Fires under power lines reduce the breakdown strength of air insulation due to the effect that heat and particles have on the electrical field surrounding the conductors. To the support staff in the corona cage, thank you for the support given and the extra work done in recording a lot of data.
CHAPTERl Introduction
THE PROBLEM
The burning of sugar cane is a common practice among the farmers of Kwazulu Natal because it helps with the harvesting process. Changes in the composition of the air around the conductors imply a change in the corona activity around the conductors.
RESEARCH OBJECTIVES
During the winter months the fields become increasingly drier and pose a risk to the transmission network as fires in the region become more likely. Previous research showed that the disruptions were due to the loss of air around the conductors due to the proximity of the fire [1].
SCOPE OF DISSERTATION
The signals that have been detected from this environment are presented in this chapter. This chapter covers a number of alternative detection algorithms considered to detect and handle fire-related high-frequency signals.
Background Information
- ESKOM'S NETWORK
- MEASURES TO MINIMISE IMPACT ON ESKOM CUSTOMERS
- PREVIOUS RESEARCH
- HIGH FREQUENCY NOISE AND HOW TO GET ACCESS TO IT
- CORONA, CONDUCTOR DIMENSIONS AND FIRES
- OTHER SOURCES OF CORONA
- RESEARCH PRIORITIES
- RESEARCH LOCATION
- SOURCE OF FIRES FOR THE TEST OBJECTS
The main differences observed in the time domain measurements were related to the observed pulses and thus to the physical appearance of the streamers at the conductors. The two main differences between the fire-induced corona noise pulses and the rain-induced corona noise pulses were the amplitude and the time between the pulses.
Basic insulation breakdown process
BREAKDOWN OF THE AIR INSULATION
W obtained by the electron as a direct result of the fires under the conductors can be expected. Therefore, the larger the radius of the conductor or bundle of conductors, the smaller the field will be for the same applied voltage, and the lower the energy~W obtained by the electron.
THE CORONA PHENOMENA
The avalanche process will result in filamentary streams propagating beyond the boundaries of the critical volume. Further increases in voltage lead to "negative streamers" and small twigs begin to develop from the spherical glow - pulses with rise times of the order of 0.5 IlS.
ONSET OF NEGATIVE AND POSITIVE CORONA
- RAIN AS A CATALYST FOR CORONA
- FIRE AS A CATALYST FOR CORONA AND BREAKDOWN
- Particles
- POLLUTION ON INSULATORS
- SURGES
- INFLUENCE OF OTHER SPURIOUS SIGNALS AND NOISE
The radio noise component of the corona is a measurable entity and is both transmitted radially from the conductors (causing radio and television interference) and propagated along the conductors (to terminate in the substation at one or both ends of the line). In the vicinity of the conductors, the particles become dipoles (figure 3.4), which are ionized by the electric field and become part of the field lines emanating from the high-voltage conductors.
Laboratory - Large Corona Cage: Set-up and method
- BACKGROUND
- OBJECTIVES
- OPERATION
- CONDUCTORS UNDER TEST
- TEST METHODOLOGY
- TEST PROCEDURES
- MEASUREMENT METHODOLOGY
In the photo, a double Zebra wire bundle was in place strung between the two sets of corona rings at either end of the cage. From the capacitance and the inductance, the characteristic impedance of the conductor bundle can finally be calculated:
CHAPTERS
RESULTS - REFERENCED TO GAS FIRE
In fire conditions the RIV levels approach the extreme levels reached in dry and rain conditions almost from the beginning at the surface gradient of 9kV/cm with an RIV value of about 80dBJlV. As per the objectives given in section 4.2, (a) the noise levels generated by the gas fire induced corona are 90dBIlV at the surface gradient of 18kV/cm. However, as previously discussed for the Twin Dinosaur bundle, the gas fire was carefully controlled to produce a stable flame at the conductors.
For all other surface gradients, dry conditions increased from about 1OdBJ.lV at 11kV/cm 65dBJ.lV at 25kV/cm and a steep increase at about 15kV/cm to 50dBJ.lV. The RlV for rainy conditions climbed from 52dBJ.lV at 11kV/cm to 80dBJ.lV at 26kV/cm.
ANALYSIS OF THE QUASI PEAK DATA
The coronal noise activity caused by the fire showed a gradual increase in the noise level from the lowest surface gradient recorded at 9 kV/cm. The noise generated above 11 kV/cm is higher than the rain induced corona noise at any gradient during the test. A maximum separation of 21 dBIlV between rain- and fire-induced corona noise was recorded at 12 kVfcm.
The fire-induced corona noise data was consistently higher than the rain-induced corona noise data (changing the 24 kVfcm result as for the Twin Zebra test as illustrated in the graph below).
DIMENSION COMPARISONS
With equal spacing between the two conductors (ie 380mm) in each case, the Dinosaur twin conductor package has lower RIV signal levels than the Zebra twin conductor package. When the double set of dinosaurs were tested, conditions were ideal for testing with very little wind. At higher surface gradients, the Twin Dinosaur scores can be seen to increase relative to the slopes of the Zebra and Wolf twins, implying that at higher gradients the effects of dust particles, etc.
The results were less predictable, but show that for the same surface gradients, the Dinosaur twin conductor bundle generates the highest level of corona noise.
Dinosaur ~Zebra -+-wolfl 120
- CONCLUSIONS: QUASI-PEAK EXPERIMENTS
The excitation function is related to the quasi-peak measurements performed here by a constant which is dependent on the dimensions of the test environment. For all five bundle configurations, the quasi-peak values for fire-induced corona are larger than the corona for heavy rain conditions at all surface gradients. An increased number0 f discharges, even a low amplitude, will increase the quasi-peak values measured on a power line.
On this basis alone, the number of electrical discharges due to the presence of fires will be lower in the operating environment, reducing the recorded quasi-peak values.
Laboratory - Large Corona Cage
Time Domain Experiments
RESULTS - REFERENCED TO GAS FIRE
At the same time, large electrical discharges also began in the negative half cycle (see the last half cycle). With the exception of the last half cycle, the corona on the negative half cycles remains much smaller in amplitude than that in the positive half cycle. In the negative half cycle, small discharges can be seen at the beginning of the half cycle, these may be Trichel pulses.
The electrical discharges in the negative half-cycles were here again seen to be very small even.
RESULTS - REFERENCED TO SUGAR CANE FIRE
During heavy rain, even at 15 kV/cm, we can see in Figure 6.10 the familiar envelope in positive half-cycles with marginal pulses in negative half-cycles. The corona was visible in both half-cycles, in contrast to it being primarily visible in the positive half-cycle for the corona caused by the gas fire. Here, the coronal pulses in the positive half cycle and the negative half cycle have similar repetition rates.
The streamers in the positive half cycle have a higher amplitude and may be the onset of breakdown streamers.
ANALYSIS OF THE TIME DOMAIN DATA
In Figure 6.1 on page 85, a magnified portion of the waveform in the last full positive half cycle showed two random pulses. The data shown in Figure 6.22 contains a significant amount of background noise, with the two streamers taking up only a small fraction of the total data. The end result was that the information in each of the two streamers was minimized due to the accumulated influence of the background noise (Figure 6.23).
The total number of sample points taken to calculate the frequency spectrum shown in Figure 6.23 was 16002 points.
800 kHz
MEASUREMENTS COMPILED FROM POLLUTED INSULATOR TESTS
The salt spray test involves spraying a salt solution onto the insulators, which are loaded to the normal operating gradient of the conductor bundle with the particular number of insulators in the string. The insulators were then baked to dry the deposit and assembled in place for the test. The insulators were then wetted with the insulators loaded to the normal operating gradient of the conductor bundle with the required number of insulators in the string.
A finer array of nozzles was chosen to reduce the droplet size arriving on the isolators.
CORONA CAGE RESEARCH RESULTS IN PERSPECTIVE
- No Flashover Due To The Sugar Cane Fire
Several tests were performed with sugar cane stacked below the test line in a corona cage and high voltages were applied in accordance with the appropriate normal operating gradients for a given conductor bundle. Energy was required to initiate the disruption of air insulation and it was reasonably safe to assume that additional energy was required to increase this disruption as well. The gas fire emphasized the effects of increased heat and reduced air density in a very small portion of the conductor bundle.
The result in the gas fire was a "single" path to ground with only a small amount of current flowing in the path as graphically represented in figure 6.70 (state A) by the current flow "i".
Ground Plane
CONCLUDING REMARKS ON LARGE CORONA CAGE RESEARCH
Time-domain measurements have recorded several useful features of rain- and fire-induced corona noise patterns. The amplitude of the corona noise pulses produced by the fire was primarily influenced by the changing environmental conditions - changing air density and joint activity - within the secondary influence of the applied voltage. The amplitude of fire-induced corona noise pulses has been recorded at values up to 48 dB higher than rain-induced corona noise pulses.
Therefore, the repetition rate of the fire-induced pulses was also less predictable than the repetition rate of the rain-induced corona noise.
Laboratory - Modelled Corona Cage
BACKGROUND
CONCLUSIONS REACHED
RELATING MODELED CORONA CAGE TO LARGE CORONA CAGE
Operational Conditions
- BACKGROUND
- POWER LINE CARRIER CIRCUITS
- ACCESS TO FIRE-INDUCED CORONA
- FILTERING
In CVT, stacked capacitance with respect to ground is also used for high-speed protective communications. The filters used by the carrier system, which are essential for measuring high-frequency noise and signals on the transmission line, are similar in concept to those used in the corona cage power circuit. Fires are rarely planned and when they are planned sugar cane fires in the region where it is.
34; unplanned" fires cause the most problems, these power lines are de-energized to ensure the highest possible quality of supply to major electricity consumers in the entire region.
Operational Line - Controlled Fuel
- OVERVIEW
- NORMAL CONDITIONS AND FILTERING AFFECTS
- GAS FIRE UNDER THE TRANSMISSION LINE
- SUGAR CANE FIRE UNDER THE TRANSMISSION LINE
- SUMMARY
However, the ratio between the noise levels at 52 kHz and the noise level at 324 kHz could be compared with the result in Figure 9.13. The total increase in noise was the sum of the dB difference (dBI) of the carrier signal to the background noise and the dB difference (dB2) of the fire noise to the carrier signal. The components above 324 kHz were harmonics of the 52 kHz signal added to the carrier.
The remainder of the carrier signal that was not attenuated by the 428 kHz filter clearly broke through the background level in Figure 9.14.
Operational Lines - Sugar Cane Plantation Fuel
BACKGROUND
SUGAR CANE FIRE TESTS - MARATHON / KOMATIPOORT
Of the three photographs taken at the time of the flashover, only the photograph in Figure 10.6 was evidence of the failure. A limited percentage (up to 100%) of the first channel was available for pre-trigger data. An expanded view of the background noise in Figure 10.12 below revealed the envelope effect of a carrier signal.
However, it was clear from Figure 10.13 below that the observed "carrier" was actually the influence of the bandpass filter circuit.
200oJJ U
SUGAR CANE FIRE TESTS - AVON /IMPALA
