EARLY DETECTION OF WEAK POINTS IN MEEC ELECTRICAL POWER NETWORK

Transcription

1 EARLY DETECTION OF WEAK POINTS IN MEEC ELECTRICAL POWER NETWORK M. Abdel-Salam, Electrical Engineering Dept., Assiut University, Assuit, Egypt S.Abdel-Sattar, Electrical Engineering Dept., Assiut University, Assiut, Egypt. Y.Sayed, Electrical Engineering Dept., Menia University, Menia, Egypt. M. Ghally, Middle Egypt Electricity Company, Menia, Egypt. INTRODUCTION Not only the pollution of the transmission-line insulators but also other types of weak points in the electrical power network of MEEC, "Middle Egypt Electricity Company" are investigated. These include poor connections, loose hardware with subsequent arcing, polluted insulators with subsequent tracking "baby arcing" and corona at micro-roughness (sharp edges) on line conductors and insulator hardware, where air ionization occurs to form a blue or purple glow. Measurement of ultrasound emissions from the above-mentioned weak points is an avenue to detect serious weak points in MEEC electrical power network such as arcing, tracking and corona. Thus, these emissions can be utilized to warn of impending failure, equipment damage and supply interruption. Once the weak points are identified and located, they are easy to quiet. This is achieved by hot washing of lines and insulators, short-circuiting the gaps by better bonding or tightening the connections, and by smoothening the coronating points to suppress corona activity. This will reduce the outage time and add incentives to the Revenue Recovery Program sought by MEEC administration. In this paper, the expected weak points in MEEC electrical power network including sharp edges, polluted insulators and loose contacts were simulated and stressed in the laboratory by AC voltage. Discrimination among these weak points was pursuded through three different approaches; by listening to the sound pattern emitted from each weak point; by performing a frequency analysis of these patterns and by recording the patterns on an oscilloscope. The last approach was extended in field (outdoor) testing for discrimination between corona at sharp edges on the conductors and baby arcs on the suspension insulators of a 33-kV transmission line, one of the MEEC electrical power network. The results obtained in the field are correlated with those recorded in the laboratory. All the obtained results are discussed in the light of gas-discharge physics. II. METHODOLOGY When electricity "jumps" cross a gap forming an arc or corona in an electrical connection, it disturbs the air molecules around it and generates ultrasound. The tool for such sound measurements is ultraprobe 2000 [1], which detects ultrasonic frequencies between 20 khz and 100 khz and convert them to 100 Hz to 3 khz audio. II. 1. Ultraprobe Circuit The ultrasound is detected by receiving crystals and amplified by a preamplifier with a variable gain to allow for a wide dynamic signal range. The heterodyned signal is then routed to an audio amplifier to drive headphones and to produce a conditioned output suitable for further signal processing, such as spectrum analysis and signal recording. The signal amplitude is displayed in numeric Decibels. II. 2. Recording circuit for acousticsignal Fig. 1: Acoustic-signal recording system in a block diagram The components of the system for recording acoustic signals are: 1-Ultraprobe. 2-Headphone jack. 3-Storage oscilloscope. 4-Interface card. 5- Cable 132 with two end jacks: 25-pin jack to fit in the oscilloscope and 9-pin jack to fit in the computer (in place of mouse). 6-Personal computer. 7- Data cable with two 25-pin end jacks for connection between the computer and printer. 8- Printer. While it is possible to discriminate among arcing, tracking ASU_AbdelSalam_A1 Session 1 Paper No

2 or corona by the sound pattern, there can be occasions where it may prove confusing [2]. Since the Ultraprobe instrument heterodynes ultrasound down into the audible range, the headphone jack may be used to observe these sounds in acoustic signals on an oscilloscope, Fig. 1. To record these sounds, a personal computer and a printer were used as shown in Fig. 1. stage of discharge before "flashover" is arcing. The sound of arcing is erratic bursts of discharge without any steady state buzzing. III. RESULTS AND DISCUSSION III. 1. Accuracy and factors affecting probe reading 1- Voltage applied to the weak point: The readings increase with the increase of the applied voltage. 2- Distance from the weak point: The readings are reduced with the increase of the distance. 3- Operating frequency of the probe: The optimum frequency corresponds to the maximum reading. 4- Type of the weak point and the emitted audible sound. 5- Size of the weak point: The reading depends on the size (as the size of the point increases, the reading increases and occurs at a lower voltage and a larger distance). 6- Direction of probe: Ultraprobe should be directed to the weak point. For sharp points on transmission lines, this calls for scanning of line conductors from all directions. 7- Noise surrounding scanning region: Any noise recived by the ultraprobe can cause deviation of its reading from the correct value. III. 2. Laboratory (Indoor) testing Different weak points including sharp edges, polluted insulators and loose contacts were simulated in the laboratory, Fig. 2. Precautions have been made to make sure that the high-voltage circuit is free from partial discharges by having all circuit connections made from thick straight conductors. III Weak-point discrimination by sound Fig. 2: Laboratory simulation of different weak points (a) HV conductor with loose contact posted on insulating supports Conductor diameter = 1 cm Conductor height above ground plane = 30 cm (b) Post-type polluted insulator standing over a ground plane. Insulator height = 25 cm Insulator diameter = 10 cm (c) HV conductor with sharp edge posted on insulating supports. Conductor diameter = 1 cm Conductor height above ground plane = 30 cm Sharp-edge diameter = 1mm. These sounds have been recorded on a recorder tap and one can listen to these sounds to sense the difference. There is a difference in sound between sharp-edge corona at the one hand and baby arcs and loose-contact arcing on the other hand. However, discrimination between baby arcs and loose contact arcing is sometimes difficult and the sound pattern becomes confusing. This calls for frequency analysis of these sounds after being recording on a tap and / or recording the acoustic signal appeared at the headphone jack of the probe using an oscilloscope, Fig. 2. Such detection and discrimination between the different weak points are aimed at warning against impending failures, equipment damage and supply interruption in MEEC electrical power network. It is considered the main discrimination approach, where each weak point has different sound radiation. III. Corona 2. 2.Weak point discrimination by occurs at sharp edges [3-6] and the associated sound is a frequency analysis steady state buzzing. The next stage is tracking, and this is where there is a low current pathway to ground across an insulator. This is a combination of the buzzing sound with Analysis of the sound signals after being recorded (on a tape) for loose little "popping" sounds (baby arcs) mixed in. These popping contacts, sharp edges and polluted sounds are caused by a slow buildup of charge and a rapid insulators was made in the frequency discharge across the "pollution" on the insulator. The last ASU_AbdelSalam_A1 Session 1 Paper No

3 domain using MATLAB computers program as shown in Fig. 3. As the sound power is proportional to the square of the sound pressure [6]; thus Sound-pressure level in db = 10 log p 2 /p o 2 =20log(p/p 0 )db Where P is the sound-pressure level being measured and P 0 is the reference sound pressure of µbar, which is the minimum level that an "average" person can detect at 1000 Hz. edge. With excessive increase of the applied voltage, Trichel pulses appear again in on the negativegoing of each negative half cycle, Fig. 5, with amplitudes smaller than those on the positive half cycles. Figure 6 shows the acoustic signal observed on the scope with polluted insulator as a weak point. The insulators were polluted by being sprayed by salty water. The signal appears in-groups of pulses displayed around the peak of the successive half cycles with a time span (10ms) between successive groups. This is because dry-band flashover occurs every half-cycle of the applied voltage. Each group of pulses has large number of pluses in comparison with that of the sharp-point. The number of pulses depends on how dense the water spots on the insulator surface. Figure 7 show the acoustic signal watched on the scope with loose contact as a weak point. As shown, the signal has different pattern than other signals, and one can easily define them on spot being large in number, and almost continuous. Positive Positive Fig. 4: Acoustic-signal recorded for sound emission from a sharpedge weak point stressed by 29kV. Fig. 3: Frequency analysis of sound pressure level (db above µ bar) (a)loose contacts (b)sharp edges (c)baby arcs on insulator. Negative Negative It appears that the pattern of each case (loose contacts, sharp edges, baby arcs) is different, Fig. 3. As shown the peak pressure-level occurs at certain frequency depending on the weak point. The pattern in loose contact has a peak pressure-level higher than that of the baby arcs on the insulator, while the baby arcs have higher peak pressurelevel than that of the sharp edge. Thus, the sharp edges, polluted insulators and loose contacts are progressively ranked as regards their sound pressure level. III Weak-point discrimination by acoustic signal recording Figure 4 shows the acoustic signal seen on the scope with sharp-edge as a weak-point. A sample of the applied voltage through a voltage divider was displayed on the scope (not shown in Fig. 4). The acoustic signal appears in of pulses with a time span (20ms) between the. This is because the positive corona streamers appear first to generate the audible noise. These streamers occur in on the positive-going of every positive half cycle. Each packet has smaller number of pluses. The intensity of sound and amplitude of pulses depend on the size of the sharp Fig. 5: Acoustic-signal recorded for sound emission from a sharpedge weak point stressed by 60kV. Fig. 6: Acoustic-signal recorded for sound emission from polluted insulator as a weak point. III Discussion As the AC voltage applied to a transmission-line conductor at a given height above the ground plane reaches a critical value, the electric field at the conductor surface becomes sufficiently high enough to ionize the air in the immediate vicinity of the conductor. The corresponding value of the electric field is called the onset field, depending on the polarity. Corona starts when the electric field exceeds the ASU_AbdelSalam_A1 Session 1 Paper No

4 onset value, and charges of the same sign as that of the conductor potential are emitted into space and move away from the conductor in the form of shells. The charges in a given shell raise the field in the space outside the shell, thus absorbing a part of the applied voltage, while reduce the field at the coronating conductor so emission is regulated in such away that any increase in the space charge due to a new emission causes a reduction of the surface field and therefore a slowing down of emission. This means that the surface field of the coronating conductor doesn t change significantly from its onset value. Fig. 7: Acoustic-signal recorded for sound emission from loose contact as a weak point. The corona onset voltage V 0± is the applied voltage at which corona starts in the first positive or negative a half cycle. In the succeeding half cycles, corona starts at voltages V 1± <V 0± due to the effect of the residual space charges [7]. V 1± are termed as ionization onset voltages. It has been assumed [8] that the ionization during each cycle is terminated at the voltage peak (V p ) and starts at voltage V 1± determined by a simple relation III Weak-point discrimination by sound The sound recorded on a tape at different distances from line conductors and insulators was so weak to be listened. III Weak-point discrimination by acoustic signal recording Figure 9 shows the acoustic signal seen on the scope when the ultraprobe was directed towards the conductors of the 33-kV Abu-Teeg Bokharia transmission line. The sinusoidal waveform shown in Fig. 9 was obtained from a built-in generator in the scope. The waveform is positioned on the scope so that the positive and negative pulse of the acoustic signal terminate at the respective peak of the successive half cycles in conformity with the discussion reported in section III As observed in the laboratory testing, the signal appears in of pulses with 10 ms time span between the. The occur on the positive-going and negative-going of the respective positive and negative half cycles, Fig. 10, in conformity with those in Fig. 5 for indoor testing. V P -V 0± =V ± -V 1± Fig.9. Accoustic-signal recorded for sound emission from 33-kV Abu Teeg- El Bokharia transmission-line conductors (at mid span between towers). Fig. 8. Instantaneous applied voltage showing where corona appears during positive and negative half cycles Thus, the corona during the positive and negative half cycles occur before the respective peak of the half cycle as shown in Fig. 8. This is agreement with the observed experimental finding of Fig. 5. Moreover, the corona terminate at the peak of the successive half cycles. III. 3. Field (Outdoor) Testing Field testing was conducted underneath conductors and towers of 33-kV Abu-Teeg Bokharia transmission line, one of the overhead lines of MEEC electrical power network. Fig. 10. Accoustic-signal recorded for sound emission from suspension insulators of tower # 54 of 33-kV Abu Teeg- El Bokharia. Figure 10 shows the acoustic signal seen on the scope when the ultraprobe was directed towards the suspension insulators of tower # 54 of the 33-kV Abu-Teeg Bokharia transmission line. In Fig. 10, the sinusoidal waveform remains positioned as in Fig. 9 to serve as a reference for correlating the phase relationship of the pulse in Fig. 10 with that in Fig.9, when the probe was directed to the line conductors. As observed in the laboratory testing, the signal appears in of pulses with 10 ms time span between the. The are almost centered ASU_AbdelSalam_A1 Session 1 Paper No

5 around the peaks of the positive and negative half cycles, Fig. 10. III Discussion Transmission-line audible sound has two characteristic components, namely: (1) broadband component (variously described as frying, or crackling, or hissing) and (2) puretone components at frequencies of 100 Hz and multiples [4,5]. The pure tones are superimposed on the broadband component. The most noticeable tone is the 100-Hz "hum". The broadband component is caused by a random sequence of pulses produced by partial discharges (corona) in the air at the surface of the transmission-line conductor. Among the corona modes produced by an alternating voltage, the most important with regard to the audible-sound generation is the positive-polarity streamer [6]. These streamers occur in on the positive polarity of every cycle, Fig. 4 and consequently 50 Hz and higher harmonic components may be present in the frequency spectrum. At each point where streamer is produced, a point-source pressure wave is generated and propagates into the surrounding space. These waves are produced in different locations along the conductor and at different times, and the acoustic energy arriving at any location is randomly distributed over the entire cycle. The spectrum extends to frequencies above the sonic range forming the broadband (or hissing) component of the audible sound. The hum, on the other hand, is caused by the movement of space charge surrounding the conductor, which causes reversal of air pressure twice every half-cycle due to the movement from and to the conductor surface of positive and negative ions [6]. Space charge is created by ionization of air and this is generated by the same partial discharges causing the broadband (hissing) component. However, not all the corona modes create hissing and hum in the same proportions. As already mentioned, hissing is mainly generated by positive-polarity streamers [3] while Trichel pulses, Fig. 6 for instance, can produce intense ionization, and consequently a strong hum, with a much lower level of hissing. IV. CONCLUSIONS 1-The sound emission from weak points in electric power distribution networks proved to be a useful tool for early detection of these points. Not only detection of weak points, but also discrimination between loose contacts, polluted insulators and sharp edges is made possible. 2-Whenever these is a confusion in discrimination by sound pattern, frequency analysis of sound pressure level made it possible to discriminate between the different weak points. Sharp edges, polluted insulators and loose contacts are progressively ranked as regards their sound pressure level. 3- Whenever there is a confusion in discrimination by sound pattern, oscilloscopic recording of acoustic signals received from weak points becomes another tool for discrimination between "baby arcs" on polluted insulators and loose contact arcing. 4- Phase relationship of the pulse on the positive and negative half cycles makes it possible to discriminate in laboratory (indoor) and field (outdoor) testing between corona discharge at sharp points and baby arcs on polluted insulators. The of sharp points occur on the positive-going and negative-going of the respective positive and negative half cycles while those of the polluted insulators are centered around the peaks of the successive half cycles. Such detection of and discrimination between the different weak points are aimed to warning against impending failures, equipment damage and supply interruptions in MEEC electrical power network. V. ACKNOWLEDGEMENTS The authors wish to acknowledge Eng. M. Sayed Awad, the president of MEEC, "Middle Egypt Electricity Company", for his approval of this project as a joint research work between Assiut university and MEEC. The support they received from MEEC to conduct the project in the HV laboratory of Assiut University in Egypt and in the field underneath 33- and 66- kv electrical power network lines is highly appreciated. VI. REFERENCES [1] Ultraprobe-2000, A product manufactured by UE Systems, New York, USA [2] M. Goodman, UE Systems, New York, USA, Private Communication,?2001. [3] M. Khalifa and M. Abdel-Salam, The Corona Discharge in "High Voltage Engineering Theory and Practice", eds. M. Abdel-Salam, et. al., Marcel Dekker, New York, USA, [4] W. Weeks, Transmission and Distribution of Electrical Energy, Harper and Row publishers, New York, USA, , [5] J. C. Anderson et al., "Ultra high voltage transmission" Proc. IEEE, Vol. 59, pp , [6] EPRI, "Transmission Line Reference Book-345 kv and Above", Published by Electric Power Research Institute, Palo Alto, Calif., USA, ,1986. [7] M. Abdel-salam and D. Shamloul, Computation of ionflow fields of AC coronating wires by charge simulation technique, IEEE Transactions on Electrical Insulation, Vol. 27, pp , [8] J. J. Clade, C. H. Gray and C. A. Lefevre, Calculation of corona losses beyond the critical gradient, IEEE Transactions on power Apparatus & systems, Vol. 88, pp , ASU_AbdelSalam_A1 Session 1 Paper No