You are on page 1of 15

Oil Power Factor Mysteries

By
Russ Crutcher
Co-Author Ken Warner
Microlab Northwest

TechCon North America 2016 Albuquerque, NM


Oil Power Factor Mysteries
Author: Russ Crutcher
Co-Author: Ken Warner
Microlab Northwest

Abstract

A study of 130 transformer oil samples seems to contradict much of the accepted knowledge
regarding the relationship between power factor, moisture, oxidation, interfacial tension, and
other parameters. This paper examines many of these contradictions and offers explanations
based on the analysis of polar contaminants in the oil.

Introduction

The measurement of transformer oil power factor is listed as one of the important diagnostic tests
to be performed.1,2,3 How often is this test requested in reality? Not that often, and even then,
generally at only one temperature, either 25o or 100o Celsius. How important is the omission of
this test? What is lost by testing at only one temperature? The literature would seem to justify
its omission in the early life of the transformer, assuming that the main cause for a change in the
power factor is aging of the oil.4,5,6 For that same reason it is often omitted as a routine
diagnostic test. It is assumed that dissolved gas analysis, acid number, and interfacial tension
monitor the same processes that would cause a change in power factor. What is the basis for
these assumptions? Are these assumptions justified?

One way to test the validity of these assumptions in the real world is to plot power factor and
these other parameters to see if a pattern emerges. Scatter plots have been generated to test the
role of oxidation, acid formation, and gas generation on power factor using a data base of up to
130 oil samples. These plots study two parameters at a time. If there is any kind of
interdependence between these two parameters a pattern in the graph would show that as one
increases the other either consistently increases or consistently decreases. The quality of that
correlation will be shown by how confidently one can predict the change in one of the
parameters based on the change of the other. The results of this study indicate that the expected
patterns tend not to be confirmed. This includes the assumption that interfacial tension and acid
number correlate with the amount of oil oxidation.

TechCon North America 2016 Albuquerque, NM 1


Power Factor and Dissolved Gas Analysis

The importance of dissolved gas analysis cannot be questioned but, dissolved gas analysis tell us
nothing about the power factor of the oil. It provides information on the release of energy that is
resulting in chemical changes but not about the chemicals being formed. The gases or the ratios
of the gases provide valuable information but the other part of that chemical reaction is the
modified molecule of oil left behind. The chemicals being produced, the modified oil, are
different for each type of oil because each type of oil is composed of different starting chemicals.
Pulling a hydrogen atom from each of two different chemicals is going to have a different result
depending on the shape, molecular weight, and which hydrogen was removed from the starting
molecule of oil. There will be a change in power factor over time as a result of these reactions
but the change may be negligible if the result of the chemical reaction does not create a
compound that is strongly polar.

The two scatter plots shown below do not show any recognizable pattern that would indicate a
relationship between the total combustible gases generated in these transformers and the power
factor.

Scatter Plot: PF25 vs Total Combustible Gases


0.500
Power Factor at 25 Celsius

0.400

0.300

0.200

0.100

0.000
0 200 400 600 800 1000 1200 1400 1600
Total Combustable Gases

TechCon North America 2016 Albuquerque, NM 2


Scatter Plot: PF100 vs Total Combustible Gases
14.000
Power Factor at 100 Celsius

12.000
10.000
8.000
6.000
4.000
2.000
0.000
0 200 400 600 800 1000 1200 1400 1600
Total Combustable Gases

These graphs show that for the oils for which we had dissolved gas results and power factor
values there was no evidence of a correlation. The fragments of the original oil molecules and
any particles generated had little effect on the power factor.

Interfacial Tension and Power Factor

What about interfacial tension? Interfacial tension tells us about the hydrophobic nature of the
oil, how much force is required to move the water through the interface between the water and
the oil. The greater the force required the more hydrophobic the oil. Clean, new mineral oil is
very hydrophobic. The force required is relatively large. The addition of polar compounds to
the oil, such as hydrocarbons with a bonded oxygen atom, will reduce the force required to move
through the oil/water interface. The amount the force is reduced is dependent on the specific
chemical added and how much of that chemical was added.

For real world samples, as the oil ages the interfacial tension tends to become less. That is
because the oil oxidizes over time. The oil molecule oxidized at any given time can be any of
the thousands of molecules that make up the mineral oil that was placed in the transformer. The
result over time is a reduction in the interfacial tension. It has been suggested that polar
compounds created by oxidation that reduce the interfacial tension also contribute to the increase
in the power factor. If oxidation was the main cause for a decrease in interfacial tension and an
increase in power factor then there should be a relationship shown in the scatter plot between
interfacial tension and power factor. That doesnt seem to be the case. The two charts below
show no consistent relationship between either the power factor at 25o or 100o and interfacial
tension.

TechCon North America 2016 Albuquerque, NM 3


Scatter Plot: PF25 vs Interfacial Tension
0.300
Power Factor at 25 Celsius

0.250

0.200

0.150

0.100

0.050

0.000
20.0 25.0 30.0 35.0 40.0 45.0
Interfacial Tension

Scatter Plot: PF100 vs Interfacial Tension


14.000
Power Factor at 100 Celsius

12.000
10.000
8.000
6.000
4.000
2.000
0.000
20.0 25.0 30.0 35.0 40.0 45.0
Interfacial Tension

These charts seem to show a complete independence of these parameters or that any relationship
between power factor and interfacial tension is trivial compared to the controlling factors. So
why do laboratory tests performed in the past seem to support a relationship? The laboratory
tests done in the past involve adding polar compounds to new oil. When the oil was then tested
both interfacial tension and power factor were affected. That is not what happens in a real
transformer over time. Some of the materials produced over time in a transformer have little
effect on power factor but a significant effect on interfacial tension. Other materials in the aged
oil may have a significant effect on the power factor but little effect on the interfacial tension.
This was demonstrated by T. V. Ooman and others a number of years ago when it was shown

TechCon North America 2016 Albuquerque, NM 4


that at small parts per million levels certain metal-organic chemicals from paints, polymers, or
other materials could have dramatic effects on power factor.7,8,9 At those low levels there are no
effects on interfacial tension.

That would seem to be supported by laboratory tests where polar compounds added to a mineral
oil decrease interfacial tension and increase power factor. The change in these two parameters is
different depending on the polar compound added. Since different oils generate different
compounds as they age it would be expected that some oils change power factor more rapidly
than interfacial tension and vis-a-versa. Changes in interfacial tension suggest a change in power
factor but provides no information on the amount of change.

Acid Number and Power Factor

What about acid number? Maybe it is just one type of polar compound that effects the power
factor. Acids are certainly polar compounds but they are only one of the many types formed in a
transformer. The acids themselves vary greatly in molecular weight and relative polarity. The
acids formed in a transformer will be different depending on the type of mineral oil in the
transformer and the operational history of that transformer. An increase in the acid number of
the oil might be expected to decrease the interfacial tension. That seems to be true based on the
scatter plot below.

Scatter Plot: Acid Number vs IFT


45.0

40.0
Interfacial Tension

35.0

30.0

25.0

20.0
0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160
Acid Number

As the interfacial tension decreases the acid number tends to increase. There is a lot of scatter in
the data so one could not measure the interfacial tension and then calculate or predict what the
acid number might be. An interfacial tension of 30 could be found with an acid number
anywhere between 0.02 and 0.14. There is a relationship but it is not a close fit. It is
significantly better that any of the charts comparing power factor to total combustible gases or
interfacial tension.

TechCon North America 2016 Albuquerque, NM 5


Back to acid number and power factor, the charts below show that the power factor is
independent of the acid number. There is no tendency, regardless of how vague, for power
factor to increase as the acid number increases. The materials that drive the power factor are not
the acids created in a transformer.

Scatter Plot: PF25 vs Acid Number


0.300
Power Factor at 25 Celsius

0.250

0.200

0.150

0.100

0.050

0.000
0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160
Acid Number

Scatter Plot: PF100 vs Acid Number


14.000
Power Factor at 100 Celsius

12.000
10.000
8.000
6.000
4.000
2.000
0.000
0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 0.160
Acid Number

Power Factor and Moisture

Another polar chemical in transformer oil is water. Is there a correlation between how wet a
transformer oil might be and power factor? There are two ways to look at the moisture level in
an oil. One is simply measure the amount of water in the oil and the other is to calculate the

TechCon North America 2016 Albuquerque, NM 6


degree of saturation with water in the oil. Both were examined and neither showed any
correlation. The scatter plots for the moisture content and power factors are shown below.

Scatter Plot: PF25 vs Moisture


0.300
Power Factor at 25 Celsius

0.250

0.200

0.150

0.100

0.050

0.000
0 5 10 15 20 25 30 35 40
Moisture in Parts per Million

Scatter Plot: PF100 vs Moisture


14.000
Power Factor at 100 Celsius

12.000
10.000
8.000
6.000
4.000
2.000
0.000
0 5 10 15 20 25 30 35 40
Moisture in Parts per Million

Power Factor and Sample Temperature

A potential source of bias in the sampling process could be the temperature at which the
transformer was sampled. Polar compounds tend to migrate into the paper as the temperature of
the transformer oil decreases. Is there a tendency to have oil with a lower power factor if the oil
is sampled cold rather than when the transformer is still warm? For the oils included in this
study there was no strong evidence for that effect when the moisture content of the oil was
plotted with the temperature of the oil at the time of sampling. None of these oils were near their

TechCon North America 2016 Albuquerque, NM 7


saturation point so that might have affected the results. The two charts below show that there is
no indication of such a pattern in the case of power factor.

Scatter Plot: PF25 vs Sample Temp


0.500
Power Factor at 25 Celsius

0.400

0.300

0.200

0.100

0.000
0 10 20 30 40 50 60
Temperature at time of Sampling, Celsius

Scatter Plot: PF100 vs Sample Temp


14.000
Power Factor at 100 Celsius

12.000
10.000
8.000
6.000
4.000
2.000
0.000
0 10 20 30 40 50 60
Temperature at Time of Sampling, Celsius

Power Factor and Polar Weight

When the power factors were plotted by the weight of polar compound removed from the oil
there was a correlation. The weight of the polar compounds was based on passing a five
milliliter sample of oil through a column of silica gel and then extracting the silica gel with
electronic grade dichloromethane. The dichloromethane was then evaporated off at a
temperature of 70o Celsius and the resultant volume of polar compounds was weighed. There
were some polar compounds that were not extracted from the silica gel by dichloromethane.

TechCon North America 2016 Albuquerque, NM 8


That has been determined by subsequent extraction using methanol and acetic acid but the
amount recovered in those instances was minor. It was decided that the dichloromethane
extraction would suffice for this study. It is also possible that some of the higher vapor pressure
polar compounds were lost in the process of evaporating the dichloromethane. With these
caveats the charts shown below do show some correlation between power factor and the weight
of the polar compounds recovered from the samples.

Scatter Plot: PF 25 vs Polar Wt.


2.5

1.5

0.5

0
1 10 100 1000
Weight in Micrograms per 5 Milliliters

Scatter Plot: PF 100 vs Polar Wt.


18
Power Factor at 100 Celsius

16
14
12
10
8
6
4
2
0
1 10 100
Weight in Micrograms per 5 Milliliters

The correlation is not so good that polar weight or volume alone would be a good indication of
power factor but it does show a rough correlation. It also documents that the identification of the
polar compounds is as important, if not more, important than the total amount of polar
compounds.

TechCon North America 2016 Albuquerque, NM 9


Power Factor 25o vs 100o

One additional correlation was investigated between the power factor measured at 25o and that
measured at 100o. This was one of the better correlations observed but still shows a bit of
scatter. It is interesting to note that the best fit line has a slope of about 30. That means that if
the alarm level for a power factor measured at 100o of 5 should correspond to an alarm of 0.15 if
measured at 25o Celsius rather than 0.5. In other word, if the 25o power factor is 0.15 or more
then the oil should be tested at 100o Celsius to make sure the risk at operational temperatures is
still low.

Scatter Plot: PF100 vs PF25


14.000
Power Factor at 100 Celsius

12.000
10.000 y = 29.701x + 0.428
R = 0.7481
8.000
6.000
4.000
2.000
0.000
0.000 0.050 0.100 0.150 0.200 0.250 0.300
Power Factor at 25 Celsius

Chemistry and Polar Compounds

A preliminary examination of the polar compounds recovered from these oils offers some
indication of why power factor does not correlate with interfacial tension or with acid number.
The set of images below show the thin layer chromatograms of the polar compounds extracted
from a high power factor oil and a low power factor oil. Each panel shows a single
chromatogram photographed with three different forms of illumination. The first image is the
chromatogram illuminated with ultraviolet light with a wavelength of 254 nanometers. The
second image is the same chromatogram illuminated with ultraviolet light having a wavelength
of 365 nanometers. The final image is taken with visible light illumination.

The same amount of material was applied to the base of each chromatogram so the materials
present are shown with respect to their quantity to one another from that oil. For example, the
yellow band at the top of each chromatogram when viewed with visible light is an oxidized
antioxidant. In the case of the high power factor oil there are a number of other yellow

TechCon North America 2016 Albuquerque, NM 10


compounds visible over the entire chromatogram. In the case of the low power factor oil the
chromatogram below that yellow band looks relatively clean in visible light. The low power
factor oil contained over twice as much polar material as the high power factor oil as can be seen
by the weight of the polar compounds removed from five milliliters of each oil. Clearly, the
amount of polars was not the controlling factor. It is also of interest that the oxidized antioxidant
had little effect on the power factor, being well represented in both oil samples.

Photographs 1, 2, 3: PF at 100o = 16.5 Photographs 4, 5, 6: PF at 100o = 0.69


Polar Weight = 17.8 milligrams Polar Weight = 39.8

One of the bands that is evident on the chromatogram of the high power factor oil and that is
absent on the low power factor oil appears on many of the high power factor oils (see red arrow).
The infrared spectrograph of this material is shown in the figure below. We are in the process of
identifying this material. The fact that a similar polar compound appears in many high power
factor oils suggests that it may come from a component used in building the transformer or some
accessory equipment attached to the transformer. Materials used in transformer construction are
all tested for stability in mineral oil but not in the mineral oil modified by use. Some of the

TechCon North America 2016 Albuquerque, NM 11


oxidation products created over time by how the transformer is used may attack one or more of
these organic components. The apparent correlation between the total polar content and power
factor may be the result of the time and conditions necessary to form a compound that will attack
paints, sealants, and gaskets.

100

95

90

85

931.66
3619.06

814.26
80

685.36
3421.87

75

968.90
1594.44
70

1240.27

1112.96

753.46
65

1459.06
60
%T

1291.89

887.75
1158.33
55

1688.53
50

45

1378.03

1197.42
2869.03

40

1737.09
35

1429.88
30

25
2954.91
2925.76

20

15

400 0 350 0 300 0 250 0 200 0 150 0 100 0 500


W av e numb er s (c m-1 )

Figure 1: Infrared Spectrum of Unique TLC Band Seen on Many High Power Factor Oils, Power Factor 16.5

Conclusion

The power factor of the oil in a transformer is an independent parameter important to the
diagnostics of a transformer. There are a number of issues suggested by this study. One
interesting suggestion is that high power factor is the result of the formation or extraction of
specific types of compounds that may enter the oil at any time in the history of the transformer.
They may appear within the first year or may take years to enter the oil. Their relationship to
aging may be the result of the time required to form oxidation products that will degrade paints,
sealants, or gaskets in the transformer. This can occur within months or it may take years.

Another interesting feature of the relationship between the power factors measured at 25o and
100o is that they seem to differ by a factor of 30. That would suggest that the alarm levels in
IEEE C57.106-2002 might need adjusting.

TechCon North America 2016 Albuquerque, NM 12


Power factor is an important parameter that should be part of the routine analysis. The
independence of power factor and interfacial tension suggests that field tests that rely on surface
tension as an indication of power factor may be susceptible to significant errors.

This has been a limited study of about 130 oil samples from 130 different transformers but the
implications are significant. A much larger study is needed to verify the findings over a larger
population of transformers and over the same transformer through time. As our data base grows
we will continue to make the result generally available.

Future Related Studies

Two other parameters currently being investigated are the relationship between power factor and
soluble copper compounds and copper corrosion potential in these oils. The powerful effect
metal-organic compounds have on power factor have been well documented. The formation of
copper-organic compounds that are oil soluble in transformers has also been demonstrated. The
relationship between power factor and the formation of these copper organic compounds have
not been studied.

References

1. IEEE C57, 106-2002, The Institute of Electrical and Electronics Engineers, Inc., 2002
2. Duval, M and T. O. Rouse, Physical and chemical properties of mineral insulating oils,
Chapter 4, ENGINEERING DIELECTRICS VOLUME III: ELECTRICAL
INSULATING LIQUIDS, ed. R. Bartnikas, ASTM, Philadelphia, PA, 1994.
3. U.S. Dept. of the Interior, Bureau of Reclamation, Transformer diagnostics,
FACILITIES INSTRUCTIONS, STANDARDS, AND TECHNIQUES, VOLUME 3-31,
June 2003.
4. Woodcock, David J. Maintenance strategies for the new millennium, Cinergy
Conference, Indianapolis, IN, November 6-7, 2000.
5. Hamrick, Lynn, Transformer oil sampling, NETA WORLD, Summer, 2009
6. Savio, Leo J. and Ted Haupert, Insulating media, ELECTRIC POWER
TRANSFORMER ENGINEERING, ed. James H. Harlow, CRC Press, 2004.
7. Oommen, T.V. Flow Electrification Properties of Certain High Charging Polymeric
Materials in Transformer Oil, IEEE International Symposium on Electrical Insulation,
Arlington, VA. pp. 664668, June 710, 1998.
8. Crutcher, E.R., Phil Hopkinson, Ray Rettew, Steve Smith, Ken Warner, and Jaime
LaFave. Paint Induced Elevation of Oil Power Factor in Electrical Equipment, Doble
Conference, Boston, MA, April 11, 2006.
9. Crutcher, E.R. and Ken Warner, Organic Contaminants and Premature Failure in
Transformers and Bushings, TechCon US, Glendale, AR, February, 2014.

TechCon North America 2016 Albuquerque, NM 13


Biography

Ernest Crutcher, best known as Russ, an analytical specialist has over thirty-five years as the
owner of Microlab Northwest, twenty-nine years as the lead engineer for the Microanalysis
Laboratory of Boeing Aerospace's Analytical Engineering Group, eight years as an instructor at
the University of Washington in Materials Sciences Engineering, and three years as a criminalist
for a large city police department. Russ has spoken at technical conferences worldwide and has
authored over seventy papers and chapters in numerous books on microanalytical techniques.

TechCon North America 2016 Albuquerque, NM 14