspection also revealed a flash across the bottom porcelain of the " B " phase high-
voltage bushing. This was unlike the Renner failure. However, oscillograph records indicated
the fault started on " C " Phase. Carbon deposits and fire damage to the exposed top section of
the shell-form transformer windings made it impossible to find evidence of static electrification.
Prior to failure all eight pumps were running (mistakenly left in the manual "ON" position)
with the oil temperature in the intermediate [313 to 318°K (40 to 45°C)] range. The pumps had
been on for only twelve days. The oil had a charging tendency measurement of 970 /iC/m-' prior
to failure and was from the same oil supplier as in the previous failure. In both cases, the oil was
furnished by
Downloaded/printed by the transformer manufacturer in accordance with their specifications.
140 INSULATING OILS
In an attempt to prove or disprove the static electrification theory as a failure cause, a pump
was removed in an attempt to see the bottom of the winding through the pump mounting area.
This did not work; however, a hole approximately 61 by 38 cm (24 by 15 in.) was cut in the side
of the transformer at the pump mounting area. With mirrors, electronic flash, and a telephoto
lens, pictures of the bottom of the " C " phase coil package were taken. These pictures showed
the same type of tracking as was seen on the previous failed transformer.
The failed Anna unit was returned to the manufacturer for untanking and investigation. A
layer-by-layer inspection of the " C " phase winding showed evidence throughout of static electri-
fication, apparently once again the cause of the failure. There was tracking on coils, insulating
washers, oil ducts, and joint closures. Numerous locations of incipient or potential failures were
seen throughout the unit. Evidence of subsurface paths ("wormholes") in the pressboard and
insulation punctures were also found.
The bottom end of "A" and " B " phase packages had tracking similar to "C" phase. The
tracking was most evident in the areas where there are changes in oil flow parameters. Also, the
tracking was very similar to that found on the bottom end of the phase packages on the Renner
transformer. In both the Renner and Anna failures, the static electrification damage was so
severe that eventual transformer failure would have occurred exclusive of the flashovers above
the windings. This represents several failure modes operating simultaneously.
Union Electric—On 24 June 1984, a 345/138-kV, 560-MVA autotransformer owned by
Union Electric Co. failed at the Belleau Transmission Substation. This transformer is a shell-
type design with a total of five oil cooling pumps and failed less than one year from energization
while carrying 150 MVA of load. All cooling was running in the manual "ON" position. An
internal field inspection indicated the failure was initiated between the high-voltage (HV) wind-
ing of "A" phase to a test tap lead. This represents an arc path of 24 cm (9.5 in.) through oil
above the windings. The oil was tested for static charging tendency and measured 510 /iC/m^.
Damage was not extensive, and the unit was repaired in the field and returned to service [14].
CentraisElectricasDoSulDoBrasilS.A.—ELECTROSUL—lnFehma.iy 1985, ELECTRO-
SUL suffered the failure of a 200-MVA, 525/230-kV autotransformer. This transformer is of
shell form design and failed violently in service after four years of operation. The unit failed
between the outer first turn of the HV winding and the tertiary winding connecting lead. In June
1985, an identical transformer failed in service. The unit also failed between the outer part of
the HV winding and the tertiary winding connecting lead. ELECTROSUL personnel feel cer-
tain that these two failures are due at least partially to static electrification.
Alabama Power—On 10 Sept. 1985, a single-phase, 500/230-kV, 448-MVA autotransformer
owned by Alabama Power Co. failed at the South Bessemer Substation. The unit is a shell-type
design and failed several hours following energization without load. The pumps were running
continuously in the manual "ON" position. Static charging tendency of the oil was 227 ^C/m-'.
The oil was from a different supplier than that used in the failed TU Electric and Union Electric
transformers. These units, however, were built by the same manufacturer. The Alabama unit
and the two TU Electric transformers were disassembled in the factory. The manufacturer dis-
closed the pressboard to be from the same supplier of insulating paper (screen-like surface tex-
ture).
Factory inspection of the Alabama unit revealed that the fault was initiated between the "A"
phase HV series lead and the "A" phase LV series lead located 33 cm (13 in.) apart above the
windings. The normal operating potential between these points is approximately 288-kV. Signs
of tracking were also found in the bottom section [18].
Another identical unit in the same three-phase bank at Alabama Power was given electrical
field tests since undergoing similar field operating conditions. The unit failed the field test and
was returned to the factory where the field test failure was duplicated. Teardown of the unit
disclosed a manufacturing defect to be the failure cause. Layer-by-layer disassembly of this unit
did not show
Downloaded/printed by any evidence of static electrification activity. The pressboard insulation, however.
was smooth and from a different supplier than other-mentioned failures, as per the manufac-
turer [27].
American Electric Power (AEP)—In November 1985, AEP experienced a failure of a single-
phase, 500-MVA, 765/345-kV autotransformer. This transformer failed during insulation test-
ing. The point of failure was from the high-voltage winding to the neutral lead. In December
1985, an identical transformer failed violently and caught fire after 7.5 h of service at no load.
The failure was again from the high-voltage winding to the neutral lead. It is surmised that
these two transformers failed due to static electrification.
In January 1986, a single-phase, 500-MVA, 765/26-kV generator step-up transformer failed
in service while at minimum load. The failure occurred between a high-voltage series crossover
and top leads at the neutral end of the windings. This failure is believed to have been possibly
caused by static electrification.
South Africa—In 1986, a South African utility also experienced a failure of a large power
transformer of core form design. Unconfirmed reports have indicated design concerns were
included in the manufacture in an attempt to prevent static electrification flashovers at the
upper winding exits. The failure, however, was located near the bottom. The unit was also
equipped with a conservator. Excessive cooling was apparently included in the design of this
unit [28]. At the April 1987 Doble Conference, it was inferred that a similar failure of a previous
unit was probably due to electrification rather than resonance as previously postulated.
Australia—In 1987, the Electricity Council of New South Wales suffered two large generator
step-up transformer failures. These transformers are rated 23/330 kV, 390 MVA and are of the
same design and manufacturer. They are of core form design with concentric windings built on
five horizontal and two vertical legs. The transformers are outfitted with two oil pumps rated
5.68 m-'/s (1500 gal/min) that are wired for automatic control.
One of the failed units has been untanked and inspected at the factory and found to have an
apparent turn-to-turn failure in the center portion of the HV winding of one phase. Inspection
did not reveal, however, any evidence of tracking or "wormholes" that are typically found in
failures attributed to static electrification. The other failed unit, at the time of this writing, has
not been untanked, but the field internal inspection reveals evidence of similar damage.
Tests, including dissolved gas in oil, made previous to the failures did not indicate any abnor-
malities. Since these two failures, the utility has performed partial discharge tests on identical
in-service units, but, for the most part, all of the investigative effort is devoted to the failed
units. The utility is investigating all possible failure modes, such as excessive stress distribution
and static electrification or a combination of the two.
Other—Several other transformer failures that are unexplained have also been suspected to
be a result of static electrification. Unfortunately, the evidence was destroyed long before static
electrification modes were well known.
Failure Scenario—The following scenario is offered in an attempt to rationalize how static
electrification failures may have occurred at the top of the windings. Electrification failures,
however, can occur anywhere within an operating transformer:
For cooling purposes, the forced oil flow in power transformers is directed into the bottom of
the windings and out the top. Charged are generated, carried, and distributed by this flow of oil
across insulation in the oil path. Through this process, and other methods of charge generation
or injection, an impaired dielectric path could be created over which a power discharge arc
could occur where the oil exits at the top of the windings. The impaired dielectric path may be
only clouds of charges impairing dielectric strength in the oil space above the windings. The
power follow current would then cause the failure of the transformer. The previously mentioned
impaired dielectric path could also result from d-c potential differences allowing d-c flashovers,
a corollary to those occurring in nature during a thunderstorm. Prior researchers of the static
electrification
Copyright by ASTM Int'l (allphenomena have
rights reserved); Thu May 6hypothesized
14:02:13 EDT 2021 d-c potential differences as high as hundreds of
142 INSULATING OILS
thousands of volts or more. Also, they have actually observed static discharges in power trans-
formers (See Fig. 4).
The South African failures occurred in the bottom of the windings of core form transformers.
Perhaps a similar scenario can be developed for these failures [30].
Conditions Influencing Static Electrification

In the late 1970s and early 1980s, the Japanese conducted detailed research of the static elec-
trification phenomena. Much of the research work was published in IEEE transactions [3-8].
Since then, much has been learned from other researchers with many areas not yet clearly un-
derstood.
Accordingly, static electrification in power transformers is influenced by several factors in-
cluding:
1. Oil temperature. Charging tendencies of the oil increase with temperature. However, in a
transformer, the charging level reaches a maximum due to a charge leakage process referred to
as "relaxation." The maximum charging tendency apparently occurs at some intermediate tem-
perature [variable between 303 and 333°K (30 and 60°C)] [1.3-5,7,10]. Current research has
not verified that the maximum hypothesis is germane. This may be due to the physical limita-
tions of the research models.
An Example of the Discharge Caused

by Streaming E l e c t r i f i c a t i o n .
FIG. 4—From Japanese research.
2. Oil moisture content. Charging tendencies of insulating oils increase as moisture levels
decrease. Oils with moisture content in the range of 15 ppm or less can have higher charging
tendencies [10],
3. Flow rate. Charging tendencies increase with greater flow rates. Japanese research that
displayed the increase varied somewhere between the second and the fourth power of the veloc-
ity of the oil flow. Some of the Japanese research was done at high flow rates, higher than most
flow rates in transformers manufactured in the United States. For example, the average flow
rate involved in our failures was 21.3 cm/s (0.7 ft/s) over an average typical cross section. This
flow rate increases to 45.7 to 61.0 cm/s (1.5 to 2.0 ft/s) in coil areas and to greater than 457
cm/s (15 ft/s) near pumps and pipes in the heat exchangers. Ongoing research has not con-
firmed electrification activity being proportional to as high as the fourth power of the velocity
[24].
4. Turbulence of oil flow. Japanese research indicates: "the charge motion or generation
does not depend on diffusion, but turbulent motion of oil" [5,7].
5. Contaminants or surface active agents. Charging tendencies greatly increase with trace
quantities of surfactants such as sulfonates present in the oil. Interfacial tension can be greater
than 40 dyne/cm (satisfactory under present standards) with surfactant contamination of less
than 50 ppm. This concentration can create undesirable charging tendencies in the oil [10],
6. Charging tendency of oil. Some oils characteristically have higher charging tendency than
others. This could possibly be based on the source of crude, the refining process, and other
variables or contaminants yet to be identified. Additional research is needed for establishment
of levels of concern for and the importance of charging tendency measurements. We have as-
sumed, however, that ranges for typical new oils may have been in the order of 1 to 150 fiC/m^
(through mid-1986) with less than 25 /iC/m^ for low-charging oils and as high as 1730 ^C/m^ on
our highest test for aged oils. The latest tests on new oils have shown that the refiners have
apparently addressed the charging tendency. Oils available for use in 1987 appear to test in the
order of 25 /iC/m^ or less. As strange as may seem, the refiners have not shared the reason for
this apparent decrease.
7. Surface conditions. The charge generation/separation process is enhanced with increases
in roughness of the solid insulation. The texture of the insulation surface apparently affects
charge generation/separation. Insulation materials from different suppliers have different sur-
face characteristics. The pressboard used in our transformers and the Alabama unit involved in
failures was apparently from the same supplier, according to the transformer manufacturer.
The surface of the pressboard was not smooth (screen-like texture) as a result of the manufac-
turing process of the paper.
8. Energization. Static electrification intensity is proportional to increasing a-c field
strength, probably due to charge injection processes. Current research in flowing oil/press-
board systems has shown the magnifying effect of an alternating current field can enhance the
localized charge density perhaps as much as five-fold.
9. Dielectric strength of oil in motion. The dielectric strength of flowing transformer oil var-
ies with the velocity of flow and is not necessarily the same as when tested at rest. With due
respect to model dependency, a sharp dip has been demonstrated at flow rates of near 50 cm/s
(1.6 ft/s). Obstructions to oil flow paths and flow direction modifiers can create cellular convec-
tion. Likewise, cellular convection has appeared as flow rates leave laminar and approach tur-
bulent velocities. These convection cells have been observed with Reynolds numbers in the 97 to
2500 range. These measurements were made using a particular laboratory model and are di-
mensionally dependent. There may not be an absolute parallel comparison in a dimensionally
different operating transformer. Nevertheless, such convection cells are formed. The range of
97 to 2500 is similar to the Reynolds numbers calculated for flow through the insulating washers
of many transformers. Cellular convection in itself apparently enhances the charge injection
process into the oil stream [23,25]. The dielectric strength of oil also varies with temperature,
moisture
(UFRN) Universidadecontent,
Federal do Riogas bubble
Grande do Nortepresence, contamination,
((UFRN) Universidade anddoparticulate
Federal do Rio Grande Norte) pursuant tomatter.
144 INSULATING OILS
10. Migration of moisture. In an oil/paper operating transformer system, there is a migra-

tion of moisture between the oil and the insulation. As temperature and pressure changes occur
in a transformer, the moisture migrates to reestablish equilibrium conditions. As the moisture
leaves or enters the surface of the insulation, there is a corresponding change in the conductivity
of the fluid at the interface. This condition likewise affects the charge separation process.
11. Particulate matter. Particulate matter in the transformer oil from the manufacturing or
installation process has an effect on electrification activity in addition to the previously men-
tioned dielectric characteristics.
12. Charge injection. Upstream charge injection has been demonstrated to affect down-
stream dielectric integrity [29].
13. Pumps. Pumps are suspected as being a substantial source of charge generation. This
presumption requires confirmation and accurate quantification on an operating transformer or
acceptable semblance [29].
14. Orifices. Orifice effects have also been demonstrated to generate charge.
Transfonner Characteristics Influencing Static Electrification

Not unlike most products, a transformer is designed to operate safely and reliably for a rea-
sonable length of time and be produced at a price that is affordable to the consumer yet profit-
able to the manufacturer. As technology has advanced, design improvements have been imple-
mented and manufacturing techniques and materials have been improved to effectively produce
a product that will meet today's growing power requirements at a minimum cost. Utilities de-
mand larger transformers yet; conversely, shipping limitations impose dimensional and weight
restrictions. As an option, many manufacturers pursued improved heat transfer and computer-
refined design. In the challenge of meeting these requirements, some factors involving static
electrification were neither adequately recognized nor appropriately considered. As the many
static electrification variables entered the picture, something finally exceeded the dielectric
withstand and failures occurred.
These factors include the design and selection of solid insulation materials, liquid insulation
cooling system design, pump design, charge relaxation consideration, application of charge
collectors, use of additional conductor clearances, addition of conductivity modifiers, lack of
adequate design information, and absence of methods of monitoring.
Insulating Fluid—As mentioned previously, studies have shown that the charging tendency
of the liquid insulation is influenced by many variables. Insulating oil standards do not ade-
quately address the influence of oil in electrification activity.
A test of transformer oil that is most commonly associated with static electrification is static
charging tendency. Liquid insulation possessing a high static charge density has been related to
increased potentialfor static electrification activity. Static charging tendency generally has been
measured using the "Mini-Static" testing device (see Fig. 5) adapted by Westinghouse, which
has yet to be accepted by industry as the standard test procedure [/]. The true significance of
this test remains unresolved, but who can offer a better method of forewarning? Has anyone yet
developed an improved method of measuring the accumulation of charges within the oil? Let us
concentrate our efforts on formulating a better method rather than criticizing the innovative
efforts of others.
Other properties which may identify the liquid insulations charging tendency, such as con-
ductivity and resistivity, are being investigated with favorable optimism.
Why, I ask, have existing oil test methods only considered static oil? Oil in an operating
transformer is constantly in motion. Why then haven't standards addressed the characteristics
of oil in motion?
Insulating Material—Previous study has shown that the physical properties of the solid insu-
lation material
Downloaded/printed by will influence static electrification activity. The type of material used along with
its Universidade
(UFRN) texture and Federalsurface patterns
do Rio Grande will influence
do Norte ((UFRN) Universidade the degree
Federal of static
do Rio Grande electrification
do Norte) activity.NoGener-
pursuant to License Agreement. further reproductions auth
CROFTS ON POWER TRANSFORM ERS 145
ally, the charge separation will be enhanced as the roughness of the solid insulation material
increases. The surface characteristics can be greatly influenced by the manufacturing tech-
niques of the insulation items. Unfortunately, however, the removal of insulation for testing
purposes is not even practical for consideration.
Operation of Transformer Ancillary Systems—Since the liquid insulation in a power trans-
former also serves as a coolant, it is imperative that a means of efficient heat transfer be de-
signed. This effective heat transfer process usually takes the form of numerous directed cooling
ducts, forced oil flow pumps, fans, coolers, and radiators. Historically, cooling systems have
been designed to assure that the transformer will operate at a certain temperature rise over
ambient temperature for a given load capacity. As a result, transformer cooling systems were
designed and built with precarious consideration given to influence on static electrification.
Users should operate existing units in a manner to avoid increased exposure to static electrifica-
tion problems, that is, automatic modes for cooling system operation as was intended in the
design.
As mentioned previously, the oil velocity or flow rate, oil temperature, surface conditions,
and vorticity of oil flow all influence the existence and degree of static electrification activity.
Flow considerations can be controlled by altering transformer design without sacrificing trans-
former load capacity.
Prudent operation of transformers necessitates understanding of the conditions that mini-
mize static electrification activity. Currently, the state of the art involves operating cooling con-
trols in the automatic mode to avoid susceptible temperature ranges, monitoring of the insulat-
ing fluid and modifying if necessary, implementing field modifications as agreed to by the
manufacturer, and participating in investigations to better understand static electrification. All
of this is being done without the existence of a recognized good test or indicator to give fore-
warning of distress.
Monitoring Methods
Rather than employing "ostrich" techniques and ignoring static electrification failure modes,
many have chosen to explore the feasibility of finding a system to provide forewarning of dis-
tress. Certain tests are available, but, in themselves, they do not offer adequate monitoring to
prevent the "blindsiding" failures that have and will continue to occur.
Insulating Oil Tests

Static Charging Tendency—Currently, the most popular means of identifying transformers
exhibiting potential static electrification problems involves the measurement of static charging
tendency of the oil (test method not yet approved by ASTM). This test, originally developed for
petroleum fuels and later adapted for electrical insulating oils, measures the charge density
generated in a volume of oil passed through a filter [4,10]. Some researchers use the term elec-
trostatic charging tendency (ECT). The apparatus used, described in Fig. 5, is an adaptation of
the measuring device developed by Exxon in 1972 known as the "mini-static tester" [10].
In principle, an oil sample is forced through a specified filter and a charge separation occurs.
The charge on the filter is measured by an electrometer and is usually negative. Charging ten-
dency is defined as the amount of charge generated per unit volume of oil in the flow. The
practical unit for charge in the present measurement system is microcoulombs (^C); therefore,
charging tendency is usually expressed as nC/rn^.
TU Electric and companies such as Doble Engineering, Shell, Exxon, Gulf, Westinghouse,
and Union Electric built "ministatic" testers and began testing and investigating oils following
the failures.
CopyrightAfter a period
by ASTM of reserved);
Int'l (all rights severalThuyears
May 6 and many
14:02:13 tests, factors that affect the measurement of charg-
EDT 2021
ing tendency have been found. Test parameters such
(UFRN) Universidade Federal do Rio Grande do Norte ((UFRN) Universidade Federal asGrande
do Rio typedoofNorte)
sample
pursuantbottle, method Nooffurther
to License Agreement. sam- reproductions auth
146 INSULATING OILS
TO GAS SUPPLY
.RUBBER
STOPPER
CLAMPS •
SYRINGE ELECTROMETER
/
FILTER
VALVE
HOLDER-
- TO CHART
RECEIVER
T RECORDER
. TEFLON BLOCK
FIG. 5—Electrostatic charging tendency measurement device.
pling, effects of storage time, light, exposure, moisture, temperature, agitation, type of filter
paper, flow pressures, number of runs, and measurement interpretation are variables needing
to be addressed by standards to develop approved methods for consistent results; then maybe
guidelines can be established for a tool to monitor electrification activity, perhaps one of several
tools.
Oil Diagnostic Tests—Other tests on transformer oil, such as ASTM Test Method for Inter-
facial Tension of Oil Against Water by the Ring Method (D 971), ASTM Test Method for Wa-
ter in Insulating Liquids (Karl Fischer Method) (D 1533), ASTM Test Method for Dielectric
Breakdown Voltage of Insulating Oils of Petroleum Origin Using VDE Electrodes (D 1816),
Percent Power Factor at 100°C, Conductivity (D 3114), and ASTM Test Method for Specific
Resistance (Resistivity) of Electrical Insulating Liquids (D 1169-80), have been performed to
determine their relation to static charge density. Results of a TU Electric study on in-service oils
indicate:
1. A correlation between increasing conductivity (greater than 1.0 pS/m) and decreasing
resistivity (less than 10 times 10''' fi-cm) exists with increasing charging tendency of oil (increas-
ing charging tendency is negative). See Figs. 6 and 7.
2. Service-aged oils with high static charging tendency (150 /^C/m-' or greater) in our tests
exhibited high power factor at 375°K (100°C) (greater than 1.0%) and low interfacial tension
(less than 40 dyne/cm). The converse did not necessarily hold true.
Dissolved Gas In Oil—Dissolved gas-in-oil analyses from power transformers are excellent
tools utility maintenance personnel use to evaluate transformer condition. It is assumed that
static discharge activity is very similar to partial discharge activity and will generate a similar
gas profile. In terms of percentage of the total combustible gas produced by static discharge
activity, hydrogen (H2) should represent 75% or greater, carbon monoxide (CO) 10 to 15%,
methane (CR,) 5 to 10%, and ethylene (C2H4) 0 to 5% [5].
Prior to the Renner failure, a dissolved gas-in-oil analysis was performed 22 days after the
unit was energized carrying no load. Results of the dissolved gas-in-oil analysis were:
Carbon dioxide (CO2) 167 ppm

Hydrogen
Copyright by ASTM Int'l (all rights reserved); Thu May 6 (H2)
14:02:13 EDT 2021 10 ppm
Carbon monoxide (CO) 7 ppmto License Agreement. No further reproductions auth
(UFRN) Universidade Federal do Rio Grande do Norte ((UFRN) Universidade Federal do Rio Grande do Norte) pursuant
CHARGE DENSITY VS. CONDUCTIVITY

OF WKSTINGHOUSB TRANSrORlBH OILS
-300 -MO -100
CHABci Dimmr (uc/m3>

FIG. 6—Conductivity versus electrostatic charging tendency.
Methane (CH4) 2 ppm

Ethylene (C2H4) 2 ppm
Due to the low concentration of these gases, little, if any, significance was given to these
results before the failure. Even if existing technology hydrogen monitors were installed, these
CHARGE DENSITY VS. VOLUME RESISTIVITY

OF TPL mSTINGBOUSI TIUNSFORICIB OILS
-400 -300 -MO -IM
Copyright by ASTM Int'l (all rights reserved); Thu May

dUBoa oiNstrr (acvms)
6 14:02:13 EDT 2021
Downloaded/printed by FIQ. 7—Resistivity versus electrostatic charging tendency.
148 INSULATING OILS
values would have been ignored and would have been well below the alarm point chosen for the
monitors. Since most units normally exhibit higher concentrations of combustibles, it may be
difficult to assess how much gas is generated due to static electrification discharges alone. We
do, however, continue to monitor each suspected unit for signs of distress by using both dis-
solved gas in oil tests and installed monitors on suspicious units. Such tests, however, have not
detected any conditions anticipating electrification troubles.
External Tests
Acoustical Emission—Partial discharge signals have some similarity to static discharge sig-
nals [15]. With this in mind, partial discharge detectors have been used with the intent of identi-
fying static electrification activity. The acoustic emission partial discharge detector (PDD 101)
developed by Electrical Power Research Institute (EPRI) under RP 426-1 has been used by TU
Electric to detect static discharges. Our experience, thus far, has indicated that higher levels of
activity (measured in counts) occur near the transformer pumps. Whether this activity is actu-
ally static discharges or possible pump noise remains undecided. Research in this area, such as
EPRI RP 426-2, studied the effects of oil pump performance on static electrification and at-
tempted to identify static discharge signals [15]. The results failed to answer the questions.
Electrical Emission—Another instrument designed to measure partial discharges, developed
by Biddle Instruments, was used by TU Electric to measure the partial discharge energy in a
deenergized transformer. Partial discharge energy measurements were made continuously for
3 h on a deenergized 800-MVA EHV transformer stored at Martin Lake SES. Results from
these tests indicated a gradual increase in partial discharge energy when the pumps were turned
on and a gradual decrease when the pumps were turned off. This energy was believed related to
static electrification. This is further verified by the elevation of charging tendency measure-
ments of the oil following pump operation.
Measurement of Ground Current—After our Renner failure, a Stoddart EMI/Field Intensity
Meter (Model 17/27) was used in conjunction with a RF current transformer to measure current
activity in the ground connection of several power transformers. Measurements on our trans-
formers with pumps running ranged from a few hundred microvolts to 10 000+ MV. On units
that had the pumps mounted directly adjacent to the tanks and with the pumps running, the
Stoddart instrument indicated high RF activity on the units with high charging tendency oil.
This instrument has been used in conjunction with other identification methods. There is poten-
tial application in use of this device together with an acoustical device as inputs to a dual-trace
scope to identify static discharges. More work is needed in this area.
Wave Guides—The use of wave guides for acoustical monitoring of partial discharge activity
is included in the scope of our ongoing research activities. Experience from application of these
devices could prove their effectiveness.
Charging Tendency Reduction Options

As documented in earlier papers, static electrification activity can be minimized or elimi-
nated when the oil flow velocity is reduced by altering pump design or cooling system operation.
Transformer retrofill using the original oil treated with Fuller's Earth or using new low charging
tendency oil is another effective method of reducing the level of static electrification activity.
Conductivity Modifiers or Charge Suppressant Additives—A possible static charging ten-
dency reduction method needing to be further investigated involves the addition of charge sup-
pressants to the oil. The effectiveness of using charge suppressants, such as BTA (1,2,3-benzo-
triazol) and AB (alkylbenzene), in transformer oil will require further study. The potential
detriments for the use of charge suppressant additives should be further explored. For example,
it has been suggested that these additives could produce high-resistance coatings on copper or
other
(UFRN) construction
Universidade Federal do Riomaterials which
Grande do Norte ((UFRN)could leadFederal
Universidade to tapdo Riochanger contact
Grande do Norte) pursuantproblems, etc. No
to License Agreement. Another
DISCUSSION ON STATIC ELECTRIFICATION 149
potential detriment is the suggested possibility that some additives are consumed in the oxida-
tion process and produce water as a by-product. Literature on BTA indicates dielectric strength
is reduced somewhat if BTA concentration is greater than or less than approximately 10 ppm
[17]. Proposed Institute of Electrical and Electronic Engineers (IEEE) transaction papers dis-
agree with this hypothesis.
Following static electrification flashovers, EPRI RP 1536-7, RP 1536-9, RP 1207, and RP
1291-5 studied the interaction of Mobil DCA48 additive in Refrigerant 113 used as a coolant in
the converter valves of a prototype high voltage d-c terminal. Preliminary study demonstrated
that a thin organic film was deposited on the electrodes due to electrolysis. This work concluded
that the DCA48 addition did not create undue complications in this application and the Refrig-
erant 113/DCA48 mixture would be an acceptable design parameter for eliminating electrifica-
tion problems [9,20,26].
It may be entirely possible that a contaminant or "gremlin" in the oil may be responsible for
the oil's unusual ability to produce static charge. On the other hand, the lack of an important
charge suppressant or "guardian angel" in the oil (due to severe refining, etc.) may be responsi-
ble as well.
Conclusion
As a doctor examines his patient, so do we monitor and examine a transformer. Much can be
learned from the chemistry of body fluids. Likewise, knowledge can be gained from examining
the chemistry of transformer fluid.
Some have claimed the transformer oil is pristine, innocent, unquestionable, noncausative to
static electrification problems, and not an indicator of static electrification activity. Many oth-
ers, including myself, must disagree. What if physicians had adopted this philosophy with re-
gard to body fluids? Therefore, rather than ignoring the transformer's oil as a means of poten-
tial detection and letting the patient die, why shouldn't the industry consider and pursue
monitoring and testing of the fluids as one of the indicators of deleterious electrification activ-
ity? Other methods of monitoring may employ systems other than fluid monitoring or testing.
Like it or not, the fluid in a transformer is easily accessible without the necessity of removing
an operating unit from service. Therefore, if the fluid is readily accessible for testing, why then
has there not been an industry accepted and approved method of testing and monitoring the
fluid for electrification activity?
Let us get off dead center and be open minded. We should quit blaming one manufacturer of
transformers, one oil refiner, one manufacturer of insulation, one method of operation, etc.
The answer is not that simple! Static electrification is a highly complex phenomenon that occurs
to some extent in any transformer. Let's simultaneously give our designers, manufacturers, and
refiners the basics they need to design around the deleterious effects while providing the users
with the needed tools to prevent catastrophic failures.
When the final truth is known, the problem will involve combinations of designs, operating
philosophies, characteristics of oils and insulating materials, and a complex combination of
other postulated failure modes. The failure mode may involve bubble evolution, moisture, flow
rates, design, resonance or other voltage excursions, preexisting conditions to which the trans-
former has endured, manufacturing techniques, installation and maintenance procedures, as
well as static electrification.
It is doubtful that we will be able to distinguish between some of the causative conditions.
Why then doesn't the industry diligently pursue all avenues to define the safe operating and
design parameters for all of the failure causes? Once this is done, appropriate monitoring de-
vices can determine, with high degrees of sophistication, when we are operating (or designing)
outside of a safe parameter envelope. To do this we must, by necessity, look at all facets includ-
Copyright
ing bythe
ASTM Int'l (all rightsoil.
insulating reserved); Thu May 6 14:02:13 EDT 2021
(UFRN) In conclusion,
Universidade Federal doletRio the
Grandepreceding be Universidade
do Norte ((UFRN) considered asdoinput
Federal fordoASTM
Rio Grande consideration.
Norte) pursuant WeNodo
to License Agreement. not
150 INSULATING OILS
have an approved and acceptable test method for insulating oil electrification potential and we
cannot dissect operating transformers to provide answers. Let us not resort to surgery if other
methods are available. Please, ASTM, give us a test method—don't ignore it and let the patient
die. Likewise, be cognizant to the use of conductivity modifiers—even doctors find it necessary
at times to inject additives into the body fluids and yes, many of the patients live.
Acknowledgments
The efforts and diligence of the Electric Power Research Institute (EPRI), Exxon, Shell,
Gulf, Union Electric, Consolidated Edison, Alabama Power, Massachusetts Institute of Tech-
nology, Rennselaer Polytechnic Institute, General Electric, McGraw Edison, Harley Pump
Works, Doble Engineering, Westinghouse Electric, and K. D. Mills of TU Electric are sincerely
appreciated and herein acknowledged.
Notice
Neither TU Electric, employees of TU Electric, nor any person acting on their behalf:
a) makes any warranty or representation, expressed or implied with respect to the accuracy,
completeness, or usefulness of the information contained in this report, or that use of any
information, apparatus, method or process disclosed in this report may not infringe pri-
vately owned rights,
nor
b) assumes any liabilities with respect to the use of, or for any damage resulting from the use
of any information, apparatus, method, or process disclosed in this report.
References
[/] Leonard, J. T., "Pro-Static Agents in Jet Fuels," NRL Report 8021, Naval Research Laboratory,
Washington, DC, August 1976.
[2] Takagi, T., "Reliability Improvement of 500 kV Large Capacity Power Transformer," International
Conference on Large High Voltage Electric Systems (CIGRE), Paper 12-02, Paris, France, 30 Aug. to
7 Sept. 1978.
[3] Higaki, M., "Static Electrification and Partial Discharges Caused by Oil Flow in Forced Oil Cooled
Core Type Transformers," IEEE Transactions on Power Apparatus and Systems, Vol. 98, No. 4,
IEEE PES Winter Power Meeting, New York, NY, February 1979.
[4] Okubo, H., "Charging Tendency Measurement for Transformer Oil," Transaction A 79 051-4, IEEE
Winter Power Meeting, New York, NY, February 1979.
[5] Shimizu, S., "Electrostatics in Power Transformers," IEEE Transactions on Power Apparatus and
Systems, Vol. PAS-98, July/August 1979.
[6] Tanaka, T., "Model Approach to the Static Electrification Phenomenon Induced by the Flow of Oil in
Large Power Transformers," IEEE Transactions on Power Apparatus and Systems, Vol. PAS-99, No.
3, Vancouver, British Columbia, Canada, July 1980.
[7] Tamura, R., "Static Electrification by Forced Oil Flow in Large Power Transformers," IEEE Trans-
actions on Power Apparatus and Systems, Vol. PAS-99, No. 1, IEEE PES Summer Meeting, Vancou-
ver, British Columbia, Canada, July 1980.
[8] Okubo, H., "Suppression of Static Electrification of Insulating Oil for Large Power Transformers,"
IEEE PES Winter Meeting Transaction, IEEE PES Winter Power Meeting, New York, NY, February
1982.
[9] Abuaf, N., "Forced Vaporization Cooling of HVDC Thyrister Valves," EPRI Publications EL-2710,
Electrical Power Research Institute, Palo Alto, CA, October 1982.
[10] Oommen, T. V., "Electrostatic Charging Tendency of Transformer Oils," IEEE PES Winter Meeting
Transaction, IEEE PES Winter Power Meeting, Dallas, TX, February 1984, IEEE, New York, NY.
[//] Crofts, D. W., "Failure of a 450 MVA, 345/138kV Autotransformer from Static Electrification,"
Electric Power Research Institute (Electrical Systems Division), AC/DC Transmission Substations
Copyright byTask
ASTM Int'l (all rights
Force, Newreserved);
Orleans,Thu May
LA,6 14:02:13
May 1984,EDT 2021
EPRI, Palo Alto, CA.
[721 Crofts, D. W., "Failure of a 450 MVA, 345/I38kV Autotransformer from Static Electrification,"
DISCUSSION ON STATIC ELECTRIFICATION 151
1984 Transmission Substation Design and Operations Symposium Transaction, University of Texas at
Arlington, September 1984.
[13] Crofts, D. W., "Static Electrification Phenomena in Power Transformers," Minutes of the Fifty-Sec-
ond Annual International Conference of Doble Clients, 1985, Sec. 6-1001.
[14] Ditzler, D. A., "Failure and Repair of a Westinghouse 345/138kV Autotransformer," Minutes of the
Fifty-Second Annual International Conference of Doble Clients, 1985, Sec. 6-1101.
[/5] "Acoustic Emission Detection of Partial Discharges in Power Transformers," EPRI Project RP-426-1,
final report, EPRI, Palo Alto, CA, August 1985.
[16] Crofts, D. W., "An Update on the Static Electrification Phenomena in Power Transformers," The
Pennsylvania Electric Association, Hershey, PA, September 1985.
[17] "Effects of BTA on Transformer Materials," Mitsubishi Electric Corp., October 1985.
[18] Brunson, P. W., Jr., "Investigation of the Failure of a 500-230kV, 480 MVA Westinghouse Shell-Type
Power Transformer," Minutes of the Fifty-Third Annual International Conference of Doble Clients,
1986, Sec. 6-1101.
[19] Crofts, D. W., "Methods of Controlling Static Electrification in Large Power Transformers," Minutes
of the Fifty-Third Annual International Conference of Doble Clients, 1986, Sec. 6-1001.
[20] Gasworth, S. M., "Electrification Problems Resulting from Liquid Dielectric Flow," EPRI Publica-
tion EL-4501, EPRI, Palo Alto, CA, April 1986.
[21] Crofts, D. W., "The Static Electrification Phenomena in Power Transformers," 1986 IEEE Confer-
ence on Electrical Insulation and Dielectrics Phenomena, King of Prussia, PA, November 1986.
[22] Crofts, D. W., "Transformer Characteristics that Influence Static Electrification Development,"
EPRI Workshop on Static Electrification in Transformers, Monterey, CA, November 1986, EPRI,
Palo Alto, CA.
[23] Nelson, J. K., "Electrokinetic Phenomena in Transformer Oil," EPRI Workshop on Static Electrifica-
tion in Transformers, Monterey, CA, November 1986, EPRI, Palo Alto, CA.
[24] Oommen, T. V., "Static Electrification Control in Power Transformers," EPRI Workshop on Static
Electrification in Transformers, Monterey, CA, November 1986, EPRI, Palo Alto, CA.
[25] Zahn, M., "Flow Induced Electrification in Transformer Oil/Cellulosic Systems," EPRI Workshop
on Static Electrification in Transformers, Monterey, CA, November 1986, EPRI, Palo Alto, CA.
[26] Damsky, B. L., "Design, Installation, and Initial Operating Experience of an Advanced Thyrister
Valve," IEEE 87 WM 184-5, February 1987, IEEE, New York, NY.
[27] Brunson, P. W., "The Second Failure of a 400 MVA, 500-230 kV Shell-Type Auto Transformer at
South Bessemer Transmission Substation," Minutes of the Fifty-Fourth Annual International Confer-
ence of Doble Clients, 1987, Sec. 6-1401.
[28] Hochart, B., "Oil Charging Tendency Effect on Large Power Transformers," Minutes of the Fifty-
Fourth Annual International Conference of Doble Clients, 1987, Sec. 6-1501.
[29] Lindgren, S., "Progress in the Control of Static Electrification in Power Transformers," Minutes of
the Fifty-Fourth Annual International Conference of Doble Clients, 1987, Sec. 6-1601.
[30] McNutt, W., "Discussion of the B. Hochart and J. P. Grandjeon Paper 'Oil Charging Tendency Effect
on Large Power Transformers'," Minutes of the Fifty-Fourth Annual International Conference of Do-
ble Clients, 1987, Sec. 6-1501A.
[31] Crofts, D. W., "The Static Electrification Phenomena in Power Transformers," IEEE Transactions
on Electrical Insulation, in press.
DISCUSSION
C. L. S. Vieira^ (written discussion)—Today there are some utilities that recondition the oil
with thermovacuum equipment in the normal operating transformer and maybe with the p u m p s
working. W h a t is t h e potential effect of this t r e a t m e n t in the charge density or in t h e
electrostatic charging of the system? Should this procedure be revised due to this new view of
charging?
D. Crofts (author's closure)—This method is employed with the transformer deenergized and
there is no potential for a-c power follow current. Perhaps we should be more cognizant of the
electrification problem and watch p u m p operation in deenergized modes since d-c potential
differences still can exist and allow d-c dielectric breakdowns of low-energy magnitude.
'CEPEL,
Downloaded/printed by P.O. Box 2754, Rio de Janeiro, Brazil.

Stp998-Eb 2198

Uploaded by

Copyright:

Available Formats

Stp998-Eb 2198

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Stp998-Eb 2198

Uploaded by

Copyright:

Available Formats

Copyright by ASTM Int'l (all rights reserved); Thu May 6 14:02:13 EDT 2021

Electrical Insulating Oils

Herbert G. Erdman, editor

Copyright © by AMERICAN SOCIETY FOR TESTING AND MATERIALS 1988

Peer Review Policy

Printed in Ann Aibor, Ml

SECTION I—REFINING AND SPECIFICATION LIMITS

Mineral Insulating Oil Manufacture and Safekeeping—T. G. LIPSCOMB, II 5

PCBs: The Beginning, the Muddle, the End?—H. C. MANGER 25

SECTION II—SIGNIFICANCE OF APPLICATION

Insulating Oil Qualification and Acceptance Tests from a User's Perspective—

Significance of Transformer Oil Properties—H. A. PEARCE 47

SECTION III—ANALYSIS OF OIL

Newly Developed Analytical Techniques for Characterization of Insulating Oils—

SECTION IV—DISSOLVED GAS IN OIL

Transformer Fault Gas Analysis and Interpretation—A User's Perspective—L. J. SAVIO 83

A Review of the Operating Principles and Practice of Dissolved Gas Analysis—

SECTION V—ELECTROSTATIC BUILDUP IN TRANSFORMER OIL

An Engineering Model for Streaming Electrification in Power Transformers—

Mineral Insulating Oil Manufacture and

REFERENCE: Lipscomb, T. G. II, "Mineral Insulating Oil Manufacture and Safekeeping,"

Gravity of Key Fraction, "API

FIG. 2—Definitions of naphthenic oil and paraffinic oil.

Atmospheric Atmospheric Vacuum Vacuum

TABLE 1—Comparison of naphthenic and paraffinic distillates

Initial boiling point, °C 283 239

Molecular Components of the Distillate

1 - Ring Naphthene 2 - Ring Naphthene

FIG. 9—Polars. oxygenates.

Refining the Distillate

'^ " t "

Mixer Centrifug B Mixer - stripper Perc

FIG. 11—Acid treating/day treating.

Furnace Reactor Separator Stripper Drier

Extraction Solvent Driers

Chillers Solvent Rotary Stripper Drier

Furnace Reactor Separator Stripper Drier

FIG. 15—Catalytic dewaxing

Physical and Chemical Effects of Refining

'Small Amounts May Be Solubilized Into The Extract Phase.

FIG. 17—Hydrocarbon reactions: iso-paraffins, naphthenes.

Clay Treating - Limited Adsorption

4> At Treating Conditions

FIG. 20—Hydrocarbon reactions: aromatics (continued).

Clay Treating - No Effect

•At Treating Conditions 9*f"""f"!*u'^

rSff^v, ^_R ^ Partial Removal Into

FIG. 22—Hydrocarbon reactions: benzothiophenes (continued).

09 >• Adsorbed On Clay

FIG. 23—Hydrocarbon reactions: polar nitrogen compounds.

0 0 * 7 "2 "EiT^ Q-CH2-CH2-CH3+NH3t

r ' ^ ^ l r ^ + Solvent >• M o d e r a t e T o Substantial

FIG. 25—Hydrocarbon reactions: polar sulfur-mercaptans and disulfides.

00 — COOH • Adsorbed On Clay

OSHA Regulation Effects

Handling and Storage

PCBs: The Beginning, the JVIuddie,

TABLE 1—Molecular composition of typical Aroclor.

1221 1016 1242 1248 1254 1260

•At Treating Conditions 9f"""f"!u'^