Premature Aging of the Transformer Due to Inadequate Drying of the Insulating System

Presented By:
Brian Sparling, SMIEEE
TechCon 2024


Moisture accumulation in the solid insulation of the transformer, is one of the key contributors to a transformer’s premature end of life. A wet transformer can fail at any given time even when operating well under its nameplate loading. The traditional method for transformer dry-out is to circulate filtered and dried oil through the unit. This process will remove the moisture from the oil, but what about the transformer insulation? To have complete peace of mind, both the oil and the insulation must be dried.

If not removed periodically, moisture can lead to partial discharge, catastrophic dielectric breakdown, accelerated aging, and bubble formation during overload situations.

The story begins with the review of the principal components involved, the cellulose insulating paper and the dielectric fluid (oil in this case). For decades performance and costs of the paper/oil insulation system has worked well. One aspect of this system that has a large impact on both materials is the presence of water.

The traditional method of measuring water in transformer oil is the ASTM D-1533 test, well known as the Karl Fischer reaction test. When used with new, filtered, and processed oil (before it goes into the transformer tank), this test gives an accurate measurement of the absolute water content of the sample in parts per million (ppm).

Unfortunately, when used to measure the water content of an in-service transformer, the Karl Fischer test provides virtually no indication of the moisture content of the transformer (i.e., including the paper insulation where 99% of the moisture resides) and may even give a false sense of security. In fact, many technicians still believe that if the transformer oil contains less than 20 ppm water, the transformer can be considered good, and dry. This false conception may have led to many transformer failures.

Why is this?

This method fails for several reasons, a primary one being that aging of oil has a major impact on the moisture titration assay. Furthermore, the Karl Fischer Titration (KFT) method suffers from moisture ingress into the sample during transportation to the laboratory, and different processes utilized during transportation and improperly sealed sample containers can also lead to the release of water from the sample leading to unsatisfactory comparability of the results. The most important piece of information needed (and often ignored at the time of sampling) is the temperature of the oil sample when it is taken. Without this temperature being noted, the value of the moisture-in-oil test is meaningless.

Figure 1 Relative Saturation Curve for Transformer Oil at different Sample Temperatures graph

Figure 1 Relative Saturation Curve for Transformer Oil at Different Sample Temperatures

To illustrate this point, the graph shown in Figure 1 reveals the situation for a hypothetical Karl Fischer test result of 50ppm. The curves at different temperatures represent the saturation curves up to 100%. Assuming the oil temperature of the sample was 300C then the RS is 60%. This result indicates the transformer unit needs to be dried. If the sample temperature was instead 630C, then the RS would be 20%, which is considered acceptable for service. If the sample temperature was 200C. the RS is 100%, which suggests a high moisture content symbolic of “raining” inside the transformer. In this case, the unit should be removed from service, and consideration for a complete and effective dry out of both solid and liquid insulation should be mandated.

Water can exist in transformer oil in three states: dissolved, bound, and free. Refer to Figure 2.

Figure 2 Three levels of water saturation in oil image

Figure 2 Three Levels of Water Saturation in Oil

Dissolved water is water in simple solution; bound water is chemically bonded to oil molecules, particles (dust, fibrous particles, as well as acids), and more likely to be present in aged oil. In other words, old, deteriorated oil has a greater capacity to absorb water than new oil. This is one reason why an absolute measurement of water in oil in ppm, is NOT very useful.

The third form, free water, exists when the oil is completely saturated. It can be in the form of droplets or found in liquid form at the bottom of the tank. This implies that it has precipitated (or ‘rained’) inside the transformer.

The bigger problem is not limited to the water in the oil; it is water held in the solid cellulosic insulation around the transformer windings. All transformer oil contains a certain amount of dissolved and/or bound water with little to no effect on its dielectric properties. However, water in the transformers’ solid paper insulation causes its insulating properties to deteriorate rapidly, not to mention the accelerating impact it has on the aging of the paper wrapped around the conductors of the windings.

Cellulosic insulation can absorb about 400 times as much water by weight as transformer oil can at any given temperature, with the effect that any water entering the transformer (via leaks/seepage), will find a ready recipient in the paper and not in the oil.

The other source of water in the tank is from the aging of the paper itself. Cellulose insulation when heated in the presence of water and oxygen (already dissolved in the oil) will gradually depolymerize, releasing byproducts that dissolve into the oil, primarily water, carbon monoxide and carbon dioxide, and Furanic compounds. The chemical reactions governing aging of the paper accelerate when higher temperatures and more moisture are present, hence the need to monitor and manage moisture in the transformer.

Even more important to note, whereas it is relatively easy to remove water from transformer oil, it is very difficult to remove it from the solid cellulosic insulation.

Relative Saturation of Moisture and the Temperature of the oil is the key.

While the absolute measurement of water in the oil tells virtually nothing about the dryness of the transformer insulation, measurements of the relative saturation (%RS) does provide a very clear indication of the danger of water entry to the solid insulation system.

Moisture measurement in oil as a means of estimating the moisture content in paper using equilibrium curves and moisture monitoring in transformers have been studied extensively. It has been found that the Karl Fischer titration method used for moisture determination in oil may not be all that reliable. In contrast a capacitive moisture in oil sensor used for on-line monitoring. Figure 3 provides a comparison of the different methods used to assess the average moisture content in the solid insulation system. This previously in service transformer was left indoors in a controlled environment, with only the oil pumps running thus circulating the oil. After 18 months we can conclude that there was an equilibrium condition met. This implies that the amount of moisture held in the oil, is equal to the amount of water held in the paper. In other words, the water had STOPPED moving between the oil and the solid insulation.

Dielectric properties as measured by PDC, FDS, and Dirana, had some differences. The equilibrium approach using the absolute water content in oil measured (KFT), and the measurement of the %RS reveals a significant difference. The final test of taking paper samples from the unit, and weighing them before and after reveal the most accurate method for the water content in the paper (wt./wt.)

The %RS method is very close the actual water content in the paper.

Figure 3 Comparison of Techniques used to assess moisture in the solid insulation system image

Figure 3 Comparison of Techniques Used to Assess Moisture in the Solid Insulation System

The relative saturation of moisture in oil can be measured with a sensor permanently installed on the transformer, in an area with good oil circulation. It measures directly and continuously the % RS and temperature of the oil at the sensor location.

The question then becomes at what point should I seriously consider drying my transformer?


  1. When the % RS moisture in oil consistently exceeds 18%. At this point, the dielectric breakdown strength of the oil has decreased by 20 to 30%.
  2. When the moisture content in paper consistently exceeds 2% wt./wt. for units < 69 kV class.
  3. EHV transformers the limit for moisture in paper may be lower at 2% wt./wt.

The moisture in power transformers using dielectric response methods and equilibrium diagrams can provide the user with information on the condition such as.

  • Dielectric diagnostic methods deduce moisture in the solid insulation from dielectric properties like polarization/depolarization currents and dissipation factor as a function of frequency of applied voltage.
  • Improved technology for de-energized testing combines time-domain (PDC) and frequency-domain (FDS) measurements and thus substantially shortens the measurement duration.
  • The analysis algorithm compensates for the influence of conductive aging by-products.
  • The updated software has been successfully utilized (1) for onsite measurements that compare favorably to other measurement and analysis methods.
  • To exclude the interference due to oil aging, the moisture in oil relative to saturation level is more appropriate than moisture in oil in ppm.
  • Updated equilibrium curves using % RS of dissolved moisture in oil have been produced, and when used provides a realistic view of the overall (therefore average) moisture condition of the transformer.
  • One step forward constitutes the use of moisture relative saturation in oil and cellulose. This measurement is easy to perform on a continual basis, with accurate and repeatable measurements. These data points should be then integrated into an on-line monitoring system, where the values can be computed, and alarms set accordingly.

Where does all this water come from?

A primary source for this accumulation of moisture inside the tank is via leaks due to deteriorated gaskets. When oil leaks out from gasketed joints, it is possible for moisture and ambient air to enter. Another major source of moisture is from the paper itself as it naturally ages due to the chemical reactions induced by heat, moisture in the paper and acids found in the oil. The byproducts of this continuous chemical reaction are water, carbon monoxide and carbon dioxide, as well as Furanic compounds. All these chemicals dissolve into the oil. The oil gives up the water, and the paper absorbs it, and feeds the chemical reaction.

Why should I be concerned about this?

Excessive moisture in the oil and paper insulation can lead to the following.

  • Dielectric breakdown of the oil and paper leading to partial discharges and dielectric breakdown.
  • Accelerated aging of the winding insulation, reducing its tensile strength and hence its useful remaining life.
  • The risk of ‘bubble formation’ between turns and sections of the winding insulation under heavy loading leading to sudden failure.

Figure 4 Moisture Distribution in the Solid Insulation System diagram

Figure Moisture Distribution in the Solid Insulation System

Moisture in the solid insulation system is greatly affected by temperature of the oil it is in contact with. Figure 4 represents a cross section view of the transformer from a) the temperature gradient the of oil and the winding structure, bottom to top of the active part and b) the moisture held in the solid insulation is a ‘mirror’ image to the temperature gradient.

It reveals that the moisture in paper related to the winding hotspot temperature is 1.2% wt./wt., yet at the bottom of the winding it is 2.2% wt./wt. The pressboard barriers have a similar profile of 1.7% wt./wt. at the top, and 3.3% wt./wt. at the bottom.

The degree of ‘wetness’ at the bottom of the active part should be the determining decision point as to when to consider drying out the transformer. The higher the moisture in the pressboard at the cooler regions of the transformer set up the possibility of surface partial discharges in these areas of high electrical stress. The same is true for insulation components such as cap and collar rings and hi-lo barrier wraps which are outside the main heat source of the windings and thus are more prone to moisture pickup.

Physics of Drying

Three moisture-moving forces are at play when drying a porous material such as paper.

  1. Moisture gradient (as shown in Figure 4)
  2. Temperature gradient (as shown in Figure 4)
  3. Pressure gradient (under oil, or exposed to a vacuum)

On-Site Drying Methods

The traditional method of drying using hot oil spray and hot oil circulation in conjunction with periodic exposure to vacuum (Hot Oil / Vacuum, HOV) is well known and has been practiced for many years.

The main drawbacks of this method are.

  • Limited Temperature
  • No heating during the vacuum phase with oil circulation method
  • Difficulty to obtain uniform temperature distribution with the spray method.
  • Long drying times
  • Limited drying effectiveness

Low-Frequency Heating (LFH) Drying Method

The technique differs in two important ways. Firstly, vacuum can be applied to the paper insulation at the same time as heating of the windings is taking place. Secondly, the winding insulation is heated directly by using circulating current at low frequency (~ 1 Hz.). This method allows a higher temperature to be achieved in the winding paper than is possible using an oil processor as a heat source.

It is the simultaneous application of heat and vacuum that results in the exceptional depth of dryness and speed of dry-out with the LFH process.

Using a 1 Hz. circulating current, the impedance voltages present inside the tank are much reduced such that the applied voltage is very low. The LFH circulating current is applied when the oil is removed from the unit, but the applied voltage is thus low enough to eliminate any risk of a flashover. Refer to Figure 6.

The leakage flux is negligible so the temperature across the windings is uniform. Under normal AC operation, leakage flux causes uneven winding heating. Thus, the low frequency current allows a higher temperature to be safely achieved across the entire winding during the dry out.

Due to the much higher temperatures achieved compared with traditional hot oil treatments (1100C versus 800C) greater amounts of moisture are removed in a much-reduced amount of time.

Overall Results LFH versus HOV

  • Moisture Results – LFH achieved below 1% which is not possible with HOV.
  • Overall time reduction (8 weeks vs 12 weeks)
  • Reduced labour & equipment
  • Reduced outage requirements

Figure 6 Set Up of LF System diagram

Figure 6 Set Up of LF System

Comparison of Different Drying Technologies Versus LFH

The time needed for the LFH drying is much shorter, that it could fit in with most service shutdowns of any in-situ installation. Refer to Table 1.

An additional benefit since the LFH vacuum requirements are not necessarily as low as those required for the HOV method, the stress applied to the transformer tank is reduced. Even if a tank can sustain a deep vacuum (which can be questionable for many older units), significant effort must be exerted to locate all leaks.

Table 1 Comparison of Drying Times and Effectiveness


A high moisture content in the transformer accelerates the ageing of the solid insulation which in turn leads to a reduced lifetime of the active part of a transformer. High moisture content can also lead to the immediate failure of a transformer by causing the formation of water vapour bubbles in critical regions of the windings or when free water appears. Therefore, keeping the solid insulation system of the transformer dry is a major factor when wanting to maintain a high level of reliability and possible life extension, avoiding situations such as that shown in Figure 7.

figure 7 400 MVA, 1 phase, GSU at a nuclear power plant that experienced dielectric failure due to moisture in the insulating paper image

Figure 7 400 MVA, 1 phase, GSU at a nuclear power plant that experienced dielectric failure due to Moisture in the insulating paper.


1) M Kock, M Kruger, S Tenbohlen. “Comparing Various Moisture Determination Methods for Power Transformers”, Cigre 6th Southern Africa Regional Conference 2, August 2009.

2) Cigre Technical Brochure No. 741, “Moisture Measurement and Assessment in Transformer Insulation”, August 2018.

3) Cigre Technical Brochure No. 445, “Guide for Transformer Maintenance”, February 2011.

4) B Sparling, J Cross. “The Threat of Transformer Failure & Premature Aging Caused by Moisture in The Paper Insulation”, Cigre Canada 2023 Conference, Vancouver BC. September 2023.

5) E. Figueroa, T. Kalicki, E. teNyenhius, “Low Frequency Heating Dry-Out of a 750MVA 500kV Autotransformer”, Electricity Today Magazine January/February 2009.


Join our email list

We use cookies to give you the best online experience. By using this website you agree with our cookie policy.