Alistair Reid

Partial discharge current pulses in SF6 and the effect of superposition of their radiometric measurement

The practical advantages of employing non-contact radio frequency (RF) methods for detecting partial discharges (PDs) in high voltage equipment have led to significant effort being focused on the diagnosis of electrical plants using RF techniques. This has particularly been the case for gas insulated substations, which use sulphur hexafluoride (SF6) as an insulating medium. One of the most important challenges facing RF diagnostics is the problem of relating the RF emissions to some measure of severity of the PD. Previous work has established that the amplitude or energy of RF signals radiated from a PD source is strongly dependent on the rate of change of current in the PD pulse. In this paper, measurements of PD current pulses in SF6 are presented for a point-plane configuration using an extremely wide bandwidth (13 GHz) measurement system. By this means, PD pulse shapes have been recorded with better resolution than has previously been possible and rise times have been measured with a high degree of accuracy. The results show a considerable variation in pulse shape, with the minimum rise time measured being 35 ps. With this high time-domain resolution, we have been able to distinguish features within the PD pulses that will affect the energy of the radiated RF signal. In particular, the current pulses tend to occur in bursts of up to ten individual pulses in as little as 1 ns, which will excite multiple RF signals in rapid succession. The effect of superposition of RF waveforms has been investigated by studying the variation in detected RF energy with respect to the time delay between PD pulses. It was found that when two PDs occur within a short period (< 150 ns) the combined energy of the resulting RF pulse has the potential to vary by ±30% of that resulting from two equivalent PD pulses with a wider pulse spacing ( 150 ns). In terms of a practical monitoring system concerned with order-of-magnitude variations; this is not considered to pose a major problem for the RF technique.

Sensor response characteristics for UHF location of PD sources

Ultra-high frequency (UHF) systems are well established for monitoring partial discharges (PD) in gas-insulated substations (GIS) and are increasingly being used on power transformers. Different sensor designs such as disk sensors, spiral sensors, monopoles and dipoles have been used depending on the application. A primary concern is to ensure that the sensitivity of these different sensors can meet the CIGRE requirement for detecting 5 pC partial discharges GIS. Recent work on locating PD in power transformers has focused attention on recording the arrival time of the received UHF signal, which is important for locating insulation defects in three dimensions. An accuracy of 1 ns or better is desirable for time-of-flight measurements to locate a PD source. In some situations, a clear arrival time for the signal can be identified but in other cases the arrival time can be uncertain within a tolerance of several nanoseconds. Therefore, a number of sensors were tested for their speed of response using a transient calibration system. Results suggest that more complex sensor structures have a slower build-up of energy at their output, while simple sensors exhibit less of a dasiablurringpsila effect on the leading edge of the signal.

FDTD modeling of UHF partial discharge sensor response

Ultra high frequency (UHF) partial discharge sensors are valuable tools for condition monitoring of high voltage equipment. This paper explores the effectiveness of the Finite Difference Time Domain (FDTD) numerical modelling technique as a means to support the design and optimisation of these sensors. The responses a monopole, a planar logarithmic spiral and a novel helical spiral sensor to a step electric field have been obtained both empirically, using a gigahertz transverse electromagnetic test cell, and through software modelling by means of the FDTD technique. Results indicate that FDTD modelling can achieve a reasonably accurate prediction of UHF partial discharge sensor response provided specific environmental electromagnetic conditions are met and the model geometry is an acceptable trade-off between accuracy and simulation time. This software-based approach may prove a useful additional tool in UHF sensor design process.

Frequency distribution of RF energy from PD sources and its application in combined RF and IEC60270 measurements

Using an integrated measurement system, partial discharge activity has been recorded simultaneously in terms of radiated energy and IEC60270 apparent charge. Correlation between these two quantities can yield important information regarding the shape of the underlying PD pulse. It is proposed that this new measurement strategy can be further enhanced by additionally calculating the relative spectral energies of the measured RF signals. The spectral energies of radiated RF PD pulses vary depending of the topology of the defect source and therefore may serve to enhance the quality of diagnostic information. Results indicate that the frequency distribution of RF energy measured simultaneously along with apparent charge could prove useful as an additional means of defect identification and characterisation.

Investigation of radiometric partial discharge detection for use in switched HVDC testing

This paper reports on initial trials of a non-contact radio frequency partial discharge detection technique that has potential for use within fast switching HVDC test systems. Electromagnetic environments of this type can arise within important electrical transmission nodes such converter stations, so the methods described could in future be useful for condition monitoring purposes. The radiometric technique is outlined and the measurement system and its components are described. Preliminary field trials are reported and results presented for a discharge detected in part of the HV test system during set-up for long-term testing of a reactor. The calculated and observed locations of the discharge were in agreement to within 60 cm inside a test housing of diameter 5 m and height 8 m. Techniques for improving the location accuracy are discussed. The issue of data volume presents a considerable challenge for the RF measurement techniques. On the basis of observations, strategies for moving towards automated interpretation of the partial discharge signals are proposed, which will make use of intelligent software techniques

Fault location and diagnosis in a medium voltage EPR power cable

This paper presents a case study on fault location, characterization and diagnosis in a length of shielded 11 kV medium voltage ethylene-propylene rubber (EPR) power cable. The defect was identified on-site as a low resistance fault occurring between the sheath and the core. A 43 m section was removed for further analysis. The fault resistance was characterized and the location of the defect pinpointed to within a few cm using a combination of time-difference-of-arrival location and infra-red imaging. A combination of X-ray computed tomography, scanning electron microscopy and energy dispersive X-ray spectroscopy were then applied to characterize any abnormalities in the dielectric surrounding the breakdown region. A significant number of high density contaminants were found to be embedded in the dielectric layer, having an average diameter of the order of 100 μm, a maximum diameter of 310 μm and an average density of 1 particle per 2.28 mm3. Scanning electron microscopy and energy-dispersive X-ray spectroscopy were used to determine the geometry and elemental composition of some initial contaminant samples. It was concluded that contamination of the EPR layer, combined with an observed eccentricity of the cable¿s core and sheath resulting in a reduced insulation gap, may have led to an electric field concentration in the region of the defect sufficient to initiate breakdown. Preventative strategies are discussed for similar families of cables, including more stringent dielectric testing requirements at the manufacturing stage and PD monitoring to detect incipient failure.

Simultaneous measurement of partial discharge using TEV, IEC60270 and UHF techniques

This paper explores a means of quantifying the relationship between transient earth voltage (TEV), ultra-high-frequency (UHF) and IEC60270 partial discharge (PD) measurements on a pulse-by-pulse basis for well-defined, laboratory-based discharge sources. Since each technique responds differently to the same PD event, there is no theoretical relationship between the measured quantities; TEV and UHF techniques respond to the rate of change of charge movement, while IEC60270 responds to its integral. Empirical measurement is therefore necessary. Discharge pulses for each system were captured simultaneously using a 3 GHz, 20 GS/s digital sampling oscilloscope. PD amplitudes recorded by the respective systems were analysed and their relationships evaluated. Although these relationships are complex and bear characteristics particular to the PD geometry, the order-of-magnitude variation in amplitude between the sources allows approximate linear interpolation of the relationships when pulses are plotted on a logarithmic scale. The TEV/IEC correlation approximated to 1 μV/pC. The TEV/UHF correlation approximated to 0.05 mV/mV (TEV/UHF). Using TEV data measured at various points at an on-line medium voltage substation, approximate corresponding pC levels have been estimated based on the above relationships. The implications for PD severity judgment are discussed.

Energy scavenging technique for powering wireless sensors

A capacitive scavenging device is proposed to extract energy from the electric field that surrounds the equipment in a substation. Related research has shown that switching techniques can be used to increase the useful power obtained. An experiment has been set up by putting two parallel plates of 10 cm radius in a uniform AC electric field up to 20 kV/m. A mechanical reed relay was used at this stage to provide a high dasiaOFFpsila resistance. Tests were carried out by switching the relay at different frequencies and varying the length of the switching pulses. Initial results show that introducing switching could improve the scavenging efficiency compare with the basic linear circuitry. The proposed device could enable simple sensors to be energy self-sufficient, opening the way for wireless sensors to be installed widely in the power grid.

Identification of Multiple Defects in Solid Insulation Using Combined RF and IEC60270 Measurement

Using an integrated measurement system, partial discharge activity has been recorded simultaneously on a pulse-by-pulse basis in terms of radiated radio-frequency (RF) pulse energy and the apparent charge measured according to IEC60270. Plotting the energy of the RF pulses against the corresponding apparent charge produces a correlation that will be sensitive to the shape of the underlying current pulse, which would be expected to vary depending on the type and severity of the defect. This paper presents combined measurements for several resin insulation samples containing a number of voids of known dimensions. Results indicate that using a combined measurement strategy, it may be possible to discriminate between discharge activity occurring in each void based on the correlation between apparent charge and RF energy.

Comparing IEC60270 and RF partial discharge patterns

Defect classification using partial discharge (PD) measurement has historically been achieved by representing the data in phase-resolved form. This paper presents phase-resolved PD patterns obtained using radio frequency (RF) and conventional IEC60270 techniques for a number of different defects including free particles, protrusions, floating electrodes and voids. Oil, SF6 and resin were used as insulating media. Each PD geometry was placed inside an aluminium test chamber fitted with an internal UHF antenna. Around 2000 RF and IEC PD pulses have been recorded simultaneously along with their phase position for each PD source configuration using a 3 GHz digital oscilloscope. Results indicate that due to the responses of each system, clear differences exist in the phase-resolved patterns, implying that more accurate defect classification should be achieved if a system is trained exclusively using either IEC or RF data. In other words, if a classification algorithm has been trained using measurements from a conventional IEC system, the accuracy of diagnosis will be reduced if RF sensors are then used to measure on-line data without re-training the system.