Advanced Transformer Diagnostics

Jill C. Duplessis, Omicron

Enormous economic pressures exist in the electric power industry to maximize asset life. Utilities can’t count on an asset achieving or exceeding its average life expectancy. Because an asset issue, like a transformer failure, can negatively impact a utility, diagnostic activities are more important than ever. Utilities recognize that testing can prevent issues, avert imminent failures and even reduce maintenance costs.

A diverter switch contact shows its advanced age.

Clearly, testing can save money. It is likely that unrealized value can be gained from testing dollars that are already being spent by asset owners. Modern power amplifiers enable wide frequency range measurement, which provide a way to execute today’s widely-embraced diagnostic field tests and perform advanced diagnostic tests at the same time. From an engineering perspective, these advanced diagnostic tests provide new and critical information about an asset’s health. From an economic point of view, the asset owner’s testing dollars are working harder. Knowing and incorporating innovative test techniques is a step toward improving the performance and reliability of power system transformers and achieving greater economic efficiencies. Following are three advanced diagnostics tests for transformers:

Advanced Diagnostic Test No. 1: Power Factor vs. Frequency

For decades utilities have used a single power factor measurement at line frequency to determine an insulation system’s integrity. The resultant calculated power factor obtained from values measured during this test is an index that represents the test specimen’s average condition. An elevated power factor means the insulation system is contaminated or has deteriorated. By comparing a specimen’s power factor with its previous power factor result, or with an expected value obtained from similarly tested apparatus, it is easy to determine when a system no longer tests well and its losses have increased. It is impossible, however, to differentiate and characterize these losses, which may indicate moisture, aging, contamination, oil conductivity or a combination of these conditions. These increased losses may be caused by external environmental conditions. Methods to confirm whether these effects are at play exist.

In contrast, techniques that provide power factor measurements across a frequency band help identify the loss agent(s). In addition, and quite overtly in some cases, these advanced techniques alert the tester to an insulation system in distress even though a single power factor measurement at line frequency does not indicate such.

Plotting power factor, or dissipation factor, as a function of frequency is essentially the same thing as performing a frequency domain spectroscopy (FDS) test, which belongs in the family of methods more commonly known as dielectric frequency response (DFR) tests.

The FDS test measures and models the properties of insulation systems across a wide frequency range (1,000 Hz to 0.1 MHz), which discriminates between polarization losses, conductive losses and aging by-products within the overall insulation system. Analysis algorithms are then applied to determine levels of moisture, conductivity and insulation geometry.

In an insulation system consisting of mineral oil and cellulose, which is common in most large power transformers, polarization and conductivity phenomena occur. These two phenomena occur simultaneously and superposition must be applied to discriminate between their effects. Moisture, temperature and aging by-products influence both polarization and conductivity domains. It is impossible to separate polarization losses and conductive losses at an arbitrary frequency (60 Hz).

In the frequency range neighboring 60 Hz, namely from direct current at 0 Hz to 10 kHz, two types of polarization losses exist; interfacial polarization at 0.0003 Hz and molecular polarization at10 kHz.

When dissimilar materials, such as cellulose and oil, are combined, an interfacial polarization process materializes. Interfacial polarization is typical for non-homogeneous dielectrics with different permittivity or conductivity. Here space charge carriers such as ions accumulate at the interfaces, forming clouds with a dipole-like behavior. The interfacial polarization is the resonance that occurs between the propagation speed and distance traveled of the space charge carriers as a function of the insulation geometry (ratio between oil, barriers and spacers). Interfacial polarization between cellulose and oil occurs at lower frequencies; 1 MHz for dry and cool insulation systems and 10 Hz for wet and hot insulation systems.

In cellulose and oil insulation systems, the individual molecular structures produce polarization losses. These molecular losses can peak around 10 kHz. At or near 60 Hz, these losses cause the power factor values to slightly increase and decrease proportionally with frequency for healthy insulation systems. Figure 1a illustrates this behavior in the 10 Hz to 1 kHz range.

Both cellulose and oil exhibit conductive losses, however, oil is unique in that, by itself, it produces only conductive losses. Figure 1b illustrates contact conductive losses in oil. While conductive losses are seen in both cellulose and oil, losses associated with the oil dominate. At very low frequencies occurring below the interfacial polarization range, however, oil’s conductive properties are minimal compared to cellulose’s.

Figure 2 displays the dielectric behavior of paper, pressboard and oil having 1.0 percent moisture content at 20 C. The frequency range of 10 Hz to 1 kHz is dominated by the cellulose insulation, however, measurement cables and connection technique also influence this region. Oil conductivity causes the steep slope at 0.01 Hz to 1 Hz. Dissolved conductive aging by-products increase the oil’s conductivity and influence this area. The interfacial polarization (insulation geometry, ratio of oil to pressboard) determines the local maximum or “hump” in the 0.003 Hz range. The higher the ratio of oil to pressboard, the more dominating this effect. Finally, the moisture effects within the cellulose appear again at frequencies below 0.5 MHz

Advanced Diagnostic Test No. 2: Frequency Response of Stray Losses

Variable frequency measurements during a leakage reactance test give rise to an advanced diagnostic method known as frequency response of stray losses (FRSL). The principle focus of FRSL is the resistive part of the short circuit impedance, Rsc, measured from 15 to 400 Hz. The FRSL is an indicator for short-circuited parallel strands of transposed conductors, a failure mode that often generates combustible gases but until now has been undetectable by other diagnostic methods, including exciting current tests. Although short-circuited parallel strands will cause gassing, the resultant gas signature is not unique to this failure mode so a method to identify the root cause of such a condition is necessary. Often, knowledge of the combustible gas pattern in a gassing transformer simply narrows the list of possible things gone wrong; additional testing is required to determine the underlying problem. Numerous recounts exist of asset managers investigating the reason their transformers are gassing even though all standard electrical field diagnostics come back with acceptable results. This illustrates that these tools do not cover all trouble and failure possibilities. FRSL is another tool that can be used to identify another failure mode.

FRSL analysis criteria is weighted on the expectation that the response of the stray loss measured for each phase will be nearly identical, as illustrated in Figure 3. An increase in frequency will result in an increase in impedance as the skin effect becomes more pronounced.

Figure 4 shows how short-circuited parallel strands will manifest in the measurement. In this case, C-phase is affected, whereby the higher losses are seen prominently at high frequencies but none are seen at line frequency.

Advanced Diagnostic Test No. 3: Dynamic Winding Resistance

A static winding resistance test is a standard field diagnostic measurement used to verify the integrity of connections, prove electrical continuity of windings, and locate high contact resistance in tap changers. The idea behind a dynamic resistance measurement is to study the dynamic behavior of the diverter switch as the tap changer is moved from one position to the next. The diverter switch transfers the load current from the tap in operation to the preselected tap with the expectation that the load current will not be interrupted or change appreciably.

Comparison to benchmark or with the other two phase results or both facilitates an effective analysis of dynamic winding resistance test measurements. A glitch detector measures the peak of the ripple (Imax – Imin) and the slope (di/dt) of the measuring current, as both are important for correct switching. If the switching process is interrupted, even for less than 500 microseconds, the current’s ripple and the slope change significantly.

For tap changers in good condition, the ripple and slope measurements for all three phases should be comparable while the tap changer is moved through all positions in a raise direction, and similarly should compare favorably while it is moved through all the positions in the opposite direction. Figure 5 shows a ripple measurement for a diverter switch in good condition.

Figure 6 shows the ripple measurements for the three phases of an aged diverter switch. The differences in ripple values were due to the diverter switch contacts’ advanced age (see photographic lead art for this feature, page 34).

The advantage of these advanced diagnostic tools is that they can be conducted at the same time as their respective “brethren test” (e.g., FRSL with leakage reactance tests, dynamic winding resistance tests with static winding resistance tests), which are already being performed. This results in cost benefits, as well as an engineering advantage. For example, the power factor measurement is suitable as a routine field diagnostic test. Concurrent with this measurement, additional power factor measurements should be made across a frequency band. Variable frequency power amplifiers allow these extra measurements to be performed with the same test equipment, requiring no additional test preparation or connections and with minimal added time.

In today’s world, achieving the average life expectancy from major power assets is becoming less than a realistic expectation and more a reflection on how well the asset was managed. Therefore, it’s important for engineers and technicians to be familiar with all available diagnostic tools and implement them appropriately to maximize economic and engineering gain.

Jill Duplessis is primary manager— transformers, North America with OMICRON Electronics. She specializes in diagnostic testing of primary apparatus (assets involved in the flow of energy) with a specific focus on power transformers.

More POWERGRID International Issue Articles
POWERGRID International Articles Archives
View Power Generation Articles on


Previous articleNew Products
Next articleThe Power Play of IP in the Smart Grid Arena

No posts to display