Optical Acoustic Monitoring of High-Voltage Transformers

D. H. Smith and J. W. Spencer

Introduction

High-voltage transformers are an essential part of high-voltage electric power transmission systems. Their manifestation involves the use of wire windings immersed in electrically insulating oils, and they can be automatically switched to meet power demands using switches called tap changers. Continuous use over long time periods can lead to gradual deterioration, so monitoring them can assist in their maintenance and the avoidance of unnecessary power supply failures. Acoustic techniques have been used for such purposes (Cichon et al„ 2011). However, monitoring such transformers is difficult because of the high voltages involved and the radio frequency active environment in which they operate making conventional electronic methods difficult to use (Warren et ah, 1999; Gungor et ah, 2010). The use of optical techniques based upon optical fibres provides a means for overcoming such difficulties provided suitable sensing approaches can be evolved (Jones and Spencer, 2013).

The use of optical fibre-based chromatic methods has the potential for overcoming such difficulties and for monitoring to be implemented without the need to intrude into the manufactured structure of the transformer (Deakin et ah, 2014). The use of such an approach based upon optical chromaticity of acoustical signals is described, along with results obtained from in-service high-voltage transformers.

Chromatic Monitoring System

Optical sensing is based upon the propagation of monochromatic light from a laser diode being propagated through a length of unjacketed multimode optical fibre attached to the outside wall of a tap changer tank of a transformer (Figure 16.1).

Acoustic signals from the transformer and tap changer mechanism produce small changes in the refractive index of the core and cladding of the fibre, which changes the propagation mode of the monochromatic light in the fibre. The output from the fibre is detected electronically and processed chromatically in the time- or frequency-transformed mode. A schematic diagram of the monitoring system is shown in Figure 16.2.

This system is an advanced form of the system mentioned in the previous chromatic monitoring book (Jones et al., 2008).

High-voltage transformer with optical fibre sensor attached

FIGURE 16.1 High-voltage transformer with optical fibre sensor attached.

Schematic diagram of chromatic monitoring system

FIGURE 16.2 Schematic diagram of chromatic monitoring system.

Chromatic Monitoring Principles

A typical amplitude versus time signal over a period of five minutes obtained from such an optical fibre system addressing a high-voltage transformer is shown in Figure 16.3. The acoustic signal received includes the continuous background transformer operating hum, the mechanical action of the tap changer and other acoustic events generated by the transformer.

Typical time variation of transformer acoustic signal

FIGURE 16.3 Typical time variation of transformer acoustic signal.

A set of chromatic filters (R, G, B) are shown being applied to the time domain signal in Figure 16.3, one set (Rl, Gl, Bl) being applied to the background signal and a second set (R2, G2, B2) being applied to the tap changer signal. The outputs from these filters are chromatically processed (Chapter 1) to yield H, L, S values for each of the two signals. The time interval covered by each of the two sets of R, G, В filters is chosen to be 0.5 seconds (i.e., 25 cycles of the alternating current). The resulting H, L, S values may be displayed in H:L and H:S chromatic maps.

A form of secondary chromatic processing may be deployed by applying a set of R. G. В filters to parts of a longer time interval signal (e.g., 5 minutes), as shown in Figure 16.4a. The filters may then be applied to a series of time intervals during the 5-minute period, each period having values of chromatic parameters FI, L, S. These H. L, S values may then be individually plotted as a function of time, as shown in Figure 16.4b. Applying a second set of chromatic filters to these primary H-S and FI-L parameters yields secondary H. L, S parameters H(t), L(t), S(t) which may be used to provide additional chromatic information.

Test Results and Their Chromatic Interpretation

In situ test results have been reported (Deakin et al., 2014)) for 6 transformer units over an 11-month period during which about 13,000 tap changes occurred. These results show that the chromatic monitoring system enables two main aspects of the transformer condition to be monitored (Figure 16.3), namely the continuous operation of the transformer and the operation of the tap changers. Different forms of chromatic addressing have been described for each of these two aspects.

Continuous Transformer Operation

More details of the normal operation of a transformer (Figure 16.3) are given in Figure 16.5a and b. Figure 16.5a shows a time-varying signal obtained during a normal power-demand condition (e.g., night-time), whilst Figure 16.5b shows a signal obtained during a heavy load operation. Fourier- transformed forms of these two signals show that the heavy load condition (Figure 16.5b) is noisier than the normal load condition (Figure 16.5a). Primary chromatic processing of these two signals yields H:S and H:L chromatic maps, shown in Figure 16.6a and b. These maps show a wide distribution of points for both H:S and H:L and for the low and high power demand conditions. However, the maps suggest a more contracted distribution of both S and L for the high power load condition (Figure 16.6b).

Secondary chromatic processing, (a) Primary chromatic processing of time varying signal, (b) Time variation of primary chromatic parameters H, L, S

FIGURE 16.4 Secondary chromatic processing, (a) Primary chromatic processing of time varying signal, (b) Time variation of primary chromatic parameters H, L, S.

Time varying signals of 5-minute duration, (a) Signal with distinct harmonic peaks, (b) Signal with noisy harmonic peaks

FIGURE 16.5 Time varying signals of 5-minute duration, (a) Signal with distinct harmonic peaks, (b) Signal with noisy harmonic peaks.

Trends in the transformer operation may be displayed in a convenient and less complex manner using secondary chromatic processing. Figure 16.7 shows a graph of the secondary chromatic parameters H2(H-L) versus H2(H-S). This shows results from the low and high noise levels (Figure 16.5a and b) varying linearly on the same locus but with the low noise signals having outliers.

Tap Changer Operation

Operation of individual tap changers may be distinguished from the background transformer signals by following the primary chromatic processing with secondary chromatic processing (Figure 16.4) to yield a secondary H(t): L(t) polar chromatic map. Figure 16.8 presents such a H(t): L(t) polar map for background and tap changer signals. This shows the transformer background results lying at a fixed value of H(t)=0 but with some variation in strength L(t) (consistent with the background results; Section 16.4.1) The results for four different tap changes are also shown - two each with tap up and two with tap down. All tap changer results have a higher value of H(t)= 10-120 than the background results, so tap changing is chromatically distinguishable from the background. In addition, the tap down condition has substantially different values of H(t) compared with the tap up condition.

Primary H-S and H-L chromatic maps, (a) Distinct harmonic peaks, (b) Noisy harmonic peaks

FIGURE 16.6 Primary H-S and H-L chromatic maps, (a) Distinct harmonic peaks, (b) Noisy harmonic peaks.

Secondary chromatic processed results of H2(H-L) versus H2(H-S)

FIGURE 16.7 Secondary chromatic processed results of H2(H-L) versus H2(H-S).

Secondary chromatic map of L(t)

FIGURE 16.8 Secondary chromatic map of L(t): H(t) for tap changer and transformer background signals.

SUMMARY AND OVERVIEW

The test results obtained over a period of ll months with 6 units and 13,000 tap changes have confirmed the viability of the optical chromatic approach for monitoring non-invasively with retrofitting and RF immunity of high-voltage transformers. Overall acoustic transformer operation can be monitored for extended time periods as well as trends in individual tap changer events. The use of primary and secondary chromatic processing of data has extracted useful trend information.

Acknowledgements

The involvement and facilities provided by Electricity North West, UK and Western Power Distribution UK are acknowledged.

References

Cichon, A.. Fracz, P. and Zmarzly, D. (2011) Characteristic of acoustic signals generated by operation of on load tap changers. ACTA phys. PolouicaA, vol. 120. no. 4, 585-588.

Deakin. A. G., Spencer. J. W., Smith. D. H., Jones, D.. Johnson, N. and Jones, G. R. (2014) Chromatic optoacoustic monitoring of transformers and their onload tap changers. IEEE Trans. On Power Delivery, vol. 29, no. 1, 207-214."

Gungor, V. C., Lu. B. and Hancke, G. P. (2010) Opportunities and challenges of wireless sensor networks in smart grid. IEEE Trans. Ind. Electron., vol. 57, no. 10, 3557-3564.

Jones, G. R., Deakin. A. G. and Spencer (2008) Chromatic Monitoring of Complex Conditions.

Jones, G. R. and Spencer, J. W. (2013) Intelligent monitoring of high voltage equipment with optical fiber sensors and chromatic techniques. In: High Voltage Engineering and Testing, Ryan. H. M. (ed.), London, U. K. Chapter 22.

Warren. C. A.. Ammon. R. and Welch. G. (1999) A survey of distribution reliability measurement practices in the U S. IEEE trans. Power Del., vol. 14, no. 1, 250-257.

17 Chromatic Monitoring

Chromatic Monitoring of Arc Electrodes

Z Wang

Introduction

High-voltage circuit breakers (HVCBs) are key equipment to ensure the safety and reliable operation of the power transmission and distribution grid and therefore the reliable supply of electricity.

Switching high currents at high voltages involves mechanically separating two metallic electrodes to form an electric arc plasma which is subsequently quenched. The arcing process can cause the electrode materials to melt and evaporate, causing electrode wear which ultimately leads to the failure of the switch. There is therefore a need to monitor the extent to which the mass of the electrode is being lost during the arcing process.

The behaviour of the arc plasma and highly luminous evaporated electrode material may in principle be monitored by observing in situ the optical spectra of the evaporated material. However, such spectra, which evolve with time, are complex and not easily interpreted. However, chromatic analysis of such spectra can provide chromatic parameters which relate to the mass loss of an electrode during the plasma arcing process.

Instrumentation

A system for obtaining optical spectra of an arc plasma in situ during a current switching operation is shown in Figure 17.1. The system was mainly composed of three optical fibre sensors, one ND filter unit, one high-speed spectrometer (HSS), one photo-diode detector (PDD) unit, one digital oscilloscope, one computer and one main control unit (MCU). The material of the arcing contacts was Cu/W, and the quenching medium inside the arcing chamber was SF6 at a pressure of 1 bar (absolute pressure) (Wang et al., 2017).

Figure 17.1 shows the electrode of an enclosed switch being addressed by the three optical fibres located at 120° to each other around the arcing contact and connected to a high-speed spectrometer. Optical signals were captured from the electrode and plasma during the current-interrupting operation.

By using this optical measuring system, the visible spectra of the arc were captured through the highspeed spectrometer. Meanwhile, the trajectory of the arc root on the plug contact surface was monitored

Schematic diagram of arc between arcing contacts plus optical measurement system (Wang. 2018)

FIGURE 17.1 Schematic diagram of arc between arcing contacts plus optical measurement system (Wang. 2018).

Time variation of trigger pulses, arc current, arc voltage and arcing contact displacement

FIGURE 17.2 Time variation of trigger pulses, arc current, arc voltage and arcing contact displacement: (a) moving contact displacement, (b) arc voltage, (c) arc current, and (d) trigger pulses - (i) trigger for hydraulic mechanism, (ii) trigger for DC current, (iii) trigger for half-cycle AC current and (iv) trigger for dump ignition (Wang, 2018).

quantitatively via the photo diode detector. Figure 17.2 shows the time variation of the switch operation for a half cycle of alternating current plus the separation of the electrode to form the arc plasma.

Test Results

Mass Loss of Plug Contact

Following each operation of the switch, the mass of the plug contact was mechanically measured to yield the mass lost from the electrode due to arcing. The mass loss so determined is shown in

Directly measured mass loss from plug contact as function of peak current (Wang. 2018)

FIGURE 17.3 Directly measured mass loss from plug contact as function of peak current (Wang. 2018).

Figure 17.3 as a function of the peak alternating current (AC) and following each of five operations of the switch at eight different peak currents (5-40 kA).

Time-Resolved Spectra

Typical spectra captured with the system (Figure 17.1) are shown in Figure 17.4 for different peak currents of 5 (i), 20 (ii) and 40 kA (iii). Figure I7.4a shows the time-wavelength variations of approximately ten spectra in the time window from 46 to 56 ms during the positive half cycle of the alternating current. Figure 17.4b shows the wavelength variation at the time of peak current for 5, 20 and 40 kA.

Figure 17.4a (i) shows that the arc spectra at 5 kA were mainly composed of copper line emissions from the contact surface region (i.e., 510.6, 515.3 and 521.8 nm)

Figure 17.4a (ii) and (iii) show that the profiles of the arc spectra for 20 and 40 kA (wavelength domain) were similar to each other but that their time variations were more diverse.

Figure 17.4b (i), (ii) and (iii) show more clearly the extent to which the wavelengths of the arc spectra captured at peak current time for 5, 20 and 40 kA differed.

Thus, Figure 17.4a and b illustrate the complexity of the spectra and their time variation.

Chromatic Processing

Spectra of the form shown in Figure 17.4 were chromatically processed in the wavelength domain (primary chromatic processing) for each time interval to produce wavelength domain chromatic parameters H,v, Lw, Sw, xw, yw, zw (Chapter 1; Jones et al., 2008) using the chromatic processors R,v, Gw, B,v (Figure 17.5a). Each wavelength domain chromatic parameter was then subjected to secondary chromatic processing in the time domain using time domain chromatic processors R,. Gt, B,

Optical spectra for different peak currents of 5 (i). 20 (ii) and 40 kA (iii). (a) Time-wavelength variations, (b) Wavelength variation at peak current (Wang. 2018)

FIGURE 17.4 Optical spectra for different peak currents of 5 (i). 20 (ii) and 40 kA (iii). (a) Time-wavelength variations, (b) Wavelength variation at peak current (Wang. 2018).

Chromatic processors R, G

FIGURE 17.5 Chromatic processors R, G. В superimposed on wavelength and time domain signals, (a) Wavelength domain - Rw, Gw, B„ processors superimposed upon intensity versus wavelength (w) at peak current. (b)Time domain - R,, G,, B, processors superimposed upon xw(t), yw(t) versus time (t).

(Figure 17.5b). This yielded a total of 36 chromatic parameters (Wang, 2018), each representing a different signal feature. Empirical test considerations were then used to select the most sensitive parameters for representing particular signal trends.

Wavelength Domain Chromatic Processing

A primary chromatic analysis was undertaken of the arc spectral emission at the time of peak (tp) for each peak current investigated (e.g.. Figure 17.5a). The processor outputs (Row, Gow, Bow) were converted into six wavelength domain chromatic parameters, Hw(tp), Lw(tp), Sw(tp), xw(tp), yw(tp) and zw(tp) (Wang, 2018; Chapter 1) Two of these parameters, yw(tp) = (GoJ/3Lw(tp), xw(tp) = (Row)/3Lw(tp) [representing relative magnitudes of medium yw(tp) and long xw(tp) wavelength components, respectively] were used to form a chromatic map of yw(tp) versus xw(tp) (Figure 17.6).

The shadow level of each point in Figure 17.6 represents the mass loss of the contact (according to the shadow scale shown in Figure 17.3). This shows two different data trends which correspond to low and high currents.

At high currents (>10 kAp, tests hl-m5), the yw(tp) parameter reduced (0.51-0.43) as the mass loss increased, with a lower trend for xw(tp) (0.37-0.325), providing an approximate indication of mass loss. This trend corresponded to the relative optical emission in the B„ band increasing.

At low currents (5-10 kAp, tests f 1—g5), the data points had a wider spread of xw(tp) (0.7-0.26) at values of yw(tp) between 0.475-0.5. The spectral line emission from the metallic atoms (Figure 17.4) was not predominant, implying that the vaporisation of contact material was less prominent.

Time Domain Chromatic Processing

The primary chromatic parameters [xw(t), yw(t)] at various times (t) during a current half cycle were each addressed by three time domain chromatic parameters (R,. Gt, Bt) as shown in Figure I7.5b. (These were empirically chosen from the possible 36 secondary chromatic [time domain] parameters; Wang, 2018.) The outputs from R,. Gt, B, were then used to determine the effective spectral intensity LUw and nominal time domain spread 1 — S, , w'hich were displayed on a time domain chromatic map of L, ,w versus 1 — Styw (Figure 17.7). This map showed a linear trend locus for the two parameters which distinguished various levels of contact mass loss (Figure 17.7). Consequently, it led to two chromatic calibration graphs 1 - S, versus mass loss; L,Jw versus mass loss (Figure 17.8a and b) which showed that 1 - S, decreased linearly w'ith mass loss, whilst L, lw increased linearly wuth mass loss.

Primary chromatic parameters x(t) versus y(t) (Wang, 2018) (mass loss indicated by shadow on each point)

FIGURE 17.6 Primary chromatic parameters xw(tp) versus yw(tp) (Wang, 2018) (mass loss indicated by shadow on each point).

Secondary chromatic parameter (time domain) map of L, versus 1 - S (Wang. 2018)

FIGURE 17.7 Secondary chromatic parameter (time domain) map of L,Jw versus 1 - S(yw (Wang. 2018).

Secondary chromatic parameters (time domain) as a function of mass loss, (a) Effective spectral intensity L and (b) nominal spread (1 - S) (Wang, 2018)

FIGURE 17.8 Secondary chromatic parameters (time domain) as a function of mass loss, (a) Effective spectral intensity LUw and (b) nominal spread (1 - Styw) (Wang, 2018).

Summary

Time-resolved arc spectra processed using chromatic methods applied in both wavelength and time domains have been shown to provide a good correlation w'ith arcing contact mass loss.

Wavelength domain chromatic parameters [representing relative magnitudes of medium yw(tp) and long xw(tp) wavelength components, respectively] of the arc spectra at the peak current time showed a dependence upon the mass loss of an arcing contact. Selected wavelength domain chromatic parameters reflect certain characteristics of the arc spectra w'hich are directly related to the interaction between the arc and arcing contact.

Time domain (secondary) chromatic parameters (representing the effective spectral intensity LtJw and nominal time domain spread l - S, .) have been shown to have features which improved the prediction of arcing contact mass loss and w'hich therefore are more suitable for online monitoring of arcing contact erosion. Both selected time domain chromatic parameters (L, lw, 1 - Styw) show- strong linear correlation with the mass loss of an arcing contact.

The procedures followed to produce chromatic results from the spectral data are summarised in Figure 17.9.

Flow chart for processing spectral data using primary and secondary chromatic methods (Wang, 2018)

FIGURE 17.9 Flow chart for processing spectral data using primary and secondary chromatic methods (Wang, 2018).

References

Jones, G. R., Deakin A. G. and Spencer. J. W. (2008) Chromatic Monitoring of Complex Conditions. CRC Press, ISBN 978-1-58488-988-5.

Wang. Z. (2018). Study on the Mechanisms and Prediction Methods of Arcing Contact Erosion of High-Voltage SF6 Circuit Breaker, Ph.D. Thesis. University of Liverpool.

Wang. Z., Jones, G. R.. Spencer. J. W„ Wang. X. and Rong, M. (2017) Spectroscopic on Line Monitoring of Cu/W Contacts Erosion in HVCBs Using Optical Fibre Based Sensors and Chromatic Methodology. Sensors, 17 (3), 519.

 
Source
< Prev   CONTENTS   Source   Next >