See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/286926525

The Premier Electrical Maintenance and Safety Event THE SCIENCE OF

SAMPLING INSULATING LIQUIDS

Article · March 2013

CITATION READS

1 349

2 authors, including:

Lance Richard Lewand

Doble Engineering Company
20 PUBLICATIONS   144 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Lance Richard Lewand on 15 December 2015.

The user has requested enhancement of the downloaded file.

 The Premier Electrical Maintenance and Safety Event
Hosted by

THE SCIENCE OF SAMPLING

INSULATING LIQUIDS
Lance R. Lewand and David Koehler
Doble Engineering Company, USA
OUTLINE

This paper will discuss the procedures and techniques used to retrieve a representative
sample of dielectric liquid from an electrical apparatus. Also presented is the rationale for these
recommendations along with the consequences of retrieving a poor sample. The main discussion
is based on sampling transformers, however oil circuit breaker (OCB), and load tap changer
(LTC) sampling are also briefly discussed.

INTRODUCTION
The results obtained from the analysis of an insulating liquid sample can provide
unparalleled information concerning the condition of the insulating materials within electrical
apparatus, life assessment, and the operating condition of the device. However, a sample that is
not representative of the bulk liquid insulation can provide erroneous information which can
easily mislead maintenance personnel to incorrectly assess the condition of the liquid insulant or
the electrical apparatus. The classic example of this is given in Reference 1 where a sample
taken from two transformers showed a very high concentration of hydrogen and no other gases.
This prompted maintenance personnel to give these units priority for diagnostic surveys. It was
later found that the cause of the high hydrogen was a galvanic reaction occurring in the drain
valve in which water was converted to hydrogen and oxygen because of the interaction of a
galvanized fitting with a dissimilar metal.

TERMINOLOGY

Throughout this paper, the liquid inside a transformer or other liquid-filled electrical device
is referred to as an insulant, insulating liquid or dielectric liquid. These terms are
interchangeable. Only insulating liquids will be discussed in this paper. The use of the term
“fluid”, which is sometimes employed, refers to liquids and gases. Sampling of gases such as
sulfur hexafluoride (SF6), nitrogen, or air is not discussed. Sampling of gases is a complex
subject and requires a separate discussion to cover it properly.

1
 Insulating liquids are a group of liquids that have been used to provide insulating qualities to
the dielectric system and dissipate heat. Most are familiar with transformer mineral oils which
are used about 90% of the time in all electric apparatus especially large power transformers.
Transformer mineral oil has also been referred to as mineral and oil, insulating oil, dielectric oil,
transformer oil, 10C oil and oil and other terms for this product most likely exists. This material
is refined from crude which is mineral oil based and is a combination of naphthenic, paraffinic,
and aromatic molecules. It has been used in transformers since the 1890s and possibly earlier
but that was when the first patent for the use of mineral oil was filed.

Over the years, other liquids were also integrated into the dielectric system. The first were
vegetable oils (natural esters) in the early 1900s but soon fell out of favor because they oxidized
very quickly and polymerized. The oxidation of these vegetable oils was fortuitous for other
reasons as it led to the development of varnishes and other coatings used in electrical apparatus.
In the mid-1990s, vegetable oils re-emerged as an insulating liquid, refined and processed
differently and are currently in-use and gaining favor. Askarels appeared around 1929. Askarels
are synthetic insulating liquids which are highly stable. Askarels that are chlorinated on a
biphenyl ring were commonly referred to as polychlorinated biphenyls (PCBs). These were
exceptionally stable liquids and coupled with very high fire points could be used inside buildings
whereas mineral oils could not. Because these liquids were so stable, they were prone to bio-
accumulation. With the environmental and health concerns that followed, the use of these
liquids was outlawed by the United States in 1976. Most pure PCB-filled transformers or those
diluted with a chlorinated solvent have been removed from service and disposed of but some still
remain. Because PCBs were so prevalent, a large number of mineral oil transformers are also
contaminated with PCBs.

Synthetic hydrocarbons such as poly alpha olefins (PAOs), polybutenes and alkyl benzenes
were also used but because of the high cost of manufacturing were not usually found in
transformers and other large liquid-filled electrical apparatus (although some installations do
exist). Most of the synthetic hydrocarbons found a market in capacitors, small electronics and
cables. Another synthetic liquid, silicone, was introduced into transformers in the 1960s and
found favor as a replacement liquid used in retrofills of askarel-filled transformers or in new
transformers used to replace askarel-filled transformers. Silicone is a very stable liquid but can
be cost prohibitive. High molecular weight hydrocarbons (HMWH) also came to favor at that
time mostly under the trade name R-Temp and could be used in the same capacity as silicone.
HMWHs were mineral oil based just more viscous and could meet the fire point requirements
necessary for liquid filled electrical apparatus used inside buildings. Synthetic (polyol) esters or
pentaerythritols were used mainly in Europe as retrofill liquids for askarel-filled transformers
during the same era. Trade names of these liquids were Midel 7131, Reolec 138 and Envirotemp

2
200 (later in the US). These synthetic esters are now predominantly used in electrical apparatus
in ice-breaking ships and in traction transformers used on rail systems across the world. Other
types of insulating liquids do exist and new ones are being worked on currently.

In spite of all the insulating liquid development over the years, transformer mineral oil has
remained the most cost effective solution and is the dominant product used in the industry today
and thus will be the main focus of this sampling discussion although sampling techniques will be
similar for the other insulating liquids.

REASONS FOR SAMPLING

WHY SAMPLE?
For in-service liquid filled electric apparatus, sampling of the liquid dielectric provides a
method to determine the condition of the solid (paper, pressboard, etc.) and liquid insulation
system as well as the operating condition of the apparatus without opening or de-energizing the
apparatus in most instances. This is especially important in the present utility climate, as
equipment outages for out-of-service testing have become very limited. Sampling provides a
means to check the condition of oil in storage whether it be new or used, to determine if it
complies with specifications such as Doble TOPS [2], ASTM D3487 [3], IEC 60296 [4], IEEE
C57.106 [5], or some other manufacturer or user specifications. Sampling can aid in
determining: 1) if accidental mixing of different insulating liquids has taken place; 2) if the
method of transportation contaminated the insulating liquid; and 3) if the handling equipment to
transfer the insulating liquid contaminated the product.

WHAT IS A GOOD SAMPLE

Simply put, a good sample is one that is representative of the content of the bulk liquid
insulation. Since samples are usually retrieved from a drain valve or the attached sampling cock,
preparation of that area is important to obtain a good sample. Cleaning the drain valve inside
and out and the sampling cock is the first step in avoiding sample contamination. Cleaning the
outside of the drain valve is just as important as cleaning the inside. The atmospheric dirt and
debris that adheres to the surface of the valve can and will fall off into the sample container
during the sampling process resulting in sample contamination. Contamination in the apparatus
consists mostly of water and particles (paper fibers, metal particles, etc.) and over time will settle
out on the bottom of the apparatus near the drain valve. This material needs to be flushed out of
the system to get to the bulk liquid insulation. It is necessary to remove at least 1 to 2 liters of
liquid from the drain valve, cap the drain valve, and then flush out the sampling cock before
proceeding with sampling. On occasion, 2 liters is not sufficient, especially when sampling a

3
non-energized transformer or certain OCBs and LTCs and will be discussed in detail. Specific
sampling techniques and precautions, especially those dealing with low volume electric
equipment, are thoroughly reviewed in the Doble Reference Book on Insulating Materials [6],
and ASTM Practice D923 [7].

GOOD SAMPLE VERSUS POOR SAMPLE

It is sometimes very clear to the laboratory performing the analysis on the dielectric liquid
that the sample was taken improperly. For example, the presence of free water or foreign objects
such as insects, pipe sealing tape or putty are strong indicators that the drain valve was not
adequately cleaned and flushed prior to sampling. Once analysis has begun and it is determined
that high moisture or free water content is present coupled with a low dielectric strength with all
the other test results being acceptable, strongly suggests that the proper sampling technique was
not adhered to. It may even imply that some chemical reactions were taking place in the drain
valve that was not representative of the bulk liquid insulation.

LAB TESTS MOST EASILY AFFECTED

As indicated previously the analytical tests most easily affected by sampling are dielectric
strength and water content. This is due to the fact that an apparatus drain valve is located near
the bottom of the tank, where debris and water accumulation occurs. Water would be present as
a result of condensation that occurs in the drain valve, and is most often due to the position of the
drain valve on the tank. In most cases the drain valve protrudes 15 to 30 cm (6 to 12 inches)
away from the main tank (refer to Figure 1 – valve on right).

Drain Valves on Main Tank for Transformer

Figure 1

4
 The valve on the right in Figure 1 is a 2-inch Globe valve which is used to take samples.
From experiments conducted at Doble Engineering, the dielectric liquid in many of these valves
is cooler than the bulk liquid insulation. Often the temperature of the valve is 10 to 15°C cooler.
As described in Reference 8 when oil or air has an elevated relative saturation or humidity and
there is a significant cooling, condensation of water will occur and the same phenomena will
occur in a drain valve (refer to Figure 2). Other analytical tests easily affected by sampling are
dissolved metals, particulate metals, particle counts, dissolved gases-in-oil, and power factor.

Water and Oil Emulsified in Valve

Figure 2

The concentration of metals, whether dissolved or in a particulate state, are especially

influenced by the amount of cleaning performed on the drain valve and the amount of flushing
that is performed. Debris that settles to the bottom of the apparatus and subsequently into the
drain valve can consist of metal particles, fibers, water and other debris. In addition, just the
simple fact of removing the drain-valve plug or opening the sampling cock will create particulate
metals. This is due to the grinding of the surfaces between the threads of the valve body and the
drain plug or sampling cock. In fact, it is becoming more apparent that that these types of
samples should only be retrieved after a minimum of 2 liters, and sometimes 3 to 4 liters, of
dielectric liquid have been passed through the drain valve in order to adequately flush it. The
following case illustrates this point.

Samples were taken from a large power transformer with cooling pumps to determine if
pump cavitation was impacting the integrity of the bearings or possibly visa-versa. Particulate
metals analysis, which provides results measured in the parts per billion range (ppb) and is

5
utilized for this type of condition, are easily influenced by any sort of metallic contamination
such as the grinding described in the previous paragraph. The results from the first sample are
reported in Table 1.

Table 1
Particulate Metal Concentrations from 1st Sampling, (ppb w/w)

Metal Metal
Content
Copper 63
Lead 10
Iron 33
Zinc 6

For this analysis, results over 20 ppb are considered abnormal and anything over 50 ppb
requires immediate attention. Contamination from valve components and other metals used in
the construction of a transformer, are a source of metallic contamination. Cooper is a major
component of valves constructed out of brass or bronze, and will also have some level of nickel,
lead and zinc depending on the specific alloy that is used. The transformer tank and all the core
steel is composed of iron. In discussing the results with the customer a plan was formulated to
resample only after flushing 5 gallons of oil from the drain only after first cleaning real well the
inside of the valve with a lint free cloth and using a toothbrush to remove debris from the female
threads of the valve. The transformer was resampled in this manner and the results are provided
in Table 2.

Table 2
Particulate Metal Concentrations from 1st and 2nd Samplings, (ppb)

Metal 1st Sampling 2nd

Metal Sampling
Content Metal
Content
Copper 63 6
Lead 10 1
Iron 33 4
Zinc 6 14

As shown the metals concentrations were reduced considerably except for nickel. This was
not the only lesson learned from this experience. The second lesson was that if the utility had

6
acted upon the first sample to drain the transformer and dismantle and rebuild the pumps, a
substantial cost would have been incurred were no trouble actually existed.

The same is true of retrieving a sample for particle count where valve debris, whether inside
or outside, can severely skew the results. Of serious consequence are the debris, soot and grime
that is present on the outside of the drain valve especially in industrial locations. This debris can
be easily transferred to the sample bottle while the sampling process is taking place especially
when manipulating the valve hand wheel and thus validates the importance of cleaning the
outside of the valve prior taking the actual sample.

Dissolved gas-in-oil analysis is another test impacted by sampling, drain valve components
and sampling materials. When galvanic fittings (zinc coated) are used in the drain valve
assembly such as the drain plug, galvanic reactions with condensed free water can cause very
high levels of hydrogen to be produced. If this residue is not flushed out adequately then it will
be transferred to the sample and included in the analysis, causing a level of concern that is not
warranted. In addition, galvanic plumbing fittings such as nipples can have the same effect. In
one such case, two large power transformers were transferred to a shop for repairs where none
was actually needed [1]. In this case, not enough oil was flushed out and no confirming samples
were taken.

Cast iron drain valves can cause a similar issue. This was well documented by Sheppard
who showed that free water in contact with iron in the presence of oxygen from air would yield
hydrogen and iron oxide [6]. Christensen and Ohlsson showed that free water introduced into a
transformer could indeed produce such reactions. This was detected using a continuous
hydrogen detector when circulating oil that unintentionally contained free water in the
conservator tank into the main tank of a transformer [9]. They proposed a corrosion mechanism
of: 3H2O + 2Fe → 3H2 + Fe2O3.

Brass, bronze, or stainless steel should be the only materials used for transformer valve
construction. Unfortunately, as valves become larger such as the metric DN50 and DN80 valves,
the cost of these metals and alloys becomes prohibitive, and valve manufacturers tend to use iron
to lower the cost. In addition, drain valve assemblies should not be composed of dissimilar
metals as corrosion can result which may slough off into the sample.

Debris, water and other ionic contaminants also affect the power factor test when these
materials increase dielectric loss thus in turn increasing the power factor. Incompatible inorganic

7
and organic materials from the drain-valve stem packing or drain-plug sealants can also have the
same effect on the power factor. Power factor, particularly when conducted at 100°C, is a very
sensitive test to low levels of polar and ionic contamination.

COSTS ASSOCIATED WITH A POOR SAMPLE

In the case of a single sample, the costs for routine oil quality analysis and DGA testing are
a small fraction of the total costs associated with taking and analyzing a sample. Some of the
items and costs associated with sampling for a single dissolved gas-in-oil analysis (DGA) are:

ITEM COST ($)

Labor to take sample 275
Materials to take sample 15
Packaging and shipping cost 8
Analysis cost 70
Engineering evaluation of the data (10-15 min) 35

TOTAL 403

It is recognized that every situation is different but this examination of the costs puts into
perspective that, in many cases, the analysis cost is only about 17% of the entire sampling and
data review process. In the situation where the sample has been determined to be non-
representative of the bulk liquid insulation, the following costs in addition to the original costs
may be incurred as well.

ITEM COST($)
Labor to take original sample 275
Materials to take original sample 15
Packaging and shipping cost for original sample 8
Analysis of original sample 70
Engineering evaluation of the data (10-15 min) of original sample 35
Additional engineering time to confirm sample was non-representative 35
Labor to take 2nd sample 275
Materials to take 2nd sample 15
Packaging and shipping cost for 2nd sample 8
Analysis cost of 2nd sample 70
Engineering evaluation of the data (10-15 min) of 2nd sample 50

TOTAL 856

8
 As described above the cost of taking a poor or non-representative sample has more than
doubled from the original total. This is in part due to the fact that review of data from the second
sampling takes longer as there is a more critical and thorough review.

If the original sample was not recognized as being a poor sample then the costs associated
with that sample can be staggering. For example, if a poor sample caused a utility to react in
such a way as to provide a remedial effort to improve the condition of the insulating liquid, such
as processing the liquid through clay or vacuum processing a transformer to remove moisture,
then those associated costs may skyrocket to between $10,000 and $30,000. This is part of the
reason Doble always recommends taking a second sample to confirm the results of the first
before any remedial activities begin. Other factors such as accidental sample switching or
misidentification either in the field or the lab can also be the source of an erroneous assessment.

PRACTICES FOR SAMPLING INSULATING LIQUIDS

There are a number of practices that clearly define the proper way to retrieve samples from
electrical apparatus or storage containers. These practices have been developed over many
years and have incorporated the expertise of many individuals. These practices are to be
followed as closely as practical recognizing that not all sampling situations are the same. Special
attention should be paid to sampling low volume apparatus, apparatus under negative pressure
and those with associated electrical hazards. Some of these standards can be found in the
following:
• Doble Reference Book on Insulating Liquids and Gases [6]
• ASTM D923: Standard Practice for Sampling Electrical Insulating Liquids [7]*
• IEC 60475: Method of Sampling Liquid Dielectrics [10]
• IEC 60567: Guide for the Sampling of Gases and of Oil from Oil-filled Electrical Equipment
and for the Analysis of Free and Dissolved Gases [11]

*NOTE; In 2007, The ASTM sampling practice for DGA and water, D3613, was incorporated
into D923 so that there was a single document where all sampling information could be found.
There were also other substantial changes in the main body of D923.

SCIENCE OF SAMPLING
It is important to know the reasons why the above procedures are used and how samples are
affected when deviation from the written procedure occurs. Already discussed were the tests that
were most affected by poor sampling but there are other factors that are extremely important to
consider prior to taking a sample and these include:

9
 • Sample containers
• Sampling technique
• Sample storage and transport

PURPOSE OF A SAMPLE CONTAINER

The proper use of sample containers to retrieve and store samples prevents contamination
and provides the best sample for a specific test. ASTM Practice D923, the Doble Reference
Book on Insulating Liquids and Gases, or IEC Methods and Guides 60475 and 60567 are to be
consulted when choosing a sample container. Sample containers should have the following
characteristics:

• Large enough to hold the volume of liquid necessary for analysis

• Do not impart any contamination (chemical or particles) to the sample
• Seal the sample from external contamination
• Shield the sample from direct sunlight to prevent photo-degradation. This can be done either
by having a dark container or by having a covering for that container once the sample is
taken.
• Prevent the loss or gain of gases or water when testing for these properties

The volume of the sample is of the utmost importance, as various analytical tests require very
different sample volumes. For example, a test for inhibitor content (ASTM D2668) requires only
a few milliliters of dielectric liquid whereas the test for impulse breakdown (ASTM D3300) may
require as much as two to four liters. In general it is good practice to provide the sample volume
required for each test plus 10%. If unsure of the sample volume required for specific test or
tests, laboratories such as the Doble Materials Laboratory have reference lists that detail such
information.

For general oil quality tests, glass bottles, either amber or clear, function well (Figure 3).
Amber bottles provide protection against photo-degradation while clear bottles enable visual
inspection of the sample. To prevent photo-degradation in clear bottles, shielding from direct by
storing them in cardboard or some other type of container works well. This is important in a
laboratory setting as well as high intensity lights or sitting samples on a windowsill can cause
degradation of the sample in just a few hours. An inspection of the laboratory conditions in
which your samples are held for analysis is recommended and encouraged. Bottle caps (aka
closures) must be constructed from a compatible material which will not contaminate the sample.
Bottle caps with liners composed of paper or having glue that is soluble in the dielectric liquid
are not appropriate. Liners made out of foil, Teflon or polyethylene are usually safe to use. If
sealed tightly, the glass bottle is an appropriate container from which to draw a sample for water

10
analysis. Problems however have been encountered with caps working loose over time. Glass
bottles are more apt to break, if not properly protected, than other containers and they are not a
suitable container for dissolved gas-in-oil analysis, as gases such as hydrogen and carbon
monoxide will be lost. Glass bottles can be cleaned to remove most of the particulate debris and
are best container used for particle count testing. It cannot be assumed that all glass bottles are
particle free and most are not so it is recommended to check with the provider before they are
used. Glass bottles not processed to be particle free have many particles in them from the
manufacturing, storage and shipping processes and can will impact tests such as dielectric and
particle count tests. For this reason, the Doble laboratories have used only particle free glass
bottles since 2001.

Occasionally, in order to prevent the cap from coming loose, electrical tape or some other
type of adhesive tape will be wrapped around the bottle neck and cap. This is not a good practice
as the adhesive on the tape is not compatible with the oil and by the time it gets to the lab it is
usually a gooey mess. In order to get to the sample, the lab needs to remove the dripping
adhesive and the tape and there is a very high potential for contaminating the oil. Currently,
there are cap designs that provide a more positive seal and heat shrink tubing can also be used
but in the application of the heat shrink tubing care has to be used so that the sample is not
overheated and thus some of the oil qualities altered.

Example of Amber 1-liter Glass Bottles

Figure 3

11
 Canning jars (Mason, Bell, Etc.) are never to be used as the natural rubber liner of the lid
imparts corrosive sulfur compounds to the oil that will contaminate the oil and affect not only the
corrosive sulfur tests but the power factor test also even from brief contact.

Metal cans have become popular because they are more resistant to breakage than glass
bottles. Cans with soldered seams prepared with some fluxes will contaminate the sample and
therefore welded seams are preferred. Cans made of tin, aluminum, and stainless steel have been
used and are good containers especially to hold samples for oil quality tests. Extruded aluminum
bottles, although expensive when compared to glass and plastic, have served admirably well in
retaining the properties of the oil samples over long distances and handling during shipping
(Figure 4).

Examples of Aluminum Bottles

Figure 4

The use of plastic bottles has grown significantly in the past decade (Figure 5). Like cans,
they resist breakage and shield the sample from sunlight when dark plastics are used. They are
appropriate containers for samples in which oil quality tests are to be performed. Samples being
tested for water content and dielectric tests (D877, D1816, IEC 60156) should not be stored in
plastic bottles as water ingress or egress can occur in just a few hours.

12
 Examples of Plastic Bottles
Figure 5

Not all plastics are compatible with oil so selection of the construction material is important.
Tests such as power factor will be affected by incompatible plastic bottles because of dissolved
components transferred to the sample that will cause increases in dielectric loss. Bottles made of
high density polyethylene have been found suitable and are one of the preferred sample
containers when electrostatic charging tendency or furanic compound analysis is to be
performed. In the case of furanic compounds, high-density polyethylene does not have the
silanol groups found in glass that attracts the semi-polar furanic compounds to the glass walls
and thus remove them from the sample. Samples with a low concentration of furanic compounds
are not as affected as samples with larger amounts of furanic compounds.

The most appropriate container for taking samples for dissolved gas-in-oil (DGA) and water
content analyses and the easiest to manipulate is the ground glass syringe whose barrel and
plunger have extremely tight tolerances (Figure 6).

13
 Example of Glass Syringe
Figure 6

This type of syringe has been found to satisfactorily prevent the ingress of gases and water
into the sample and in turn prevent the egress of water and dissolved gases from the sample over
a period of time. Care should be taken with glass syringes to ensure the stopcock is tight and in
the closed position once the sample is taken. The samples collected in syringes must be quickly
protected from photo-degradation (degradation by light) by immediately placing them in the dark
or in their shielding containers (syringe boxes). Stainless steel cylinders may also be used but
are sometimes more difficult to manipulate or to determine when all the air has been removed
from the cylinder, thus contaminating the sample. The metal cylinders will add significant
shipping costs because of weight but are definitely more durable in transit. Flexible metal cans
and aluminum bottles have also been used when they are filled to overflowing and capped.
Although some laboratories use these exclusively they are not recommended for DGA. In most
cases, samples are retrieved from operating transformers or ancillary devices where the oil is
hotter than the outside environment. As the sample cools down in the container after the sample
is taken, some gases may dissociate from the oil and form a gas space in the container.
However, this gas is still part of the samples and must somehow be incorporated back into the
sample once it reaches the laboratory. In the case of syringes, this dissociated gas can clearly be
seen as a gas bubble in the syringe (Figure 7). The laboratory can either process the entire
sample including the bubble if the entire volume of the syringe is used or can be re-dissolved
back into the oil so that the sample can be processed directly.

14
 Gas Bubble

Gas Bubble in Syringe

Figure 7

For cans and bottles especially those that are made of metal, it is not known if there is a gas
bubble or not. Glass bottles cannot be filled to overflowing for they may implode or explode due
to pressure and temperature changes and are purposely not filled completely. Whatever the case,
as soon as the cap is loosened on one of these containers, whatever gas has disassociated from
the oil will escape to the atmosphere and cannot be captured. These samples do not represent
what was produced in the electrical apparatus and are to be treated with caution. Hydrogen and
carbon monoxide represent the most easily lost gases as the solubility of these in oil is very low.

Water Ingress into Various Sample Containers

Four (4) different types of sample containers were evaluated in this testing: high density
polyethylene (HDPE), glass bottles, glass syringes and aluminum bottles over 9 weeks.

Transformer oil (22 liters) was dried with nitrogen to a water content of less than 5 ppm.
This dried oil was transferred to 40 containers representing 10 each of four different styles: 1-
liter HDPE bottle, 1-quart glass bottle with a hard plastic cap and paper liner, 50cc glass syringe
and a 1-liter aluminum bottle with sure-lok cap fitted with an aluminum liner. The containers
were uncapped and aspirated with compressed air to remove unwanted particles and then
transferred to a glove bag previously filled with nitrogen. All containers were filled with oil
inside the glove bag.

15
 Each container was filled to overflowing with oil and capped. Syringes were filled to the
40cc mark and all gas bubbles were expelled. The stopcocks and caps were then checked for
tightness on each container. After filling, the first and last samples of each category were tested
within 1 hour for water content. The rest of the samples were maintained in a glove bag under
conditions of room temperature (approximately 22.5°C) and high humidity (83% to 88%). Due
to different construction of an additional humidity chamber, the aluminum bottles were exposed
to humidities between 71 and 80%. One sample for each category was tested weekly and the
results recorded. See Tables 3 through 4 for results.

Table 3
Water Ingress into HDPE and Glass Bottles
Sampling Container Type: HDPE Bottle Container Type: Glass Bottle
Avg. of 3 Chamber Avg. of 3 Chamber
determinations, Humidity determinations, %RH
mg/kg (ppm) mg/kg (ppm)
Initial, Bottle 1 3.8 88 % 3.4 88 %
Initial, Bottle 10 5.0 88 % 5.7 88 %
Week 1 14.4 88 % 7.6 88 %
Week 2 20.9 88 % 5.5 88 %
Week 3 30.4 88 % 9.5 88 %
Week 4 29.2 88 % 8.7 88 %
Week 6 37.8 88 % 10.4 88 %
Week 7 40.9 88 % 9.6 88 %
Week 8 44.7 88 % 12.3 88 %
Week 9 45.0 88 % 17.1 88 %

16
 Table 4
Water Ingress into Glass Syringes and Aluminum Bottles
Sampling Container Type: Glass Container Type: Aluminum
Syringe Bottle
Avg. of 3 Chamber Avg. of 3 Chamber
determinations, Humidity determinations, %RH
mg/kg (ppm) mg/kg (ppm)
Initial, Bottle 1 3.4 88 % 4.8 75 %
Initial, Bottle 10 5.9 88 % 9.8 75 %
Week 1 5.5 88 % 5.0 75 %
Week 2 4.4 88 % 4.3 75 %
Week 3 9.7 88 % 6.2 75 %
Week 4 7.5 88 % 4.6 75 %
Week 6 5.9 88 % 4.7 75 %
Week 7 7.7 88 % 5.5 75 %
Week 8 7.2 88 % 6.1 75 %
Week 9 8.0 88 % 6.4 75 %

The graph provided in Figure 8 summarizes the results from Tables 3 and 4.

Water Ingress into Various Containers Under Controlled Conditions

Figure 8

17
Discussion of Water Ingress Experiment:

The results over the nine week period indicate that HDPE bottles were not suitable for long
term storage when water content is a test to be performed. As Figure 8 indicates the maximum
water content at the experiment's end was 45 ppm for the HDPE bottle whereas the other
containers were much lower. In addition, that caps worked loose on both the HDPE and glass
bottles over time. If sample integrity is to be maintained for a long time, then periodic resealing
of the caps is necessary. Because of this experiment, Doble employed the use of a special cone
shaped cap to minimize this occurrence.

The glass bottles, syringes and aluminum bottles provided the best environment for
maintaining the water content of a sample over a prolonged period. Figure 8 indicates that there
was a small increase of water in oil in the syringe over time. However, it is not known if that
contamination is due to external or internal sources. For example, water adhering to the inside of
the glass barrel may have contaminated the sample after the oil was placed in it. When the
syringes were allowed to sit for a period of time the water attached to the surface of the barrel
may eventually contaminate the oil, as evidenced by the slight increase in water content in all the
syringes after the first four were tested. This was not the case for the aluminum bottles where
the water content essentially remained the same for the full nine weeks. The sure-lok caps with
the aluminum liner provided an effective sealing mechanism to keep water out.

Figure 8 depicts how the oil in each container fluctuated in water content in relation to the
maximum water content that could be dissolved in oil using 88% or 75% relative humidity and a
laboratory temperature of 22.5°C. At 22.5°C the 100% saturation of water in oil is 61.2 ppm.
Eighty-eight percent of 61.2 ppm is 53.8 ppm (75% of 61.2 ppm is 45.9 ppm for the aluminum
bottles). Those were the concentrations used to calculate to what percentage the oil in each
container rose in relative saturation over time. As shown in Figure 8 the Nalgene container was
almost 84% (45 ppm/53.8 ppm) of that value by experiment's end whereas the glass bottle rose to
32% and the syringe was only about 15%. The aluminum bottle stayed fairly constant at 10 to
12%. It is obvious that the external relative humidity affects the moisture content of the sample
in certain containers (i.e. Nalgene bottles). Testing the sample quickly and keeping the
laboratory's relative humidity low would negate some of these effects. However, keeping the
laboratory's relative humidity low may also have the opposite effect. Equilibrium with the
relative humidity of the ambient environment would also tend to dry the sample if the
concentration of water on the outside of the sample was less than the water in the oil in the
sample container.

18
 These effects do not seem to manifest themselves in the syringes and the aluminum bottles.
This probably would also hold true for the glass bottles if the caps where able to stay on tight.
The syringes and aluminum bottles seem to be the best containers to store oil for the
determination of water content in transformer oil samples. The aluminum bottles provide
additional advantages of holding a larger volume of oil for other testing, shielding the sample
from sunlight and providing a less breakable container for shipping.

SAMPLING TECHNIQUE
Sampling technique involves much more than just taking the sample. It involves a more
thorough knowledge of the information to be gained from taking a proper sample and includes
sample site preparedness and site cleanup after sample retrieval. Some the items that are a part
of sample technique are:
a. Materials used to aid in retrieval of a sample
b. Safety precautions to adhere to
c. Environmental concerns
d. Identification of the sample and apparatus information
e. Final checks prior to sampling
f. Taking the sample (cleaning and preparation of valves)
g. Cleanup after sample has been retrieved

Materials Used
Whoever takes the samples must be fully prepared for most eventualities that will occur at
the sample site. For example, items such as sheet plastic, plastic bags, absorbent materials, flush
oil containers, and catch pans are all important materials to have to prevent or clean up liquid
spillage. It must be remembered that, in order to take a proper sample, some liquid waste will be
generated. Of course, bottles and syringes will be needed as sample containers and must be of
sufficient size to hold the volume of dielectric liquid necessary for the desired tests. Labels are
required to sufficiently and correctly identify those containers. Make-up oil and bottled nitrogen
may be necessary to add oil to low volume devices and nitrogen is used to pressurize a
transformer to relieve a negative pressure in order to get the dielectric liquid out of the apparatus.
Tygon tubing or other compatible tubing is necessary to direct the dielectric liquid from the
drain valve to the flush container, sample bottle and syringe. Tubing should only be used once
and then discarded as the walls of the tubing have memory (can hold gases, water and other
chemical compounds in the walls of the tubing) which can then be transferred to the next sample.

19
 The most common example are situations in which workers take a sample from a break-in-
oil LTC and then using the same tubing for the main tank sample. Although the run of tubing
can be very short (9 inches (20cm) or so), the chemistry allowing the tubing to be flexible also
allows gases, PCBs, water, etc., to diffuse into the walls of the tubing. Then, when oil flows
through the tubing from an apparatus that does not have any of these contaminants, diffusion of
the contaminants from the tubing to the oil will occur in an attempt to maintain an equilibrium
between the contamination in the tubing and the contamination in the oil.

Any tubing will not suffice. Incompatible tubing such as natural rubber or PVC tubing will
contaminate a sample with unwanted materials. Natural rubber tubing is particularly problematic
as it imparts corrosive sulfur compounds to the oil.

The appropriate tools and plumbing accessories must be on-site to manipulate the drain
valve in order to retrieve the sample. Personal protective equipment such as nitrile gloves is
used to protect personnel from the liquid dielectric and/or polychlorinated biphenyls (PCBs).
Personal protective equipment and safety practices to protect against electrical or physical
hazards must also be present and observed.

Safety Precautions
There are safety precautions that must be adhered to in order to secure the well-being of the
equipment as well as the personnel retrieving the sample. Death is not normally associated with
sample taking but it has occurred when electrical hazards have not been observed. Routine and
complacency often contribute to a lax adherence to safety precautions. Some of the more critical
safety precautions are:
• Make sure there is positive pressure on the electrical apparatus (even if it is a free
breathing apparatus as air vents or desiccant dryers are often blocked)
• Take into consideration the remaining volume in a low oil volume apparatus (this
may require de-energizing the equipment and adding oil when finished)
• De-energize instrument transformers and network transformer switches before
sampling
• Secure electrical dangers
• Make sure Occupational Safety and Health Administration (OSHA) requirements are
adhered to

Confirming there is positive pressure on the electrical apparatus prior to sampling is the
single most critical factor in assuring that the equipment and sampling personnel survive the
sampling procedure. Sampling of electrical equipment while under negative pressure will allow

20
atmospheric air to be drawn into the equipment through the drain valve and will rise through the
transformer as bubbles. These bubbles are areas of weak dielectric strength and can easily cause
failure of the apparatus through flashover. Most transformers have pressure gauges that allow
determination of the actual pressure. If positive or negative pressure cannot be determined, then
follow the procedure in ASTM D923 to determine the pressure condition which involves using a
slug of oil in clear tubing attached to the sampling cock. If negative pressure does exist then no
samples are to be drawn until that negative pressure is relieved. Sometimes this is as simple as
adding dry nitrogen to the headspace of a transformer to pressurize the units or waiting until
ambient temperature has increased to a sufficient degree to cause the expanding dielectric liquid
to pressurize the apparatus. Nitrogen must be added in such a way as not to set off the sudden
pressure relay device.

The remaining volume in some electrical apparatus especially OCBs, LTCs, bushings and
small instrument transformers is of serious concern. Electrical components are positioned at
critical distances taking into consideration that this distance is determined with liquid insulation
present. If the liquid insulation level drops too low the insulation between these components is
now air instead of the insulant and thus the dielectric integrity has been compromised. This is
why it is important to check the liquid levels not only before sampling but after in order to
maintain a safe operating environment. Instrument transformers and network transformer
switches must be de-energized prior to sampling in order to secure the electrical hazards.

Electrical hazards are especially prevalent in small distribution pole and pad mounted
transformers. For pole transformers, these units do not usually have a sample valve so the lid of
the transformer must be removed to take a sample. In addition the primary and secondary
voltages terminations are extremely close to personnel. This is also true of pad mounted
transformers where the secondary and sometimes primary cable or busbar is within feet of the
sample valve. Any wrong move by sample personnel and serious injury or death can result.
Because of such concerns, OSHA instituted lockout/tag procedures that must be adhered to
secure against such dangers.

Environmental Concerns
Dielectric liquid spillage as a result of sampling is a main environmental concern, as some
of these liquids may still contain PCBs. The United States as well as many other countries have
very strict guidelines for spill cleanup and notification of PCB materials. In the United States
even one drop of liquid containing more than 50 ppm PCB is considered “improper disposal”.
Even if the oil does not contain any PCBs, many states within the United States as well as other
countries have regulations dealing with the release of dielectric liquids even if it is small and
accidental. It is therefore easier to prevent against spillage then to cleanup after the spill has
occurred. This is why many sampling personnel lay down plastic and absorbent materials under
the drain valve prior to sampling and then use a catch pan to trap larger volumes of liquid. There

21
is always some oil remaining in the valve after the sampling process is completed and will be
present the next time a sample is retrieved.

Identification of the Sample and Apparatus Information

Sample identification is an extremely vital aspect of the sampling process. Many
laboratories receive samples that cannot be related to a specific device. Even if the sample was
taken properly, the lack of proper identification makes the sample useless and it therefore is a
poor sample. When sampling personnel retrieve samples from electrical apparatus, it should be
done in a prescribed sequence so nothing is forgotten.

Case in point. LTCs can contain the selector and transfer switch components in one tank or
they can be in two separate tanks. In such cases where they are housed in two separate
compartments, the gases produced in one compartment should be completely different then in the
other compartment. The transfer switch houses the arcing contacts and the thus the DGA
analysis should indicate large amounts of gases such as ethane, ethylene, methane and acetylene.
If this sample is mislabeled and is termed the selector, interpretation of the results will be vastly
different and can result in an overhaul of a compartment that does not need it. Although rare,
there are conditions that exist where the selector can have high gas contents and thus diagnosis of
this condition is very important and thus the need for proper identification.

The lack of information concerning an apparatus severely limits the laboratory in its ability
to provide an in-depth diagnosis. Apparatus information such as the core construction, age, type
of preservation system, insulating liquid type, any previous incipient fault conditions or oil
reclamation activities can alter a diagnosis. For instance, the Doble Insulating Materials
Laboratory relies heavily on the type of preservation system that is part of the transformer to
provide a diagnosis based on oil quality, DGA and furanic compound results. In transformers
that have a sealed conservator preservation system, oxygen and nitrogen values are expected to
be below certain levels. If test values are above those levels, there may be several causes for this
such as a breach in the bladder or diaphragm, another leak elsewhere on the transformer or poor
sampling. All these items would concern the operator of the equipment. However if no
information is provided or the information is incorrect then no or an incorrect diagnosis will be
provided by the laboratory. Age and the type of insulating paper of a transformer are
exceedingly pertinent information when providing a diagnosis based on furanic compound
results.

An example of a sample identification (ID) card is shown in Figure 9. Not all sample ID
cards are the same and the layout of the card really depends on the laboratory issuing the ID
cards. Doble requires more information than most as that information is used for diagnostic

22
processes after the sample is tested. Most ID cards require 3 subsets or sections of information,
the:

Section 1: Apparatus information

Section 2: Information to be gathered at the time of sampling
Section 3: Tests that are to be performed

Section 1

Section 2

Section 3

Example of Sampling ID Card

Figure 9

23
 Section 1 of the Sample ID card is just relaying information about the apparatus to the
laboratory that will be performing the analysis. Most of the fields should be self-explanatory as
most of the fields are directly from the nameplate mounted on the apparatus. The sample ID card
shown above is for a transformer or LTC. The serial number is very important as it is unique to
each transformer and the easiest way to identify the apparatus. This is necessary to track the
apparatus over the life of the unit. Many of the tests performed such as DGA, furanic
compounds, certain oil quality tests such as interfacial tension, neutralization number and power
factor will be used as a trending tool over the life of the unit and this is the best way to track that
information. It is analogous to the blood work of a patient over the life of that patient where
items such as blood sugar, triglycerides, total cholesterol, LDL an HDL levels are followed.

Equipment numbers are useful as most utilities identify the apparatus by this number as it
defines a certain asset. It is also important in some oil tracking software such as Delta-X TOA3®
or TOA4® or SAP® where it is a required field to identify the apparatus. Not everyone uses the
Transformer name and most names relate to the position of the apparatus in a plant or substation
but not always. Maximum kV and MVA is based on cooling. For example a transformer
nameplate may indicate three levels of cooling and as such the MVA rating will be listed as
40/50/60 MVA. In this example the maximum MVA rating would be 60 MVA. The kV rating
is the voltage rating of the highest side. The mistake that is repeated most often is when small
transformers are being sampled and the rating is in kVA instead of MVA. Sometimes the kVA
value is written into the area where the kV should be displayed and this is incorrect. To obtain
the MVA rating from a transformer that is listed in kVA, divide that value by 1000. For
example, a 1000 kVA transformer would be listed on the ID card as 1 MVA, or a 500 kVA
transformer would be listed as a 0.5 MVA.

If the transformer does not have a load tap changer or if the sample is not coming from the
LTC then the information in that field does not need to be completed. If the LTC is being
sampled, then the manufacturer of the LTC, model number and preservation system are of the
utmost importance in order to provide a reliable diagnosis of the unit from the testing performed.
As discussed previously if the LTC has multiple compartments then the one being sampled needs
to be clearly identified.

A large portion of, or all of Section 1 can be completed beforehand. Most laboratories will
have a majority of this information in their software system if a sample has been provided to
them before and the nameplate information has already been provided. For repeat sampling of
the same equipment, usually only the serial number and equipment numbers are necessary.

24
 Section 2 can only be completed at the time of sampling. Again, most of the fields should
be self-explanatory. The first item “Syringe and/or container ID” is absolutely necessary to
identify the sample especially if there are samples from multiple apparatus in the same shipment.
Most, but not all, syringes will have some sort of ID number on them, Refer to Figure 10.

Syringe Identification Numbers (circled area)

Figure 10

On occasion a syringe will not have any sort of marking and the sampler will need to find a
way to identify the syringe. Writing the serial number of the apparatus on the syringe box that
the syringe came in is one such method. Tying a tag with the information around the syringe
plunger is another method.

Bottles used to take samples for oil quality are usually equipped with an identification
number. These are usually identified with tie-on tags or adhesive labels such as the one shown in
Figure 11.

25
 Customer Name:________________________
Serial Number:__________________________
Date:___________________________________
Top Oil Temp:_________°C

Syringe No.:_________________________
Please complete the Sample ID sheet with the
rest of the necessary information
Lab Use Only:
Report No.:______________ Sample No:______

Example of Bottle Label

Figure 11

Not all samples are performed by utilities but by service organizations such as those
belonging to NETA. The “Secondary Name (Job)” was added to aid these organizations identify
the samples which belonged to a certain customer mainly for cost and billing reasons.

Sample Point and apparatus type need to be defined. “Main Tank” or “bottom” are often
used for transformers. For LTCs, “LTC” or if it has multiple compartments then “Transfer” or
“Selector” are used. For Apparatus type, the use of “XFMR” or TRN” denotes a transformer
whereas LTC denotes a tap changer and OCB refers to an oil circuit breaker. The phase
designation is especially important for OCBs that have 3 separate tanks and each phase needs to
be identified.

The time field is especially vital if a tanker, oil processing or insulating filled electrical
apparatus will have multiple samples taken over the course of a day or several days. Since all
the samples will have the same serial number, and/or tanker number, there is really no other way
to identify it other than sample date and time.

26
 Ambient temperatures and humidities can affect the quality of the sample. If there is a need
to sample a very dry transformer on very hot and humid day then the water content and possibly
the dielectric values can be influenced particularly if plastic bottles are used. This gives the
laboratory or the reviewing engineer information that may explain a less then desirable result.

The last item in Section 2 can be one of contention between laboratories and/or engineers.
The water value obtained from an analysis is used to determine the wetness or dryness of an oil.
If the temperature is elevated within the transformer and it is known then the reasonable
assessment of the wetness or dryness of the paper insulation can also be made. The temperature
needs to be determined at the time of sampling as the water content in the oil will change as the
transformer temperature changes in particular for those transformers that cyclically pick up and
shed load. A brief discussion of the three ways to determine the temperature in provided in
Table 5 and can also be found in ASTM D923-07.

TABLE 5
Sample Temperature Recordings

Location of Temperature Measurement Advantages and Disadvantages

Retrieval
Apparatus Top Oil Temperature
Gauge
Temperature of sample as it
exits the sampling valve
(thermometers)
Infrared point and shoot
thermometer (small handheld
units)
Type
Advantages
The thermowell of the gauge is installed directly into the oil of the
apparatus thus eliminating external influences.
Direct Disadvantages
Gauges need to be kept in calibration. Very small transformers and
OCBs usually do not have temperature gauges. Convection cooled
transformers can have a large temperature differential between the top
and bottom of the apparatus.
Advantages
Have the ability to always retrieve a temperature no matter what
apparatus is being sampled.
Disadvantages
Indirect
The thermometer used may be out of calibration. The drain valve is a
large thermal sink and most often will cool the exiting oil from 2 to
25°C thus not representing the bulk oil temperature. There can be a
response time issue with the use of a thermometer and the oil may even
cool down more before the reading stabilizes.
Advantages
Have the ability to always retrieve a temperature no matter what
apparatus being sampled. Depending on the model, various features
are available including a scanning type that can record the temperature
from the top of the tank to the bottom.
Indirect
Disadvantages
Accuracy of the measurement is based on how close the user is to the
apparatus being surveyed. The emissivitty of the apparatus material of
construction and color add additional error to the measurement.

27
 All temperature measurements are an average as the temperature profile within a
transformer is not uniform due to heating and cooling activities that are constantly occurring
inside and outside of the transformer. Doble would recommend the use of the top oil
temperature as it the temperature that is the least impacted by external influences.

Doble undertook a study to understand temperature gradients within transformers. A

summary of the methodology and parameters employed was as follows:

 Used a laser guided infrared thermometer to take readings, error was ± 1°C
 Top Oil Versus Bottom Oil
 Tank Versus Drain Valve
 Convection Cooling Versus Pumped Cooling
 Different Manufacturers
 All transformers were power plant transformers.
 Transformers located in Ohio, Pennsylvania, New Jersey and Connecticut,
 61 transformers were surveyed during sampling

Temperatures were taken from areas shown in Figure 12.

Transformer Tank and Oil Temperature Profile Locations

Figure 12

28
 Observations from the temperature profile survey showed that top oil gauge read very close
to laser guided infrared thermometer, usually within 1 or 2°C and the largest difference was 5°C.
Table 6 summarizes the data from the study.

Table 6
Temperature Averages from 61 Transformers, °C

Averages Top Bottom Oil Delta from Top Delta from Bottom
Oil XFMR Tank Sample to Bottom Tank Tank to Sample

All XFMRs 43.82 31.54 33.79 12.28 +2.25

No Pumps 40.45 26.92 21.00 13.53 -5.92

Pumps On 60.00 53.42 49.14 6.58 -4.28

Largest delta 73.0 31.7 41.8

Smallest delta 41.0 44.4 -3.4

Some individual profiles are shown in Figure 13 but are representative for the population of
transformers tested.

Common Temperature Profiles of Transformers Studied

Figure 13

29
 In Figure 13, the blue line represents temperatures of transformers with operating pumps and
for this population of transformers these were generator step-up (GSU) transformers that were
under constant load. The green and red lines represent the temperatures from unit auxiliary and
transmission transformers. In general, transformers with operating pumps show the smallest
change from top oil temperature to the bottom oil temperature (“Tank Behind Drain Valve”)
shown in Figure 13 with the difference being about 4 to 5°C. Generally, the convection cooled
transformers show a much larger temperature delta from top oil temperature to the bottom tank
temperature with the examples in Figure 13 exhibiting deltas of 15 and 23°C.

Lessons learned from this study include the following:

 Different profiles between convection cooled and pumped cooled units
 There was a very small change in sample valve temperature during sampling
 There was a significant cooling effect from top to bottom oil and then again from bottom
to final oil sample
 Top oil will still be a little cooler than the upper portion of the transformer windings
 Top oil is a better proxy than bottom oil or sample temperature

The most significant issue with using the bottom oil or sample temperature is how it used to
calculate the percentage of water in the cellulose insulation. Figure 14 shows the temperatures
that were recorded from one of the study transformers, where the temperatures were taken on the
transformer and the calculated relative saturation and water in paper values from the water
content, in this case 10 mg/kg.

The calculated values depicted in Figure 14 indicates that a doubling of the calculated paper
water content can occur depending on the use of top oil versus other temperature measurements
and thus will impact the population of transformers to be considered wet. In certain cases, there
would be an overestimation of the water content in paper insulation which may lead to drying an
already dry transformer.

30
 Calculated Relative Saturation and Water in Paper Concentrations
Figure 14

In referring back to Figure 9, the third section of the sample ID card allows the sampler to
indicate which tests are required for the sample. This section can be confusing to those that do
not understand the significance of each test or do not know what tests are included screen
packages. The person requesting the sample or the laboratory should be consulted in order to
determine which tests are appropriate.

There is no official definition of what constitutes a screen. Over time, certain screen
packages have been developed by utilities or laboratories to easily identify a set of tests that are
commonly performed. A very popular screen is a “7-part” screen (Table 7) that incorporates
different chemical, physical and electrical tests to determine the quality of the oil being tested.

TABLE 7
Tests Found in a 7-part Screen

Test Name ASTM Test Name ASTM

Method Method
Color/Visual D1500/D1524 Interfacial D971
Tension
Dielectric Strength D877 or D1816 Specific Gravity D1298
Water Content D1533 Power Factor at D924
25°C
Neutralization D974
Number

31
Laboratories offer many types of screens and the amount of tests in a screen can range anywhere
from 2 to 35 different tests and it can be daunting task to select the correct one, so consultation
with your laboratory is necessary to determine which one is best for a particular apparatus or
situation.

Final Checks
Adhering strictly to safety and environmental concerns will assist personnel in adequately
preparing the site around the electrical apparatus to be sampled. However, specific attention
should center on some final checks before the sampling commences. These include:
• Confirmation of positive tank pressure – again!!!
• Using the correct sample containers both in size and compatibility
• Labeling and sample information completed
• Protecting the sample from outside contamination
• Relative humidity of ambient air less than 50%. (Avoid rain or snow conditions.)
• Temperature of dielectric liquid higher than or equal to the ambient air

Checking for positive pressure is so critical that it requires a second check and must not be
overlooked. As mentioned previously, it is very difficult to perform the requested analysis when
not enough sample volume is present. Therefore, it is very important to make sure the correct
size sample container is used. If unsure of the sample volume requirements, it is better to
provide more sample than to re-sample the apparatus at another time. It is recommended to label
the samples with complete sample information prior to the actual taking of the samples. In this
way, sample containers will not go unmarked and confusion can be avoided especially when
multiple pieces of equipment are being sampled.

Some geographic locations are inherently inhospitable to providing an environment where a

good sample can be taken. Locations such as those in salt spray areas, high wind gusting areas
that naturally kick up sand, dirt, soot and other debris can easily deposit foreign matter into an
open sample container while it is being filled. In these instances protecting the sample from
outside contamination may require special precautions. Really adverse environments may
require totally enclosed sampling systems such as stainless steel cylinders to protect against
outside contamination.

The ambient environment in which a sample is taken can contaminate the sample with
excessive amounts of moisture. Taking samples during rain, hail or snow conditions should be
avoided. If the situation is completely unavoidable then necessary precautions must be enacted

32
to keep the sample from absorbing any external moisture. It is recommended that sampling be
performed only when the relative humidity conditions are less than 50%, so as to minimize the
amount of external moisture the sample will absorb. It is recognized that this may be
unavoidable in certain geographical locations and during certain periods of the year but, every
effort must be made to have a sample that reflects the bulk liquid insulation and to minimize
external contaminants.

Taking the Sample

Different dielectric liquids require sampling from different locations on the electrical
apparatus based on their relative density (specific gravity). In general, dielectric liquids with a
relative density less than 1 should be sampled from the bottom drain valve whereas dielectric
liquids with a relative density greater than 1 should be sampled from the top fill valve as long as
it is below the liquid level. There are exceptions to this and the sampling point can change
throughout the life of the transformer. For instance, mineral oil transformers that have no drain
valve are usually accessed and sampled through the top. Another example involves retrofilling
of askarel transformers. When a transformer is filled with askarel, sampling should be
performed from the top fill valve because the relative density is greater than 1. However, many
askarel transformers have been retrofilled and the askarel insulating liquid replaced with silicone
which has relative density less than 1. In this case the transformer should now be sampled from
the bottom. Table 8 is a list of sampling points for various dielectric liquids in apparatus during
routine testing. Sampling of drums, tankers and other types of storage containers is performed in
a different manner. Consult the referenced guides for specific procedures.

Table 8
Sampling Points for Various Dielectric Fluids in Apparatus

Sample from Bottom Sample from Sample from Top Not Recommended
Bottom to Sample
Transformer Mineral oil Biotemp Polychlorinated Freon filled
Biphenyls transformers
Silicone Biotrans Trichlorobenzene Sulfur Hexaflouride
(SF6)
R-Temp ECO Fluid Tetrachlorobenzene
Beta Fluid Edisol TR Wecosol
Shell Diala HFX Envirotemp® FR3 Perchloroethylene
WEMCO-FR Envirotemp 200
Opticool Midel 7131
ALPHA-1 FLUID Reoloec 138
Polyalphaolefins (PAOs) Nycodiel

33
 Once the correct valve from which to retrieve the sample has been determined, that valve
should be prepared for taking of the sample. As mentioned previously, check for positive
pressure on the apparatus before opening the drain valve. Adequate preparation of the valve for
sampling consists of the following:

• Clean the outside of the valve to remove any loose debris that may fall into the
sample
• Make sure the valve and sampling cock are closed before removing the drain plug
• Prepare the area under the valve with absorbent materials and a catch pan
• Slowly remove the drain plug
• Clean the inside of the valve with a lint free cloth
• Reinsert the drain plug and then purge the sampling cock.
• Close the drain valve and remove the drain plug again remembering to be prepared to
catch left over oil from the sampling cock purge
• Clean the inside of the valve again
• Install brass, bronze, or stainless steel adapters to the drain valve and then to a hose
barb so that tubing can be attached

A diagram of a 2 inch globe valve is shown in Figure 15. Globe valves are used most often
in transformer construction as they provide the best seal against pressure and vacuum.

Globe Valve Diagram

Figure 15

34
 Although the procedure listed above sounds like a lot of extra work it is necessary in order
to retrieve a sample free from outside contamination. Doble Engineering recommends that
samples be retrieved from the main drain rather than the sampling cock. Although convenient,
the sampling cock is connected by a very small hole to an area between the drain plug and the
valve seat. This is the area that accumulates all the debris and water as shown in Figures 2 and
16. Special care must be taken to purge this area.

Valve Seat
Transformer Tank

Drain Plug

Sampling Cock Water and Debris Accumulation

Debris and Water in Sampling Cock

Figure 16

Even after repeated flushings, the sampling cock is rarely totally clean, and water and debris
will break free and subsequently contaminate the sample. However, flushing of the sampling
cock is important as it does remove a large portion of the water and debris prior to taking the
sample through the main drain valve. Once the adapters are all installed with the hose barb, the
final assembly may resemble the shown in Figure 17.

35
 Adapter

Hose Barb

Valve Body

Tubing

Drain Valve with Adapter, Hose Barb and Tubing

Figure 17

Once the valve is totally prepared sampling can commence. The practices as referenced
above all provide very detailed information concerning taking the actual sample. However,
listed below are additional points to remember.
• Flush at least 2 to 4 liters of dielectric liquid through the valve prior to taking a
sample
• If taking both syringe and bottle samples, take bottle samples first and syringe
samples last
• Rinse bottles 2 to 3 times with about 1/3 of their volume prior to taking actual
samples
• Rinse syringes 2 to 3 times prior to taking actual samples
• Fill the bottles without causing aeration or turbulence of samples
• If using glass bottles, fill to the shoulder of the bottle and secure the caps
• If using metal cylinders, metal cans or plastic bottles fill to overflowing and close or
cap

36
 The flushing procedure is very important in order to remove debris and water from the
valve, in order to get a sample that reflects the bulk liquid insulation. Cast iron valves tend to
retain more moisture on valve walls then do brass, bronze or stainless, so more flush liquid may
be required. A good quality control check, to determine if all the water has been removed from
the drain valve, is to install a portable water sensor such as the DOMINO USS MIO onto the
drain valve. In this way the water content can be monitored while flushing is taking place. Once
the water sensor has reached a stable ppm value it can be assumed that all of the excess or
unwanted moisture has been removed from the valve (see Figure 18).

Water Sensor

Drain Valve

Water Sensor Manifold

Tubing for liquid

sample

Water Sensor attached to Drain Valve

Figure 18

37
 De-energized equipment may require more flush liquid (8 to 15 liters) to clean the valve as
more condensation of water and settles to the bottom occurs as the apparatus cools. On low
volume apparatus this should be monitored closely.

As mentioned earlier, when multiple samples are required, Doble recommends taking the
bottle sample first and syringe sample last for several reasons. One reason is that the water
content is usually performed on the syringe sample so in addition to the original flushing that is
performed, taking the bottle samples first provides additional flushing. The other reason is the
syringe sample is used for the DGA test, which is the most critical of all the tests, as it provides
information on the operating condition of the transformer. Hopefully, the additional flushing
caused by the filling of the bottle sample will provide a syringe sample that is best representative
of the bulk liquid insulation.

Rinsing a bottle several times removes any debris remaining from the bottle manufacturing
process, and conditions the container to receive the sample by warming the walls of the
container, so water condensation does not occur during sampling. The same is true of the
syringe, where flushing and purging helps to remove any debris and moisture, coats the plunger
to create an adequate seal and helps to remove air bubbles. Once the syringe is filled, any air
bubbles remaining must be quickly removed. However, if gas bubbles appear after the dielectric
liquid has cooled then do not release those bubbles as they are gases that have just come out of
solution but still comprise the sample. Syringe samples must also be shielded from the sunlight
to prevent photo-degradation of sample.

When filling bottles with the dielectric liquid sample, aeration and turbulence must be
avoided. Aeration and turbulence will cause air and water to be entrained in the sample thus
increasing the water content and possibly affecting some of the other properties of the oil. Glass
bottles are not filled, to the very top, to avoid breakage due to the expansion or the contraction of
the liquid. Metal cylinders, metal cans and plastic bottles do not suffer from this problem and
therefore may be filled to overflowing and sealed.

Cleanup
Cleanup is a necessary step of the sampling activity. The area should be left cleaner than
found so that possible hazards can be minimized for the next sampling crew. Make sure that
drain valves and sampling cocks are wiped clean of oil and closed tightly. Replace drain plugs
using the correct type of pipe sealant so that they are easily removed the next time the apparatus
is sampled. Remove all debris and tools from the area, and cleanup any liquid spillage. In

38
addition, record the top oil temperature of the apparatus so that the relative saturation can be
calculated, and make sure that the samples are adequately labeled.

SAMPLE STORAGE AND TRANSPORT

Samples should not be stored longer than a few days before shipping to the laboratory for
analysis. The properties of the dielectric liquid tend to decay over time, and gases for DGA can
be lost or gained if the sample is held too long. Remember that the sample is supposed to
represent the bulk liquid insulation and storing it too long will change those properties. Store
samples in the dark to prevent from photo-degradation, and keep them away from temperature
extremes and humid environments.

Package the samples to protect from leakage and breakage, this is especially true for samples
in glass. The United States Environmental Protection Agency Resource Conservation and
Recovery Act (RCRA) listed in 40 CFR Part 261.4(d) requires that all samples shipped must be
packaged so they do not leak, spill or vaporize from their packaging. Other countries may have
similar requirements as well. Use methods to prevent breakage such as the use of cardboard,
Styrofoam, or other protective materials. Package the samples in such a way as to avoid sample
to sample contact. Also, some shipments become too heavy and unwieldy and the shipping
company may not be able to handle it safely. In this case, it is advisable to separate the shipment
into several packages. It is also advisable to send in multiple shipments for analysis when the
associated sampling process occurs over a period of time.

OBSERVATIONS AND RECOMMENDATIONS

Sampling of the liquid dielectrics from electrical apparatus is a non-intrusive way to permit
access to helpful information regarding the operating condition of the apparatus. Physical,
electrical and chemical tests performed on the sample can permit an educated determination of
the condition of the solid and liquid insulation system in the apparatus; or the existence of an
incipient fault condition.

Tests of the liquid dielectric usually supplement electrical tests performed on the apparatus
further assisting in the interpretation of such tests. These tests are especially helpful in cases
where apparatus cannot be removed from service for complete testing or where complete testing
can only be performed infrequently.

39
 The first requirement of liquid dielectric testing is a representative sample of the material in
question, what has been referred to as “Good” samples. A “Poor” sample is not a “Sample”. A
“poor” sample represents the loss of time and expense, and the possible overlooking of a
hazardous condition in its incipient or developing stage. Poor sampling practices resulting in
poor samples are to be avoided.

This paper has attempted, sometimes repetitiously, to summarize reasonable and effective
sampling procedures. It has discussed sampling containers, their labeling, sampling care,
sampling techniques, cleanliness before and after sampling, sample storage and transport, and
safety and environmental concerns. A review of practices on each client system, in light of these
recommendations, may result in some “tightening up” of routine, familiar practices and lead to
fewer “Poor” and hopefully all “Good” samples in the future.

REFERENCES
1. “Items of Interest” in The Doble Exchange, The Doble Engineering Company, Watertown,
MA, USA, Volume 11, Number 3, September 1993, Page 4.

2. Transformer Oil Purchase Specification (TOPS), edited by the Doble Oil Committee,
Revised 2008, Doble Engineering Company, Watertown, MA.

3. “ASTM D3487-09: Standard Specification for Mineral Insulating Oil Used in Electrical
Apparatus” in Electrical Insulating Liquids and Gases; Electrical Protective Equipment,
Annual Book of ASTM Standards, Vol. 10.03, ASTM, West Conshohocken, PA, 2012.

4. “IEC 60296: Specification for Unused Mineral Insulating Oils for Transformers and
Switchgear” International Electrotechnical Commission, 3, rue de Varembe, Geneva,
Switzerland, 2012.

5. “IEEE C57.106-2006: IEEE Guide for Acceptance and Maintenance of Insulating Oil in
Equipment”, IEEE, 345 east 47th Street, New York, NY, 2012

6. Reference Book on Insulating Materials, edited by the Doble Client Committee on Liquid
Insulation, 2006, Doble Engineering Company, Watertown, MA.

7. “ASTM D923-07: Standard Practice for Sampling Electrical Insulating Liquids” in Electrical
Insulating Liquids and Gases; Electrical Protective Equipment, Annual Book of ASTM
Standards, Vol. 10.03, ASTM, West Conshohocken, PA, 2012.

8. Griffin, P. J. “Water in Transformers – So What!”, National Grid Condition Monitoring

Conference, May 1996.

40
9. Sheppard, H. R., “The Mechanism of Gas Generation in Oil-Filled Transformers”, Minutes of
the Thirtieth Annual Conference of Doble Clients, 1963, Section 6-60

10. “IEC 60475: Method of Sampling Liquid Dielectrics”, International Electrotechnical

Commission, 3, rue de Varembe, Geneva, Switzerland, 1974.

11. “IEC 60567: Guide for the Sampling of Gases and of Oil from Oil-filled Electrical
Equipment and for the Analysis of Free and Dissolved Gases”, International Electrotechnical
Commission, 3, rue de Varembe, Geneva, Switzerland, 1992.

41
 Lance R. Lewand
Lance Lewand is the Laboratory Director for the Insulating
Doble Insulating Materials Laboratory and is also the Product
Manager for the Doble DOMINO, a moisture-in-oil sensor.
The Insulating Materials Laboratory is responsible for routine
and investigative analyses of liquid and solid dielectrics for
electric apparatus. Since joining Doble in 1992, Mr. Lewand
has published over 75 technical papers pertaining to testing and
sampling of electrical insulating materials and laboratory
diagnostics.

Mr. Lewand was formerly Manager of Transformer Fluid Test Laboratory and PCB and Oil
Field Services at MET Electrical Testing Company in Baltimore, MD USA for seven years. His
years of field service experience in this capacity provide a unique perspective, coupling
laboratory analysis and field service work.

Mr. Lewand received his Bachelor of Science degree from St. Mary's College of Maryland. He
is actively involved in professional organizations including the American Chemical Society, a
representative of the U.S. National Committee for TC10 of the International Electrotechnical
Commission (IEC) and ISO TC28, ASTM D-27 since 1989 and is the sub-committee chair 06 on
Chemical Tests, secretary of the Doble Committee on Insulating Materials, and the newest
recipient of the ASTM Award of Merit for Committee D-27.

David Koehler
David Koehler received his Bachelor’s Degree in Chemistry
from Indiana University and his M.B.A. from Indiana
Wesleyan University. Mr. Koehler has over 14 years of
experience in the testing of insulating fluids and management
of analytical laboratories. He has provided numerous technical
presentations at various power industry conferences. Mr.
Koehler is a proud member of various industry groups
including Chair of the IEEE-Central Indiana Section, IEEE
Region 4 East Area Chair, ASTM D-27 Technical Committee
on Electrical Insulating Liquids and Gases, and in 2011 was an
Executive Committee Member of the Indiana American
Chemical Society and remains active within ACS.

42

View publication stats