KEYWORDS
* Swedish Transmission Research Institute, P.O. Box 707, S-771 80 Ludvika, Sweden
** Presently with ABB Power Systems AB
the coast 23 km from Kilroot.) Both substations lie within bushings was removed by manual washing from elevated
1 km from the sea. platforms. The bushings had relied on natural cleaning by
rain which created problems. The deep under-rib profile
The recommendations of the panel included: and the angle of installation of the bushings tended to
• monitoring of the bushings, concentrate a higher level of pollution on the lower edges.
• appraisal of pollution detection systems (at that time The recommendations of the panel of inquiry were then
only installed at Ballylumford), investigated in parallel and these are now discussed.
• appraisal of water washing systems (at that time only
installed at Ballylumford), 3.2. RTV Coating
• extension of the 275 kV bushing insulator creepage
distance, Extensive investigations were carried out to identify a
• application of grease or other coatings on the insula- suitable coating and pollution monitor. Sylgard [1], an
tors. RTV silicone rubber coating, was applied to two of the
four circuits in February 1995 as a trial to ascertain its
Ballylumford had a pollution monitor based on the performance on the existing bushings. The principle of
principle of immersing a dummy insulator into a water the silicone coating is to ensure that moisture filming on
bath and then measuring conductivity. This substation the insulator surface is minimised by having a water
also had water washing facilities on the bushings. Kilroot repellent (hydrophobic) surface. The silicone coating at
had neither washing nor pollution monitoring. this stage was intended as a temporary measure as the
issue of creepage distances required to be addressed.
The first action taken after the incident in 1994 was to The connection to the bushing consists of a thin copper
clean the existing bushings at Kilroot. Salt residue on the band fixed 2-3 cm from the lower end of the porcelain
surface and connected by a wire of small cross sectional A new technical specification was drafted. The key points
area to an interface box. The input resistance is low of the specification were as follows:
(approximately 5 Ω) and is protected by Zener diodes and • 300 kV, 2000 A, capacitance graded, oil impregnated
a 100 V spark gap. The lowest part of the porcelain paper type with porcelain insulators generally in
surface is greased to prevent leakage current shunting accordance with IEC 137 [4].
directly to ground. The surface leakage currents are • Minimum nominal specific creepage distance of out-
passed to earth via a charge measurement section - the door porcelain to IEC 137 and 815 for the case of
value of charge being integrated over a fixed period (10 'heavy' pollution levels (25 mm/kV at highest system
s) and passed to the datalogger for storage. voltage and appropriate shed diameter factor).
• Preference for alternate long-short (ALS) shed pro-
The datalogger software is arranged to store all sensor file.
values in battery backed memory at 10 s intervals and can • Mounting arrangement - eighteen intended for
be downloaded either locally or remotely via an attached mounting at 15 degrees to the horizontal, twelve for
modem. Leakage currents above a set threshold will mounting vertically.
trigger an alarm card to generate a system alarm for the • Type and routine tests according to IEC 137 and IEC
individual 275 kV circuits. 60-1 [5].
• Pollution performance tests:
The present trigger logic demands that the average 1. Salt fog test in accordance with IEC 507 [6], (3
leakage current exceeds the set threshold in 3 consecutive out of 4 tests without flashover to constitute a
measuring periods (each 10 s long) in order to produce a pass) withstand salinity of 80 kg/m 3 followed by,
valid alarm signal. 2. Wet test with artificial rain according to IEC 60-1
[5] falling at 90 degrees to the bushing (rain con-
ductivity 100 µS/cm, duration 60 s).
4. SPECIFICATION OF REPLACEMENT • Additional pollution tests to be agreed if required.
BUSHINGS
4.2. Testing
4.1. Dimensions
Although, composite type bushings were investigated, it
The old bushings at Kilroot and Ballylumford had a was decided to restrict the enquiry to porcelain bushings
creepage distance of 6985 mm which equates to a for several reasons. Firstly, NIE has little or no
minimum nominal specific creepage distance of 19,4 experience of composite insulators, particularly at this
mm/kV. From IEC 815 [3], voltage level. Secondly, although the pollution
performance of composite insulation may be considered
L = d sc ⋅Ur ⋅k D (1) superior initially, there were concerns over the lifetime of
the hydrophobicity of the composite material. Thirdly,
L = creepage distance (mm) the current standard in relation to insulator pollution
dsc = minimum nominal specific creepage distance (mm/ testing, namely IEC 507, (BSEN 60507, 1993), only
kV) applies to ceramic/glass type insulators and not to
Ur = highest system voltage (300 kV) composite materials and NIE was seeking test assurance
kD = average diameter correction factor (1,2 for the old of the pollution performance of the replacement
bushing) bushings.
NIE currently purchases outdoor insulator equipment to Discussions with manufacturers and other utilities on the
Class III category (heavy pollution) in accordance with pollution performance of different shed profiles resulted
IEC 815, which specifies a minimum nominal specific in an ALS profile being specified. In the vertical position,
creepage distance of 25 mm/kV. The old bushing there is shielding of the short shed from the long shed. In
creepage distance would therefore place them in Class II the horizontal position, the ALS has better self cleaning
category (medium polluted). It was decided to bring these properties than the anti-fog design. It was however the
bushings in line with the requirements of Class III as a opinion that testing would be required to determine the
minimum. pollution performance, in both orientations, of the ALS
design.
Initially, bushing refurbishment was considered but
turned out not to be a feasible option due to the internal The IEC 507 Salt fog test was specified as a pass/ fail test
construction of the bushing. Current bushing designs for the bushing design. The successful Tenderer would be
were investigated and an enquiry was placed for the required to test the offered bushing in both vertical and
supply of 30 wall bushings as replacements for both horizontal (15 degrees) positions - a requirement of the
Kilroot and Ballylumford substations. standard to test in the in-service orientation. A value of 80
kg/m 3 was selected as the pollution severity. This was
deemed to be a reasonable figure based on available Table I: Parameters of the tested wall bushings.
information and is stated as a reference value for ‘heavy’
polluted locations in Elektra [7]. The Salt fog test was to
Bushing
be followed by an artificial rain test. Additional pollution
Parameter (mm)
tests were envisaged, primarily for calibration of the new
Replacement Old
pollution monitor.
Creepage distance
The successful Tenderer for the supply of these bushings 10290 6985
outdoor
constructed one unit to undergo the Salt fog tests. The
parameters of the new bushing are as follows: Insulation length
3225 2134
• Rated voltage: 300 kV outdoor
• Shed profile: ALS
• Outdoor creepage distance: 10290 mm Core shape Straight Conical
• Minimum nominal specific creepage distance: 31,2 Average core
mm/kV 355 406
diameter
The bushing passed the Salt fog tests without a single Average diameter 406 535a
flashover. NIE decided to pursue further pollution testing
in order to provide calibration information for the Shed profile Long-short Underrib
installed pollution monitor. This was to attempt to set a Shed spacing 70 152
pre-warning threshold level, based on on-line leakage
current measurements which could be used to initiate a a. This is approximated by the average small
course of action, in advance of dangerous conditions at shed diameter.
either substation. The Dry Salt Layer (DSL) test, which
more closely resembles the natural pollution conditions
experienced on site was chosen as the additional pollution
test. The tests and the suggested pre-warning threshold
level are described in sections 6. and 7.
5. REPLACEMENT BUSHINGS
5.1. Design
5.2. Salt fog performance An estimate of the average surface conductance, required
to perturb the capacitive field significantly is
12 ⋅10 Ω − 1. It is assumed that the surface conductivity
–9
A rough estimation of the performance of the insulator
for Salt fog testing can be made by equations (2) and (3). is equal all over the surface, which of course in reality is
never really true.
σ
S = --- (2)
F
S = surface conductance 6. VALIDATION TESTS
σ = surface conductivity
F = form factor of insulator 6.1. Salt fog
The dimensionless form factor, F, can be estimated by The Salt fog tests were conducted according to IEC507
equation (3). [6] including preconditioning with repeated flashovers.
Each individual test takes one hour.
L
F ≈------- (3)
πD 6.2. Dry Salt Layer
L = creepage distance
D = average diameter In the DSL test, salt water spray is generated by nozzles
and dried in the air to give small salt particles. These are
From equations (2) and (3) and the values from table I the blown towards the energized test object by fans and result
surface conductance can be estimated thus: S=0,124σ for in a salt accumulation on its surface. The salt is then
the replacement bushing and S=0,241σ for the old wetted by steam fog during which the test object is
bushing. The surface conductivity is roughly proportional energized at constant voltage. The test object fails the
to salt fog salinity if the whole surface is wetted and the individual test if it flashes over during the test time. The
test objects are hydrophilic. This means that the test is otherwise continued until the leakage current has
withstand salinity roughly can be a factor of two (0,241/ decreased to such a level that the risk of flashover is
0,124=1,94) higher on the replacement bushing and still deemed negligible, typically 1-2 hours.
give the same surface conductance as on the old bushing
for a given salt fog salinity. This is largely in agreement During the wetting of the polluted test object the leakage
with the flashover results in section 7. The old bushing current was measured at 1 kHz sampling frequency. In
withstands 28 kg/m 3 and flashes over at 40 kg/m 3. The some of the tests the NIE leakage current monitor was
connected in series with the laboratory shunt to
replacement bushing withstands 80 kg/m 3.
demonstrate that the current signals could be compared.
The higher average axial electric field (applied phase-
Tests showed that the output signal from the NIE leakage
ground voltage divided by the insulation length) of the
current monitor could be simulated by simply calculating
original bushings, 81 kV/m at service voltage compared
the average current from the laboratory measuring system
to 54 kV/m for the replacement bushings, and the slightly
over 10 s intervals. The difference in values for each 10 s
non-linear relation between salinity and water conductiv-
interval was not more than a few percent and can be
ity could further explain the difference in outcome of the
understood from the difference in exact sampling time
Salt fog tests between the two different bushings.
and absolute calibration. In addition:
5.3. Distortion of the capacitive field grading
• Flashovers were registered.
• The visible discharge activity was recorded by a
The capacitive current from condenser core to surface de-
video camera.
termines the field distribution on the insulator surface of
• For the DSL tests the Salt Deposit Density (SDD)
the condenser bushing during dry conditions. During wet
was measured on small parts of the bushing sheds
and polluted conditions, this distribution is distorted by
after the polluting, but before the wetting.
the surface currents on the insulator. The field distribu-
tion on the insulator surface is dominated by the pollution
7. RESULTS For comparison of the different types of leakage current
variables, figure 3 and figure 4 have been included. By
7.1. Flashovers comparing these figures to each other it is possible to
more easily grasp their different influence on the raw data
The flashover results of the tests with Salt fog and DSL signal. The ‘Minimum of three’is just a filtering of peaks.
are summarized in table II and table III respectively. Single high readings do not influence this variable and
consequently will not trigger the alarm. The Imean100 is
Table II: The Salt fog flashover results an averaging of all the absolute values measured in 100 s
intervals. Peaks are not filtered out but are instead
smoothed out and influence the result with equal weight
Salinity Flashovers
Bushing Position as all other values. From a comparison of the curves in
(kg/m 3) (No. of tests) figure 3 and figure 4, it is seen that for this test object the
‘Minimum of three’ and the largest Imean100 do not
vertical 28 0 (3)
Old deviate very much from each other. The difference is a
vertical 40 2 (2) few mA only. The same applies to the other tests
performed on this bushing, both the DSL and the Salt fog
vertical 80 0 (3) tests.
Replacement
horizontal 80 0 (3)
0.035
0.03
Table III: The DSL flashover results for equal polluting 0.025
times. I (A)
0.02
0.015
Average
SDD Flashovers 0.01
Bushing Position
(No. of tests)
(mg/cm 2) 0.005
0
Old vertical 0,96 2 (4) 0 500 1000 1500 2000 2500
Time (s)
Replacement vertical 1,38 0 (4)
Figure 3: Leakage current information from DSL test 1
on the old bushing. The dark curve is Imean10
7.2. Leakage current signals and the lighter curve is the ‘Minimum of
three‘. The time 0 is counted from the start of
During each test the leakage current was measured and the wetting.
the following quantities were calculated from the
individual values sampled at 1 kHz: 0.035
0.03
• Imean10 is the 10 s average of the rectified leakage
0.025
current. This is the same quantity which is measured
Imean100 (A)
For the DSL tests the average SDD (salt deposit density) 7.3. Leakage current pre-warning signal
after salt application was calculated from eight individual
measurements along the bushings on both the bottom and In addition to the tests mentioned in table II and table III
top sides of the sheds. some DSL tests with shorter polluting time, giving lower
SDD levels, have been performed as pre-tests. These tests Which is then the best choice of current variable, Imax,
have been included in the following analysis. Imean10, Imean100, the ‘Minimum of three’ or some
other? Obviously the Imax is attractive because it is the
In order to find a suitable leakage current pre-warning highest currents which cause the flashover and Imax
signal the following treatment of the measured leakage therefore has a strong coupling to flashover probability.
current has been adopted: Because of Imax’s potential sensitivity to electronic noise
and the fact that the NIE leakage current monitor does not
1. For each individual test (both DSL and Salt fog) the measure Imax, one of the other current variables should
highest value of Imean10, Imean100, the NIE trigger be considered. A demand should be that the current
and Imax has been determined. variable should show a strong coupling to the flashover
2. The current values for an individual test have been probability or pollution severity. A current variable
omitted in case Imax did not exceed 100 mA. Imax which is more or less constant independently of different
above 100 mA is here regarded as one criterion for pollution conditions is of little use since it will be difficult
dangerous pollution event. to set a distinct pre-warning signal. The signal may either
3. From all the remaining values the smallest one has suffer from many false alarms or will not give any
been picked for each of the two test objects and for warning at all.
each of the two test methods separately.
Because of the similarity in values of the largest
These values are shown in table IV and constitute a Imean100 and the ‘Minimum of three’mentioned in table
suggested upper limit of pre-warning signal for the IV only the Imean100 is included in the following
corresponding current variable. Generally the DSL tests analysis.
show lower values than the Salt fog tests in table IV. This
is probably caused by the larger span of severity covered 7.4. Correlation between leakage current and SDD
in the DSL tests since also the pre-tests with low SDD
levels were included. In the comparison of the different leakage current
variables an independent parameter is necessary with
Table IV: Suggested upper limits of pre-warning signals which to judge the severity of the test. For the Salt fog
for the two test objects and for the two tests the salinity is such a parameter, but the salinities
different test methods based on different were limited to the few pre-defined levels used in the
current variables. The lowest value for the tests. For the DSL tests, however, the SDD can be used as
respective current variable has been severity parameter and it furthermore covers a wide range
shadowed. of values.
Imean100 11 9 25
Imean10 13 24 rain
20
Largest Imean100 (mA)
Replacement
Imean100 11 13
15
COM DSL Imean100 (m
a. A flashover occurred during this test.
BC DSL Imean100 (mA
10
0.8