Professional Documents
Culture Documents
OF
SUGAR CANE FIRES
ON
AC CORONA NOISE
DEVELOPMENT OF A LABORATORY
SIMULATION AND MEASUREMENT PROCESS
January 1992
DECLARATION
INTRODUCTION . , .. .. .. ..........
, 1
IMPACT.... , ........ 19
2.2,4.3 AC CORONAMODES . . . m
3.3.2 RAIN . . e r n . . . . . . . . e .
3.6.2 ...........
DIRECT (RIV l a )
3.6.3 QUASI PEAX (RIV l b . c) . . . . . . . .
3.6.4 TIME DOMAIN (RIV 2) . . . . . . . . .
3.6.5 E-FIELD . . . . . . . . . . . . . . .
3. 7 TEST PROCEDURES TO SIMULATE ENVIRONMENTAL
CONDITIONS ..................
3.7.1 NO FAIR WEATHER CONDITIONS * .
3.7.2 RAIN ................
3.7.3 FIRE ................
3.7.4 FIRE AND SMOKE . . . . . . . . . . .
3.7.5 SMOKE . . . . . . . . . . . . . . . .
3.7.6 INSULATOR POLLUTION . . . . . . . . .
3.7.6.1 STEAMFOGTEST . . . . . . . . .
3.7.6.2 SALTFOGTEST . . . . . . . . .
3 , 8 PERFORMED TEST PROCEDURES TO OBTAIN DATA . . .
3.8.1 DIRECT . . . . . . . . . . . . . . .
3.8.2 QUASI PEAK . . . . . . . . . . . . .
3.8.3 TIME DOMAIN . . . . . . . . . . . . .
4 . 1 INTRODUCTION . . . . . . . . . . . . . . . . . 130
4 . 2 QUASIPEAKDATA . . . . . . . . . . . . . . . . 130
Eskom has for several years experienced on its 275 and 400 kV
networks flashovers and resulting v o ~ t > ~ dips
e caused by the
presence of fires, suc as sugar cane fires, bush or grass fires.
- air pressure
- temperature
- air motion
- humidity
- electrode material and shape
- irregularity of electrode surface
- presence of water vapour
- presence of an aerosol (smoke) or other gases
- incident photo-ionization
- type of voltage.
- light
- audible noise
- radio noise
- vibration due to 8gelectronicwindw
- a deterioration of insulating materials due to ion
bombardment
- ozone, oxides of nitrogen, and in the presence of moisture,
nitric acid
- dissipation of energy (corona losses).
2.2.1 BASICS
pV = C = c o n s t a n t ... (1)
T
= C = constant .. . (2)
where
According to Clausius the free path (X) of molecules and
particles are defined as the distance they can travel
freely before colliding with another particle. The free
path is a random quantity and its mean value depends upon
the concentration of particles or the density of the gas.
The mean free path is given by
P e r m i t t e d energy s t a t e s ( o r b i t s )
outer orbits
@ I
Figure 2.1 Energy states of hydrogen atom and its
schematic representation.
e- + A" - A .. . (18)
energy level o f
electron outside
t h e metal
electron
XI metal s u r f a c e
positive
image
energy level o f
electron inside
t h e metal
where
- direct attachment
- or dissociative attachment
ultraviolet
/ light electrons
v photon
avalanche avalanche
FLASHOVER
SELF-SUSTAINED /L/
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
NON-SELF-SUSTAINED
Uo U1 U2 Udc
VOLTAGE
y ead = 1 , .. (42)
we obtain
electric
field
emitted photon
field
l - y e I.*= o ..=
L
do d
distance
anode (+>
*:C*
cathode
. W
cathode (-1
(3
anode
0'0,
q s n s ~
*Q,
@+
produced:
electron
do d
distance
- burst corona
- onset streamer discharge
- positive glow discharge, Hermstein glow
- breakdown streamer discharge
Corona on a positive conductor has the appearance of a
uniform bluish-white sheath over the entire surface of
the conductor,
CAPACITIVE
CHARGING
CURRENT
, RESIDUAL
IONISATION
CURRENT
POSITIVE
CYCLE
CORUNA CURRENTS
NEGATIVE
CYCLE
- air pressure
- temperature
- air motion
- humidity
- line fittings
- insulator pollutio
- presence of water vapour or rain
- the presence of particles and smoke.
Rain and snow are the cause of high corona activity on the
surface of a conductor. The water droplets or snow which
collect on a conductor surface during a storm, modify the
electric field significantly, creating a large number of
corona sources [18].
Water droplets deform on a conductor surface when voltage is
applied (usually becoming conical in shape). The electric
forces can sometimes overcome the surface tension of the
drop, and the tip then erupts into minute charged droplets.
Boulet and Jakubezyk [l8 1341 found that the drop
disruption occurs when
- particle size
- number of unit charges and sign (plus, minus or bipolar)
- and particle mobility,
where
and
anode (+l
particle
ions
m
-' -
electric
f ietd
cathode (-)
particle f
Thus from the above we can conclude that for the periods
(b-c) and (e-f) in figure 2-18, a change in corona current
and even flashover could occur due to the presence of
aerosol particles in the air gap between the electrodes.
Bird [22] observed that the intense heat from a fire, with
no smoke present, caused no flashover. The only effect
observed was an increase in the audible corona noise.
%v=
-
-
v .. . (72)
n r In (-1 R
GMR
2 % ~ ~
C =
R
. .. (75)
In (->
GMR
25c
In(-) R
GMR
... (76)
z =-1
2n GMR
... (77)
where
and
2.4.2 RADIO NOISE GENERATION FUNCTION
at a point of radius r e S e p
at a point of radius p C S C R
Substituting for
n
t
_ INTERMEDIATE
DETECTOR STAGE
DETECTOR OUTPUT
WEIGHTING CIRCUIT
OUTPUT
I FILTER I
l
W
AMPLIFIER
7 JTER
H*RO
DRIVE
It has been found that the presence of sugar cane fires under
transmission lines has an effect on discharges and causes an
increase in radio noise. The sugar cane fire leads to possible
flashovers between phases and phase to ground causing unwanted
power line outages,
The maximum design voltage gradient for the affected lines are
given as [l]:
The three main sources of radio noise on the affected 275 and
4 0 0 kV transmission lines in eastern Natal and the line
parameters under normal conditions have to be simulated and are
discussed below.
3.2,1 NORMAL FAIR WEATHER CONDITIONS
3.2.2 RAIN
c o r o n a dome
//
Figure 3.2 Support system for the strung conductor at both
ends of the test cage.
The radio noise due to metal protrusions, caused mainly
during stringing of a line, reduces as the line ages (ionic
bombardment erodes away the metal protrusions). In the
laboratory the conductor under test is sandpapered with fine
water emery to get rid of most of these metal protrusions.
3.3.2 RAIN
The effect of rain on clean insulators was left out for the
purpose of this study.
r a i n nozzle
n
flames
S
I corona
cage
damping
1 conductor
fan heater
flap \
The first section seen in figure 3.4 is the air supply fan
with speed control used to generate a controllable air
burner
r
metal p l a t e
flame
. . . . . . . __II_
. . plasma
The flame plasma, smoke and particle plume are then directed
to the conductor under test by the directing plate into the
enclosed test cage. The temperature of the smoke and
particle plume can be controlled by switching on or off the
gas flame and by controlling the air supply.
The basic corona cage from UCT, with a diameter of 1.5 m and
an active length of 1 m, was adapted and modified to
minimise any obstruction to the flow of the flame plasma and
smoke plume by replacing the centre test cage with one where
the top and bottom sections where removed, leaving us with
1 m of cage on either side.
- the steam fog test, where 4 standard glass discs and line
fittings are fitted between the conductor and the cage as
shown in figure 3.7. The top three insulators are shorted
out, by soldering a copper braiding across the insulator
from cap to pin, A pollution layer is applied to the
fourth insulator, in accordance to the IEC recommendation
507 (Artificial pollution tests on HV insulators)[28].
conductor
and
line f i t t i n g s
steam
l----l corona
cage
L
aJ
f.'
aJ
E conductor
0
C
d
E
particle
content
atmospheric
pressure
3.4.1 TEMPERATURE
The flame temperature and the resulting hot air plume are
monitored using a thermal-video camera (Videotherm 94, Eskom
WCR), The camera detects thermal energy and transforms it
into a black and white television image. The relative
temperature of any object can be observed as shades of grey
- the warmer an object the lighter it appears. The
instrument has a temperature range of -30 to 1100 degrees C.
The emissivity of the reference object has to be taken into
account to obtain correct readings. An object with maximum
emissivity of 1 is called a black body, and is a good
absorber of radiation thus also a good emitter. An object
with an emissivity of zero is a perfect reflector, and is a
poor absorber of radiation thus also a poor emitter. The
emissivity of a fire at 1000 degrees C is given by the
manual of the apparatus as 0.75.
- Stellenbosch, 75.4 cm Hg
- Halfway house, 64 cm Hg
- Durban botanical gardens, 75 cm Hg
3,4.6 OPACITY
The glass fibre filter discs are weighed before and after
the test. Knowing the difference in weight and volume of air
sampled leads us to the particle content in mg/m3.
3.5 MEASURING CIRCUITS AND METHODS
1 I
r 4 p ~ e p ~ - l - .y -L-
IP~F-I- 44p~
t I -T- -T - -T- Il
r i t- = CORONA CAGE
I
r---lmn
l I I CORONA CAGE I I I
I I I l I I
d
ANTENNA
t-300 OHM
RIVlCa)
DIRECT
I I CURON4 CAGE I I
I l I I I I
~r I I I I I I ,,
l I I I I I
I I I
l
thus
CORONA CAGE
292.5 OHM
oscilloscope 0 0
CH 2 Vout(f r o n o u t p u t o f circuit)
Frequency (kHz)
Frequency (kHz)
-C Measured Calculated
Frequency (kHz)
[ -'Measured 4 Calculated I
0 100 200 300 400 500 600 700 800 900 1000
Frequency (kHz)
: : ~ 1 1 , , , , , , 1 ~
0
0 100 200 300 400 500 600 700 800 900 1000
Frequency (kHz)
-60 I t I I I
0 100 200 300 400 500 600 700 800 900 1000
Frequency (kHz)
3.7.2 RAIN
3.7.3 FIRE
3.7.5 SMOKE
polluted H
insuiators
3.8.1 DIRECT
- Supply on
- take voltage up slowly to max
- keep at max for 3 min
- bring voltage down in desired steps or to desired value
and take readings
- take voltage down to zero and switch off supply.
- Supply on
- take voltage up slowly to rnax (120 kV)
- keep at max for 3 min
- bring voltage down in steps of 10 kV (3 kV/cm) and take Q-
Peak readings (max, average and min)
- take voltage down to zero and switch off supply.
- Supply on
- take voltage up slowly to max (100 kV)
- keep at max for 3 min
- bring voltage down to 80 kV (24 kV/cm) and take all time
domain readings
- take voltage down to zero and switch off supply.
4 EXPERIMENTAL RESULTS
4.1 INTRODUCTION
The flow rate in the outlet stack is omitted from the results
due to inaccuracy (possible reasons are given on page 107 in
.
section 3.4.4) During all the sample runs the pressure between
the inner cage and the atmospheric pressure of the laboratory
remained the' same.
Average (abc)
64 64 64 59 59 59
65 65 65 36 60 60 60
65 66 65 61 61 60
Average (abc)
-
E
I
0
80
70
60
Average (obc)
50
5 40
9 30
20
10
0
12 15 18 21 24 27 30 33 36
E (kV/crn)
4.2.3.1 F I R E (HOT A I R )
min
+
Average (abc)
+4+
max
78 77 76 70 72 71
79 78 77 36 71 73 72
79 79 78 72 74 74
40
q 30
20
10
0
12 15 18 21 24 27 30 33 36
E (kV/cm)
-I-
min
+
Average (abc)
+
max
The scaling of the data in the figures can be seen in the top
left hand corner of each sample. Two y axis ranges can be seen
in the figures, (0 mV to 2000 mV) or (0 mV to 100 mV). Due to
the restriction of the analog to digital converter the ranges
should be read as (-1000 mV to 1000 mV) and (-50 mV and 50 mV)
respectively.
The data obtained for the steam fog test can be seen in
figure 4.21. The samples are all taken during the same
pollution test run. The creapage length of the insulator is
calculated as 23 mm/kV. The ESDD (equivalent salt deposit
density) was determined as 0.4 mg/cm2.
The data obtained for the salt fog test can be seen in
figure 4.22. The samples are all taken during the same
i
pollution test run after the 20th minute. The creapage
length of the insulators are calculated as 22 mm/kV.
5.1 INTRODUCTION
Note that all the sugar cane fire tests are done far below the
flashover point for these conditions. This is done to ensure
that the measuring equipment is not damaged due to the
occurrence of a flashover. Closer to flashover point
conditions, higher noise values can be expected.
J? = i.-
2 a ~
C
. S m (l)
and the capacitance per meter between Corona cage and conductor
v, = i . 2
but
2010g(T) = 201og(-1v m + 75 1)
20log(ln( 0.7125
Z
thus
Now
Z = 25 ohm . . . (8)
substituting (8) into (7) gives
Stellenbosch
From the above results it can be seen that the two corona cage
results (Stellenbosch and Megawatt Park) compare well and fall
within a few dB from each other.
5.3 POWER SPECTRAL DENSITY
At this frequency the noise due to sugar cane fires are in the
order of 20 ~ B ( v ~higher
) than those during rain conditions.
0.0
0.6
0.4
0.2
-0.2
-0.4
-0.6
-0.8
-1
0 200 400 600 800 1000
N e g s r l v e cycle n o l s e pulse ( R a l n ]
0.05
0.04
0.02
0.01
*
-
0 0
>
-0.01
-0.02
-0.03
-0.04
-0.05
0 200 400 600 800 $000
4 5 6
Froauency CH11
0.045
0.04
-
N
0.035
-c
"
0
>
,0 . 0 3
-
.-
C
0
g 0.025
-
EI
.L
d
0.02
a
Y1
g 0.015
P
0.01
0.005
0
0 1 2 3 4 5 6 7 E 9 10
Negatlve c y c l e n o l s e p u l s e (FlreJ
0.05
0.04
0.03
0.02
0.01
-
*
0 0
-0.01
-0.02
-0.00
-0.04
-0.05
0 200 400 600 800 1000
0 i 2 3 4 5 6 7 B 9 10
Power S p e c t r a l d e n s l t y o r N e g a t l v e c y c l e n o l s e p u l s e ( F l r e ]
0.05 1
0
0 l 2 3 4 5 6 7 B 9 to
N e Q a t l v e c y c l e n o l s e p u l s e (Smoke]
0.05
0.04
0.03
0.02
0.01
-+
0 0
-0.01
-0.02
-0.03
-0.04
- 0 . OS
0 200 400 600 800 1000
PO I n t s
0 1 2 3 4 5 6 7 B S 10
Power S p e c t r a l a e n s l L y o f N e g a t l v e c y c l e n o l s e p u l s e (Smoke)
0.05
0.045
0.04
-c 0 . 0 3 5
0
z
-
C
,0 . 0 3
-
U)
g 0.025
-
m
L
0.02
a
V)
t 0.015
.c
0.01
0.005
0
0 1 2 3 4 S 6 7 a 9 10
400 600
Polnts
N e g a t l v e c y c l e n o l s e p u l s e [ F l r e & Smoke)
0 . OS
400 600
Polnts
0 1 2 3 4 5 6 7 9 10
Frequency ( H z ) X 10'
o . 0 5 5
4 5 6
Frequency ( H z )
BOO
0.0
0.8
0.4
0.2
-0.2
-0.4
-0.6
-0.0
-1
0 200 400 600 800 1000
450
400
350
300
250
200
150
900
so
0
0 l 2 3 4 5 6 7 E 9 10
Frequency ( H z ) X lol
~(posdata=
) ( p o s d ata)
lb 2b 30 4b 50 60
Positive half cycle (VoltA2)
NORMAL
From the Power Spectral density results it looks evident that for
the laboratory setup the presence of sugar cane fires could be
detected in the frequency range of 50 to 500 kHz (section 5.3).
The noise due to heavy insulator pollution would thus mask out
all the other sources of power line corona noise. Cleaning the
insulators would solve this problem (could be a good way of
warning when cleaning has to commence).
The question is whether the above would work in the real world
of power line carrier noise?
Work on this subject has been done in Natal and it was found that
insulator pollution could be detected [l]. However, interpreting
their results representing the noise due to other environmental
conditions was found to be suspect as these were not sufficiently
monitored and documented.
TRANSMITTER
RECEIVER
POWER SUPPLY
L
V.
APPENDIX C: METHODS USED TO CALCULATE AND MEASURE THE
CAPACITANCES IN FIGURE 3.9 (PAGE 110)
C c i = 12 pf
SIGNAL 1 F I N E G A I N ADJUST
(SOH2 + SIGNAL) FOR T O T A L AV = l 0
R7
COARSE N U L L I N G
I
SIGNAL 2 & VRI
( 5 0 H z ONLY)
10 TURN P O T
F I N E NULLING
10 TURN P O T
% load the data from data file and assign to the variable y
load filename.mat;
y = filename;
B = ..a ;
y = y (A : B);
To determine the most probable centre point for each power line
noise condition the method of "Dynamic Aggregationw is used.
Plot out the calculated data points for the various power
line conditions.