Conference Agenda

The increasing integration of power electronic converters across low-, medium-, and high-voltage systems introduces new challenges for the design and evaluation of electrical insulation. Fast-switching semiconductor devices operating with pulse-width modulation (PWM) produce complex voltage stresses that differ significantly from traditional sinusoidal excitations. These stresses—characterized by high rise times, repetitive impulses, and wide frequency content—can accelerate insulation degradation and promote partial discharge (PD) activity.

Due to the nature of the inverter voltage signals, the classic measurement circuit according IEC 60270 for PD-detection at sinusoidal voltages (50-60 Hz) is not suitable in this case.

Therefore, we developed and tested a measurement system especially for PD-detection under inverter stresses with rise-times around down to 50 ns and repetition frequencies up to 20 kHz. The PD-probe can be operated at temperatures up to 155 °C.

This paper presents the process of designing, manufacturing and commissioning of the special PD-antenna for our demands, which are defined by our inverter generators used in the laboratory.

In a first step the PD-measurement system is tested at 50 Hz sinusoidal voltages on different test objects in parallel to the classic PD-measurement system acc. to IEC 60270. By doing so, the determined results, i.e. PDIV and PD-magnitude can be compared.

In a second step, the PD-measurement system is tested at inverter operation. In this case, the determined PDIV values can be compared with the values obtained by visual measurements with an UV-camera and the results are discussed.

In a final step, first results of PD-behaviour, i.e. PDIV and PD-magnitude depending on inverter rise-time, repetition frequency and temperature are presented and discussed

Metal-clad switchgears distribute electrical energy in Medium Voltage (MV) distribution networks and play an important role in the reliability and safety of the power supply to customers. According to statistics, 40% of metal-clad switchgears faults are attributed to insulation failures originating from cracked supporting insulator, loose electrical connections or vibrating components, flawed insulation material, cable terminations defects or surface contamination combined with moisture. These defects can excite thermal degradation or Partial Discharges (PDs) under normal working stress which are hazardous to switchgears’ insulation systems and may eventually result in catastrophic failures. Therefore, online Monitoring and Diagnostics (M&D) techniques have been widely recognized as an effective, nondestructive, and noninvasive predictive maintenance tool, which provides an early indication of potential deterioration or failure in the switchgears. In this paper, an advanced diagnostic technique based on online Partial Discharge (PD) and ultrasonic testing using portable devices was applied to detect and localize abnormal sound source in 13.8 kV Medium Voltage (MV) metal-clad switchgear. Although PD magnitude was high, however, developed Phase Resolved PD (PRPD) patterns and waveforms didn’t confirm presence of any PD activity. Consequently, ultrasonic measurements were conducted to check if the abnormal sound was due to electrical or mechanical issue such as, corona, tracking, arcing or loose components in the switchgear. The observed time series and Fast Fourier Transform (FFT) waveforms were compared with the established patterns and it was revealed that abnormal sound can be due to mechanical fault caused by loose or vibrating components in tie-breaker compartment of the switchgear. It was recommended to create outage on the switchgear and visually inspect tie-breaker compartment thoroughly to find the source of abnormal sound and fix it. The loose nuts and bolts of Mechanism Operated Cell/Truck Operated Cell (MOC/TOC) cover plate were found which triggered the vibrations in the plate to cause abnormal noise. These nuts and bolts were tightened and a gasket was installed to fix the plate to minimize the vibrations and diagnostic testing were repeated to check the integrity and quality of rectification work. The PD measurements were in acceptable level, and ultrasonic measurements in terms of magnitude, time and frequency domain didn’t match with the pattern of loose or vibrating components and were observed in acceptable level. The acceptable levels of diagnostic measurements confirmed that the source of abnormal sound was correctly identified and successfully rectified. In the future, a permanent advanced online fault diagnosis system is proposed to be installed in switchgears for effective condition assessment and enhanced predictive maintenance to reduce equipment damage, flashovers, and personnel injuries, while enhancing the reliability and safety of power systems in the future smart grid.

Ultra-high voltage gas-insulated switchgear (UHV-GIS) is a critical component in power grids, where insulation stability plays a pivotal role in ensuring safe and reliable grid operation. This study constructs a full-scale 126 kV busbar cavity experimental platform to simulate partial discharge (PD) signals induced by four types of metallic contaminants: linear, spherical, block-shaped, and plate-shaped. A time-frequency feature recognition method based on Beq-Deq is proposed to enable accurate identification of discharge signals. This approach offers a technical solution for fault diagnosis and condition monitoring in GIS equipment, thereby improving the safety and protection level of power monitoring systems.

Sudden surface flashovers along Gas Insulated Switchgear (GIS) insulators during long-term operation pose a significant threat to power grid security. This study established a long-term voltage application test platform and a surface charge measurement system to characterize the surface charge distribution on specimens subjected to varying durations of voltage stress. The trap energy distribution within the specimens under prolonged AC voltage was obtained using the isothermal surface potential decay method. By combining the Weibull distribution analysis of flashover voltage, the evolution of surface flashover performance in an SF6 environment under long-term AC voltage was revealed. The results indicate that surface charges exhibit negative polarity under long-term AC voltage, primarily due to the higher mobility of electrons compared to positive ions. Prolonged voltage application increases the density of surface traps. Consequently, negative charges tend to accumulate near the grounded electrode due to enhanced electron injection, leading to severe distortion of the spatial electric field during the positive half-cycle of the AC voltage, which ultimately triggers flashover.

Electronic devices and power systems are vulnerable to transient overvoltage caused by lightning strikes, switching operations, and system faults. Metal oxide surge arresters (MOSAs), composed of multiple metal oxide varistors (MOVs), protect power systems by absorbing surge energy and limiting overvoltage. However, MOVs may exhibit non-uniform V–I characteristics due to manufacturing tolerances, cooling performance differences, and progressive aging under repeated surge current and overvoltage stresses. Such non-uniformity can lead to current imbalance within a MOSA bank, resulting in uneven energy absorption and degradation of overall overvoltage protection performance. In this paper, the impact of current imbalance in a MOSA bank is investigated using a PSCAD/EMTDC-based impulse test-bed model. A virtual MOV impulse test environment is implemented to analyze current sharing behavior under different MOV degradation conditions. The degradation level of individual MOV units is represented by controlled variations in their V–I characteristics to reproduce non-uniform degradation scenarios. The resulting current imbalance is quantitatively evaluated in terms of transient interruption voltage (TIV) and absorbed energy. The results demonstrate that progressive MOV degradation significantly exacerbates current imbalance and degrades the overvoltage protection capability of the MOSA bank.

Nanocomposite insulation has been extensively studied due to its superior insulating properties. The dispersion of metal-oxide nanoparticles within a resin matrix significantly enhances partial discharge endurance. Mica–epoxy insulation, which is specifically engineered to resist partial discharge, is commonly used in high-voltage rotating machinery.

Over the past two decades, considerable research has focused on the incorporation of nanoparticles into mica–epoxy insulation systems, typically by adding them to the impregnating varnish. In contrast, this report investigates the integration of nanoparticles directly into mica tape.

The global vacuum pressure impregnation (VPI) process, widely adopted for medium-voltage motors, enables the production of high-quality insulation within a shortened processing time. In this process, the impregnating varnish is reused multiple times, circulating between the storage tank and the VPI tank. When nanoparticles are added to the varnish, maintaining resin stability becomes a critical concern. However, this issue does not arise when nanoparticles are incorporated into mica tape.

The electrical, mechanical, and thermal properties of the developed nanocomposite insulation have been thoroughly evaluated. In electrical endurance testing, the slope n in the lifetime equation (t=kV^(-n)) was approximately 1.4 times greater than that of the conventional insulation, indicating a significantly extended lifetime at rated voltage levels.

Furthermore, a 6-pole, 800 kVA rotating machine utilizing this advanced material was successfully manufactured, featuring an increased design electric field.

Distribution transformers play a critical role in the conversion of electrical energy in the power grid. The complex temperature rise characteristics during transformer operation are crucial, particularly hot spot and average temperature rise. High temperature rise increases the susceptibility to age-related failures. However, the hotspot and temperature distribution can't be obtained in the conventional heat run test. Therefore, studying the hotspot and temperature distribution of the transformer via numerical and experimental methods can greatly improve the design for reliability and safe operation.

In this study, a numerical and experimental investigation of the winding temperature rise of a 72.5 kV dry-type transformer was carried out, with foil winding for the low-voltage winding and disc winding for the high-voltage winding. The aim of the study is to investigate the winding temperature distribution and identify the hotspot of the windings.

Computational fluid dynamics (CFD) simulations were carried out firstly to calculate the temperature distribution of the windings. A symmetrical geometry model, as shown in Figure 1(a), was representative considering the symmetry of the transformer structure and used for simulation to reduce computational complexity. The temperature distribution of the one phase winding can be reflected by setting symmetrical boundaries. However, in this simplification process, the influences of the low-voltage winding lead side and the high-voltage winding dome side were ignored. A semi-symmetrical geometry model as shown in Figure 1(b) was established to figure out the temperature difference between the dome side and the opposite side of HV winding.

The temperature distribution of HV winding is shown in Figure 2. The simulation results show the hotspot position of the winding and indicate that the HV winding temperature of the dome side is about 5.5 K higher than the opposite side.

Unlike the conventional transformer heat run test, the thermocouples were installed inside the low voltage winding and the fibre optic temperature sensors were installed inside the high voltage winding during the manufacturing process. The sensors were installed in the upper winding of Phase B as shown in Figure 3. These sensors were used to detect the temperature of the disks. Besides these sensors, there were also several sensors installed on the outer surface of each HV and LV windings to detect the temperature difference between the surface and the centre of the disks. The sensor positions were set according to the simulation results, which showed where temperatures were higher. All sensors were calibrated before testing.

The experimental results are in good agreement with the simulation results. The temperature difference between the dome side and opposite side of HV winding measured in tests is almost the same as the results calculated by CFD simulation. The trend of the temperature fits well, and the location of the hotspot is captured accurately.

This study provides a certain reference value for improving the structure and thermal performance of 72.5 kV dry-type transformers and ultimately aims to provide engineers with more detailed data to design the most efficient and reliable transformers for different applications.

The ferroelectric ceramic composition Ba0.89Ca0.11Ti0.87Zr0.13O3 (BCZT) was doped with Niobium, synthesized via the conventional solid-state reaction method, to investigate its structural, microstructural, and nonlinear properties for applications in nonlinear transmission lines (NLTLs). X-ray diffraction confirmed a single-phase perovskite structure for all compositions, indicating effective Nb incorporation. Scanning electron microscopy revealed high densification and significant grain size reduction with doping (from 7 µm to 1.1 µm), suggesting that Nb inhibits grain boundary migration. Dielectric measurements showed that Nb addition shifts the ferroelectric–paraelectric transition toward room temperature and increases the room-temperature dielectric permittivity. Tunability measurements revealed that the 0.5 mol% Nb-doped sample exhibits the highest sensitivity to the applied electric field, achieving an optimal balance between high permittivity and field response. These results demonstrate that controlled Nb-doping is an effective strategy to optimize the nonlinear performance of BCZT ceramics, making them promising candidates for NLTL-based devices.

This study aims to advance insulation diagnostic techniques for rotating machines based on partial discharge (PD) measurements. The objective is to develop a high-frequency equivalent circuit simulation method capable of reproducing PD current propagation and the corresponding detected waveforms within form-wound stator windings. In our previous work, an experimental setup was constructed by introducing artificial defects into form-wound stator windings, and waveform reproduction was examined through analysis of the propagation characteristics of PD currents generated at those defects. In this paper, by changing the interconnections between coils of the same phase, the PD occurrence positions were varied equivalently to investigate their influence on the detected PD current waveforms. The motor investigated in this study is a three-phase induction motor with a rated output of 75 kW and a rated voltage of 3 kV. The winding consists of a two-parallel structure with ten turns and sixteen coils per phase, featuring four poles. To investigate the propagation path of PDs generated in the winding, the wedge of the upper coil of the first winding in phase W was removed to create an artificial defect, allowing PDs to occur from the winding to ground. The neutral point was disconnected, and coupling capacitors were placed at each phase terminal to electrically isolate the windings. PD current waveforms flowing in the ground wire were detected using four high-frequency current transformers (CTs) under an applied voltage of 3.0 kV at 60 Hz. The PD current waveforms were simulated using the transient analysis software LTspice® (Analog Devices Inc.) and compared with the measured results. As a result, it was found that the farther the PD occurrence position is from the voltage terminal, the more the waveform becomes distorted and the lower the signal intensity. These results indicate that PD currents generated near the voltage terminal are less affected by propagation through the winding and can easily flow into the external circuit, whereas those generated farther away undergo significant attenuation during propagation and are mainly transmitted through stray capacitances between coils of the same or different phases. In future work, the developed high-frequency equivalent circuit simulation will be utilized to establish diagnostic methods for insulation degradation mechanisms.

This study investigates the development of dielectric cellulose pressboard modified with additives to improve long-term durability of power transformers under severe operating conditions. TiO₂ nanoparticles and polyamide–epichlorohydrin (PAE) were incorporated into cellulosic pulp via vacuum filtration and hot-pressing. Characterization through SEM, XRD, and impedance spectroscopy revealed that the additive-containing material achieved a 14.6% increase in electrical resistivity and a 12% reduction in dielectric permittivity. Notably, TiO₂ incorporation led to a 28% improvement in dielectric strength (10.1 kV/mm) without altering cellulose crystallinity. While electrical conductivity in aqueous extracts rose slightly, all parameters remained within technical standards. These results demonstrate that controlled nanoparticle addition is an effective strategy for developing more robust and reliable high-voltage insulation components.

Electrets are dielectric materials capable of maintaining quasi-permanent electric charge or polarization and offer a promising approach for suppressing partial discharge (PD) in medium and high-voltage power electronics. This work develops

and evaluates a conduction-charging methodology for forming fluorinated parylene (AF-4 type) or also known as Parylene HT electrets compatible with insulated power modules. The proposed system replaces conventional corona-needle charging

with a cylindrical electrode conduction charger, demonstrating a safe electret charging process without the risk of damaging semiconductor chips, wire bonds, sensors, and overcoming limitations in packaged or multilayer substrates. Aluminum nitride substrates coated with Parylene HT are investigated with and without a silicone polydimethylsiloxane (PDMS) encapsulation. PD performance is evaluated under a 60 Hz unipolar square-pulse voltage with slew rates up to 150 V/ns. Results show that conduction-charged electrets with PDMS significantly reduce PD activity and maint PD-free operation up to 2.9 kV. This validates conduction charging as a scalable alternative for PD mitigation in medium-voltage power modules.

Cellulose nano-fiber (CNF) has attracted considerable attention as a low environmental impact material. CNF boards, fabricated through compressive forming of cellulose nano-fibers, consist of a complex network of fibers with diameters on the order of several nanometers. This structure incorporates air into the inter-fiber gaps, effectively reducing relative permittivity. Moreover, the extremely narrow spacing between fibers suppresses electron avalanches under high electric fields, thereby enhancing dielectric strength. Applying these spacers to the end-winding region in rotating machines is expected to reduce electric field concentration between adjacent out-of-phase coils even in narrow gaps, thereby suppressing partial discharge and improving the overall field distribution.

In this study, two critical factors for practical implementation—resin penetration and moisture absorption behaviors—were evaluated. Resin penetration during vacuum pressure impregnation (VPI), a conventional manufacturing process, was examined through cross sectional observation, mass variation, and relative permittivity measurements. Moisture absorption behavior was evaluated using saturated salt solutions.

To control the size of internal voids, samples were prepared by varying the mixing ratio of pulp (which has a larger fiber diameter) and CNF in 25% increments. The sample thicknesses were set to 0.2, 0.4, and 0.8 mm. In addition, considering the line to line voltage applied during operation, the dielectric strength was measured for 100% CNF boards with the same thicknesses.

The results showed that higher CNF content led to lower resin penetration, indicating that internal air was retained within the structure. The relative permittivity exhibited almost no change before and after the impregnation process, demonstrating that the dielectric properties were maintained regardless of the resin treatment. The equilibrium moisture content increased with relative humidity. In practical environments, applying a surface resin coating was found to be effective in suppressing moisture absorption. During impregnation processes in rotating machines, resin is expected to adhere to both surfaces of the spacer, resulting in a total thickness of approximately 2–3 mm including the felt. When the CNF boards were coated with an equivalent amount of resin, they exhibited almost no moisture absorption even at 76%RH, confirming that moisture uptake does not pose practical concerns. Furthermore, CNF boards in the dry state exhibited dielectric strength exceeding 30 kV/mm, outperforming conventional epoxy glass cloth and polyester glass mat laminates. These results demonstrate that CNF boards possess excellent dielectric and insulation properties, indicating their potential for use as spacers installed in the end-winding region of electrical insulation systems in rotating machines.

With the expansion of renewable energy sources, the importance of voltage-source HVDC systems has been increasingly emphasized. To ensure system reliability and grid protection, DC Gas-Insulated Switchgear (DC GIS) has become an essential power apparatus. During DC GIS operation, various overvoltages such as impulses caused by lightning strikes or circuit breaker operations may occur. Therefore, considering insulation design under DC rated operating conditions is indispensable, and analyzing surface charge accumulation behavior under DC pre-stress conditions is a fundamental research task for DC GIS insulation design. In particular, long-term DC application can lead to surface charge accumulation, which causes local electric field distortion and consequently reduces insulation strength. In this study, DC/LI superposed breakdown tests were conducted at the gas–electrode interface based on IEC 62271-5:2022 to support insulation design for DC GIS. Furthermore, to quantitatively understand the charge accumulation mechanism induced by DC voltage application, surface charge accumulation under DC pre-stress conditions was analyzed using COMSOL Multiphysics.

The electrification of aircraft is driving an increase in voltage ratings of printed circuit boards (PCBs) used in their power converters. This means that they need to be designed appropriately to reduce high voltage risks such as partial discharge (PD) and electrical breakdown. The PCBs used in aircraft power converters are often multi-layered to allow for the transfer of high current. Conductors on both external and internal layers need to be designed with adequate spacings between one another. Design standards such as IPC-2221C provide conductor clearances both internal and external of the PCB to ensure safety and reliability. However, these values are considered conservative by the sector. As a result of the strict weight and size restrictions in aerospace a reduction of these clearances would be valuable. In this investigation, internal conductor spacings are explored. Data is collected from parallel internal conductor clearances and clearances around a via or plated through hole (PTH). Due to high switching frequencies used in power converters, which are known to reduce breakdown strength of insulation, testing was also performed to understand the impact of increased frequencies.

This paper examines the use of CIM, TPSD, and PPDA algorithms, recently introduced at international conferences. The online CIM detects partial discharges within a frequency range of 1 MHz to 50 MHz. In the setup, partial discharges are measured with an RFCT mounted near the end winding of a rotating machine rated below 3 MW. Considering the PD detection bandwidth, the online PD calibration signal is a 20 MHz sinusoidal wave capable of producing an integrated charge of up to 50,000 pC. Testing a 200 kW motor shows that the PRPD pattern measured with the standard PD measurement unit, as outlined in IEC 60270, matches the pattern obtained using CIM. The primary benefit of the online CIM method is that it eliminates the need to trend data over time to evaluate stator winding health; instead, a single measurement provides a PRPD pattern comparable to offline measurements, utilizing a standard PD measurement unit. The TPSD technique differentiates between phase-related PD, phase-to-phase PD, and external noise, allowing PRPD data collection for each phase. It compares the magnitudes of three-phase signals in real time, with the relative peak magnitudes serving as criteria to accept or reject a PD signal. The threshold can be adjusted from 0% to 100%. Field experience has shown that it is especially effective in detecting corona discharge between phases in rotating machinery. The PPDA Cross and Cactus technologies offer a standardized numerical method for assessing insulation health, eliminating the need for expert interpretation. Typically, specialists identify the peak PD magnitude, analyze different PD behaviors based on the applied voltage polarity, examine pattern shapes to locate discharge sources, and assess the overall condition. However, these diagnostic results still rely on expert judgment, which can be a limitation. The PPDA Cross algorithm extracts six parameters from the PRPD pattern in real time. These parameters form the basis of a machine learning system whose results have been proven to accurately identify the source and activity level of partial discharges. Additionally, the PPDA Cactus provides a graphical output based on the same PRPD pattern, which can be interpreted by experts using the initial results from the PPDA Cross.

POWERGRID, India’s central transmission utility, manages one of the largest EHV transformer fleets in the world, which has grown from approximately 782 units in 2010 to 4,030 units by 2025 (about 12% CAGR). The reliability of the power transmission system strongly depends on the safe and reliable operation of transformers. Reliable performance is achieved through proven design criteria and high production quality. Long-term performance of transformers can be ensured with high-quality insulation components. Furthermore, the design and workmanship of insulation materials play a vital role in the trouble-free operation of transformers.

Most problems are identified during factory acceptance tests. Additionally, during site operation, insulation is stressed due to variations in load, temperature, and ingress of moisture. This paper discusses the approach taken by POWERGRID with respect to problems observed in transformer insulation systems and the remedial measures adopted. A few recent case studies are also presented in this paper, related to insulation system failures during factory acceptance tests.

In a 500 MVA, 765 kV transformer, partial discharge (PD) measured was in the range of 7 to 8 nC at 1.2 Um (PD level) on the HV side of the tank. Acoustic sensors indicated the fault location on the outer side of the HV winding towards the auxiliary limb side. It was decided to drain the oil and un-tank the active part for further investigation. Burn marks were observed on the spacer between discs 80–81 of the HV winding. Microscopic imaging after delaminating the faulty outer duct piece, spacer, and outer barrier piece revealed burnt fibers and pinholes. Surface contamination was found to be the root cause of the problem. Necessary rectification measures were taken at OEM works for dust free environment to prevent such failures.

In another case, a 500 MVA, 765 kV transformer failed during a lightning impulse test. Finite element method (FEM) calculation carried out between tap conductors (Regulating winding) with application of LI on IV terminal at maximum tap. With 0.8 mm paper Insulation on regulating winding conductor, actual stress occurring in oil was 19.0 kVrms/mm while maximum allowed stress in oil was (as per OEM standard) 16.7 kVrms/mm.

As a rectification measure, Paper insulation was changed to 2.4 mm. With this insulation, actual stress calculated is 14.0 kVrms/mm which is within the allowed stress in oil of 16.7 kVrms/mm. Regulating winding design was changed to accommodate the paper insulation thickness change.

This paper discusses transformer failures involving insulation systems and the root cause analyses carried out by the manufacturer in collaboration with utility engineers, ensuring that such problems are not repeated in future transformers. As a result of these investigations, significant improvements have been observed in subsequent supplies of 765 kV class transformers in India.

This paper discusses various transformer failures involving insulation systems and the root cause analyses carried out by the manufacturer in collaboration with utility engineers, ensuring that such problems are not repeated in future transformers. As a result of these investigations, significant improvements have been observed in subsequent supplies of 765 kV class transformers in India.

The tri-post insulator is a critical component of gas-insulated transmission lines (GILs). Discharge and even explosion failures caused by internal defects frequently occur, severely threatening equipment reliability. In this paper, a 90-day long term test is conducted for a 500 kV GIL tri-post insulator with an internal defect. The development trend of partial discharge (PD) signals is studied and the degradation of the internal defect is also analyzed. The results show that for the first 84 days, discharge gradually intensified. On the 84th day, discharge became sporadic, with intervals between adjacent discharges lasting several minutes or even a few hours. As discharge develops, the maximum hourly apparent charge first slowly increased from 5 pC to 15 pC, then sharply rose to 50 pC. The discharge repetition rate first increases slightly, from 120 times/s to 190 times/s, and then sharply decreases, with a signal number of less than 40 times per minute. The trend of amplitude and repetition rate changes in ultra-high frequency (UHF) signals is similar to that of pulse current signals. However, the UHF signal repetition rate remained consistently lower than that of pulse current signals. after the 90-day test, the original defect inside the insulator have developed and expanded, with the maximum size increasing from 1.39 mm to 1.89 mm. Moreover, some new voids, cracks, and uneven material defects have also formed inside the insulator. The results of this study provide valuable insights for PD monitoring and diagnosis of GIL insulators in field applications.

Electrical insulation materials often face severe mechanical, electrical, thermal and environmental stresses. Over time, these multiple stresses can deteriorate or cause complete failure of electrical insulation. The insulation materials need to have a high degree of stress tolerance and, in case, the ability to self-heal. Self-healing materials are increasingly explored to enhance the durability and reliability of the insulation material.

At present, there are two types of self healing mechanisms: intrinsic and extrinsic. In extrinsic healing, there are microcapsules embedded in the materials which act as healing agents when stress is applied. In contrast, in the intrinsic healing is triggered under specific conditions due to the reversible dynamic covalent bonds, hydrogen-bond network, or supramolecular interactions within the polymer structure. Intrinsic healing can undergo multiple healing cycles.

In this study, Natural Rubber/Polycaprolactone (NR_PCL) and Natural Rubber/Polyethylene oxide (NR_PEO) blends are prepared in a 75:25 ratio. Natural rubber provides high dielectric strength and elasticity where thermoplastic acts as a healing agent. The compression molding was used to make samples in this dimension (25x25x5mm) for electrical treeing tests.

The NR_PEO and NR_PCL samples have been analyzed for their self-healing capability in the bulk. The samples were preheated up to 90°C for 1 hour to ensure smooth inserting of needles with a low mechanical pressure on the sample. The needles with tip radius of 5µm were inserted in the samples. The 3mm gap between the surface in the sample and the needle tip was maintained for all insertions. The samples are tested in silicon oil at room temperatures. The needles insertion point was covered with silicon sealant to prevent oil penetration inside the samples.

The PD measurement is performed according to IEC 60270 standard. The ramp voltage is increased in the setup till PDIV is reached. PD patterns are acquired at the PDIV and higher voltage is applied for aging. A few of the samples broke down due to aging under high voltage stress.

The optimal healing protocol is applied for all samples. The optimal self healing protocol was previously defined by mechanical scratch testing and recovery at different temperature to find the highest healing efficiency in the least amount of time. The optimal healing protocol for NR_PCL is 110°C for 3 h and 90°C for 2 h for NR_PEO, respectively.

After the breakdown event and the application of the self-healing protocol, both the NR_PEO and NR_PCL samples recover their initial ability to withstand electrical voltage stress, matching the performance they exhibited prior to the first breakdown.

Due to the increase in power demand in the last couple of decades, insulation materials are required to operate under high electrical stress. The application of a high electric field can cause severe damage to insulation materials which can result in degradation or failure. If damaged, the insulation needs to be repaired or replaced, this brings additional costs.

Self-healing materials can be introduced with two methods: intrinsic or extrinsic. Intrinsic self healing is activated by using reversible dynamic covalent bonds or supramolecular interactions which can reform the microstructure. The extrinsic self healing process depends on microcapsules or microvascular fibers which are added to the material as healing agent and released upon rupture to heal damage surface. Intrinsic methods have several advantages over extrinsic ones. They are reversible, easy to manufacture and can perform multiple healing cycles.

Natural rubber has great elasticity and high dielectric strength. Rubbers and thermoplastic polymers can blend together to form thermoplastic elastomers (TPEs). The thermoplastic acts as a healing agent when triggered above its melting points. Thermoplastic elastomers have good chemical resistance, mechanical properties, dielectric performance and recyclability. In this work Natural rubber/Polyethylene oxide (NR_PEO) and Natural rubber/Polycaprolactone (NR_PCL) blends are prepared in ratio of 75:25 in as thin films (~200 µm).

The thin films were subjected to high electrical stress resulting in localized surface discharges on the surface of the samples. In order to activate the healing process, the optimal Self-healing protocol was applied. The optimal self healing protocol was previously defined by mechanical scratch testing and recovery at different temperatures to find the highest healing efficiency in the least amount of time.

Post healing characterization revealed a significant reduction in discharges marks, that confirms a significant restoration for both Natural rubber/Polycaprolactone (NR_PCL) and Natural rubber/Polyethylene oxide (NR_PEO) blends. The results demonstrate that these rubber/thermoplastic systems show strong potential self healing characteristics to be applied for high voltage applications.

Detecting partial discharge in the insulation of high-voltage rotating electrical machines is a fundamental aspect of ensuring the reliability, safety, and longevity of power systems. Partial discharge represents one of the earliest indicators of insulation deterioration, and its accurate detection and classification are critical for preventing catastrophic failures and costly downtime in industrial applications. This study presents a comprehensive machine learning framework for diagnosing insulation conditions by classifying phase-resolved partial discharge patterns obtained from high-voltage rotating machines. The proposed approach integrates ensemble learning algorithms, specifically Extreme Gradient Boosting and Random Forest, which have shown strong capabilities in handling complex data and improving prediction accuracy. Texture-based statistical features were extracted from partial discharge images using the Gray Level Co-occurrence Matrix method, which captures spatial dependencies and distributional variations within the discharge patterns. When only the features derived from the Gray Level Co-occurrence Matrix were used, the classification results demonstrated considerable accuracy, reaching 89.4 percent for the Extreme Gradient Boosting model and 87.15 percent for the Random Forest model. To further enhance model performance, the Gray Wolf Optimization algorithm was incorporated for feature selection, effectively identifying the most significant and discriminative attributes while removing redundant or less relevant ones. The inclusion of this optimization technique led to substantial improvements in model accuracy, achieving 93.7 percent for the Extreme Gradient Boosting model and 94.7 percent for the Random Forest model. These results highlight the effectiveness of combining advanced feature extraction and intelligent optimization with ensemble learning methods for high-precision insulation fault detection. Furthermore, the proposed framework demonstrated strong robustness and efficiency, making it suitable for real-time monitoring and predictive maintenance applications. The findings of this research indicate that integrating Gray Level Co-occurrence Matrix-based texture analysis with Gray Wolf Optimization and ensemble learning classifiers can significantly enhance the reliability of partial discharge detection systems. This improvement directly contributes to more efficient maintenance planning, reduced operational risk, and extended equipment service life. Overall, this study provides valuable insights into the development of intelligent diagnostic tools for high-voltage rotating machinery, advancing the use of machine learning techniques in electrical insulation health assessment and predictive maintenance.

The rapid integration of artificial intelligence into global infrastructure is driving a fundamental transformation in power generation systems, which are increasingly designed to operate at higher voltages and elevated temperatures. This evolution imposes unprecedented demands on electrical insulation systems (EIS), particularly on coating-based insulation materials. Coatings—encompassing both semi-conductive and dielectric formulations—must maintain mechanical and electrical integrity under severe thermal aging conditions to ensure long-term operational reliability.

This paper presents a comparative analysis of established thermal aging standards, including ASTM D2584 and D257, IEEE 117 and 1776, UL 1446, and IEC 60085, with emphasis on their applicability to coating systems. The study identifies limitations in existing methodologies when applied to thin-film and surface-applied insulation technologies. To address these gaps, a refined testing protocol is proposed to more accurately simulate in-service thermal aging behavior of coating materials. The method integrates both mechanical degradation assessment and electrical performance evaluation. Validation will be performed through accelerated aging experiments, with benchmarking against conventional standards.

By advancing thermal aging evaluation methodologies for coating systems, this work contributes to the development of more robust insulation strategies for next-generation power generation applications, thereby enhancing material suitability and system resilience under elevated electrical and thermal stress conditions.

Power transformers are critical assets in power systems, and their reliability depends strongly on the mechanical and electrical integrity of winding insulation and structure. Transformer windings can be affected by internal faults such as short circuits, partial discharges, or mechanical deformations. If such defects fail to be detected at the early stages, they can evolve into catastrophic failures. Therefore, there is a need for accurate diagnostic methods to enable early detection, fault type discrimination, and risk assessment.

Internal fault detection in power transformer windings has been extensively investigated in recent years through both online and offline monitoring of transformers. Frequency Response Analysis (FRA) is one of the most common methods to interpret inter-disk faults. Other methods include differential admittance monitoring and fault-related incremental current techniques, which are useful for fast online protection. Some researchers suggest learning-based methods to detect faults, including feature extraction and machine learning or image processing and deep learning. However, most existing studies either do not use detailed transformer winding models or focus only on low-frequency models, which makes it difficult to simulate and distinguish different types of winding faults, especially high-frequency issues such as those due to partial discharges.

To address this, a multiconductor transmission line (MTL) model has been implemented to simulate internal faults in transformer windings in the frequency domain and the time domain. In this paper, a simulation model for a transformer winding is presented where each turn of the winding is modeled as a conductor of a multiconductor transmission line topology. To solve the resulting coupled transmission line equations, a SPICE-based model is implemented that performs phase-to-mode and mode-to-phase conversions by employing dependent sources. Using this model, two common defects in transformer winding are investigated and identified using FRA analysis and MTL-based model of the winding. MTL-based model is the most accurate model of white box modelling among the other existing ones. The results show that each fault creates a noticeable change in the resonant frequency and admittance curve. This means that the FRA can be used to identify the type and location of internal faults in transformer windings.

The integrity of power transformers is critical for the reliability of electrical systems. These devices can suffer damage from the moment they are manufactured until they are put into operation. Mechanical deformations in the windings are one of these possible defects. The Sweep Frequency Response Analysis (SFRA) technique has proven effective at detecting such alterations. However, it is still difficult to accurately recognize the patterns of change and their location within the winding.

The study investigates the electromagnetic behavior of a power transformer winding by experimentally obtaining its transfer function and its spatial distribution.

Although it is not possible to measure directly inside the windings in real transformers, local variations in inductance and capacitance are reflected in the terminal SFRA response. Therefore, an approach is proposed to infer and visualize, in three dimensions, how internal alterations would modify the local transfer function, thereby relating the patterns observed in the terminal response to the affected areas within the winding.

The study was carried out by applying a low-magnitude, variable-frequency signal (10 Hz to 1 MHz) to an experimental transformer winding. Controlled winding faults were simulated by connecting a capacitor across sections of the winding to emulate local variations in inter-turn capacitance. The output measurements were recorded progressively at different points on the coil, allowing the local transfer function to be obtained on each disc. The amplitude and phase data were processed to construct three-dimensional plots with axes of frequency (Hz), winding position (disc), and amplitude or phase, enabling spatial visualization of the winding electrical behavior. Once these plots were obtained, a classification method was applied to distinguish the winding regions into three zones, where potential faults can be reliably localized.

These 3D representations allow us to observe how variations in capacitance or local mechanical displacements modify the distribution of resonances and how these effects propagate toward the winding ends, affecting the overall SFRA response. The experimentally obtained patterns enable identification of the regions most affected by internal changes, providing information to evaluate the winding condition and analyze fault locations.

The presence of power transformers is essential for the supply of electrical energy; their construction commonly adheres to requirements that depend on the system into which they are incorporated, which can lead to non-conventional designs. In this situation, the commercialization of “resilient” mobile transformers has become popular, with flexible and cost-effective designs capable of operating at different voltage levels.

However, their adaptability makes the winding designs more sophisticated. The transformer’s leakage impedance is a fundamental parameter for design validation and operational diagnostics; although its adaptable nature makes it difficult to estimate. In this context, a reliable calculation method that considers the equipment’s geometric, electrical, and magnetic characteristics is essential.

In this work, a segmented model of the windings was developed using the Finite Element Method, implementing series and parallel arrangements according to the internal design configuration, enabling the accurate calculation of the leakage impedance for all possible connection combinations.

The numerical results are compared with conventional analytical formulas, showing their limitations in the presence of multiple discontinuities and a non-uniform distribution of ampere-turns arising from the switching in and out of series and parallel arrangements; additionally, the model enables a study of the magnetic flux behavior and of the distribution of eddy-current losses in the windings, estimated from the electromagnetic loss density, revealing loss non-uniformities that depend on the reconnectable arrangement (series/parallel).

The model is validated by comparing the numerical results with those from manufacturer experimental tests, achieving a minimum per-unit impedance error of ±2%. This methodology not only provides improved estimates of leakage impedance in non-conventional transformer designs but also, through the multifunctional program developed, enables the study of the distribution of eddy-current losses in the windings. The simulations prove valuable for identifying risk areas and estimating their values, highlighting the arrangements with the highest concentration of losses, which facilitates identifying areas prone to hotspots; they can also be used as a diagnostic tool for evaluating changes in impedance due to local winding deformations.

Partial discharges (PD) in high-voltage power equipment can occur under intense electric fields, resulting in insulation degradation, diminished reliability, and potential failure. Polydimethylsiloxane (PDMS), a silicone-based polymer characterized by a siloxane backbone, exhibits excellent thermal and electrical properties. Upon exposure to PD, oxidation reactions occur on its surface, which may enhance its resistance to its degradation. Alkoxysilane-terminated PDMS can form cross-linked structures through hydrolysis and condensation during thermal curing, allowing for the regulation of network density and enhancement of durability.

This study investigates the influence of molecular weight on the PD resistance of PDMS films made of alkoxysilane-terminated PDMS with different molecular chain lengths. Three distinct PDMS types, characterized by molecular weights of 6k, 20k, and 40k, were applied to silicon wafers, thermally cured at 250 °C for 2 h, and subsequently fabricated into films with a thickness of 150μm. The PD testing utilized an electrode system comprising a spherical electrode, glass plate, air gap, PDMS film, silicon substrate, and metal base electrode. An alternating current (AC) voltage of 240 Hz, corresponding to twice the PD inception voltage, was applied for a maximum duration of 9 h. The discharge energy was assessed through V–Q Lissajous figure analysis, while chemical alterations on the film surface were examined using Fourier-transform infrared (FTIR) spectroscopy both prior to and following PD exposure.

A strong electric field concentrated near the spherical electrode caused localized PD activity. The region directly beneath the center of the spherical electrode was defined as the discharge center, the surrounding luminous zone as the discharge-active region, and the outer area with negligible PD effects as the non-discharge region.

In the case of the 6k PDMS, the FTIR spectra revealed an increase in carbonyl (C=O) peaks, a decrease in methyl (Si–CH₃) absorption, and the emergence of SiOx-related peaks at the discharge center, indicating surface oxidation. These spectral changes diminished progressively with distance from the center and were absent in the non-discharge region, suggesting that degradation was confined to the discharge-active region.

In contrast, the 40k PDMS demonstrated a significant reduction in siloxane (Si–O–Si) absorption at the discharge center, indicating main-chain scission. Evidence of oxidative degradation was observed through increased carbonyl peak. Similar, albeit less pronounced, changes were also detected beyond the discharge-active region, suggesting that low-molecular-weight degradation products diffused and redeposited on the surrounding surfaces. The 20k PDMS exhibited behavior that was intermediate between those of the 6k and 40k samples.

The results indicate that the molecular chain length of PDMS significantly influences both its chemical degradation and the diffusion of degradation products when exposed to PD. Specifically, short-chain PDMS experiences localized degradation, whereas long-chain PDMS undergoes more extensive degradation owing to the diffusion and redeposition of degradation products. Consequently, the molecular chain length is a critical determinant of the spatial distribution of PD-induced degradation. This finding offers valuable insights for the molecular design of silicone-based insulating materials to enhance the durability and reliability of large motors and high-voltage equipment.

Cold sintering (CS) has been shown to improve microstructural homogeneity in ZnO metal oxide varistors (MOVs). A digestive liquid phase is critical for the diffusion of Zn ions during the low-temperature, high-pressure CS densification. This work investigates the impact of the digestive liquid on the microstructure of CS ZnO and current nonlinearity. CS samples were sintered at 300°C, 300 MPa, using either acetic acid, or citric acid. DI water is used as the control group. Scanning electron microscopy (SEM) was used to observe the microstructure while J-E curves were taken to determine the nonlinear coefficients. Here, it is shown that despite ZnO having a lower solubility in acetic acid than citric acid, samples made with acetic acid have superior relative density of 90.4%. Using acetic acid produces a more uniform microstructure with an average grain size of 0.11 µm, and an increased current nonlinear coefficient of 1.48. In comparison, citric acid samples have a relative density of 74%, an average grain size of 0.07 µm, and a nonlinear coefficient of 1.08. Clearly, acetic acid is the digestive liquid phase of choice when cold sintering ZnO to improve MOV energy absorption performance.

Cast resin transformers employ a solid insulation structure, in which internal defects such as voids and delaminations can become major causes of insulation breakdown. Partial discharge (PD) measurement is an effective method for the early detection of such defects; however, technologies for accurately locating PD sources have not yet been fully established. Diagnostic methods for PD detection in cast resin transformers are still under development and often rely on empirical techniques.

Previous studies have investigated PD characteristics in microscopic voids within insulating resin, methods for estimating defect locations, factors influencing the frequency characteristics of PD signals, and the propagation characteristics of PD signals between the winding and ground. Based on these findings and the structural configuration of the voltage transformer (VT), a simulation model was developed.

In this study, with the goal of establishing a method for identifying PD occurrence locations within cast resin transformers, we aim to construct a more detailed high-frequency equivalent circuit model of a 6.6 kV instrument VT. In this paper, the previous simulation model was refined to more closely resemble the actual transformer by increasing the number of turns and adjusting the parameters of the circuit components.

To verify the validity of this high-frequency equivalent circuit model, the analytical results were compared with experimental results obtained from (1) the frequency characteristics of the impedance between both ends of the primary winding in an unsealed VT skeleton model, and (2) measurements of transient currents (simulated PD currents) generated when switches were inserted and shorted between specific layers at multiple positions drawn out from the primary winding of the VT model. These measurement results were compared with the analytical results of corresponding inter-layer short circuits simulated using the circuit model.

As a result, the measured and calculated frequency characteristics of the VT primary winding impedance Z(f) showed good agreement over a wide frequency range up to 200 MHz, demonstrating the validity of the proposed high-frequency equivalent circuit model. Furthermore, the relationships among the switch short-circuit location, inter-layer short-circuit current waveforms, frequency characteristics, and current distribution were clarified. Based on these results, it was shown that the PD occurrence location within an actual VT can be estimated through the detection and analysis of PD current waveforms.

Insulation degradation in medium-voltage (MV) switchgear and cable systems is a major cause of equipment failure and reduced power system reliability. This paper presents a field-based diagnostic assessment of insulation degradation using practical case studies from MV cable and switchgear installations.

The investigation includes partial discharge (PD) activity in cable terminations, switchgear surface tracking under humid conditions, and thermal deterioration caused by improper cable lug connections. Diagnostic techniques including phase-resolved partial discharge (PRPD) analysis, infrared thermography, insulation resistance testing, and visual inspection were applied to evaluate insulation condition and degradation severity.

The results show that PD magnitude increases significantly during insulation deterioration and that PRPD analysis effectively identifies localized defects. Corrective maintenance substantially reduced discharge activity and restored stable operating conditions. The findings demonstrate that integrated condition monitoring improves early fault detection, supports predictive maintenance, and enhances the reliability of MV insulation systems

The challenge to develop more reliable electric motor design solutions capable of operating under stresses induced by the recently introduced new semiconductor technologies increase constantly. This paper continues previous efforts to compare and elaborate methods to identify and quantify PD activities in different motor insulation configurations within highly challenging voltage stress. Approaches for making suitable sensors for on- and offline PD diagnostics are proposed. Specifically, an electrical method based on high-pass filter connection, two different current transformers as well as two antenna solutions are compared when utilized on two types of test objects. Their performance is discussed focusing on strengths, disadvantages and limitations. It is found that the method based on high pass filtering combined with the modified signal processing method has shown the best performance, however the introduced UHF solution performed well also for shielded motor geometry. The results constitute a solid base for insulation system diagnostics and sensor performance.