A Key Tool for Transformer Asset Health Management
By Paul J. Griffin, Doble Engineering Co., and Stephan F. Brauer, Morgan Schaffer Ltd.
Dissolved gas analysis (DGA) is one of the most powerful tools asset managers have to determine the health of their transformers. It is a cost-effective approach that can be used to detect problems in the early stages and manage them as the condition evolves. The amount and relative composition of gases detected can be used to categorize developing problems. Categories of problems include: overheating of paper (cellulosic materials), overheating of the dielectric liquid (mineral oil or other), partial discharge activity or arcing. With this information, the data can be analyzed to determine the severity of a fault condition by examining changes in the relative composition and rate of generation of dissolved gases. A test program with more frequent DGA samples, on-line DGA monitors and other tests can be used to further elucidate the nature of the condition and the necessary urgency of response. The results can be used to manage risks and decide if an asset should remain online. In some cases, consideration might be given to controlling operating conditions, such as load and operating temperature to limit the risk of an in-service failure, which could have wider impacts with greater costs. The results might serve to avoid a costly unplanned outage.
Managing Transformer Problems
As an asset manager, it is useful to sort transformers into two main categories. For most, available data should indicate that the transformer is operating normally and aging at an expected rate without developing faults. Transformers in this category can be tested at established routine test intervals based on voltage class, power class and importance to the power system. The other category would be transformers exhibiting problems such as accelerated aging or development of fault conditions. For these units, more frequent DGA samples, or on-line monitoring can be applied. Additional test methods might also be used to refine the condition assessment. For example, if the DGA indicates partial discharge (PD) activity, then electrical PD tests can be performed. If PD is confirmed, acoustic sensors can be applied to the transformer walls to locate the fault(s).
figure 1: Fault Detection Using a Hydrogen and Moisture DGA Monitor
Depending on the nature of the gassing, other on-line tests such as infrared thermography can provide useful complementary information, as can off-line tests such as power factor and capacitance, leakage reactance, winding resistance, exciting current, sweep frequency response analysis and winding turns ratio. These tests can be used by the asset manager to determine the probable cause of the gassing and the appropriate urgency of response.
When to Apply Dissolved Gas Analysis
One challenge in applying DGA methods is that gases can sometimes be generated when no transformer fault is present. The generation of abnormal amounts of gases under normal operating temperatures is known as “stray gassing,” which can vary with different oils and other dielectric liquids. There are test methods including ASTM D 7150 to help distinguish stray gassing from fault gassing.
For expensive transformers, those in critical applications, or units with possible problems, on-line DGA monitoring is recommended. These monitors provide a wealth of information, updated every few hours or less, so changes in the gassing pattern or rates can be tracked and any sudden changes automatically reported. DGA monitors allow the informed management of critical transformers while they remain in operation. They have proven useful in preventing unplanned outages and avoiding costly transformer damage as fault conditions begin to deteriorate.
As an example, Figure 1 (previous page) shows how an on-line DGA monitor can reveal intermittent fault activity that would be difficult to detect using annual laboratory DGA testing alone. This transformer is located at a hydroelectric generation station in a tropical location. Following many years of service, the transformer insulation had absorbed a high level of moisture, resulting in water-in-oil content averaging from 45 to 55 parts per million (ppm), which is shown in blue in the lower panel. Short periods of high load resulted in increased heating, which drove moisture out of the cellulose insulation, rapidly increasing water content in the oil. Following these events, most of the additional moisture was slowly re-absorbed by the cellulose over several days. During these short periods of high-water content in oil, large increases in dissolved hydrogen (H2) were also recorded (top panel in green), with a sudden onset and a slow decline. These H2 spikes might have indicated short-lived thermal fault behavior, or partial discharge activity. To investigate, a more complete DGA analysis was performed with a portable DGA analyzer. Methane (CH4) was present at 46 ppm and ethane (C2H6) was present at 59 ppm, but no ethylene (C2H4) or acetylene (C2H2) were detected. As the high concentration of hydrogen was relatively short lived, another possible source of hydrogen formation seemed likely, specifically the electrolysis of free water condensing in the valve. If the hydrogen was being formed in the main tank and mixing in the bulk oil, the concentration would not decay so rapidly. If moisture was condensing in the valve area, electrolysis of free water could result in hydrogen formation, which would cease when the free water was dissolved. A multi-gas DGA monitor could have been installed to study how other gases evolved during these transients. If the issue was thermal, then methane and ethane would be generated at the same time as the hydrogen. On the other hand, if the issue was electrolysis of free water, only hydrogen and oxygen would be formed. The on-line monitoring indicated this transformer’s reliability would be improved by drying the insulation system.
figure 2: Multi-gas DGA Monitoring of a Defective New Transformer
As another example, Figure 2 shows data recorded by a multi-gas DGA monitor during the first days of service of a new 900 megavolt ampere (MVA) transformer in the southern U.S. After installation on site, the unit was filled with new, degassed mineral oil. Within a few hours of being energized, the transformer began to exhibit significant fault gassing. The monitor readings rose steadily over 24 hours, including an increase in acetylene (C2H2) from 0 to 6 ppm, and the transformer was de-energized later that day after which the gas levels quickly stabilized. Laboratory DGA samples confirmed the concentrations reported by the monitor. The gassing behavior indicated a “T3″ thermal fault with temperature exceeding 700 C, which could have quickly escalated to a catastrophic failure. The comparatively-constant CO readings suggest the cellulose insulation was not directly involved. The transformer was returned to the manufacturer for root cause analysis and repair under warranty.
Dissolved Gas Analysis:
A powerful asset management tool
Because DGA methods are sensitive to such a wide variety of problems that may arise inside power transformers, they now play a central role in many asset management programs. The primary categories of faults that may be identified using DGA methods include thermal faults affecting oil, overheating of paper (cellulosic materials), partial discharges and arcing. Recent DGA diagnostic tools are also able to identify various subtypes of these faults including carbonization of paper and nonhazardous stray gassing that may occur when new oils are put into service (per M. Duval, DGA Challenges at ASTM, IEEE, CIGRàƒâ€°, presented at IEEE Power and Energy Society meeting in Dallas on May 2, 2016). These methods are also increasingly capable of diagnosing faults in transformers filled with insulating fluids other than mineral oil.
Software tools are available to conveniently manage DGA data, and to apply sophisticated DGA diagnostic algorithms to interpret and present the results in intuitive ways. When these software tools are used in conjunction with DGA monitors and other on-line sensors, asset managers can oversee asset health in real-time and leverage health-index methods to prioritize which transformers most merit further investigation. Such condition based maintenance programs offer the possibility to make more efficient use of maintenance teams and operating budgets, while at the same time treating transformer faults as early as possible, before costly transformer damage and outages occur. | PGI
Paul J. Griffin is the vice president of professional services at Doble Engineering. He leads Doble’s consulting, testing and laboratory services groups. Griffin has been with Doble since 1979 and has published over 50 technical papers pertaining to testing of electrical insulating materials and electric apparatus diagnostics. He is a Fellow of ASTM and a member of Committee D-27 on Electrical Insulating Liquids and Gases. Mr. Griffin is a member of the IEEE Insulating Fluids Subcommittee of the Transformer Committee.
Stephan Brauer has managed product development and technology strategy at Morgan Schaffer since 2005. Brauer previously led technology development at small, medium and large organizations in both the private and public sector, specializing in precision measurement instruments. He holds a degree in engineering physics from Queen’s University and a Ph.D. in experimental physics from McGill. Brauer is the author of numerous publications and is a member of IEEE, APS, ASTM and CIGRàƒâ€°.