With an increasing failure rate of substation electrical equipment, utilities and heavy industry must focus on preventive and predictive maintenance to ensure power system integrity and stability, says Damon Mount of Megger

Electrical insulation is a common source of failures and dissipation factor (DF) testing, which is also known as tan δ testing and, especially in the USA, as power factor testing, is a popular way of estimating the condition of insulation as it ages. There are, however, issues relating to DF testing that are not nearly as widely understood as they should be.

DF testing is widely used on electrical equipment such as power transformers, circuit breakers, generators and cables. DF values, trended over time, can help in detecting problems like contamination, high moisture content and the presence of voids in insulation. Excitation current tests, along with DF tests, performed on power transformers, can also help in detecting turn-to-turn insulation failure.

Dissipation factor vs. voltage
DF tests are usually performed at 10 kV or the readings are converted to 10 kV equivalent. The best voltage for DF tests is frequently debated as instruments are now available that allow tests to be performed at voltages from 27 V to 12 kV. The voltage that is good enough for accurate and reliable measurements depends on the test specimen and the test conditions.
Most power transformers have oil-paper type insulating systems that exhibit a flat response when DF is measured as a function of test voltage. However, motors and generators typically have dry or solid insulation whose DF value typically increases with increasing test voltage.
One reason that industry has standardised on a 10 kV test voltage is for immunity against electrostatic interference. An HV test signal provides better signal to noise ratio, giving more accurate measurements. Test instruments with very high noise suppression capability are required for measurements in HV substations as noise levels can be as high as 20 times the test current.

Negative dissipation factor values
For perfect insulation, the DF should be zero. In practice, any value close to zero is considered to indicate good insulation. DF test sets try to measure a single capacitor, but if the test object has additional phantom circuits, the results look strange. For example, with bushings, three-winding transformers and the inter-phase insulation of rotating machinery, measured DF values are sometimes negative. Since DF is a measure of watts loss, negative DF corresponds to watts generation. As insulation cannot generate power, negative DF values are not real.

Phantom circuits introduce a current Is which changes the phase angle of the measured test current (IT). The surface loss current (Is) is predominantly resistive (Rs) and has a very small phase angle with respect to the applied voltage. Capacitive coupling (Cc) may be present as a result of this parallel path of Rs to main insulation under test.

Smaller phase angles for surface loss current (Is) can lead to negative DF values. Measured test current (IT) is the vector difference of total current (INET) and surface loss current (Is). In UST or GST configurations, this surface loss makes the measured test current (IT) phase angle greater than 90º, leading to negative DF values.

It is important to decide where negative DF values come from. For some specimens it is a result of design - for example, the presence of electrostatic grounded shield between the inter-windings of a transformer. In other cases, users should consider eliminating external effects by following best testing practices such as verifying proper grounding, cleaning bushing surfaces and using guard circuits effectively. Repeated negative values after taking these precautions could mean contamination or a bad insulation system.

Excitation current vs voltage
Excitation current testing is commonly performed along with DF testing. It is a voltage dependent test and is always performed in UST mode. Like DF tests, the excitation current readings are normalised to 10 kV equivalent values, using a linear approximation. With highly inductive specimens like power transformers, the relationship between voltage and current is, however, not linear. Assuming a linear relation to determine 10 kV equivalent excitation gives only very approximate values. Tests should therefore be performed at the same voltage if the data needs to be trended. This is important, as trending data is critical when evaluating turn-to-turn insulation problems.

When performing excitation current measurements on delta windings, the third leg of the delta configuration must be grounded. This eliminates the current flowing in the other two windings from the measurement circuit. If the third leg is not grounded, the results will be approximately 30 to 50% higher than true readings.

A transformer with magnetised core can exhibit higher excitation current measurements than normal. IEEE 62-1995 section 6.1.3.4 states, "If a significant change in the test results is observed, the only reliable method of excluding the effect of residual magnetism is to demagnetise the transformer core."

Temperature correction factors for DF readings
DF values are highly dependent on temperature. IEEE C57.12.90 section 10.10.4 Note 3 (b) states "Experience has shown that the variation in power factor (dissipation factor) with temperature is substantial and erratic so that no single correction curve will fit all cases." Nevertheless, correction factor tables are traditionally used to correct data to 20°C. It is, however, imperative only to compare a specimen's DF values taken at a similar temperature or accurately corrected to the same temperature.

For different specimens, changes in temperature affect DF values differently, and even the same specimen will become more temperature dependent as it ages. Temperature correction factors are influenced by many factors, but temperature correction data is based on average values. Since each test object is unique, using these average corrections introduces errors.
New transformers have relatively weak temperature dependence and standard tables over-compensate. With ageing, however, the same average correction factors under-compensate. Trending of DF values becomes more critical in the second half of the life cycle, where correction factors should be larger because of the increased effect of temperature on the insulation. Using average factors can lead to incorrect trending and inaccurate estimation of the remaining healthy life of the object.

IEEE Std. 62-1995 states, "Testing at temperatures below freezing should be avoided, since this could significantly affect the measurement. Among the primary reasons for performing this test is the capability of detecting moisture in insulation. The electrical characteristics of ice and water are quite different and it is much more difficult to detect the presence of ice than it is to detect water; sometimes it is impossible."

Measuring DF at too high or too low a temperature can introduce errors, and the IEEE recommends performing DF tests at or near 20 °C. However, this is not always practical.
Fortunately, new technology of the type used in Megger's Delta4000 insulation diagnostic systems makes it possible to accurately correct DF values to 20 °C without using correction factor tables based on averages. Using dielectric frequency response (DFR), a unique temperature correction factor for each test object can be determined. This is possible because a DF measurement at a certain temperature and frequency corresponds to a DF measurement made at different temperature and frequency. Therefore by measuring DF at different frequencies, it is possible to determine the supply frequency DF at any temperature [5-50°C]. With this technique, DF can be measured at any insulation temperature [5-50°C] and then accurately corrected to 20°C.

Conclusion
Electrical apparatus has failed and will continue to fail because of insulation deterioration. A proactive approach to monitoring the integrity of the insulation system is the key to preventing or at least anticipating such failures. Dissipation factor testing is an important tool in determining insulation quality and estimating remaining life.

DF readings depend on various factors and it is important to be aware of these. Test voltage, electrostatic interference, temperature, humidity, surface losses and other parameters can greatly influence DF measurements. A better understanding of the impact of these parameters will help in obtaining accurate measurements that can be relied upon in the decision making process.