Electrical Tester online
August 2021
Diagnostic testing of bushings

Diagnostic testing of bushings

13 August 2021

By Sanket Bolar and Ankit Porwal

For almost a century, capacitance and power factor tests have been routinely performed on transformer bushings. In more recent times however, dielectric frequency response (DFR) testing has been recognised as an expanded form of power factor testing that offers significant benefits, as Sanket Bolar and Ankit Porwal from Megger explain.


In the early 1900s it was shown that adopting a capacitance-graded design for transformer bushings provided better radial voltage distribution, thereby allowing smaller bushings to be used for a given working voltage. Today, condenser bushings are almost universal for applications at 25 kV or higher. These bushings can be classified according to the materials used in the insulation system: oil-impregnated paper (OIP), resinimpregnated paper (RIP), resin-bonded paper (RBP) and resin-impregnated synthetics (RIS). OIP bushings are the most widely used.

In substations, bushings provide the means of making connections to transformers. They are designed to withstand high voltage stresses during operation, to carry large currents and to operate at high temperatures. They are typically used outdoors, where they are exposed to potentially harsh atmospheric conditions. Furthermore, before installation, they can be easily damaged during transportation and, after installation, they are a target for vandalism.

For these reasons, bushing problems are among the most common causes of transformer failure, and transformer failure can be very expensive. It is important, therefore, to monitor the condition of bushings throughout their service life to ensure that defects and deterioration can be dealt with before they lead to major failures. Traditionally, condition monitoring has been carried out using capacitance and power factor tests, but now dielectric frequency response (DFR) testing – an advanced form of power factor testing – is growing in popularity either to supplement or in place of traditional tests. To find out why, let us take a closer look at each of the techniques.

Line frequency power factor testing (LFPF) on OIP bushings.

The term ‘power factor’ is typically used in the USA, but in other parts of the world references to tan delta and dissipation factor are more usual. Note that power factor and dissipation factor are calculated in different ways but, for our purposes, they are numerically equal, and the terms can therefore be used interchangeably when values are below 10 %.

The insulation system of a condenser bushing has two capacitive components: C1 and C2 (Fig 1).

In an OIP bushing, capacitance grading is achieved by wrapping Kraft paper multiple times around the conductor core and placing conductive foil inserts at specific intervals during wrapping; afterwards, this main insulation system is impregnated with oil. In bushings rated 69 kV (or 350 kV BIL), the outermost paper layer is connected to a test tap. In service, the test tap is grounded via the tap cover. In bushings rated > 69 kV (i.e. > 350 kV BIL), one of the outermost layers is connected to a potential tap while the outermost layer is grounded internally. The potential tap ‘floats’ in service with the tap cover in place.

The capacitance of the multiple layers between the core and the test tap, or between the core and the potential tap, is represented by C1. The test tap is isolated from the grounded flange when the tap cover is removed, and the capacitance of this insulation is represented by C2. The C2 insulation system of a bushing with a potential tap includes the outermost, oil-impregnated paper layers of the main core as well as insulation between the potential tap and the grounded flange.

Figure 1: Insulation system of a condenser bushing

Capacitance and power factor tests measure, usually at 10 kV, the dielectric losses in the insulation represented by C1 . The values obtained are commonly expressed as percentages, and typical values for new bushings are in the range 0.2 % to 0.4 %. Tests at the factory provide reference values that are included on the bushing nameplate, and field test values are compared with these references. Any significant deviation suggests that the insulation of the bushing may have deteriorated.

Changes in the capacitance value of C1 are also important. An increase in capacitance may be the result of short-circuited layers, while a decrease in capacitance most often results from problems with the tap connection. Capacitance and power factor measurements can also be made on the insulation represented by C2, at 500 V (test tap) or 2 kV (potential tap). Particularly when a gasket fails on a bushing that, in turn, allows for moisture ingress, water typically accumulates in the tap compartment and attacks the main insulation core from the outermost layers first. A C2 test primarily includes this most susceptible insulation and, in such instances, provides notification that moisture ingress, or other contamination, is a problem before a C1 test does.

Several standards provide guidelines on interpretation and validation of power factor test results. Examples for factory acceptance testing are IEEE C57.19.01 – IEEE Standard Performance Characteristics and Dimensions for Outdoor Apparatus Bushings, and IEC 60137 – Insulated bushings for alternating voltages above 1000 V. The limits for C1 power factor prescribed by these standards are shown in Table 1. All values are either measured at 20 ºC or are normalised to 20 ºC.

Standards applicable to field testing include: IEEE C57.152 – IEEE Guide for Diagnostic Field Testing of Fluid-Filled Power Transformers, Regulators, and Reactors.

This standard states that: 

  • A change from the initial reading by 1.5 to 2 times warrants more frequent testing of bushings
  • A change from the initial reading by more than three times warrants removal of the bushing from service
  • A change in capacitance by more than 5 % is a cause to investigate the suitability of bushing for continued service

These guidelines are in line with those provided in IEEE C57.19.100 - IEEE Guide for Application of Power Apparatus Bushings.

A point that is often overlooked when analysing power factor test results is the effect of temperature. Measuring at different temperatures yields different values, so it is important to normalise measured values by correcting them to a reference temperature (20 °C). Tables of correction factors are available from various sources, but these are generic values and cannot always be relied upon. We will look at this again later.

Narrowband Dielectric Frequency Response (NB DFR) measurements 

Some 10 kV power factor test sets can be used to make power factor measurements at multiple frequencies from 1 Hz to 500 Hz. The curve obtained by plotting these measurements is the Narrowband Dielectric Frequency Response (NB DFR) and it provides additional information about the insulation condition of the bushing. This relatively narrow frequency band does not provide quantitative information about the moisture content of the insulation, but it does provide an indication that moisture and/or impurities may be present.

NB DFR testing is an advancement from measuring power factor at line frequency only and, in about three minutes, it provides valuable additional information. It can, for example, deliver an early warning of insulation ageing and degradation, suggesting that maintenance should be prioritised or that the condition should be investigated further using more advanced test techniques.

NB DFR is typically carried out at 250 V and, because it is a low energy test, it is safer to perform than the routine 10 kV LFPF test. It allows visualisation of the unique dielectric signature of the object under test, thereby facilitating graphical comparison of results from multiple bushings on the same transformer. It also provides 
a way of measuring the temperature sensitivity of a particular insulation system so that an accurate individual temperature correction factor (ITC) can be determined for normalising results to 20 ºC.

Power factor is temperature dependent and Note 3(b) of Section 10.10.4 of IEEE C57.12.90 – IEEE Standard Test Code for Liquid-Immersed Distribution, Power, and Regulating Transformers – states “Experience has shown that the variation in dissipation factor with temperature is substantial and erratic, so that no single correction curve will fit all cases”. Fortunately, the frequency response and the thermal response of a dielectric system are related and this relationship reveals a more accurate, alternative means to ‘correct’ or normalise a power factor test result to its 20 °C equivalent.

For an OIP bushing, the dielectric frequency response plot shifts horizontally with a change in temperature, but the shape of the curve remains unchanged. It is possible to determine the exact amount by which the curve is shifted horizontally for a given change in temperature, which means that power factor measurements made at the most commonly encountered bushing temperatures can be accurately and reliably normalised to 20 ºC. Figure 2 shows an example of a normalised DFR curve alongside the measured curve.

Using the ITC method to accurately normalise results eliminates the need to wait until the temperature is close to 20 ºC to test a bushing, or to rely on generic correction factors that may or may not be valid, depending on the insulation condition.

Dielectric Frequency Response (DFR) measurements

The insulation system of an OIP bushing electrically behaves as multiple series capacitors that are formed of conductive foil and oil-impregnated Kraft paper, as previously mentioned. The oil and paper form a composite dielectric system, and DFR results are a combination of the responses from the oil and paper components. The responses from both of these components vary with frequency, but the way they vary is different and this difference is exploited in DFR testing.

When NB DFR testing does not provide a definitive indication of the condition of a bushing, DFR testing, which is carried out over a wider range of frequencies, can be used for a more in-depth analysis. Analysis of DFR curves can yield important information, such as the moisture content of the paper and the conductivity of the oil. In addition, the presence of contamination or other physical issues can result in atypical responses with the most prominent deviations seen at the lower frequencies.

DFR measurements are often carried out at 140 V. Because the test covers a wide range of frequencies and uses low currents, electrical noise can adversely affect the accuracy of the measurements, particularly when testing already relatively low capacitance specimens, like bushings. The solution is to test at a higher voltage. The use of a voltage amplifier in conjunction with a DFR test set greatly improves the signal-to-noise ratio, allowing accurate and reliable measurements to be made even in noisy environments.

In the field

To illustrate and confirm the value of the tests described in this article, the same series of tests was performed 
in the field on three OIP HV bushings. The testing was carried out in three stages: PF analysis, NB DFR analysis and DFR analysis.

  • Stage 1: PF analysis

An increase in power factor, measured at power frequency, was observed on two of the three bushings (X1 and X3), as shown in Table 2.

According to IEEE C57.152, the line frequency PF values for X1 and X3 at 20 °C exceed acceptable limits, while according to CIGRE TB 445, the PF values of bushings X1 and X2 are within limits with only bushing X3 outside acceptable limits. Based on these results alone, we may say that bushings X1 and X3 are not in good condition, and, additionally, that bushing X3 should be further investigated as soon as possible.

  • Stage 2: NB DFR analysis

The NB DFR test was carried out on the three bushings at frequencies from 1 Hz to 500 Hz. Figure 3 shows an increase in % DF for bushings X1 and X3 at lower frequencies. The thermal response curves are shown in Figure 4 and are different for each bushing because of the variations in the dielectric frequency response between the bushings. Table 3 shows the temperature corrected % PF values.



For a bushing in good condition, its C1 PF test result is only slightly temperature dependent. As its insulation ages and deteriorates, the temperature correction factor increases, which means that temperature dependence is directly related to the dielectric response of the bushing. In these results, bushing X3 shows much greater temperature dependency than bushings X1 and X2. This is a good indication that there are problems with bushing X3.

On the basis of studies carried out over more than 20 years as a leading expert in DFR testing, Megger has proposed the limits shown in Table 4 for % PF values measured at 1 Hz.

The high % PF measured on bushings X1 and X3 at 1 Hz indicates insulation-related problems but gives no indication of whether the problem is in the liquid insulation or the solid insulation. To establish this, a full DFR analysis was carried out.

  • Stage 3: DFR analysis 

The results obtained from the DFR analysis are shown in Table 5. Megger’s proposed limits for % mc (moisture concentration) and conductivity of the oil are shown in Table 6. Table 7 summarises the results obtained from the collective testing carried out on the bushings and reveals their final assessments.


Line frequency PF testing, which has traditionally been used for bushing condition assessment, has limitations that can be overcome by using DFR tests. Because of the increased sensitivity of PF to moisture and contamination at lower frequencies, DFR testing allows bushing insulation problems to be detected at an earlier stage.

Any remaining uncertainty about condition after NB DFR testing can be eliminated by carrying out a full DFR test, which allows the amount of moisture in the solid insulation and the conductivity of the oil to be estimated. DFR also eliminates the problem of the temperature dependence of PF measurements. Finally, the difficulties sometimes associated with carrying out DFR testing in electrically noisy environments can now be solved by using test voltages up to 1.4 kV RMS.