Recent Trends in the Condition Monitoring of Transformers

Deregulation of electric supply around the world has led to a number of changes and new challenges for the electric utility industries. In the changed scenario electric power utilities are trying to enhance revenue streams and to reduce incremental cost per unit electrical power by various means. Power transformers are the most expensive and strategic equipment of any electrical power system. Over the years operating stresses of power transformers have increased due to load growth and increase in voltage level. Insulation failure of power transformers results in substantial costs of repair and huge financial loss due to inadvertent outage. In the context of maximum utilization of power transformers with long service life, it is necessary to determine the adequacy of the insulation system of a transformer before installation. Design acceptance tests are performed on transformers for this purpose, the most important of which is the impulse voltage withstand test. Minor local defects in transformer insulation often manifest itself as partial discharges (PDs), which may cause catastrophic failure in the long run if allowed to continue unnoticed. Thus PD tests have emerged as one of the significant tools for identifying local defects within transformer insulation. Large numbers of oil-immersed type power transformers, which are still in service all over the world, have outlived their designed service life. Studies have shown that the most common cause of failure in aged transformers is general aging of oil-paper insulation system. Age related degradation is accelerated in the presence of moisture in insulation and also at elevated operating temperatures. Thus it is economically beneficial for power utilities to assess the overall condition of transformer insulation system with the aim to minimize the risk of failures and to avoid forced outages. Practicing engineers are using chemical techniques for assessing the overall condition of transformer insulation for a long time. The most important chemical techniques for condition monitoring of transformers are: (i) Dissolved Gas in Oil Analysis, (ii) Determination of Degree of Polymerization of Paper and (iii) Furan in Oil Analysis. In the past, these methods have enabled power utilities to anticipate developing problems and plan maintenance activities accordingly. In recent past, new non-invasive diagnostic methods have been proposed which are complimentary to classical electrical measurements. These methods are based on measurement of dielectric response of transformer insulation in time domain, viz. measurement of polarization and depolarization currents and recovery voltage measurement, or in frequency domain, viz. frequency domain spectroscopy (FDS). Large mechanical forces act on the transformer windings and mechanical structure when subjected to fault currents during its service life. These mechanical stresses can lead to winding deformation which could ultimately lead to major failure. Frequency response analysis is a diagnostic approach which looks for changes in the frequency response signature of a transformer which would be indicative of winding deformation. From the point of view of asset management, an important issue before the operational and planning staff of power utilities is to make a rational decision on the remaining life of aged transformers. The best-known and most widely used relationship in this context is the one between life expectancy and thermal loading of transformers using temperature rise as the main index of aging. But other methods of remaining life analysis based on measurement of degree of polymerization (DP) and also furanic concentration in oil are being investigated to offer reliable information taking into account complex thermo-chemical degradation of transformer insulation.

This chapter presents a detailed description of different methods for impulse fault analysis of transformers. These methods form a broad spectrum ranging from the use conventional techniques to computational intelligence tools. All these aspects for impulse fault analysis are discussed in this chapter.

Partial Discharge (PD) is a common local defect of high voltage equipment, especially insulation system of transformers. PD in the insulation of electrical equipment is a sign of dielectric defects as well as a cause of further degradation of its insulation system, which may ultimately lead to failure of the apparatus. Therefore, early detection of PD sources may prevent failures and hence save revenue loss due to damage and/or interruption in service. This chapter describes various aspects of PD occurrence and the methodology to measure them. Moreover, discussions about the methods of analysis of PD data from conventional to modern approaches are included in this chapter.

A utility can cut costs well while managing aging and its effects, if it can distinguish between the equipment that needs maintenance from that which does not. This is possible through condition monitoring. Condition monitoring of major equipment such as transformers receive more attention in most of the utilities due to the fact that such failures lead to potentially widespread interruption of service and very high replacement cost.

In the past, conventional methods for condition monitoring (discussed in the previous chapter) have enabled power utilities to anticipate developing problems and plan maintenance activities accordingly. In recent past, new non-invasive diagnostic methods have been proposed which are complimentary to classical electrical measurements. These methods are based on measurement of dielectric response of transformer insulation in time domain, viz. measurement of polarization and depolarization currents (PDC) and recovery voltage measurement (RVM), or in frequency domain, i.e., frequency domain spectroscopy. This chapter elaborates the two methods of time domain dielectric response measurement—namely PDC and RVM.

Frequency domain spectroscopy (FDS) is a non-destructive method of assessing the insulation condition of high voltage equipment, especially oil-paper insulation system of transformers. It is a reliable tool for predictive maintenance of such equipment in the field. In this method each test object is considered as a black box accessible only through its terminals. Consequently this method identifies the global changes in the insulation condition and is not suitable for identifying any local defect in the insulation system. FDS method involves the application of a sinusoidal voltage across the terminals of the test object and the measurement of the amplitude and phase of the response current flowing through the insulation. From the measured voltage and current, the complex capacitance, complex permittivity and dielectric dissipation factor of the insulation are determined over a wide frequency range. Based on the fundamentals of the methodology, it is expected that the presence of polar substance such as moisture could be detected by FDS. It is also sensitive to other polar by-products of insulation aging like low molecular weight acids. FDS results are affected by the geometry of the insulation system, i.e. the relative dimensions of spacers, barriers and oil ducts. The external factors that affect FDS most are operating temperature and weather conditions like rain. For on-site testing significant errors can be introduced in FDS measurement, if certain precautions are not taken, particularly when different equipments are connected in parallel with the transformer. Mathematical modeling is important for the interpretation of FDS results for the estimation of moisture content in solid insulation. It is also relevant to analyze the relationships between time domain and frequency domain test results under practical conditions in order to determine the condition of the transformers and also to establish general validity of the test results.

Over the last few decades it has been recognized that Frequency Response Analysis (FRA) is a reliable method for detecting dimensional changes in the transformer windings. Out of the two methods of FRA, swept frequency response analysis (SFRA) has gained popularity over the impulse frequency response analysis (IFRA) with the passage of time, although the later was developed first. SFRA is an offline, non-invasive, terminal-based low-voltage measurement, the primary objective of which is to detect even small mechanical deformations accurately. In SFRA the excitation signal of a given frequency is injected to one terminal of the transformer and the response signal is measured at the other end of the winding or at another terminal. Then the transfer function of the winding is computed from the measured excitation and response signals. The frequency of excitation signal is varied over a wide range to detect the deviations in the FRA response, with the help of which the nature of the mechanical fault as well as the quality of insulation could be judged. Frequency dependent parameters of winding such as inductances and capacitances determine the resonant frequencies which appear as peaks and valleys in the FRA transfer functions. Such resonances, which are governed by the poles and zeros of the equivalent network of the transformer winding, appearing in the test case is compared to a no-fault case for identification of fault, if any. There are several comparison techniques that are being used in practice and sometimes more than one technique need to be used simultaneously to arrive at an unambiguous decision. In FRA evaluation processes, it is important to identify the factors that affect the transformer responses in different frequency ranges. There are several terminal configurations of FRA test that may be employed in practice. The sensitivity of the terminal configuration in the detection of mechanical faults is an important issue in FRA. Accurate simulation of FRA response of transformer windings, not only in isolation but also for two-winding transformers, is required for identifying FRA features that could be used for accurate fault identification within an integrated framework. Development of suitable online monitoring system based on SFRA will be helpful in identifying small mechanical faults within transformers in service before such small faults could lead to major faults with severe consequences.

Generally, the life of a transformer is equated to the life of its cellulose insulation. In other words, the aging of cellulose insulation determines the ultimate life of a transformer. It has been established that thermal degradation of cellulose insulation like paper and pressboard eventually limit transformer lifetimes. In order to develop a model for predicting life expectancy of oil-immersed transformers, it is necessary to understand the aging mechanism of cellulose insulation. Life estimation using transformer thermal model is a practical approach in predicting transformer end-of-life as detailed in IEEE and IEC guides. In such cases, calculation of life of in-service transformers is done based on temperature rise of transformers, its loading condition, and ambient temperature. But the results of such thermal models using temperature rise as the only index of aging introduce large errors in life estimation. Temperature rise, oil condition, loading history, and design parameters are all key factors that determine the loss of life of a transformer through thermochemical degradation. Degree of polymerization (DP) of cellulose insulation is an excellent indicator of insulation aging which incorporates the effect of all the key factors of aging. Hence, transformer life estimation using the DP value of its solid insulation has been explored over the years. But DP tests are destructive in nature and are very difficult to perform for transformer in service. So life estimation through non-destructive testing by measurement of furanic compounds in oil, which are present in oil in dissolved state, is an attractive alternative. Since this methodology is based on oil tests, it could be used for online monitoring of transformer life, too. A predictive model which involves reliability engineering, physical understanding of the degradation process, and actual knowledge of the present condition of a transformer is necessary for accurate estimation of remaining life, which will facilitate appropriate replacement time of transformer and planning of appropriate maintenance scenario.