Conference Agenda

Insulating systems under stress, both thermal and electric, break down yielding decomposition products, many of them in the gaseous form. These decomposition processes are chemical reactions and, as such, are subjected to its thermodynamics and kinetics laws and principles. It means that reaction products are dependent upon reaction’s temperature, pressure, and reagents concentrations, meaning that for each temperature or energy level involved in a given incipient fault in a transformer, a specific set of reaction products will be formed. Although stablishing a universal equation of gases formation for each type of transformer fault is not possible, due to differences in the other factors, pressure, and reagents concentration, several diagnostics methods have been developed based on statistic observation of failures related with gases generation. All those methods, however, are, as expected, related to basic chemical reaction thermodynamics. Therefore, transformer fault diagnostics based on mineral insulating oil (MIO) decomposition is well stablished and widely used. Knowing what fault is occurring, however, is not enough to allow a maintenance effective action; it’s also necessary to know how severe the fault is and if it does involve solid insulation. For oil/paper insulating systems, since cellulose-based papers thermal decomposition yields carbon monoxide and dioxide, these gases, and the ratio between them, are the most used criterium. Recently, the development of insulating systems for higher operating temperature transformers such as natural ester insulating oils (NEIO) with thermally upgraded kraft paper (TUK); aramid paper (AP), and hybrid paper (HP), as well as MIO with the same solid insulating options, have brought the need for further studies on incipient fault diagnostics. Several studies have been carried out to evaluate reliability of known diagnostic methods when applied to these new insulating systems and, on how to determine solid insulation involvement. In previous works, the authors have developed an experimental device able to simulate thermodynamics of a transformer thermal fault, in which a hot spot reaches high temperatures while the bulk oil temperature remains in normal transformer operating range. This setup allows heating of solid insulation to different fault temperatures keeping the bulk oil cool and, therefore, generating characteristic fault gases. In this work, the thermal fault generating device is used in studying thermal fault behavior in different insulating systems: MIO/TUK; MIO/HP; MIO/AP; NEIO/TUK, NEIO/HP and NEIO/AP. Results have shown that current diagnostic methods need to take into account the specific insulating system and that solid insulation plays an important role in fault development and characterization. As a conclusion, is suggested that diagnostic methods shall be adapted according to the specific insulating system in order to get best maintenance results.

Turn insulation of multiturn coils is often involved in machine failures. Until now, there has been no type test for detecting the long-term performance of turn insulation. Also, conventional diagostic tests such as partial discharge (DP) and dissipation factor (DF) tests are not accurate to detect degradation of the specific turn insulation.

The frequency response measurement technique, commonly called Sweep Frequency Response Analysis (SFRA), is based on the analysis of winding impedance in the frequency domain. This technique is used for the diagnosis of power transformer windings for more than 40 years and is currently the subject of research for its application in rotating machines.

The aim of this study is to evaluate the SFRA capability to diagnostic turn insulation degradation during long-term performance testing on coils by the insertion of fault resistances between turns and to develop an RLC model representative of the stator coil using MATLAB environment.

SFRA measurements were performed on a coil with inter-turn faults made by the introduction of resistances between turns at three locations. SFRA results were compared to the simulation obtained from the developed model. This validation ensured that the model was accurate and faithfully reflected the real behavior of the coil.

There are significant concerns regarding corona discharge, partial discharge, arcing, and minor arcing in medium- and high-voltage DC systems, as well as in battery energy storage systems (BESS). These issues may arise from various factors, including micro-voids in insulation, metal powder, loose connections in medium- and high-voltage buses, and internal short circuits in lithium-ion batteries. Detecting signals below 50 MHz presents challenges due to a low signal-to-noise ratio, making it crucial to differentiate between corona signals, partial discharges, arcing signals, and external noise. To address this challenge, the authors propose a novel signal differentiation analysis technique aimed at classifying signals and reducing unwanted noise. This analysis emphasizes four key parameters: repetition rate, signal frequency, signal duration, and magnitude. The authors apply high-voltage DC and overlapping harmonic voltage to a sample of epoxy insulation. To simulate the micro arc in a BESS, a positive electrode made of braided strand wire is paired with a flat metal piece as the negative electrode inside a transparent fiberglass tube. A new method classifies various signals and examines their behaviors using the aforementioned parameters. The process involves three steps. When the signal exceeds the alarm threshold, the device assesses whether to trigger an alarm by analyzing the signal’s behavior over several minutes, utilizing magnitude, time, and frequency (MTF) analysis. This paper presents the results of the experiment related to the classification of different discharges.

High-frequency, steep-fronted impulses generated by SiC inverters significantly increase the likelihood of insulation failure in hair-pin winding motors. To prevent partial discharge (PD) in Type I motor insulation systems, it is crucial to identify the maximum voltage drop in the motor stator and compare this voltage to the partial discharge inception voltage (PDIV) of the insulation. This paper presents a model of a stator with three-phase windings to analyze the voltage distribution in hair-pin winding motors. Simulation results indicate that: (1) the maximum phase-to-ground voltage occurs at the line end of the three-phase windings, accounting for approximately 70% of the input voltage; (2) the maximum turn-to-turn and phase-to-phase voltages are approximately 70% and 115% of the input voltage, respectively; (3) the error between simulation results and actual measurements is about 8%, verifying the accuracy of the calculations. The simulations identify the phase-to-phase voltage as the point where insulation failure is most likely to occur.