Epoxy resin (EP) is a typical and widely used insulating material for medium-frequency transformers (MFTs). The dielectric loss of EP is one of the key variables in the design and operation of MFT. In this article, six types of epoxy/anhydride systems are prepared, and the dielectric loss of MFT is calculated by finite element method (FEM) analysis based on broadband dielectric spectroscopy measurements. The material dependence of the dielectric loss of MFT is investigated using dielectric spectroscopy and molecular dynamics simulations over a wide temperature and frequency range. The results show that the variation of dielectric loss with output power is attributed to the temperature characteristics of the imaginary part ( $\varepsilon ^{\prime \prime }$ ) of complex permittivity. Compared to the dielectric loss at no-load state, the one at rated power is reduced by 5–10 W. The anhydride hardener containing benzene ring or a methyl substituent has an inhibitory effect on dielectric loss, and the EP cured by methyl nadic anhydride (MeNA) has the lowest dielectric loss (16.2 W) at rated power. The dielectric loss and overload characteristics of MFT are closely related to the $\beta $ -relaxation and $\alpha $ -relaxation of EP, respectively. The acid anhydride curing agent containing benzene ring or a methyl substituent reduces the temperature corresponding to the $\beta $ -relaxation peak by more than 50 °C compared to the one without. The increased rigidity of the chain segment of EP makes the $\beta $ -relaxation move to the low-temperature side, which is beneficial to suppress the dielectric loss. This article not only serves as a bridge among dielectric loss of MFT, dielectric spectroscopy, and molecular structure of EP but also lays a solid foundation for the selection of insulating materials for MFT.

Operating experience shows that residual DC voltage remains in GIS bus bar for a long time after opening of disconnectors. This DC voltage will not only cause accumulation of charges and contaminants on the insulator surface, but will also easily lead to motion of free metallic particles in SF6 gas gap, weakening the insulation ability of GIS. At the time of disconnectors reclosing, impulse voltage could be generated in the bus bar and superimpose on the pre-existing DC voltage. As a result, GIS will be under combined voltage of DC and impulse. This paper experimentally studies the insulation characteristics of GIS under combined voltage of DC and lightning impulse. It is found that, for non-contaminated insulators, when DC and LI voltage are of the same polarity, flashover voltage under combined voltage is nearly the same as the flashover voltage under LI alone. However, when they are of the opposite polarity, flashover voltage decreases with increasing DC voltage. In cases there are metallic particle or powder around insulator, pre-stressed DC voltage will cause adhesion of contaminants on the insulator surface and lead to decrease of flashover voltage. Specially, for the particle-contaminated insulator, it is found that the surface charge accumulation caused by DC voltage will have big influence on the flashover voltage, and the most critical condition is when DC and LI voltage are of the opposite polarity. In addition, the breakdown characteristics of SF6 gas gap with free metallic particle are investigated. It is found that under pre-stressed DC voltage, particle will do standing motion or bouncing motion. And breakdown voltage becomes lowest when particle is doing standing motion. The results of the experiments show that residual DC voltage in GIS could deteriorate its insulation ability and the combined voltage of DC and impulse is found to be a critical condition for GIS insulation. Moreover, the combined voltage is shown to be more sensitive to detect some insulation defects in GIS than applying impulse voltage alone, especially for free metallic particles, which are commonly found in GIS but up to now no effective detection method is proposed. Therefore combined voltage of DC and impulse could be considered to serve as a supplement of field test for GIS.

Surface charge accumulation on insulators is one of the bottlenecks for the development of DC electric power system. This paper studies a ±200 kV DC basin-type insulator designed for GIL (Gas-insulated Transmission Line). A surface charge measurement device is made consisting of a 3-dimension, 4-axes manipulating device, a HV shielding conductor installation system and data acquisition system. A new scaling method of the probe with an inversion algorithm using Cholesky decomposition method is proposed to obtain charge density distribution. The dynamic process of surface charge accumulation and dissipation are studied under DC voltages with different polarities, amplitudes and time durations. Charge distributions are obtained and theoretically analyzed. A novel gas-side normal electric field model is proposed for regular charge distribution and methods of suppressing surface charge are suggested, including a recommended ratio of surface and volume electric conductivity of the insulator at 1~10, structure design, and direct fluorination of the surface.

It is generally accepted that the presence of bubbles severely degrades the electric strength of liquid dielectrics, but the influencing factors and mechanisms have yet to be elucidated. Dynamic behavior, partial discharges (PDs) inside bubbles, and their effects on the breakdown of transformer oil are investigated experimentally and theoretically in this study. The results show that, under a nonuniform electric field, the presence of bubbles does not significantly affect the PD inception voltage (PDIV) and breakdown voltage (BV) of the liquid, as bubbles migrate to the region with lower field strength under dielectrophoretic force. In a uniform electric field, conversely, bubbles greatly reduce PDIV and BV; an increase in the number of bubbles or a slower voltage rise rate appears to exacerbate this effect. The PDIV observed in the experiment is much higher than the theoretical value calculated by the Townsend theory or streamer criterion, which may be due to the lack of free electrons inside the bubble, thus resulting in scarce internal discharge. The addition of X-rays enhances the radiation ionization and then eliminates the inconsistency between the theoretical and measured values of PDIV, which supports our hypothesis. The bubble-triggered breakdown of liquid dielectrics, namely, once a discharge occurs inside the bubble, likely triggering the breakdown of the liquid gap, is observed. The electric field enhancement caused by surface charges and the instability of the gas–liquid interface may account for this.

Contamination of free conducting particles is a major threat to gas insulated systems. In HVDC GIL, free conducting particles can be even more harmful due to the fact that particles can easily cross the gas gap under DC voltage and distort the electric field. In HVDC systems, impulse overvoltage can be generated and superimposed on the DC operating voltage. As a result, GIL will experience superimposed DC and impulse voltage. In this paper, the behavior of free conducting particles in HVDC GIL is observed and breakdown characteristics of particle contaminated GIL under superimposed DC and impulse voltages are studied. Two motion patterns of free conducting particles in HVDC GIL are mainly found: bouncing and firefly motions. Particles tend to firefly motion near the enclosure under positive DC voltage and near the high voltage bus bar under negative DC voltage. With increased DC voltage amplitude, the probability of firefly motion increases while the probability of bouncing motion decreases. It is found that the insulation strength of GIL is greatly reduced by free conducting particles under superimposed DC and impulse voltages. With increasing particle length, the breakdown voltage of GIL under superimposed voltage decreases sharply. Breakdown voltage decreases slightly with the increase of particle radius. Breakdown voltage of GIL is lower under superimposed voltage of negative DC and positive impulse than under negative DC and negative impulse with the same size of particle. The results indicate that superimposition of DC operating voltage and impulse overvoltage can be a critical condition for HVDC GIL if free conducting particles exist. Therefore, a superimposed DC and impulse voltage on-site test can be considered for HVDC GIL to check for insulation performance.