Presented By:
Chris Rutledge
GE Memphis Light Gas and Water
TechCon 2019
Abstract
Due to the many advancements made in online monitoring of substation equipment over the last decade, it is possible to have a nearly instantaneous overview of equipment condition available at all times. However, this level of monitoring is often not cost effective when the overall expense is compared to the cost and criticality of the equipment being measured. Any successful assessment of equipment condition is also reliant on a thorough understanding of the analytics applied and the intervals at which they should be measured. This paper outlines a method with which to greatly increase system reliability through the use of proper analytics in conjunction with a minimalistic and cost effective deployment of online monitoring.
Extant Monitoring Systems
Data In order to best improve system reliability through the use of monitoring, a basic understanding of the various types of monitors and the conditions for which they are designed to prevent should be examined. This understanding allows the analyst to determine the usefulness of a given monitoring scheme as it pertains to detecting critical failure of system components. Another factor that must be considered when attempting to determine overall health of equipment is the use of off line sampling and electrical testing. This is usually performed on a cyclic basis, and the trending of this data is essential for accurate assessment of equipment condition. Dissolved gas analysis (DGA) monitors are perhaps the most prevalent type of monitoring deployed in modern substations and provide valuable insight as to the internal condition of the substations transformers.
DGA monitors are available in multiple configurations ranging from nine gases to single gas detection. These monitors also employ various technologies such as gas chromatography, photoacoustic spectroscopy, non-dispersive infrared (NDIR) and many other platforms. Nine gas monitors provide the most comprehensive data, recording ppm values for both the hydrocarbon and carbon oxide gases dissolved in the equipment’s insulating medium. These monitors usually provide a ppm value of dissolved moisture also. While the data provided by this system is sufficient to project an accurate analysis of transformer condition, monitoring of this level is usually cost prohibitive, and the constant stream of data can become cumbersome and unwarranted for tracking the normal lifecycle trends required for predictive health indexes of distribution level transformers. Use of a nine gas monitor would be best justified in applications involving high system criticality and extreme loading variation such as GSU or EHV transformers, thereby warranting the need for an increase in the observation rate of gassing patterns in order to determine the real time effects of loading values on the transformers life cycle. Figure1 Illustrates the adverse effect short periods of overloading have on gas generation and cellulose degradation via the carbon oxide ratio.
The sharp decline in the CO2/CO ratio is indicative of the effect pyrolysis is having on the cellulose insulation. Since the correlation of the CO2/CO ratio to the amount of cellulose degradation is a relatively new concept to the utility industry a brief review of this process is provided.
Trending Carbon Oxide Ratios
As scission of the of the C5H10O5 molecules which make up the cellulose insulation system occurs, CO and CO2 molecules are produced. Previously, the evaluation of these gases was based on a total ppm concentration limit. Though this method provided some indication as to cellulose degradation when observing high levels of CO saturated in oil, it could often lead to an erroneous assessment due to the low solubility rate of CO and the presence of leaks on the transformer tank affecting the levels of the carbon oxides dissolved in oil. Much like hydrogen, these factors tend to cause ppm levels of the carbon oxide gases to be erratic. But when trending the ratio between CO and CO2, these same factors tend to assist in an accurate assessment of cellulose degradation, as CO is both less soluble and produces at greater temperatures than CO2. Therefore, any rise in CO generation when weighed against the production of CO2 is a highly useful predictor of cellulose degradation and also assists in determining to what degree a thermal event occurring within the transformer tank is effecting the bulk insulation system.
Monitoring Distribution Transformers
Distribution level substation transformers (120 to 25 MVA) tend to operate at a much lower and consistent level of loading than EHV and GSU type transformers. Because of this, changes in gassing rates and cellulose degradation tend to occur over a longer duration than that of transformers in a more erratic loading environment. When examining distribution transformers, trending of manually drawn annual samples is typically adequate for providing an accurate predictor of overall transformer health, as deterioration of the transformer’s condition tends to occur over a period of several years, reducing the need to have the nearly constant access to the level of data generated by a multi gas monitoring system. While proper trending of manual samples would be sufficient to ensure reliability in most instances, occasionally an incipient condition occurs between the manual sampling cycles which could lead to the onset of rapidequipment failure. This development is often the result of a through fault incident, manufacturing defect or external factors. While these incidences are not commonplace, it is recommended that measures be put in place to warn the operator that an event has occurred which could adversely affect the life expectancy of the equipment. This can be achieved through monitoring of thermal and liquid levels of the transformer in conjunction with a reliable hydrogen or composite gas monitoring system. Hydrogen/composite gas monitoring is based on hydrogen generation occurring through every temperature greater than 1500C. Hydrogen is unique in this property as the remaining hydrocarbon gases used for diagnostic purposes scission from methane, ethane, ethylene and acetylene as temperatures are increased. Therefore any pyrolysis occurring within the transformer should result in the production of hydrogen, making it the most reliable predictor of incipient faults. There are also several factors that should be considered when relying on hydrogen monitors. These monitors should only be utilized as an early warning device as the data they provide is insufficient to perform a detailed diagnosis of what is actually occurring within the main tank of the equipment. Any alarm indication should be immediately investigated via a full DGA sample. Special attention should be given to any elevation of combustible gas and trending of the carbon oxide ratio. Another consideration when implementing hydrogen monitoring is the relationship between ppm values of hydrogen detected when compared to hydrogen and hydrocarbon gas levels established through manual DGA sampling. Due to hydrogen’s low solubility rate, it is common to see elevated levels of one of the hydrocarbon gases such as ethylene or ethane, while hydrogen remains at ppm of less than five ppm. Though hydrogen was generated throughout the thermal conditions that caused the other gases to be saturated in oil, Ostwald’s coefficient requires that a ratio of about 20:1 ppm be maintained between the gas to oil partition in order for hydrogen to remain in solution. Any leaks present on the main tank of the transformer will greatly impact this solubility in oil and cause the observed ppm of hydrogen to be much lower than that which is actually generated. This effect can cause the hydrogen sensor to be inaccurate in detecting incipient faults. For this reason, close attention should be given to the tightness of the tank, which can be noted while examining the DGA results, and the placement of the sensor, which should be in an area of the insulating medium with the greatest circulation or the headspace of the transformer. It would also be advisable to set alarm limits based on rate of change as opposed to establishing a set ppm value limit when monitoring hydrogen production.
Conclusion
Through the use of the various monitoring technologies currently available to the utility industry, it is possible to have a more complete overview of equipment condition than ever before. Real-time data can be made readily available for nearly every substation asset, ranging from DGA reading for transformers, power factor and partial discharge for bushings and real time monitoring of breaker operations. However, this level of monitoring is often not an effective solution when considering cost efficiency and the level of data being provided. Through the application of proper analytics, knowledge of monitoring applications and an understanding of equipment failure modes, it is possible to provide a high level of reliability while keeping monitoring cost and data management to a minimum.
