Due to severe competition and restructuring taking place in the power industry, there is a need to reduce maintenance costs, operate transformers as much as possible, extend their life, and prevent unplanned outages. Hence, in the recent years the monitoring and diagnostics of transformers have attracted considerable attention of researchers and practicing engineers. The monitoring can be either offline or online.
The trend is more towards online techniques due to ongoing developments in computational/analysis tools and information technology. Insulation degradation is the primary reason for aging and eventual failures of transformers. It is usually the limiting factor in the longevity of any electrical equipment including transformers.
Degradation beyond a limit can become the cause of their natural end-of-life. Various impurities and mechanisms that lead to insulation degradation are moisture, conducting particles, hot-spots, oxidation and partial discharges.
The presence of moisture in the insulation system of the transformer reduces dielectric strength, leads to bubble formation at high temperatures and accelerates cellulose decomposition. Moisture in the presence of oxygen inflicts major damage to the insulation.
Moisture and oxygen degrade the paper insulation to form sludge and acids. Sludge gets deposited on windings, impairing cooling efficiency. Acids accelerate the decay rate further. Hence, moisture and oxygen lead to aging of transformers and reduction in their life expectancy.
The third catalyst for insulation deterioration is temperature. Aging accelerates rapidly with temperature in the presence of high moisture content in the paper/pressboard insulation; the general guideline that the insulation life becomes half for every 6-degree temperature rise is valid when moisture in the insulation is within the acceptable limit, i.e., less than 0.5% by weight. The aging rate is much higher if the moisture is at an unacceptable level.
An oxygen inhibitor can be added to the oil for arresting its high decay rate. The inhibitor acts as a sacrificial agent since the oxygen attacks it rather than the cellulose insulation. The inhibitor may have to be periodically replenished to make up for its falling level due to the oxygen.
Transformers are typically dried by manufacturers so that the moisture to dry weight ratio is within 0.5%. On site a ratio more than 2.5% certainly calls for a dry-out operation. The limit on moisture in oil for a dry-out operation is typically 30% oil saturation. The amount of moisture dissolved in oil increases with temperature. When the moisture content increases beyond the saturation value, oil can no longer hold it resulting in free water suspensions.
Paper insulation is much more hygroscopic in nature than the oil. Hence, the moisture distributes unequally between the oil and the cellulose insulation; the latter has a greater amount. In fact, most of the moisture inside the transformer resides in the cellulose insulation (can be as high as 99% or more). Temperature conditions predominantly decide the moisture distribution between them.
Moisture gets into the transformer during manufacturing, through various gaskets, while inspecting it at the site, via degradation of insulation due to its normal aging process, and because of rapid temperature and pressure variations. Moisture has a tendency to drift towards cooler parts (at bottom) and regions with higher electrical stresses.
Partial discharges occurring in an insulating material can be detrimental to its condition. A partial discharge is a localized breakdown that occurs without complete bridging of the insulation. The breakdown is in the form of extremely fast, nanosecond discharges. Partial discharges over an appreciable period of time, can lead to catastrophic failures.
When discharges occur in insulation voids, signals in the electrical, chemical and acoustic domains are generated. Different techniques have been developed to detect such signals for determining the extent and location of discharges. The processes of insulation degradation and the corresponding failure mechanisms in transformers are complex.
The monitoring not only detects the incipient faults but also allows a change from periodic to condition-based maintenance. It is important to identify key parameters that should be monitored to reduce the cost of the overall monitoring system.
The high cost of monitors, reliability issues of electronic equipment, performance under adverse field conditions, and inadequate expertise in interpretation are some of the challenges that need to be overcome. There are a number of offline and online monitoring/diagnostic techniques that are currently being used and developed further.
This article elaborates these techniques; a few of them, which are well established and documented in technical reports/standards, will be discussed briefly.
A concise summary of routine tests is also given here. It should be stated here that the intention of the discussions in this article is not to give limits and values (which are given in standards and technical reports), but to highlight the importance of the discussed parameters and tests, and to give a background theory of some of the diagnostic tests.
Routine Tests on Transformer Oil
Since oil comes in contact with the active parts and the solid insulation when it flows due to natural or forced convection, various tests on its samples usually reveal the status of the health of the transformer. Incipient faults can also be detected.
The tests/measurements can be done offline or by taking samples in an energized condition. A few online sensors have also been developed in recent times for monitoring oil.
Moisture: Equilibrium charts are commonly used to indirectly determine the moisture content in cellulose insulation from moisture in oil; the method can work well if they are in thermal equilibrium, which is a rare possibility during operating conditions.
There is migration of moisture between them as the temperature changes due to variations in load. Mineral oil has a low affinity to moisture, but the solubility increases significantly with temperature. Hence, as the temperature rises, the moisture in the cellulose insulation migrates to the oil.
On the other hand, when there is a drop in the temperature (say, due to a reduction in load) the moisture tends to return to the cellulose insulation, but at a slower rate. Therefore, the oil appears to have greater moisture at lower temperatures resulting in overestimation of the moisture content in the cellulose insulation when the equilibrium charts are used.
Karl Fischer titration is another conventional technique for the measurement of moisture in cellulose. It requires careful extraction of paper/pressboard samples and their storage (until the measurement is done in a laboratory). This is an intrusive technique like the DP (degree of polymerization) test.
The main limitation of these intrusive techniques is that the diagnosis has to be done by using paper samples collected from easily accessible parts (e.g., high voltage leads) as critical insulation areas between windings are not reachable for the purpose.
More sophisticated dielectric diagnostic methods have been in use recently, which determine the moisture in solid insulation from dielectric characteristics such as charging/discharging currents, recovery voltages, and the dissipation factor.
These methods, which can be used for doing on-site diagnostics, are generally more accurate than the conventional methods. Also, the data measured using them contains much more information which can give better insight into various dielectric processes.
Cellulose degradation is an irreversible process which is accelerated in the presence of moisture. Extracting moisture from the cellulose insulation by hot oil circulation may not be an efficient process. Drying through heating/vacuum cycles or vapor-phase process after a hot oil spray can make the moisture extraction faster since the sprayed oil can be at a higher temperature (up to 110°C) than the circulated oil (at 60-75°C). Exposed surfaces have higher moisture content and they are effectively dried by the spray.
However, unexposed insulation areas are not effectively dried by this method; a LFH (low frequency heating) technique can be used concurrently to uniformly heat the windings and the cellulose insulation in close vicinity of their conductors.
Site dry-out needs to be simple and has to done using one or more of the above methods depending on available resources. Molecular sieves have been used for online drying in the field; they consist of columns with molecular sieve material which binds moisture from the oil.
Vapor phase drying, is the most effective drying method. However, high costs and sophisticated processes associated with it preclude its use on site.
Dielectric strength: Determination of the breakdown voltage (BDV) of oil samples according to the test procedures outlined in standards is an important quality check before conducting high voltage tests on a transformer in the factory or before commissioning it on site.
If the strength is below a certain value, the oil needs to be processed again to improve its characteristics. The parameter is measured by applying a voltage between two electrodes having shapes as specified in the standards. The distance between them is also standardized. The method is sensitive to the presence of moisture and contaminants in the oil.
Interfacial tension: It is defined as the force in dynes required to pull a small wire ring vertically up through an oil-water interface by a distance of 1 cm. Pure oil results in a clear line between the two media requiring a higher force (40–50 dynes/cm).
Contaminated oil weakens the interfacial tension and lowers the corresponding force value, which may call for additional monitoring and diagnostic measures.
Acid number: It is the amount of potassium hydroxide in mg required to neutralize the acid present in 1 gm of oil. A higher acid number than the limit specified in standards is indicative of an unacceptable level of acid content in the oil. This number along with the interfacial tension gives a good signal about the oil condition. The limit on the acid number is 0.4 mg.
Tan delta: Its higher value is indicative of a significant loss (resistive) component of the leakage current, which may be due to various kinds of impurities that may be present in the oil (e.g., moisture, conducting particles). It should be noted the test result can be in the form of a power factor or dissipation factor value. These indices are adversely affected by aging products and soluble polar contaminants (water molecules) in the oil; a significant increase in tan δ is usually observed.
Tests on Transformer Windings
Insulation resistance test: The insulation resistance is measured by applying a DC voltage to the winding under test with its neutral isolated from the ground (if applicable) and all other windings short-circuited and grounded. The core and the tank are also grounded. The transformer must be de-energized for conducting the test.
The measured resistance then tests the condition of the insulation between the winding and the core/tank/other windings; for a good insulation condition its value is in mega-ohms. The duration of the test is 1 minute and the measured resistance value at the end of test duration is used for future comparisons.
The readings are sensitive to temperatures and the magnitude of the applied DC voltage; these parameters must be recorded and taken into account during the comparisons.
The test is performed as a part of commissioning procedures to check for any ingress of moisture during transport or storage. Since the test is done with a DC voltage, all the terminals should be grounded for a sufficient amount of time after the test to drain the charges completely to ground.
Polarization index: The ratio of the measured insulation resistance after 10 minutes to that measured after 1 minute is known as polarization index. Its value is lower (around 1.0) for moist/contaminated insulation compared to that for dry/pure insulation (around 2.0).
Power Factor Test
This test, which helps to determine the quality of the insulation between the windings and between the windings and the core/tank/ground, is conducted using a Schering bridge or transformer-ratio-arm bridge. The corresponding capacitance and tan δ values can also be determined. The measured power factor value is indicative of losses in the total insulation system including the bushings.
With aging the power factor deteriorates; the measured power factor should be compared to previous test readings to ascertain its trend. When the power factor or tan δ value is higher than the limits specified in the standards, other conventional tests, such as moisture-in-oil analysis, can be conducted to confirm the insulation status. Additionally, advanced dielectric response tests can also be performed.
Any change in the capacitance between the LV and HV windings, between the LV winding and ground, and between the HV winding and ground is indicative of aging, winding deformations, dislocation of winding support structures, or a combination of these. It should be remembered that the capacitances of bushings are also part of the measured values.
The method may not be sensitive enough to detect mechanical irregularities in the windings; advanced methods such as frequency response analysis should be used.
Dissolved Gas Analysis
Dissolved gas analysis is an established and proven method to detect incipient faults. Degradation of oil and cellulose insulation leads to evolution of the following key gases: hydrogen (H2), oxygen (O2), carbon monoxide (CO), carbon dioxide (CO2), methane (CH4), acetylene (C2H2), ethylene (C2H4) and ethane (C2H6).
Since the mineral oil used in transformers has chains of hydrocarbon molecules, its decomposition leads to formation of H2, CH4, C2H2, C2H4 and C2H6 gases. On the other hand, cellulose (paper insulation) degradation is accompanied by release of O2, CO and CO2 gases. Since the paper is oil-impregnated, gases such as H2 and CH4 are also produced due to its thermal decomposition as it ages normally.
Because of the production of O2 and H2 the degradation process of the paper insulation leads to more moisture generation which accelerates the decay further.
The dissolved gas analysis (DGA) is done using a gas chromatograph which houses different adsorption columns. The gases extracted using a suitable technique (e.g., vacuum extraction) are fed into the columns where they are adsorbed and separated. Established dissolved gas interpretation methods/ guidelines, as described below in brief, are commonly used.
Key gas method: Various types of faults produce different gases. Therefore, the analysis can be done by identifying the most prominent gas that is generated. Arcing, overheated cellulose and partial discharges produce predominantly acetylene, carbon oxides and hydrogen, respectively. The generation of ethane and ethylene is generally associated with overheating of metal parts (e.g., contacts of tap changer assembly, terminations of conductors/leads, tank plates).
IEEE method: A guide for interpretation of gases generated in oil-immersed transformers (IEEE Standard C57.104-1991) classifies conditions into four types depending upon the concentrations of individual key gases and the amount of the total dissolved combustible gases (TDCG). Severity increases from condition 1 to condition 4.
The rate of increase of different gases in a given time interval is a better indicator than their absolute levels. Acetylene gas is the exception; a small amount of acetylene gas above a certain ppm value cannot be disregarded, although the rate may be small, as it may be indicative of highenergy arcing.
Rogers ratio method: It uses three ratios of certain key gases, viz. CH4/H2, C2H2/C2H4 and C2H4/C2H6 , to classify an already diagnosed gas generation problem into a specific fault type. The method uses the fact that the gases begin to form in small amounts when the temperature reaches a certain level.
For example, hydrogen begins to form around 150°C and increases steadily with the temperature. The concentration of acetylene, which shows up beyond 500°C, increases significantly above 700°C. Thus, in a specific temperature range, the amount of one gas is more than that of another gas. For a certain temperature prevailing inside the transformer the ratios will have typical values enabling an estimation of the temperature and hence the fault-type.
IEC 60599 (Mineral oil-impregnated electrical equipment in service – Guide to the interpretation of dissolved and free gases analysis, edition 2.1 2007-05) gives a method which is similar to the Rogers ratio approach. It uses the three ratios as above and considers six types of faults: partial discharge, discharges of low energy, discharges of high energy, and thermal faults of three intensities (temperature less than 300°C, between 300°Cand 700°C, and greater than 700°C).
Duval triangle method: This is also a popular diagnostic approach. It uses concentrations of three key gases, CH4, C2H2 and C2H4, to suggest a probable type of fault. We can either use the total accumulated content of the key gases or the increase (over the previous DGA record) in each of the key gases to arrive at a point in the Duval triangle. The location of the point in the triangle indicates the type of the fault.
DGA can be unreliable if the transformer under investigation has cooled after de-energization. Also, the analysis on a new transformer or a transformer which has been recently processed for oil-filtration can lead to erroneous diagnostic conclusions.
Whenever a problem is suspected from the analysis, it is recommended to analyze another sample and compare the two results. It may not be prudent to base inferences on just one analysis; there can be errors in sampling and handling procedures.
DGA can be done online as well to detect a few key gases such as hydrogen, oxygen, methane, carbon monoxide, carbon dioxide, ethane, ethylene and acetylene along with the moisture in the transformer oil. Ambient and top oil temperatures can also be measured. Oil samples are collected and processed every one or two hours.
Advanced and compact sensors are being developed to detect the gases. There is enough experience already available with online hydrogen sensors. Portable units are available that can detect hydrogen in oil, which is indicative of mostly the occurrence of partial discharges. Investigation of dynamic behavior of the gases with changing operating conditions is possible.
Online monitoring of oxygen can give a warning of air leaks. Online instruments with lower costs, which measure only three key gases, hydrogen, carbon dioxide and acetylene, are also available.