-
IEEE Transactions on Power Delivery, Vol. 9, No. 4, October
1994
TRANSIENT ELECTROMAGNETIC INTERFERENCE IN SUBSTATIONS
C. M. Wiggins, D. E. Thomas, F. S. Nickel, Senior Member,
IEEE
T. M. Salas Student Member, IEEE
BDM International, Inc. 1801 Randolph Road, S.E. A1 buquerque,
NM 87106
Abstract - Electromagnetic interference levels on sen- sitive
electronic equipment are quantified experimen- tally and
theoretically in air and gas insulated sub- stations of different
voltages. Measurement techniques for recording interference
voltages and currents and electric and magnetic fields are reviewed
and actual interference data are summarized. Conducted and radiated
interference coupl ing mechanisms and levels in substation control
wiring are described using both measurement results and
electromagnetic models validated against measurements. The nominal
maximum field and control wire interference levels expected in the
switchyard and inside the control house from switching operations,
faults, and an average lightning strike are estimated using high
frequency transient coupl ing models. Comparisons with standards
are made and recomnendations given concerning equipment shield- ing
and surge protection.
Keywords - Electromagnetic interference, EMI, switching
transient, electric field, magnetic field, fault, lightning,
substation, shielding, surge protection.
INTRODUCTION
Increasingly, electronic equipment is being used in switchyards
and inside control rooms. Substation switching operations,
spontaneous faults, or lightning strikes inside the substation, can
cause potentially damaging levels of high frequency electromagnetic
in- terference (EMI). This EM1 can couple into low voltage control
circuits and electronic equipment unless it is suitably protected.
This transient EM1 environment needs to be fully characterized by
waveforms and spectra for the highest expected levels both in the
switchyard and inside the control house. These EM1 en- vironment
levels may then be compared with equipment suscepti bil i ty levels
(if they can be determined) for upset and damage, and also with
applicable surge withstand capability (SWC) test levels to assess
their mutual compatibility and adequacy.
Substation EM1 issues have been investigated in a number of
studies, such as [l] - [7]. Switching tran- sient currents,
voltages and fields were measured in
S. E. Wright Senior Member, IEEE
Electric Power Research Institute 3412 Hillview Avenue Palo
Alto, CA 94303
(now at Huddersfield Polytechnic School of Engineering
Huddersfield, England HD13DH)
1869
[l] - [6]. Potential EM1 impacts and suggested protec- tion
remedies on solid state relays were discussed in [l] and [7]. In
[7], several types of EM1 sources were identified including the
fast transient and the "walkie talkie" transient which have
contributed to improved test standards [81,[91.
Investigations under project RP 1359-2 by Texas A&M for the
Electric Power Research Institute [I] are particularly relevant to
the work reported here. Some of the conclusions of this earlier
work indicated: an expectation that radiated EM1 would become more
impor- tant as new distributed automation systems were intro- duced
in substations; a need for further analysis of the nature of
radiated EM1 transients in substations; recognition that no
appropriate standard existed for determining equipment
susceptibility to transient electromagnetic fields, particularly
for equipment lo- cated in switchyards; and a need for pre-purchase
sus- ceptibility testing. Improved equipment design along with the
use of surge suppression devices and shielding enclosures was
recommended to mi tigate EM1 effects.
In 1985 EPRI initiated project RP2674-1 with BDM International
to further address these concerns. Un- der RP2674-1 there has been
a strong attempt to broaden the substation EM1 environment
characterizations and their understanding by developing validated
high fre- quency traveling wave models as well as by gathering
additional detailed measurements of EM1 phenomena in substations.
This paper summarizes the major findings o f this study as reported
in [lo]. The emphasis here is on presenting the highest expected
EM1 levels in the switchyard, in the control house and on shielded
con- trol wires in substations up to 500 kV. These es- timates are
based on the results of all measurements and model predictions for
switching transients and for faults and lightning strikes occurring
in the substa- tion. The manner in which EM1 couples from sources
on the high voltage bus to wires inside shielded control cables is
discussed qualitatively and quantitatively. Control wire EM1 levels
are compared with oscillatory and fast transient SWC test waveforms
(IEEE/ANSI C37.90.1-1989). EM1 fields in substation switchyards
have now been characterized, but currently there is no standard
with which to c o m a r e them. Possible test waveforms for
switchyard fields and for control wires are discussed. 94 WM 146-1
PWRD by the IEEE Substations Committee of the IEEE Power
A paper recommended and approved
Engineering Society for presentation at the IEEE/PES 1994 Winter
Meeting, New York, New York, January 30 - MEASURED SUBSTATION EM1
CHARACTERISTICS
Measurement Techniaues
A number of different types of transient measure- ments were
required to completely describe first how typical substation EM1
arises and then how it couples
February 3 , 1994. Manuscript submitted September 1, 1992; made
available for printing December 15, 1993.
0885-8977/94/$04.00 0 1994 EEE
-
1870
to other circuits. Switching a disconnector or circuit breaker,
for example, produces a complex sequence of high frequency
traveling wave current and voltage tran- sients on each phase of
the high voltage bus. Travel- ing wave bus current transients
excite the three- dimensional bus structure which acts as a complex
an- tenna, radiating energy into the substation as tran- sient
electric and magnetic fields. Bus current tran- sients can also
couple into low voltage circuits that are connected directly to the
bus. The net transient EM1 at some point in the substation is
therefore the superposition of both conductive and radiative
coupling components. The net transient electric and magnetic fields
at a point include contributions from the ground and from all three
bus phases, as well as scattering from nearby conducting
structures.
In general, the required types of transient EM1 measurements
included currents on the high voltage bus, electric and magnetic
fields at various locations on and above ground, and currents and
voltages on substa- tion control wiring. Switching operations,
faults, and flashovers from lightning occurring on the high voltage
system all cause abrupt arcing discharges between af- fected
substation conductors when their potentials are sufficiently
different. Peak transient current amplitudes produced during arcing
depend on the surge impedances of the conductors and the peak
instantaneous phase-to-ground system voltage. Voltages in transmis-
sion substations generally range between 115 kV and 500 kV. Surge
impedances over this voltage range also tend to be relatively
constant at approximately 350 ohms. As a result, arcing discharges
produce traveling wave currents proportional to the substation
voltage. Since field and control wiring EM1 originate from the
tran- sient bus currents and voltages, their relative amplitudes
are also generally proportional to substa- tion voltage. Therefore,
the levels of EM1 in substa- tions should be expected to be higher
at higher substa- tion voltages.
Overall risetimes of transient EM1 phenomena caused by arcing
discharges on the high voltage system are governed by the effective
charging time constants of the circuit driven by the arc, typically
the entire substation bus structure, and are on the order o f 200
nanoseconds or more in air-insulated substations (AIS). Because gas
insulated substations (GIs) have relatively small dimensions, EM1
risetimes can be up to 10 times faster than in A I S . Bus charging
times should generally be expected to increase as the physical size
of the circuit excited by the arc increases, and vice- versa.
In short, substation EM1 measurements were found to require:
(1) wideband sensors for measuring bus current tran- iients at
levels from 300A to lOkA, electric and magnetic fields up to
100kV/m and 300 A/m, and control wire transients of 10 kV and 100 A
peak amp1 i tudes.
(2) at least four data channels to correlate dif- ferent EM1
measurements simultaneously.
(3) each channel should have a bandwidth of at least 100 MHz and
a dynamic range of at least 45 dB.
( 4 ) 100 m to 200 m long, remotely controllable, analog fiber
optic data links for immunity from EM1 environment and installation
flexibility.
(5) high speed, large memory transient digitizers to record
individual transients at high resolution as well as bursts of
consecutive transients from restriking switches.
selected events including the most significant.
and control system.
(6) versatile triggering t o ensure capture of
(7) fast, large capacity data acquisition, processing
For more detailed discussion of these specific require- ments,
and the mobile transient EM1 measurement system developed to meet
them, see [lo] - [12]. Characteristic EM1 Data
During project RP2674-1, detailed transient EM1 measurements of
bus currents, electric and magnetic fields, and control wiring
currents and voltages were made in 12 separate tests at 8 different
air and gas insulated substations at voltages of 115 kV - 500 kV
addressing the requirements and using the techniques outlined
above. In all, over 800 separate events of up to 4 transient
measurements per event totaling 700 megabytes of data were recorded
during the tests occur- ring between 1985 and 1991. Many results
from these tests have already been reported. For example, switch-
ing transients measured in 115 kV substation are reported in [5].
Many detailed attributes and charac- teristics of switching
transient electric and magnetic fields in 115 kV, 230 kV, and 500
kV A I S and in 230 kV and 500 kV GIs are summarized in [ 6 ] . A
complete dis- cussion of the results o f all types of substation
EM1 measurements of staged faults and switching transients,
including line energizations, can be found in [13] and [14]. These
references also describe the details of the substations, the
switching operations performed, and the various measurement
configurations used to record the data.
Here we present only representative data to characterize
switching transient EMI, emphasizing measurements in 500 kV A I S
and GIs where the levels were found to be the highest. Waveshapes
of bus cur- rents, fields, and control wire EM1 at lower voltage
substations are the same as those provided here; however, the
dominant frequency components tend to in- crease somewhat as the
substation voltage decreases r 6 1 ~ 1 3 1 .
When high voltage switches are operated in AIS, bursts of up to
5,000 to 10,000 individual transients varying 60 dB or more in
amplitude are typically produced during the complete arcing
sequence as the switch contacts open or close. Similar bursts, but
of fewer total transients, occur under similar conditions in GIs.
These complex sequences of events during a switching operation give
rise to an overall macro- scopic, or macroburst, characteristic of
a switching transient. Depending on the type of switch and how fast
it operates, macrobursts may last from 40 ms up to 2 seconds. Any
single arcing component of the macroburst produces one transient
having a specific waveform and i s termed a micropulse. The
specific micropulse waveform is mostly determined by the length of
excited bus and its traveling wave properties [IS] and typi- cally
has a time-varying amplitude duration of no more than 10 to 15
microseconds in A I S and less than 4 microseconds in GIs. The most
severe transients, those of maximum burst size and amplitudes,
occur during operation of relatively slow hand-cranked disconnect
switches. While they do generate much higher frequency components
than disconnect switches, circuit breaker operations were found to
produce much fewer and lower amplitude transients [6] .
Within a macroburst, the highest amplitude micropul se
transients are those that are produced when the switch contacts are
as widely separated as possible and still arc. Such a condition
only occurs once per half-cycle (at 60 Hz), but can occur many
times over consecutive half-cycles for a switch with relatively
slowly moving contacts. Thus, a substantial number (20 or more) o f
these highest amplitude transients can be produced during a single
switch operation, depending on its speed. When characterizing the
maximum EM1 en-
-
1871
vironment of a particular substation switching opera- tion, it
is necessary to measure these highest ampl itude transients. During
disconnect switch opera- tion, these highest amplitude micropulses
occur first on closings 'and last on openings.
Representative micropul se switching transient EM1
characteristic of the highest levels found in 500 kV substations
are presented below. Since the highest EM1 levels were recorded in
500 kV substations, these levels are also representative of the
highest levels found in all substations investigated in this
project. This is generally true of bus current, field, and con-
trol wiring transients. However, peak field amplitudes can vary
significantly within a substation due to a number of factors as
discussed below.
Electric and magnetic fields are vector quan- tities; different
orthogonal components have different peak ampl i tudes. Normally
one component will dominate the other two in amplitude when the
field is measured relatively close to its source. Thus, the
vertical electric field between the bus and ground and the mag-
netic field component perpendicular to the bus and horizontal to
the ground usually have the highest amplitudes. In this study,
these polarization com- ponents were measured on the ground
directly beneath the section of excited bus to standardize the
measure- ment at all substations for comparison purposes, but is an
arbitrary choice otherwise. Amplitudes of electric and magnetic
fields, which generally increase with in- creasing substation
voltage for a given polarization component and measurement
geometry, will a1 so increase strongly as the distance between the
field point and the excited bus (source) decreases. As a result,
electric and magnetic field amplitudes at other loca- tions within
the substation can easily be much higher or much lower than those
referenced to the ground directly below the excited bus. Moving the
field measurement point only a few meters closer or farther away
from the bus relative to the arbitrary ground reference location
can easily cause peak field amplitudes to change by a factor of 2
or more [6]. For this reason, the peak electric and magnetic field
ampl itudes reported below are only representative of those on the
ground (approximately 8 meters below the excited bus) in 500 kV
substations; higher amplitudes could be measured closer to the
bus.
Bus Current Transients. Figure l(a) shows the sensor
installation used to measure the transient bus currents from
switching operations in a 500 kV AIS. The sensor and fiber optic
transmitter can be seen on the closest phase conductor about midway
between the column CT on the left side of the photo and left side
of an .air- break motor-operated disconnect switch on the right.
The fiber optic cable providing remote control to the transmitter
and bringing the bus current transient sig- nal to the receiver
inside the measurement van is the vertical line seen dropping from
the transmitter. The fiber optic cable also provides the required
electrical isolation from the 500 kV bus potential.
A bus current transient (typical of the highest ampl itude bus
transients) produced during a disconnect switch operation that
excited a short section of 500 kV bus is shown in Figure 2. The
zero-to-peak amplitude and risetime are about 2.3 kA and 400
nanoseconds. The transient damps out to zero amplitude in about 10
to 15 microseconds; it has a dominant frequency component near 0.5
MHz, with other significant components up to about 3 MHz.
Electric and Mametic Field Transients. Figure l(b) shows the
location of the vertical component electric field sensor (on the
left) and horizontal component
Bus current transient measurement configuration.
(b) Ground-plane electric and magnetic field transient
measurement configuration under bus.
Figure 1. Switching Transient Measurement Geometry in 500 kV
AIS
magnetic field sensor (on the right) on the ground directly
below the location of the bus current sensor shown in Figure l(a).
The magnetic field transient shown in Figure 3(a) was measured at
the same time as the bus transient measurement shown in Figure 2
and has a zero-to-peak amplitude of about 92 A/m. Note the strong
similarities in the bus'current and magnetic field waveforms and
spectra. This is because this com- ponent of the magnetic field is
directly proportional to the current on the bus,
H = l / (xh) 9
where H is the magnetic field in A/m, I is the bus cur- rent in
A, and h is the height of the bus above the ground (7.58 m). The
factor of 2, normally present in the denominator of this
expression, does not appear
-
1872
+ Q
L U
ul a m
-6 -2 .50 ' ' ' ' ' ' ' '
. 0 2 . 8 4.0 6 . 0 0 . 0 10.0 *le
Time (seconds)
Figure 2. Measured Bus Current Transient from Opening a
Disconnect Switch in a 500 kV AIS. (Event C2.628)
8.9 10.0 nI0-6 .e 2.0 4 . 0 6 .0
Time (seconds)
(a) Magnetic field transient.
C4.628 Open DSS4818 Uatleu Sub S08kU OCD-60 10"s +6dB
c, y 2.6u .3 ...
.......................................................................
c ~ , , , ~ , , , , , 1 , , , , , , , , , 1 , , , , ~
-6 -2.0
.e 4.0 8 . 0 12.0 16.0 20.0 -10
Time (seconds)
(b) Electric field transient (early-time only).
Figure 3. Magnetic and Electric Field Transients Measured on the
Ground from Opening a Disconnect Switch in 500 kV AIS (Event
C3.628)
since the horizontal component ot the magnetic field amplitude
tends to double upon reflection from the ground at normal
incidence.
The electric field transient, measured simul- taneously with the
bus current and magnetic field,is shown in Figure 3 (b). The
electric field rises to a peak amplitude of about 13.2 kV/m in
about 1
microsecond. The electric field transient exhibits a waveshape
that is characteristically different from the magnetic field and
bus current transients. This is be- cause the electric field is
proportional to the charge on the bus, i.e., the time integral of
the bus current. Since the charge is equal to the product of the
peak phase-to-ground voltage (408 kV for 500 kV system) and the
capacitance of the transmission line (the latter being further
related geometrically to the surge impedance), the vertical
electric field below an ex- cited bus can be expressed as [16]:
377 E = - F [ z, h]
where E is the electric field in V/m, V is the phase-to-ground
voltage, 2 , is the surge 'Qmpedance (about 328 ohms for this
location), and h is the bus height above the electric field sensor.
If the simple, single transmission line over a ground plane is used
to calculate the line capacitance or surge impedance, rather than
the more complicated (and exact) three phase model, the simple
expression given above over- predicts the observed peak electric
field somewhat. Again, the factor of 2, normally present in the
denominator of the E-field expression, has been sup- pressed since
the vertical electric field component tends to double upon ground
reflection. The electric field waveform is characterized by damped
oscillatory and quasi-static components. The oscillatory behavior
is due to the damping of the time-varying charge, whereas the
quasi-static component results from the fact that the switched
section of isolated bus is left in a trapped charge (or non-zero
voltage) state until the next transient (arc) occurs. For this
reason, electric fields do not damp to zero amplitude over time;
rather, they step from one quasi-static voltage state to another at
the rate at which arcs occur during the switching operation. Since
the rate at which arcs occur can vary from 40 kHz to 120 Hz during
disconnect operations, the duration of the electric field quasi-
static component steps can vary from 25 microseconds up to about 10
milliseconds. The highest amplitude electric field transients, such
as shown in Figure 3(b), occur at a repetition rate 120 Hz, and
therefore typically have durations of 10 milliseconds. The
quasi-static component of the last electric field tran- sient
produced on opening a disconnect can obviously persist for longer
than 10 milliseconds. Examples of quasi -static (1 ate-time)
electric field measurements may be found in [6].
Representative electric and magnetic fields measured on the
ground beneath the gas enclosure near the gas/air bushing in a 500
kV GIs produced by a dis- connect switch operation are shown in
Figure 4. Four things are readily apparent when comparing GIs
fields with those of AIS at the same voltage and for the same type
of switching operation. First, the principal fre- quencies of GIs
transients are at least 10 times hi her than those found in AIS,
undoubtedly due to the smafler substation dimensions. Second, the
peak field amplitudes in GIs are somewhat lower (E-field a factor
of nine, H-field a factor of two) than those of AIS; the gas
enclosure probably acts as a shield. Third, the durations of the
GIs transients are much shorter than those o f AIS. Fourth, the GIs
electric field no lon er has a quasi-static component and damps to
zero ampqitude as fast as the magnetic field. Both of the last two
observations are probably due to the fact that the gas enclosure
(the source of most of observed tran- sient fields) was grounded at
many different locations in these substations.
Control Wirinq Current and Voltaqe Transients. Figure 5 presents
control wire data representative of the highest levels measured on
terminal strips near relay
-
1873
H < t > < W m > Event C3.523
20.0
10.0
-10.0
-20.0
-30.0 .e a.0 4.0 6.0 8 .0 10.0 xle-.
Time (seconds)
(a) Magnetic field transient.
E < t > Event C4.523
Time (seconds)
(a) Control wire voltage transient.
(b) Control wire current transient. (b) Electric field
transient. .
Measured on Ground Under Gas Enclosure Near Gas/Air Bushing from
Opening a Disconnect Switch in a 500 kV G I s (Event C3.529)
disconnect switching).
Figure 4. Magnetic and Electric Field Transients Figure 5.
Control Wire Transients Measured Between C Phase CT Wire and Ground
on the Terminal Strip Near Relay in a 500 kV A I S Control House
(Event C2.643, from
equipment inside of the control house of a 500 kV A I S . A
(zero-to-peak) voltage of about 4 kV measured between a CT wire and
ground is shown in Figure 5(a). This transient rises to peak in
about 200 nanoseconds and continues to ring at low amplitude for up
to 25 to 30 microseconds. Figure 5(a) shows significant frequency
components as high as 20 MHz. The corresponding cur- rent transient
measured at the same time i s shown in Figure 5(b). The current on
the C phase CT wire rises to a peak o f 10.5 A in about 3
microseconds (it ac- tually rises to an initial, slightly lower,
peak of 9.5 A in 700 nanoseconds) and damps out in 25 - 30
microseconds. Major frequency components occur near 0.25, 0.50, and
1.0 MHz. The frequencies of the two transients are very different.
This suggests that the impedance varies with frequency.
Summarv of TvDical Highest Measured EM1 Levels In the
substations examined in this project, EM1
levels were found to be the highest in 500 kV substa- tions (the
highest tested). Disconnect switching was found to produce higher
amp1 itude transients than cir- cuit breaker switching. Table I
summarizes how peak EM1 levels were found to scale at other voltage
levels of AIS for both disconnect and circuit breaker switch- ing.
The peak levels reported are nominally the highest observed for
each type of transient at substa- tion voltages of 115, 230, and
500 kV, that is, they are averages over many measurements made at
each sta-
tion under exactly the same measurement conditions while trying
to record the highest EM1 levels produced by a particular switching
operation. Table I sum- marizes the peak EM1 levels measured for
bus current transients, principal component electric and magnetic
fields on the ground beneath the excited bus, and field-induced
current on a 26.5-111 open circuited test cable. This latter
measurement was performed to aid in the validation of the coupling
models described in [17]. In Table I, measured bus currents,
magnetic fields, and test cable currents are reported in terms of
their peak-to-peak values. Zero-to-peak amplitudes may be estimated
(because of the asymmetry of the waveforms) by taking 70% of the
bus current and mag- netic field peak-to-peak values and 50% of the
reported f i el d-driven cab1 e shield peak-to-peak currents.
Electric field amplitudes are already quoted as zero- to-peak,
since their waveforms are unipolar. Peak fields for several
locations above ground are also reported in Table I and indicate
how amplitudes in- crease closer to the bus. Peak levels of control
wire current and voltage transients measured in all' substa- tions
are summarized in [17] and in Table V. Generally, EM1 levels were
found to scale linearly with substation voltage, thus EM1 levels in
substations at voltages above 500 kV are expected to be higher than
those reported here.
-
1874
SYSTEM BUS ELECTRK: MAGWETIC VOLTAGE CURRENT FIELD FIELD
(hV/mJ ( h l 0 (4
Table I
SUMMARY OF NOMINAL PEAK AMPLITUDES OF VARIOUS EM1 TYPES AS
FUNCTIONS OF SYSTEM VOLTAGE AND SWITCHING OPERATION
(measured, peak-peak)
TESTCABLE Open Circuiied
(AI
115
230
488 'h-d88m 15.4 h - 2 3 m 56.1 h-2.3m 14.0 h-l.Om 45.2
h-l.Om
1040 h - L l 8 m 82.2 h-2.3 m 7.0 h-Om 36.7 h-Om 25.8 h-Om
I 5.5 h-Om 70.4 h-Om I 52.3 h-Om
PREDICTIVE SUBSTATION EM1 MODEIS
115
230 500
Model Requirements
Substation EM1 models were developed to aid in un- derstanding
the measurements and to provide a means for estimating EM1 levels
beyond measurement limitations. The goal for models was to
replicate all details, in- cluding waveshape and peak amplitudes,
of the transient EM1 measurements. Since fields arise from
transient bus currents, and since both can produce transient EM1 in
control wiring, models describing each different type of substation
EM1 were required. Thus, traveling wave models of air and gas
insulated substations were developed to predict transient bus
currents and volt- ages and their radiated fields as a function of
a given excitation (switching operation, fault, lightning strike).
The outputs of the bus current and field models were designed to be
the inputs to various con- ducted and radiated cable coupling
models. In this way, control wire current and voltage transients
were linked back to their causes and predicted.
Initial conditions for the source excitations on the bus
determine the levels of all resulting EM1 produced. Disconnect
switching transient calculations were often made for the condition
that twice the peak phase-to-ground voltage appears initially
across the switch gaps. This situation can occur when switching an
isolated section of bus,,with trapped charge. ;n this paper, the
statement; 2PU !nitial condition , 2PU switching transient , and
2PU prediction" all
refer to this condition, not to the amplitude of a transient
that may result from it (which could be 3PU or higher). In this way
estimates of EM1 produced from, for example, 2PU initial conditions
across a switch, or from a 10 kA lightning strike to the bus, were
made relatively easily using the models, whereas their measurement
under precisely these conditions could be very difficult. All the
EM1 models were tested to verify their ability to generate known
results before using them to make estimates in regimes where
comparisons with test results were not possible. Model details and
their validations are discussed in a companion paper [17], and in
[lo] and [18].
EM1 Control Wirinq Couolina Modes
Important conductive and radiative mechanisms for coupling
substation EM1 into shielded and unshielded CT and CCVT control
wiring have been investigated and quantified in some detail using
measured data and transfer functions [17]. These couplinq
mechanisms are
11.8 h-4.88m 9.1 h-2.3m 3.7 h-Om 0.62 h-Om 0.36 h-Om
76.8 h - C l 8 m O.9h-Om 5.3 h-Om 2.4 h-Om ,132 h-8.3Bm 5.6 h-Om
18.9 h-Om 5.3 h-Om
briefly reviewed here to illustrate how estimates of control
wire EM1 levels presented later are the sum of contributions from
each distinct coupling mode. Each mode contributes only a portion
of the total control wire transient amp1 i tude and frequency
content.
Conducted EM1 CouDlinq. When a shielded or unshielded control
cable connects to an EHV CT or CCVT, there is a deliberate coupling
of the control circuit to the hiah voltaae circuit. At sufficiently
high fre- quencies, the CT and CCVT becomes conductively coupled to
the high voltage bus via, e.g., the parasitic capacitance between
the primary and secondary and their Faraday shields. When this
coupling mode is present, a portion of the bus current transient
couples directly to the conductors inside the shielded or
unshielded CT and CCVT cable. This conducted coupling mode is par-
ticularly significant because it is not reduced by shielding the
control cable. This type of coupling can be reduced using surge
suppression devices.
Radiated EM1 Couplinq. In unshielded cables, high frequency
radiated electric and magnetic fields couple directly to individual
control wires and produce cur- rent and voltage transients at cable
loads with little or no attenuation. Transient field coupling to
wires inside shielded cables is a two step process. In the first
step electric and magnetic field coupling induces voltages and
currents to flow on the shield. The tan- gential component of the
electric field over a short length of the shie)d acts as a local
voltage source driving the cables impedance [19]. Similarly, the
component of the magnetic field normal to the effective loop area
formed by the cable shield and any nearby impedance-coupled
conductors (e.g. CT and CCVT ground straps and their grounded
pedestals), causes currents to flow in the loop [lo], [17]. Once
field-induced current transients are present on cable shields, they
can then couple onto the cable conductors. Shield-to- conductor
coupling can occur several ways. One way is via the transfer
impedance that exists between the shield and each wire [19]. The
other way is through the mutual inductance between the shield
pigtail and a conductor at the shield terminations [17]. The effec-
tiveness with which EM1 couples to the conductors from both o f
these coupling modes increases as the frequency increases,
particularly at frequencies above about 1 MHz. Studies [17], [ZO],
[21] indicate that pigtail coupling will be minimized by making the
pigtail length parallel to the conductors as short as possible.
Transfer impedance coupling can be minimized by using high-quality
cable shields, but there is little ad- vantage in providing any
better cable shielding than is required to reduce transfer
impedance coupling to a level below that of pigtails.
When summed, the three control cable EM1 coupling modes just
discussed were found to generally account for the observed control
wire EM1 voltage and current waveforms and levels. It is
interesting to note that, individually, conducted and pigtail
coupling were found to each contribute up to 70 % of the observed
coupling [lo]. However, both of these when added together with the
20% transfer impedance contribution, produced 100 X of the total
wire EMI. The reason for this was that each different coup1 ing
mode contributed EM1 components with different waveshapes,
frequency content, and rela- tive phase.
ComDarina Model Predictions With Measurements
In [17] and [18], overlays of predicted waveforms with those
actually measured for transient bus cur- rents, electric and
magnetic fields, and control wiring currents and voltages for
substation disconnect switch- ing unambiguously illustrate the
capabilities of the models described above. Here, we simply compare
measured peak amplitudes with 1PU and 2PU predictions.
-
1875
(kV) 115 230
The measurement and prediction geometries are identi - cal. Bus
current transients are nominally referenced to the center of the
section of bus being excited; fields are referenced to locations on
the ground directly below the bus sensor; control cable geometry
and coup1 ing (including source impedances) are repre- sented by
the actual CT and CCVT cable installations found i n the 500 kV
substation measured and modeled; and control wire current and
voltage transients are referenced to the inputs of the actual load
impedances measured inside the control house [17]. Predictions for
230 kV and 345 kV substations are based on linear extrapolations
between 115 kV and 500 kV models, since the former substation
voltages were not modeled.
Table I1 compares predicted and measured bus cur- rent
transients produced by disconnect switching in 115 kV - 500 kV AIS.
For each substation voltage, measured bus current peak amplitudes
are given both for the average of all peak amplitudes measured at
the substa- tion and for the highest peak amplitude measured there.
Comparing the two types of measured results may indi- cate that,
while capture of the highest amplitude transient was always
attempted, it was rarely success- ful. In general, measured bus
currents tend to fall between the 1PU and 2PU model predictions and
indicate, that on average, somewhat less than 2PU is measured. Peak
measurements are much closer, but somewhat less than, the 2PU
predictions for all substation voltages. This indicates that the
2PU model predictions provide a reasonable upper bound to the
measured transient bus current level s.
(") (Nm) (Nm) (Ah) 19.2' 26.2 34.7 38.4' 40.2" 45.3 54.4
80.4"
Table I1
COMPARING 1PU AND 2PU PREDICTED WITH MEASURED BUS CURRENT
TRANSIENTS FOR SEVERAL SYSTEM VOLTAGES
(zero-peak)
I I System (1 PU (Average (Peak Voltage 1 Predicted) I Measured)
I Measured) I P s z d ) (kv) (A) (A) (A) (A) 115 295' 330 -496 590'
230 735 860 1203"
** - Interpolated
Tables I 1 1 and IV provide similar comparisons for magnetic and
electric field transients produced by the bus currents. Magnetic
field predictions at 2PU ini- tial conditions are seen to just
bound the peak measured in all substations. Electric field com-
parisons are presented slightly differently. Predicted and measured
amplitudes are reported for both the peak as well as the DC, or
late-time (quasi-static), level since they differ substantially.
The peak electric field is governed by the transient charge re-
distribution on the bus caused during arcing, and is further
affected by the way the excited bus causing the field is loaded (or
terminated). The DC (or quasi- static) electric field is governed
by the phase-to- ground bus potential. Also, a separate column of
peak measured electric field is not provided; the minimum and
maximum values are indicated as adjustments to the average peak
amp1 itudes. Peak measured electric fields are seen to fall much
closer to the 1PU model predict- ions. This suggests the
possibility that either the measurements are too low or the model
predictions are too high. Both possibilities have been considered,
but not resolved at this time. For example, it has been shown in
[6] that measured electric field amplitudes
Table 111
COMPARING 1PU AND 2PU PREDICTED WITH MEASURED MAGNETIC FIELD
TRANSIENTS FOR SEVERAL SYSTEM VOLTAGES
(Horizontal component fields measured on the ground bel ow
excited bus)
(zero-peak)
System I (1 PU I (Average I I Voltage Predicted) Measured) I
~~
I 345 I 61.2'. I 69.0 I 69 I 122.4" I _ _ _ _ - .. 5 0 0 1 89.5'
I 97.4 1 131.5 I 179.0' * - Predicted by TRAFIC model *. -
Interpolated
Table IV
COMPARING 1PU AND 2PU PREDICTED WITH MEASURED ELECTRIC FIELD
TRANSIENTS FOR SEVERAL SYSTEM VOLTAGES
(Vertical component fields measured on ground below excited
bus)
(zero-peak)
actually double at a height of just 1 meter above ground. At
1-meter above ground 2PU predictions and measurements agree much
better. This suggests that perhaps ground plane electric field
amplitudes were suppressed somehow. Since vertical metallic
structures (insulator posts, switchgear supports, etc.) were often
within a meter or so of the measurement location, it is possible
that they could have produced scattered fields which tended to
lower the vertical field component measurement. In [6],
enhancements of the electric field strength by more than a factor
of 4 were reported near grounded structures protruding 2 meters
above an otherwise flat ground plane. An insulator bus support
column is such a structure that protrudes even higher; perhaps the
field near the top of these posts is greatly enhanced and reduced
at locations further away, such as where sensors were placed.
Alternatively, only a single phase of the three phase bus was
modeled. It may be possible that the presence of other two phase
conductors could change the line capacitance (surge impedance)
enough to lower the electric field. Such a tendency was noted
earlier during the discussion of the simple scaling formula for
calculating the peak electric field. Also, the models currently
ignore bus supports and possible effects caused by them. For now,
the matter is unresolved. Using the data in table IV, the 2PU
predictions may be overpredicting ground-plane electric fields by
almost a factor of 2. On the other hand, when comparing 2PU
predictions with peak measurements at heights of 1-m and 2-m above
ground the relative agreement is much
-
1876
closer and, of course, the field ampiitudes are much higher. It
is interesting to note that magnetic field measurements and
predictions were found to agree closely at all locations, and
showed no tendency to double in amplitude over their ground-plane
value at just 1-m above ground.
Table V compares predicted and measured voltage and current EM1
transient peak amplitudes on conductors inside partially-shielded
CT and shielded CCVT control cables at a point near the equipment
load in a 500 kV control house. To validate model predictions, it
was necessary to use actual frequency-dependent load im- pedances.
This is because common-mode and differential-mode measurements at
the inputs of a solid state transformer differential relay reported
in [20] have shown that relay impedance can vary over two or- ders
of magnitude (from 10s to 1,000s of ohms) over a frequency range
from 100 kHz to 100 MHz, and gives dif- ferent frequency responses
dependent upon the input measured. The CT inputs for the relay in
[20], for ex- ample, while having an average impedance of a few
hundred ohms over all frequencies, exhibited a high Q impedance
resonance of about 2,000 ohms at a frequency of about 1 MHz. Most
control wire switching transients observed during this project have
strong spectral com- ponents near this frequency and could be
strongly af- fected by it. In general, load impedances are complex
functions of frequency. This point was made evident during the
tests by observing the dramatically dif- ferent wire transient
current and voltage waveforms that were simultaneously measured at
the rack terminal block near the relay [lo]; but, since these
measure- ments were not made at the relay inputs, one cannot
conclude that observed impedance variations represent actual relay
impedances, some of which were protected with surge suppression
devices. The effect of measured frequency-dependent loads on
predicted wire EMI, is discussed further in [lo] and 1211.
TYPE OF PREDICTION CONTROL CIRCUIT 1 p u I 2PU
MAXIMUM MEASURED
CT(1)
ccvr(2)
For model validations, the effective frequency- dependent load
impedance was obtained by calculating the ratio V(w)/I(w) where
V(w) and I(w) are the Fourier integral transforms of the
simultaneously measured voltage and current waveforms,
respectively. A smoothed fit for the calculated impedance was
developed and used for the model validations. Once models have been
validated, open or short circuit impedances can be used to generate
the extreme load conditions for control wire EMI.
5.9A 11.8A 10.3 A 2.1 KV 4.2 KV 3.9 KV 0.48 A 0.96 A 0.81 A 1.76
KV 3.4 KV 2.4 KV
Table V compares 1PU and 2PU model predictions to the maximum
measured CT and CCVT control wire current and voltage transients
for a 500 kV substation. In each case, the model predicts the total
wire transient EM1 from all important coupling modes based on the
ac- tual cable configurations (see Table V footnotes). For both CT
and CCVT cables, the 2PU initial condition only slightly
overpredicts the maximum measured control wire EMI, indicating that
this model provides a reasonable upper bound in terms of peak
amplitudes. Had the load been an open circuit, the predicted
voltage would have increased (perhaps doubled) from the values
cited in Table V, while the predicted current would have gone to
zero. In the case of a short circuit load, the effect on predicted
currents and voltages would be reversed.
ESTIMATED MAXIMUM SUBSTATION EM1 LEVELS
Using the validated models, estimates of the max- imum
(zero-to-peak) EM1 levels expected in substations from switching
operations, faults, and lightning are summarized in Table VI [lo].
Maximum levels are given for EM1 types including: bus current
transients, switchyard electric and magnetic field transients, and
current and voltage transients on unshielded control wiring as seen
from inside the control house near the cable loads. Electric and
magnetic fields are represen- tative of those found on the ground
directly beneath a section of excited bus. Both vertical and
horizontal electric field components are reported; the vertical
component gives the highest overall peak amp1 itude, but the
horizontal component is typically important for electric field
coupling to cables. Ground currents (CT and CCVT groundstrap and
pedestal currents) are very strong sources of magnetic field coup1
ing to nearby control wiring. Voltage and current EM1 estimates are
given for both open circuit and short circuit loads on unshielded
control wiring and include conducted and radiated EM1 coupling
modes.
Table VI
COMPARISON OF SWITCHING, FAULT, AND LIGHTNING TRANSIENTS
EXPECTED IN SWITCHYARDS
AND ON CONTROL WIRING INSIDE CONTROL HOUSE (zero-peak)
Switching transient predictions are given for dis- connect
operations under 2PU initial conditions across the switch gaps,
nominally the worst case. Switching transient EM1 is reported for
115 kV and 500 kV AIS and for 500 kV GIs. A phase-to-ground fault
occurring at a closed 500 kV circuit breaker, where it and the
local bus both happen to be charged to, say -1PU, at the in- stant
an incoming traveling wave arrives from having closed a remote
switch charged to, say tlPU, was simu- lated as a worst case, 2PU
initial fault condition. The fault was simulated by inserting a
short, low im-
-
1877
a r e based s o l e l y on t h e s h i e l d i n g e f f e c t i
v e n e s s measurements obta ined f o r a metal c o n t r o l
house w i t h windows i n a 115 kV substat ion [13], and do no t
inc lude any e x p l i c i t con t r ibu t ions from f i e l d s
generated i n s i d e the cont ro l house by inadver tent
(ungrounded) conductor penetrat ions. For example, penetrat ions by
bundles o f telephone cables, etc., have been observed t o cause l
o - c a l l y h i g h f i e l d l e v e l s i n s i d e of o
therwise w e l l - sh ie lded communications rooms [lo] which, i n
some cases, could be h igher than those given i n Table V I I .
pedance transmission l i n e t o ground a t the c i r c u i t
breaker. Only the h igh frequency EM1 occur r ing dur ing t h e f i
r s t 20 microseconds o f the f a u l t was predicted, r e s u l t
i n g i n a f a u l t cur ren t o f 15.9 kA; no attempt was made t
o inc lude normal load cur ren ts o r t o p r e d i c t the la te -
t ime, low frequency (60 Hz) f a u l t EM1 which may a lso have
very l a r g e amplitudes. EM1 produced by a l i g h t n i n g s t
r i k e t o a 500 kV A I S was a lso estimated. I n t h i s case, a
10 kA st roke was d i r e c t l y attached t o a transmission l i n
e j u s t outs ide the substat ion t o repre- sent t h e maximum t
h a t might be expected from a sh ie ld - i n g f a i l u r e .
This l i g h t n i n g cur ren t waveform r i s e s t o a peak o f
10 kA i n 200 ns, f a l l s t o 525 A a t 100 microseconds, f a l l
s t o 75 A a t 5 mil l iseconds, and i s f i n a l l y zero a t 10
mil l iseconds. A s t r i k e on a 500 kV substat ion was
considered t o be worse than one on a 115 kV substat ion. It could
be argued t h a t the same 10 kA on a 115 kV s t a t i o n would
produce much h igher f i e l d s on the ground due t o the shor te
r bus height, bu t the f lashovers would occur a t lower voltages,
thus lower ing the surge cur ren t on the bus which dr ives the f i
e l d s .
The da ta i n Table V I show t h a t t h e 2PU c i r c u i t
breaker f a u l t produces t h e h ighes t l e v e l s f o r a l l
EM1 types. The l i g h t n i n g s t r i k e produces l e v e l s
comparable t o those o f 500 kV A I S d isconnect swi tch ing.
Overa l l EM1 i s worse a t 500 kV than a t 115 kV, as expected.
Levels i n 500 kV A I S are genera l l y h igher than those i n 500
kV G I s , t h e l a t t e r be ing lower presumably due t o sh ie
ld inq provided by the gas enclosure. (However, the GIs models a re
a l s o more complex and n o t as w e l l va l ida ted as those f o
r AIS.) The p r o b a b i l i t y o f occur- rence o f EM1 f rom t
h e f a u l t and l i g h t n i n g s t r i k e scenarios should be
low, whereas from swi tch ing opera- t i o n s i t i s r e l a t i
v e l y h igh. These f a c t o r s should be considered i n
choosing whether t o use t h e f a u l t o r the d isconnect sw i
tch ing t r a n s i e n t as t h e upper bound on substat ion EMI.
I n a l l cases, the EM1 l e v e l s i n a 115 kV substat ion are
considerably lower than those o f 500 kV .
Table V I g ives t h e maximum est imated EM1 l e v e l s
expected i n t h e swi tchyard and on unshie lded CT w i r e s i n
s i d e the cont ro l house near equipment loads.
Peak e l e c t r i c and magnetic f i e l d EM1 l e v e l s i n
- s ide o f c o n t r o l houses r e s u l t i n g from t h e same
EM1 sources and scenar ios j u s t discussed were a l s o es- t
imated and are repor ted i n Table V I I . These estimates
Table V I 1
PEAK TRANSIENT FIELD LEVELS INSIDE CONTROL HOUSES
(zero-peak) (Based on cont ro l house sh ie ld ing ef fect
iveness only)
COMPARING MAXIMUM EM1 LEVELS TO STANDARDS
Estimates o f t h e maximum EM1 l e v e l s expected i n subs ta
t ions o f d i f f e r e n t types and vo l tages are sum- marized
i n Table V I i n terms o f peak ampli tudes. For comparing w i t h
standard t e s t waveforms, i t i s more i n - format ive t o
compare waveform ampl i tudes versus t ime o r spect ra l ampl i
tudes versus frequency.
F igure 6 i s an over lay o f three predic ted 500 kV A I S v e
r t i c a l e l e c t r i c f i e l d spect ra: 1) 2PU phase- to-
ground c i r c u i t breaker f a u l t (bo ld curve) , 2) 10 kA l i
g h t n i n g s t r i k e t o bus due t o s h i e l d i n g f a i l
u r e ( d o t t e d curve) , and 3) 2PU disconnect sw i tch ing t r
a n - s i e n t ( t h i n curve) . S ince t h e r e i s c u r r e n
t l y no swi tchyard f i e l d t e s t standard, t h e envelope o f
these spec t ra has been approx imate ly f i t t e d w i t h t h e
smooth dashed curve as a suggested upper bound. The waveform o f t
h e dashed curve i s g iven by a simple double ex- ponent ia l o f
the form:
E ( t ) = A[exp(-at) - exp(-bt ) ] where
E = e l e c t r i c f i e l d ampli tude a t t ime t A = 100
kV/m, the peak e l e c t r i c f i e l d ampli tude a = 1.OE5 Hz,
descr ib ing the l a t e t ime decay r a t e b = 3.5 E7 Hz, descr
ib ing the r i s e t ime
The (10% - 90%) r i s e t ime o f t h i s e l e c t r i c f i e
l d i s 58.6 ns and i t s ac tua l peak ampl i tude i s 99.6 kV/m.
Decay t o 50% of peak ampl i tude occurs i n about 10 microseconds
as seen i n F igure 7. A s i m i l a r waveform can a lso be
generated f o r the magnetic f i e l d simply by sca l ing the e l
e c t r i c f i e l d by an impedance o f about 300 ohms. This w i
l l generate a peak magnetic f i e l d of about 332 A/m. A method f
o r genera t ing these waveforms f o r t e s t purposes i s
described i n [lo].
1 U
8 U
N -1
\ E -'
18
\ U v) -a z l 8 = - 4
U 3
l8
-6
l8' l8. US 18' U' ;me i m
FREQUENCY (Hertz)
Figure 6. Comparing Highest Expected Levels o f E l e c t r i c
F i e l d Spectra i n 500 kV Switchyard w i t h a Suggested Test
Envelope
1) 2PU c i r c u i t breaker f a u l t 2) 10 kA l i g h t n i n
g s t r i k e (sh ie ld ing f a i l u r e ) 3) 2PU disconnect swi
tch ing t r a n s i e n t 4) Envelope
-
1878
TOTAL VERTICAL ELECTRIC FIELD by the fast transient (curve 4) at
all frequencies. The fast transient spectrum was generated using
the
V(t) = A[exp(-at) - exp(-bt)l ~ double exponential waveform:
2PU CIRCUIT BREAKER FAULT
where
V = Voltage amplitude at time t A = 4.57 kV, the peak fast
transient
a = 5.0 E6 Hz, describing the late
b = 1.75 E8 Hz, describing the rise time
SWC test wave open circuit voltage
time decay rate
With these choices of coefficients, the peak Voltage is .0 1.0
8.0 a.. 4.0 1.0 a.. 7.0 0.0 ,.. 4 kV. The rise time is about 10 ns
and the fall time
(to one-half maximum) is about 150 ns. TIME (microseconds)
Figure 7. Suggested Electric Field Test Waveform Compared wi th
Faul t , Lightning. and Disconnect Switching Transient
Waveforms
DISCUSSION AND CONCLUSIONS
The highest expected levels of several types of Similarly, 500
kV control wire Spectra Were EM1 inside substations of up to 500 kV
have been es-
compared with the fast transient and OSCillatOry swc timated
using measurements and models based on measured test waves of
IEEE/ANSI C37.90.1-1989 1151. TO make transient interference data.
In most cases, the 2PU this conuparlson, open-circuited and
unshielded cables switching transient models only slightly
overpredict were assumed. Figure 8 shows an overlay of 5 curves Of
the highest levels of EM1 actually measured. EM1 from voltage
spectral anplitudes for the fault, lightning. switching operations,
faults, and 1 ightning have been and disconnect switching transient
data summarized compared. The 2PU phase-to-ground circuit breaker
fault earlier in Table VI. The other two (solid bold) curves
occurring during remote switching produced the highest are the fast
transient and the oscillatory SWC Voltage EM1 levels overall. The
probabilities of occurrence of spectra. All five curves are labeled
and identified in EM1 from the fault and lightning scenarios
considered the figure. The model predictions do not include loss.
here are believed to be very low compared to that o f Based on
measurements in a 500 kV GIs. losses Will 2PU disconnect switching.
However, protection against would be expected to decrease predicted
Spectral EM1 effects from a 2PU circuit breaker fault may be re-
amplitudes above 7 MHz by a factor of 2 or 3. For quired. Lightning
arrestors, which may reduce EM1 control cables that are shielded,
analysis shows that levels in the bus, were not included in the
model. spectral amplitudes will also be lower at all frequencies
shown for the environments in Figure 8. Generally, EM1 levels from
routine disconnect The alnount of decrease is greatest at low
frequencies switching are shown to increase linearly with substa-
and least (factor of 2) at high frequencies, i.e.. tion voltage.
Measured and predicted bus current tran- above 1 MHz. However, even
for shielded cables, the sients from disconnect switching are
presented and fault, lightning, and disconnect switching transient
agree well at all substation voltages. Bus current spectral
voltages still exceed the envelope presented transients are also
shown to be the origin of observed
transient electric and magnetic fields; simple peak a
OIPCCIIIWIT I-- ALPONCL. mn UIWILWI, Q- amp1 itude field scaling
formulas have been provided.
cI'xalIu* u-xWGc.OI1)* lyLT(- -3. - W. SNWOLD) Maximum predicted
transient electric and magnetic field levels (from disconnect
switching) on the ground under the bus in 500 kV substations are
predicted to be 30 kV/m and 179 A/m, (zero-to-peak). (Faults
produce peak field levels about three times higher.) The ground-
plane electric field level estimate was found to be about a factor
of 2 higher than the measurements; agreement improves at
above-ground locations. Several possible explanations for the
discrepancy were dis- cussed; it was also indicated that electric
fields near grounded structures can even be higher. The 2PU pre-
dicted electric field from disconnect switching is close to the
value reported in [l], but both the measured and 2PU predicted
magnetic field levels found in this study were 80 times higher than
those reported in [I].
FREQUENCY (Hertz) Maximum control wire EM1 inside the control
house has been described as the sum of contributions from
Amplitudes for Predicted CT Cable EM1 principal coupl ing
pathways in unshielded cables: con- from a 2PU Fault, 10 duction Of
bus current transients parasitically through 2PU Disconnect
Switching with the SWC instrument transformers, and coupl ing of
radiated Fast Transient and Oscillatory Test electric and maqnetic
fields.
Fault, lightning, and disconnect switching tran- Voltage Spectra
in 500 kV Substation. sient EM1 coupling to shielded and filtered
control cable loads was also investigated using the models
described earlier [lo]. In shielded cables, fields couple first to
control cable and then to the inner conductors via piqtails and
transfer impedance.
Figure 8 . Comparing Open Circuit Voltage spectral
lightning, and
1) Predicted fault EM1 2) Predicted lightning strike EM1 3)
Predicted di"nect switching transient EM1 4) Fast transient SWC
test wave 5) Oscillatory SWC test wave
-
1879
Comparison of the data in Table VI11 with that of Table IX
illustrates the effectiveness of using shielded con- trol
cables.
Table VI11
PEAK EM1 COUPLING BY MODE IN AN UNSHIELOED CT CABLE (2PU
disconnect switching transient; 150 ohm load)
I COUPLING MODE. 0-PEAK MPLITUDE I CURRENT I VOLTAGE I
1.bb I 0 . Z W 4 I). c ! .)n I. , c ee
Table IX
(2PU disconnect switching transient; 150 ohm load)
I COUPLING MODE. 0-PEAK AnPLITUDE I CURRENT I VOLTAGE 1
PEAK EM1 COUPLING BY MODE IN A SHIELDED CT CABLE
Table X
EFFECT OF SHIELDING AND FILTER CAPACITANCE ON LOAD AND FILTER
EM1 STRESS LEVELS IN CT CONTROL CABLES
(EM1 source: 2PU circuit breaker fault; load: 150 ohms) (500 kV
substation)
I load remonse I I I
Table V I 1 1 shows that magnetic field coupling to the CT
ground strap and pedestal is the most important cou- pling mode,
nearly twice that of conducted coupling, when the CT cable is
unshielded. However, Table IX shows that once the cable is shielded
then conducted coupling dominates all other modes.
High frequency conducted coupl ing through bushing and column
type CTs and through CCVTs was found to be a significant
contributor to control wire EMI. High fre- quency transfer function
measurements of instrument tranformers are recommended to fully
understand this mechanism because coupling is directly from the bus
to the control wire and is not reduced by cable shields. In CTs and
CCVTs employing shields, their effectiveness should be evaluated
between 100 kHz and 150 MHz, unless known. From the CT transfer
function data measured in project RP2674-1 [lo], coupling
efficiency was found to increase at higher frequencies.
Furthermore, because of the short time constants associated with
arcing in the interrupter gaps, switching transients from circuit
breakers, while of much lower overall amplitude than those of
disconnects, do tend to produce EM1 having much higher frequency
components; these frequencies may couple more efficiently through
the CT to control wiring.
Surge suppressors such as filters and MOVs are also very
effective in reducing EM1 levels, especially the conducted EM1
mode. Simple capacitors of 0.5 uF, 0.05 uF, and 0.01 UF are
typically used to protect relay equipment, with the latter two
values more typi- cal for protecting solid state and digital
relays, respectively. When appropriate cable shields and fil- ters
are used, even EM1 levels from the severe 2PU cir- cuit breaker
fault can be reduced significantly at relay equipment inputs. This
is illustrated in Table X for various shield and filter
combinations. Proper sizing of EM1 protection devices such as
filter working voltage is also important to prevent damage from the
EMI. Information concerning single transient peak volt- age, peak
current, peak power, average power, and peak energy delivered to
cable shields, filters, and equip- ment loads can be easily
calculated using the models; many of these quantities are reported
in Table X.
Disconnect switching transient macrobursts imply a quasi-random
repetitive EM1 stress of up to 5,000 varying amplitude micropulses
on cable loads per switch operation. Up'to 20 - 50 of these
transients occur at maximum amplitude (2PU) at a rate of 120 Hz per
opera- tion. At a maximum control wire EM1 level of 10 A and 3 kV
per pulse, this amounts to 30 kW peak power per pulse. At 120 Hz
this amounts to about 90 watts average power. For 50 pulses of 25
microsecond duration, this represents an energy of about 38 joules
delivered to the cable load (or to a surge protection device) per
disconnect switching operation. Cable loads and protec- tion should
be designed to withstand these EM1 levels.
The shielding and protection recommendations from [l] are
reconfirmed here. Because of the possibility of conducted coupl
ing, surge protectors are recommended in addition to cable
shielding. Surge protectors may be required anyway to limit pigtail
coupling to compatible levels.
Analytic descriptions of possible test waves characterizing
bounding levels of electric and magnetic field EM1 in switchyards
up to and including 500 kV have been presented. Comparisons of
expected control wire EM1 levels with standards suggest possible
in- creases in the fast transient test wave at certain fre-
quencies below 7 MHz be considered. The oscillatory test wave was
not found to provide an effective bound on the expected maximum EM1
levels for substations greater than 115 kV.
Susceptibility levels for control-critical sub- station
equipment, not just relay equipment, should be determined, if not
known, and compared with expected maximum EM1 stress levels and SWC
test criteria to en- sure that adequate margins exist. During
substation switching transient tests, many upsets and damage o f
non-relay equipment occurred [lo].
ACKNOWLEDGMENTS
The authors are grateful for the guidance provided by F . M.
Phillips, S . L. Nilsson and L . L. Mankoff of EPRI. Public Service
Company of New Mexico and Virginia Power have been strong
supporters throughout this project.
-
1880
REFERENCES
B.O. Russell, S.M. Harvey, and S.L. Nilsson, "Sub- station
Electromagnetic Interference, Part 1: Characterization and
Description o f the Transient EM1 Problem", IEEE Trans. on Power
Apparatus and Systems, vol.
S. A. Boggs, F.Y. Chu, N. Fujimoto, A. Krenicky, A. Plessl, and
D. Schlicht, Disconnect Switch In- duced Transients andII Trapped
Charge in Gas- Insulated Substations , IEEE Trans. PA&S, vol.
PAS-101, pp. 3593- 3602, October 1982.
PAS-103, no. 7, July 1984.
D.E. Thomas, C.M. Wiigins, F. S. Nickel, C. D. K O , and S. E.
Wright, Prediction of Electromagnetic Field and Current Transipts
in Power Transmission and Distribution Systems , IEEE Trans. on
Power pelivery, vol. 4, no. 1, pp. 744-755, January 1989.
E.F. Vance, Coupling to Shielded Cables, New York: Wi 1 ey-
Interscience, 1978, pp ., 108-176.
HEMP-INDUCED TRANSIENTS IN TRANSMISSION SUBSTA-. m. Oak Ridge,
TN: Oak Ridge National Laboratory, June 1990,
ORNL/Sub-88-SC863.
[21] D. E. Thomas, C. M. k/iggins, T. M. Salas, and P. R.
Barnes, On the HEMP Environment for Protective Relays", presented
at the IEEE/PES 1993 Winter Meeting in Columbus, OH.
J. Meppelink, K.Diederich, K. Feser, and P. Pfaff, "Very Fast
Transients in GIs", IEEE Trans. on Power Delivery, vol. 4, pp.
222-233, January 1989.
S.Ogawa, E.Haginomori, S.Nisiwaki, T. Yoshida, K. Terasaka,
"Estimation of Restriking Transient Overvoltage on Disconnecting
Switch for GIs", IEEE Trans. on Power Delivery, vol.1, p. 95
(1986).
C. M. Wiggins, F. S. Nickel, A.J. Haney, "Measure- ment of
Switching Transients in a 115 kV Substation", IEEE Trans. on Power
Delivery, vol. 4, pp. 756-759, January 1989.
C.M. Wiggins and S.E. W;ight,"Switching Transient Fields in
Substations , IEEE Trans. on Power Delivery, vol. 6, pp. 591-600,
April, 1991.
W. C. Kotheimer and L.L. Mankoff, "Electromagnetic Interference
and Solid State Relays", IEEE Trans, on PAE, vol. PAS-96, no. 4,
July/August 1977.
ANSI/IEEE C37.90.1-1989 "IEEE Standard Surge With- stand
Capability ISWC) Tests for Protective Relays and Relay Systems ,
(P472/D9, January 6, 1987). ANSI/IEEE C37.90.2-1987 "Withstand
Capabil i ty of Relay Systems to Radiated Electromagnetic Inter-
ference from Transceivers".
[ 101 ELECTROMAGNETIC TRANSIENTS IN SUBSTATIONS, VOLUME I :
PROJECT SUMMARY AND RECOMMENDATIONS, Pal o A1 to, CA: Electric
Power Research Institute, April 1993, EPRI TR-102006.
[ll] C.M. Wiggins, F. S. Nickel, and A.J. Haney, "Mob- ile
Transient Measurement System", 1987 IEEE Inter- national Symposium
on EMC, vol 87CH2487-7, pp. 42- 54, August 1987.
[12] Proceedings: Telephone Lines Enterinq Power Sub- stations.
Palo Alto, CA:Electric Power Research In- stitute, August 1988,
EL-5990-SR, Section 5.
[131 See [lo], section 4. [ 141 ELECTROMAGNETIC TRANSIENTS IN
SUBSTATIONS, VOLUME
UI: TEST REPORT. Palo Alto, CA: Electric Power Research
Institute, EPRI TR-102006, April 1993.
[15] See [lo], pp 5-13 - 5-18.
[16] See [lo], section 6.
[17] D.E. Thomas, C.M. W!ggins, T.M.Salas, F.S. Nickel, and S.
E. Wright, EMI-Induced Control Wire Tra!- sients in Substations:
Measurements and Models , presented for the IEEE/PES 1994 Winter
Meeting in New York.
Carl M. Wiqqins (M'74, SM'89) was born in Jackson, MS on August
5, 1941. He received the B.S. degree from Lamar State College of
Technology, Beaumont, TX and the M.S. degree from Sam H o u s t o n
S t a t e C o l l e g e , Huntsville, TX in 1964 and 1966,
respectively, both in Dhvsics. From 1966 to 1973 he studied
postgraduate physics at New Mexico State University in Las Cruces,
NM. In 1973 he joined BDM
International, Albuquerque, NM. His work has been in the areas
of transient electrodynamics, lasers, and op- tics. Currently, he
is a senior principal scientist investigating electromagnetic
interference phenomena. Mr. Wiggins is a senior member of the IEEE
EMC and Power Engineering Societies and has authored over 38 pub1
ications.
I
7 curved mirrors was published in the December 1980 issue of
Scientific American.
Currently, Mr. Thomas is a Principal Staff Member in
the Advanced Electromagnetics Group of BDM Interna- tional, Inc.
His research areas include assessment of electromagnetic effects on
aircraft, ships, and on electric power transmission and
distribution systems.
-
1881
Frank S. Nickel was born in Salina, Kansas in 1962. He received
the B . S . degree (1984) in physics engineer- ing and a minor in
mathe- matics from Southwestern Ok- lahoma State Universitv (SWOSU)
.
Mr. Nickel joined BDM In- ternational, Inc., Albu- querque, NM
in 1984.. He has concentrated his work in the areas of transient
signal anal ysi s , data acqui sit i on and processing, electromag-
netic and electrical network
model development, systems simulation, and hardware and software
engineering. Currently, he is an engineer for an electromagnetics
test and analysis group, and is a project manager responsible for
the operation, main- tenance, enhancement, and application of a
high bandwidth, versatile data acquisition and processing
system.
and has worked in the areas of electromagnetic com- patibility
and transient analysis. He is a student
member o f the IEEE Electromagnetics Compatibility Soci etv.
Selwvn E. Wrisht (M86, SM88) received the B.S. de- gree in
physics from North S t a f f s P o l y t e c h n i c in England,
the M . S . degree in electronics in 1964, and the Ph.D. degree in
acoustics in 1 9 6 9 from S o u t h h a m p t o n University. He
became a chartered engineer in the United Kingdom in 1965.
Dr. Wright was Scientific Advisor with the French Government
(ONERA) from 1976-1978. In 1978 he joined
Stanford University to start a laboratory in acoustics. Dr.
Wright joined the Electric Power Research Institute in 1984, as a
project manager in the Electrical Systems Division. His specialties
include acoustic and electromagnetic fields, control s , and
instrumentat ion. Dr. Wright has authored over 50 principal
publications.
-
1882 Discussion
Steven A. Boggs (Electrical Insulation Research Cen- ter,
University of Connecticut, Storrs, CT 06269-3136 and Department of
Electrical Engineering, University of Toronto). The authors are
congratulated on an interest- ing and timely study of transient
electromagnetic inter- ference in substations. My comments will be
restricted to the subject as related to GIS (SF6 Gas-Insulated Sub-
stations). The authors might have done well to separate their
results into two papers, as phenomena in GIs differ SufEciently
from those in AIS that treating both in a sin- gle paper is almost
certain to result in confusion. For example, the authors state that
Voltages in transmission substations generally range between 115 kV
and 500 kV. Surge impedances over this voltage range also tend to
be relatively constant at approximately 350 ohms. This statement is
correct of AIS but incorrect for GIS, where the impedance ranges
from about 45 to 65 ohms. In the next paragraph, the authors relate
the risetimes of tran- sient EMI phenomena caused by arcing
discharges on the high voltage system to the effective charging
time con- stants of the circuit driven by the arc, typically the
entire substation bus structure. Again, this may be an appro-
priate description for AIS but is inappropriate for GIS, where the
coaxial structure forms a system of relatively clean high frequency
transmission lines capable of supporting ns risetime travelling
waves and reflections thereof. Given that the typical time for
collapse of the voltage across a disconnect switch is in the range
of 3 to 5 ns, the frequency spectrum resulting from switching of
GIS is related to details of station structure more complex than
simply overall bus capacitance.
The authors state that the peak field amplitudes in GIS are
somewhat lower (E-field a factor of nine, H-field a factor of two)
than those of AIS; the gas enclosure prob- ably acts as a shield.
Where did the authors measure the fields in GIS? One might expect
the maximum field to occur immediately under the gas-to-air
termination. The intercontact breakdown of a disconnector in GIS
creates a travelling electromagnetic wave within the GIS which
reflects and refracts within the GIS until it reaches a gas-
to-& termination, at which point part of the wave is re-
flected back into the GIS, while part is refracted out the overhead
line. The gas-to-air termination represents the junction of three
transmission lines, viz., (1) the overhead line-to-earth
transmission line, (2) the GIs conductor-to- enclosure transmission
line, and (3) the GIS enclosure- to-earth transmission line. Part
of the refracted wave is coupled into the GIS enclosure-to-earth
transmission line. The duration and waveform of this wave depends
strongly on the length and proximity of any ground straps, as the
base of the bushing is often grounded to the station ground mat
[l-31. The field below the bushing will normally be greatest when a
line disconnector is op-
erated, when the enclosure at the base of the bushing is high
above the earth, and when a ground strap is not present near the
base of the bushing. Did the authors measure the field under
bushings, and, in particular, did they measure the field under a
bushing which represented such a worst case condition?
The authors relate many of their measured data to a 2 pu
breakdown across a disconnect switch. he operation of GIs
disconnectors has been understood for over a dec- ade, and
manufacturers take great pains to minimize the likelihood of
anything approaching a 2 pu intercontact breakdown, as such a
breakdown implies both large tran- sients and a very long
intercontact arc, both of which reduce the reliability of the GIs.
The intercontact break- down voltage is basically controlled
through careful at- tention to both the rate of contact separation
and the asymmetry between breakdown voltage in the two direc- tions
across the disconnector [4,5]. Thus while relating data to a 2 pu
intercontact breakdown provides a reason- able basis for
normalizing reported data, the resulting fields are substantially
greater than can reasonably be expected in a GIs. The authors
bottom line is that the measured fields are sufficiently great that
the IEEE/ANSI standard C37.90.1- 1989 does not assure reliable
operation. However, the authors measured transients in substations
based on the standard practice which has grown up around the use
electromechanical relays. In many early GIS, control wiring
practice was so poor that even electromechanical relays could be
damaged, especially by breakdowns dur- ing commissioning tests,
which represent the highest normal exposure to EMI and control
wiring transients. The usual reason for such damage was that
control wiring shields were grounded at only one end in order to
avoid circulating currents in the shields. When the shield was
grounded at both ends, damage to electromechanical re- lays was
eliminated. However, the resulting wiring prac- tice was still
poor. Control cable shields were grounded through long pigtails,
including such pigtails within local cabinets. Such practice,
however imperfect, has been adequate for substations based on
electromechanical re- lay technology. More recently, a number of
utilities have started to use microprocessor-based relaying in GIS,
and at least three manufacturers have provided GIS which have
incorpo- rated or been used with such technology. To implement such
systems, control wiring had to beimproved substan- tially, with
coaxial termination of control cable shields on the outer surface
of wiring cabinets, elimination of pig- tails, etc. The net result
is a control wiring system in which the control wires travel
through a continuous fara- day cage from their point of origin to
their point of termi- nation. As noted above, such technology is
now well-es- tablished with at least three manufacturers having
imple-
-
mented GIs incorporating such technology on a commer- cial
basis. Thus the authors' measurements do not relate to the eas- ily
achieved and established state-of-the-art in GIS con- trol wiring
but rather to earlier generations of technology which was, and is,
adequate when electromechanical re- lays are employed but which is
clearly an inadequate basis for implementation of
microprocessor-based relay- ing. The soon-to-be-published revision
of IEEWANSI docu- ment on GIS includes a revised Specification and
numer- ous Guides, some of which cover Fast Transients in GIs,
Transient Groundrise in GIs, and Control Wiring Prac- tice for
GIs.
None of the above comment detracts from the very sub- stantial
contribution of the present authors. Hopefully, my comments serve
to place these contributions in the context of the present
state-of-the-art as it relates to GIs. 1. Ford, G.L. and S.A.
Boggs. Transient Groundrise in SF6
Substations Investigated. Transmission & Distribution
Magazine, Vol. 31, No. 8, Aug. 1979.
2. Ford, G.L., S.A. Boggs, and N. Fujimoto. Transient Groun-
drise in GIs. Transmission & Distribution Magazine Vol. 34, No.
4, April, 1982, p. 42.
3. Fujimoto, N., E.P. Dick, S.A. Boggs, and G.L. Ford. Tran-
sient Ground Potential Rise in Gas Insulated Substations-
Experimental Studies. IEEE Trans. PAS-101, October, 1982.
4. Boggs, S.A., F.Y. Chu, N. Fujimoto, A. Krenicky, A. Plessl,
and D. Schlicht. Disconnect Switch Induced Tran- sients and Trapped
Charge in Gas-Insulated Substations. IEEE Trans. PAS-101, October,
1982.
5. Boggs, S.A., N. Fujimoto, M. Collod, andE. Thuries. The
Modeling of Statistical Operating Parameters and The Computation of
Operation-Induced Surge Waveforms for GIS Disconnectors. 1984
CIGRE, paper 13-15.
Manuscript received February 15, 1994.
Carl M. Wiggins, David E. Thomas, Frank S. Nickel (BDM Federal,
Inc., Albuquerque, NM 87106) and Selwyn E. Wright (Electric Power
Research Institute, Palo Alto, CA 94303). The authors would like to
thank Mr. Boggs for his thoughtfbl comments. There are indeed
significant differences between transient electromagnetic
interference in air-insulated substations (AIS) and gas-insulated
substations (GIs). He is correct in pointing out the differences in
surge impedances in AIS and GIs. While the high fiequency traveling
waves from an arcing switch do excite the entire bus structure,
1883 we have also shown that it is the large impedance
discontinuities on either side nearest the arcing switch, and to
some extent the details of the switch itself, that are
predominately responsible for the observed transient waveshape [
101 (references in paper). Typically such discontinuities are
provided by the nearest instrument transformers and (open) circuit
breaker bushings in the case of disconnect switching in AIS which
offer low impedance paths to ground at frequencies in the MHz
range. In GIs, the bus structure is more complicated and often may
be viewed as three different intersecting transmission lines as Mr.
Boggs points out. The surge impedances for the three lines are all
different. High frequency electromagnetic interference produced by
arcing from a switch operating inside the coaxial line couples onto
all three lines. Circuit breakers and disconnects (with its
associated breaker open) were operated to produce the measured
transients; no external line switch operations were allowed. Due to
the compactness of the coaxial structure, switching in GIs produces
transients with much higher frequency components than in AIS, as we
point out in the paper (e.g., Fig. 4) and in [6,10, 141. We
measured transient electric and magnetic fields at 14 different
locations between the gas enclosure and ground and under the air-
insulated line on both the line and transformer sides [lo], but not
between the bus and gas enclosure (i.e., not inside the gas
enclosure), as mentioned in the paper. The peak fields may be much
lower at these locations than would be found inside gas enclosure.
In some cases fields were measured close to aidgas bushings [ 141
and were found to be somewhat higher at these locations,
particularly the magnetic field. It should also be mentioned that
the field measurement geometry between the enclosure and earth
ground, which is often a three- dimensional structure of steel
wakays , is not "clean". Thus, field sensors will tend to measure
the net field scattered from these structures which may cause local
peak amplitude enhancements or cancellations. Of course, the bottom
line is: how do transients arriving at equipment in GIS compare
with those found in AIS? Many results in [lo, 141 show that,
generally, voltage and current transients have about the same peak
amplitudes in GIS and AIS where the same type of switching is
performed at the same voltage levels, and similar interference
coupling to equipment (cable practices) are employed. The major
differences were that GIS control cable transients measured near
equipment loads exhibited major fiequency components up to 20 MHz
(vs. up to - 3 h4Hz in AIS), and they damped out faster ( 4 p vs.
20 p).
The 2 pu initial condition across the disconnect switch arises
when its the line-side has been left charged to +1 pu on opening
and then it is closed where the initial arc
-
1884
occurs at the instant the source side is at amhimum, -1 pu. The
correlation between peak MeasuTemenfs and travehg wave simulation
model predictions set for 1 pu and 2 pu initial conditions c~njirm
the possibility of occurrence of 2 pu transients (see for example,
Chapter 5 of [ lo]).
Our observations of cable shielding practices found in the 10
AIS and GIs visited during project RP2674-01 show: only some use of
shielded cables; sometimes cable shields were grounded; and when
grounded, pigtails were used. Rarely did we encounter a substation
where all cables were .&ielded. We have not seen a substation
using high frequency coaxial shield terminations, but we are
pleased to learn that apparently some now do. We would also like to
point out that a key conclusion in the paper is that cable
shielding alone, without proper surge
protection, will not necessarily safely limit the peak
interference levels from switching, hults, and lightning transients
that can occur at inputs to protection equipment. The paper
characterizes quantitatively some of the peak EMI levels for normal
and abnormal high frequency sources in AIS and GIs, their effects,
and the effectiveness of various mitigation procedures such as
cable shieldmg and surge suppression. The purpose of this research
is to provide high fiequency EMI data that could be used to better
understand and improve protection of modem (digital and
microprocessor) substation electronic equipment placed either
inside the mtro l house ot in the high yard (AIS and GIs).
Manuscript received April 19, 1994.