Non-Invasive Techniques for Health Monitoring of Oil-Paper Insulation in Transformers
Transformers are not only critical components but also very expensive power system equipment. Large number of power transformers, which have oil-paper as the insulation, have already crossed their design life, but are still working fine. The strategy for health management of transformers is therefore of immense economic importance to power utilities all over the globe. Such strategy depends on monitoring of health of the insulation. Health monitoring is best performed if it is non-invasive in nature and if it could identify the aging status of the most vulnerable insulation, viz. paper. The contaminant that damages paper insulation most is moisture. Hence, a good health monitoring technique of transformer is the one that could estimate moisture content in paper insulation accurately, which requires a proper understanding of moisture dynamics within the transformer insulation.
The following will be discussed in the Tutorial:
- Moisture in Transformer Insulation: Significance of Moisture Contamination of Transformer Insulation; Grouping of Transformer Solid Insulation; Sources of Moisture Build Up in Transformer Insulation – Residual Moisture, Ingressed Moisture; Average Rate of Water Contamination; Moisture Generated due to Decomposition of Insulation; Moisture within Transformer Insulation – How much and Where? Moisture in Oil-Paper System: A Case Study; Moisture Dynamics within Transformer Insulation; Moisture Migration due to Temperature Gradient; Appearance of Free Water in Transformer; Oil-Paper Moisture Equilibrium Curve; Moisture in Oil Detection by Karl Fischer Titration; Problems with Karl Fischer Titration Technique; Classes of Moisture Contamination.
- Non-Invasive Techniques for Transformer Insulation Monitoring: Dielectric Response Measurements in Time and Frequency Domains – Polarization and Depolarization Current Measurement (PDC) and Frequency Domain Spectroscopy (FDS), PDC Measurement – Test Set Up; Typical PDC Measurement Results; Precautions in PDC Measurements; Sensitivity of FDS Spectrum Sections; FDS Measurement Principle; Typical FDS Results; Critique on FDS Measurements, Field Experiences with FDS Measurements, Correlation between PDC and FDS Measurements.
- FDS using Non-Sinusoidal Excitation, Reasons behind the use of Non-Sinusoidal Excitation in FDS, Choice of Non-Sinusoidal Excitation, Experimental Results and Prediction of Insulation Condition, How to get better estimation of paper moisture by optimizing the non-sinusoidal excitation?
- Reduction of Testing Time for FDS, Practical Problems in FDS Measurement, Reducing FDS Measurement Time using Sinusoidal and Non-sinusoidal Excitations, Magnitude of Excitation Voltage vis-à-vis Measurement Frequency, Effects of noise an interference, Determination of the Optimum Span of Excitation Voltage, Reducing FDS Measurement Time using Chirp Excitation, Advantages of FDS Measurement using Chirp Signal, Application of Linear Chirp (LC) and Logarithmic Chirp (LGC) Excitation, Comparative test results.
Remote Monitoring of High Voltage Transmission Line Insulators
Starting with the causes of HV insulator failure, this talk outlines the existing monitoring tools and indices. The focal points of this presentation are the research aspects of remote monitoring of transmission line insulators, i.e. How to tackle the problem of field noise, How to solve the problem of power consumption in on-site acquisition of leakage current waveform and How to minimize the influence of voltage harmonics on leakage current characterization. As a solution to the problem of field noise identification, the use of instantaneous signal characteristics to differentiate between noise and waveforms portraying electrical activity on insulator surface are presented. A low-complexity method based on Short-Time Hilbert Transform is presented in which there is no need for defining any leakage current threshold for discarding low activity waveforms and which can identify waveforms containing multiple spikes. For the solution to the problem of power consumption in on-site acquisition of leakage current waveform, a compressed sensing based technique is presented. In this technique, data acquisition, processing and compression are combined in a single operation. The entire computational burden is shifted from the encoder (on-site) to decoder (centralized location), which results in low power requirement at the field device and reduces hardware complexity and cost of field device. As a solution to the problem of the Influence of voltage harmonics on leakage current characterization, a method based on time integral of leakage current is presented. Major advantages of this method are that apriori information on the harmonic content of voltage waveform is not required and time-integral of leakage current is found to be a stable parameter for remote monitoring under changes in voltage harmonics.
Modern Tools for Impulse Fault Diagnosis in Transformers
Insulation failure within transformers is considered to be one of the most important causes of failure of power transformers. Impulse testing of transformers after assembly is an accepted procedure for the assessment of their winding insulation strength to impulse overvoltages. In such tests, impulse voltage sequences are generated in the laboratory and applied to the transformers as per standards. For many years, the applied impulse voltage waveforms and the resulting current waveforms were analyzed manually by studying oscillographic records. Such manual interpretation of the waveform patterns for fault identification and classification is strongly dependent on the knowledge and experience of the experts performing the analysis. With the advent of digital recorders and analyzers, there has been an increasing trend to use modern signal processing tools for identification of impulse faults in transformers.
The wavelet technique allows each frequency component to be studied with appropriate time resolution. The inherent non-stationary pattern of transformer current waveforms during different fault conditions can be effectively classified using this frequency-selective feature of wavelet transform. For classification of distinctive patterns of fault responses, relevant parameters were extracted from the wavelet transform of the impulse fault currents in transformers. 3-D scatter plot of the parameters of fault classification showed that each type of fault formed their own clusters on the scatter plot. Studies on expert system based fault analysis have also been reported, where the knowledge base of the expert system were formed either by time domain or frequency domain information.
Over the years, the uses of soft-computing techniques like neural network and fuzzy systems have been employed for different types of fault diagnosis in transformers. The advantage of a fuzzy based system over a neural network based counterpart is that a fuzzy system has better user readability and analysis tractability due to its linguistic nature of knowledge encoding which is rather heuristic on many occasions. The application of a translationally adaptive multi pass fuzzy technique based on frequency domain analysis has been reported to be very accurate for fault diagnosis in a given range of transformers.
Fractals have been found to be successful to provide a description of naturally occurring phenomena and shapes, wherein conventional and existing mathematical models were found to be inadequate. In recent years, this technique has attracted increased attention for pattern classification issues. Insulation failure in transformers during impulse test is also a natural phenomenon and the resulting winding currents have complex waveforms. This complex nature of the current waveforms and the ability of the fractal analysis to discriminate complex patterns encouraged studies on its application for impulse fault classification in transformers. Investigation results have recently been reported on the use of fractal analysis for accurate feature extraction and pattern recognition of transformer winding current waveforms.
This talk is based on the experiences gathered in the course of research works over a long period of time and covers the following areas:
- Basics of impulse testing of transformers
- Expert system for fault identification using frequency domain data
- Fault identification using wavelet analysis
- Fault identification using fractal analysis
© Copyright 2023 IEEE — All rights reserved. A not-for-profit organization, IEEE is the world’s largest technical professional organization dedicated to advancing technology for the benefit of humanity.