Proceedings national Symposium on Electrical Insul Materials, September 7-11, 2008, Yokkaichi, Mie, Japan P1-36 Measurement Techniques for Transformer Diagnostic R.Schwarz , M. Muhr Graz University of Technology, Institute of High Voltage Engineering and System Management Inffeldgasse 18, 8010 Graz E-mail : sumame@hspt.tu-graz.ac.at Abstract: Power transformers are one of the most important and also expensive components of the electrical power supply. In electrical networks, power transformers are of essential importance as interface between different system voltage levels, Due to the long, manufacture process they are one of the most critical and high-priced equipment and so this is one of the reasons why condition monitoring becomes more popular. Monitoring systems as a basis for diagnosties open possibilities for expanding the operation time, reducing the tisk of expensive failures and allow several maintenance strategies. With different - monitoring techniques, detailed information about the transformer condition can be received and this helps to minimize the probability of an unexpected outage, This paper gives an overview about different ‘monitoring and diagnostic methods and techniques, ‘especially some example in the area of partial discharge (PD) and the frequency response analysis measurement, INTRODUCTION A defect in transformers can be caused by electrical, clectromagnetical, diclectrical, mechanical, thermal and/or chemical load (stress). The typical failure distribution of high voltage transformers shows that the highest risk for failures are the tap changer, windings and the core as well as the bushings. For & check of the transformer condition, different diagnostic methods are available and they use chemical, mechanical, optical, thermal and electrical methods. Fundamental different ‘evaluation principles are under development and in applications (on-line and off-line monitoring systems). In addition to the traditional method (periodie sample of the transformer oil and to carry out a dissolved gas analysis), there are also new developments for example in the field of partial discharge detection and temperature measurement ‘One of our research project at the institute deals with the possibilities of the optical detection of partial discharges inside an electrical equipment (transformer) with the use of fibre-optic cable to detect the light emission of PDs, The aim of the research is. the detection and the exact localisation of PD inside an ‘enclosed equipment BASIC DIAGNOSTIC METHODS ‘The most common insulating system for transformer is the oil-board insulating system. The oil and the board fare organic components and underlie aging, which depends mainly on the operating condition. Deterioration processes relating to aging are accelerated by voltage and thermal stress. Chemical Diagnostic Methods Most internal transformer condition problems ean be detected through oil analysis and so the chemical methods are very important. New sensor technologies with higher sensitivity and new monitoring technologies hhave become commercially available in the last years. The methods differs between the dissolved gas anaylsis (DGA = with interpretation according 0 Duval, Démenberg, [EC $99 and other methods) and the oil analysis (Furan value, moisture, neutralization value et al) A restriction by using chemical methods is that it is possible to verily failure, which has changed since the last measurements, but it is not possible to say something about the exact failure or the date of origin, There is a number of diagnostics that are used to in-service transformer oil samples, but only a small part cean be performed on-line and in service. Calor Tease a color indiates Cloadiness or sludge should be investigated Measures oi sai to slate Vial amination Dialects Brcakdows sength Sensitive wo contamiations, Power Fear Dees polar conan Dissolved Gay Analysis Detects and idee nip Tne Tension ‘Detos polar containers ane ‘Nesalizaton Number Messures acy of Tnieaor of Can deter ontanation Mois can Gamage relation —] Table | Common In-Service Oil Diagnostics [1] Speaie Gav Moise Content For example, the most efficient tool for the recognition and classification of thermal and electrical failure is the Dissolved Gas Analysis (DGA). Dissolved Gas Analysis If partial discharge, overheating, hot spots or local breakdown inside the transformer occur, several gasesare produced and dissolved in the oil. The test is made by taking an oil sample and determining the levels of different dissolved gases such as hydrogen, oxygen, nitrogen and lower-molecular or carbon” dioxide. Different levels of each gases can be correlated to a specific type of fault and also the ratio of some gases to ‘each other for a failure diagnosis of great importance. Beside periodically performed gas-in-oil analysis new systems with sensors for on-line detection have been developed. The continuous gas-in-oil monitoring may dotect the start of incipient failure conditions to allow confirmation of a suspected fault Moisture in Oit ‘The oil humidity is another important parameter in the area of oil analysis. The used cellulose insulation in power transformers produce water when it degrades, so ‘water in oil indicates paper aging, For the analysis, the Karl-Fischer titration is used as standardized test. There is a correlation between the water content in oil and in paper for the condition of equilibrium [2] Humidity in cellulose New developed sensors measure the humidity in the cellulose. The sensor consists of impregnated paper between electrodes (fig. 1) and in the case of changing. the humidity in the paper, the value of the capacity changes too [3 Fig, 1, Humidity sensor [3] Electronic Buchholz Relays ‘Analyzing the gases collected in a Buchholz relay with an additional dissolved gas-analysis, a diagnosis of the failure can be carried out, One disadvantage of the conventional Buchholz relay is its integrative measuring characteristic, No information can be given about the history of the faifure (a large fault in a short time or a simall failure during a long time period). ‘The functions of the Buchholz relay are extended by using capacitive sensors [4]. An incorrect alarm can be suppressed because the gas-rate can be detected. The centering gas changes the capacitance of the sensor. The change of the capacitance is adequately to the ‘gas-amount. After measuring the gas volume, the gases are transferred into a gas collector, where they can be stored and analyzed. Other important chemical diagnostic methods, especially for determinating the oil condition, are the valuation of the color and appearance, acidity, neutralization value, interfacial tension, sludge content, particle count and corrosive sulphur, Mechanical diagnostic methods The tension force decreases during operation soa ‘monitoring is useful for transformers where the risk of short-circuit is high. To determinate the existing tension force of the transient oil pressure is measured after applying of a current surge. Another mechanical diagnostic method is the stream analysis of the cooling system, A further possibility is to analyze the acoustic vibration signal for a transformer. Some investigations show that the emitted acoustic spectrum of a tansformer is characteristical for types of a defect. The acoustic vibrations of a transformer can be generated for example by the coil vibrations, depending on the current amplitude and winding clamping compression or core vibration (magnetostrietion), The acoustie method is also an important method for the detection and especially the localization of partial discharge. Electrical diagnostic methods For the detection of a failure, different measurement technique are in use, for example loss factor, PDC analysis, resistance, RVM, FRA, transfer function, breakdown voltage, partial discharge and other. ‘The measurement’ and evaluation of the ‘dielectric response’ is one possible way of diagnosing a transformer’s insulation condition. Moisture and ageing strongly influence the dielectric properties of oil/cellulose insulation systems. The resistance measurement of all coils on all steppings can show conclusions about turn-to-turn fault or breaks. The most famous and popular electrical diagnostic method for evaluating of the oil condition is the oil breakdown voltage. This method is standardized and delivers the resistibility ofthe oil against alternating voltage. Polarization/ Depolarization Current Analysis (PDO) The PDC analysis is for determining the conductivity and moisture content (insulation humidity) of insulation materials in a tansformer. Polarization and depolarization currents are influenced by the properties of the insulating materials as well as the geometric structure of the insulating system, For the measurement, @ DC voltage step is used for @ specific polarization time, During this time, a pulse-like current flows and falls to a constant value depending on the conductivity of the dielectric, After the polarization time the voltage is removed and the dielectric is short circuited via the amperemeter. The discharging current jumps to a negative value, which goes towards zero. ‘After this, a model for the transformer main insulating system which describes its dielectric behaviour is assigned, All parameters for the dielectric model can be calculated from the polarization characteristics of material samples of precisely specified moisture content, the geometry of the main insulation and the positions of the samples. As a result, information about the moisture content, land, the conductivity of the oil and the polarization index of the insulation system can beobtained, Fig. 2. PDC ofa transformer winding according [5] ‘The recovery voltage method (RVM) is another time domain method used to investigate slow polarization processes, Response Analysis Another diagnostic method is the response analysis by using different excitation functions. Depending on the type of application, different methods - Impulse Response Analysis (IRA), Step Response Analysis SRA) and Frequency Response Analysis (FRA) - are used, The FRA is primarily used for detection of deformation or movements of windings. Thereby, a HE signal is applied and the corresponding winding responses are recorded. ‘The measurement of the current and the vollage during, the impulse test allows the calculation of the impedance fr its reciprocal value as function of the frequency (wransfer function). The function describes the transformer behaviour independent on the input signal Fig. 3. Transfer finction describes an transformer as an electric network For On-site measurement it is possible to generate the transfer function by switching operations or by ‘measurement of lightning or switching impulse voltages and currents. Different methods to measure the low frequeney response of power transformer are used. The swept frequency method makes use of a simple truth that the sinusoidal AC impedance of a transformer ‘winding varies with frequency over the range of interest Another method is to use @ low voltage impulse for the ‘measurement and the result is calculated by an Fourier analysis. The transfer function is calculated as the FFT of the output divided by the FFT of the input (fig. 4). Modern techniques work on the basis of the low voltage impulse method with high speed, high resolution and sensitive analyzer with a superior signal to noise performance at high frequencies. The calculation for the ‘ransfer function is done by a series of impulses (an average frequency domain spectral density estimate is computed ~ transfer function calculation as ratio of the average cross spectra to the average auto spectra) [6]. Partial Discharge (PD) The partial discharge activity is the most prominent indicator for insulation degradation. For example they ‘occur in transformer in gas bubbles in the oil, voids in solid insulation material or at metallic particles. The PD pulses generate clectromagnetic acoustic signals, chemical reactions, local heating and optical signals. For the detection different measurement techniques and therefore different types of sensors were used, for example high frequency current transformer, bushing capacitive taps and piezoelectric sensors. The measurement according to the IEC 60270 is @ usefull too! for laboratory measurement because it is a sensitive method and can be calibrated, but on-site measurements can be affected by disturbances, Some signal processing methods have been used to suppress unwanted noise Another possibility is the measurements at frequencies where the S/N level is maximal - VHF (very high frequency) and UHF (ultra high frequency) PD measurement (fig. 4b), Many types of partial discharge impulses in wansformer are sufficiently short (nanoseconds) to radiate UHF Signals, ‘so they generate electromagnetic waves with a high frequency spectrum up to 3 GH. Fig. 4, a)The UHF external sensor: (Left) front and back side of the sensor alone, and (Right) mounted on a dielectric window on the tank of the transformer [7] and b) Example of a 3-Phase Amplitude Relation Diagram (3PARD) The acoustic partial discharge measurement uses the fact, that an acoustic signal (mechanical vibration) as @ result of the pressure build-up caused by a generated discharge in the insulation medium is emitted. The frequency spectrum lies in the range between 10Hz up to 300kHz, In transformers, mechanical waves generated by PD and propagate though the oil hitting the tank walls where the sensors (Piezoelectric sensors) for the acoustic signal detection are fixed externally. An advantage of the acoustic detection over chemical and electrical methods is that position information is available from acoustic systems using sensors at ‘multiple locations Beyond the electromagnetic wave emissions, each partial discharge emits radiation. The emission spectrum ff partial discharge in liquids contains components characteristic of the insulating material or of its‘chemical components (ultraviolet. visible and infrared area). For the measurement, different optical sensors ‘and techniques can be used [8]. INVESTIGATIONS An optical partial discharge measuring system was developed, which gives the basis for investigations inside of electrical equipment. The system consists of an ‘optical sensor and fibre optic cable to transport the ‘optical signal from inside the equipment outside to the ‘evaluating processor unit. A photomultiplier converts light into an electric signal. As test setup, a simple peak-plate arrangement in oil is used for the PD measurement, Investigations with two different ‘conventional detection systems (different bandwidth) in ‘comparison to the optical system are carried out, Fig. 5: Concept of the conventional and optical PD ‘measurement To -tst setup, CC soup. Capac, VD otg ii, QP System = "undvis MMle, QP2-Syten andi IK Op yi aah EOF" opti ie Fig. 6 show a result of the measurements, The detected discharges in oil are scattered in amplitude and shape. aT ~W\ ; Fig, 6: Comparison of the optical (1) and conventional (2), (3) detected single impulse PD signal in oil In oil it comes usually to a burst of discharges pulses (many discharges in very short temporal sequence), which cannot be impulse-dissolved represented with the conventional detection method (limited bandwidth system (3)). With the optical system (inclusive a photomultiplier) a detection of these single impulses with a fast rise time is possible, In contrast to the limited bandwidth, the second wide band conventional system (2) shows good impulse dissolution. So a good correlation between the results ofthe optical system and the conventional PD measurement - with high bandwidth - is given. The intensity ofthe optical signal ‘corresponds withthe level of the partial discharge. Also the phase afiliation is given ‘An additional method to the conventional measurement is the optical detection with the advantage of the galvanic separation, the insensitive to interference (EMC), and the possibility to use systems with high ‘bandwidth to analyse the PD impulse behaviour. SUMMARY For the transformer diagnostics many methods are available, The DGA is the most widely used method for investigating incipient faults with interpretation schemes according to IEEE and IEC standards. The PDC measurement is an non-destructive method which is capable of estimating the oil/paper condition by conductivity calculation. For the detection of ‘mechanical deformations or mechanical damages the FRA is a well-recognized method. ‘Another possibility for a non destructive test is the partial discharge detection, New systems for on-line ‘measurement are increasing popular in addition to the off-line measurement, The UHF PD measurement has the advantage for on-site measurements with high sensitivity and few influences from outside disturbances. ‘The possibility of the optical detection seems to be an appropriate method for the future for specific applications with the advantage of discharge location ‘and immunity to EMV. An possible application for this technology is the observation of critical geometrical areas in electrical equipment, for example a transformer, REFERENCES, [1] T. Cargol, “An overview of Online Oil Monitoring Technologies”, Fourth Annual Weidmann-ACTI Technical Conference, San Antonio, 2005 [2] J. Fabre, A. Pichon. * Deteriorating Processes and Products of paper in Oil, Application (0 transformers” International Conf. on large high voltage electric systems (CIGRE), paper 137, Paris 1960 13] H. Borsi, “Diagnose von Leistungstransformatoren” Institut flir Energieversorgung und Hochspannungstechnik der Universitit Hannover, Schering Institut [4] H. Bors; M. Urich, T. Leibftied "Das neue elektronische Buchholzrelais" Elektrizititswirtschaft, J. 97 (1998), Heft 13. [5] G. J. Pukel, M. Muhr, W. Lick, “Transformer diagnostics: Common used and new methods”, Intemational Conference on condition Monitoring and Diagnosis, CMD 2006, Changwon, Korea, April 2006 [6] J. Briton, “Transformer Maintenance and diagnostics using frequency response analysis”, Electric Energy T&D Magazine [7] PD. Agorisl, 8. Meijer, J.J. Smit, J. Lopez-Roldan, “Sensitivity Check of Internal Sensor for Power Transformer Partial Discharge Measurements” 1-4244-0189-5/06'S20.00 ©2006 IEEE. [8] R. Schwarz, M. Muhr, “Partial discharge detection in oil with optical methods” - \Sth IEEE Intemational Conference on Dielectric Liquids - ICDL2008, June 26 July 1, 2005, Coimbra, Portugal