Académique Documents
Professionnel Documents
Culture Documents
1969
Background
Relaying Philosophy
Faults must be cleared fast and selectively. To avoid generator
instability which could readily lead to a widespread blackout,
a three-phase short circuit near a modern generating station
must be isolated from the healthy parts of the system in about
0.15 second. In this interval the relays must energize the trip
coil of the circuit breaker, wait about 0.07 second to see if the
appropriate breakers open, and then trip other breakers if oine
of the proper ones fails to interrupt the fault current. Of course,
whether a stability problem exists in a given application, the
objective is to clear fast, minimizing damage and disturbance
to loads.
Fast clearing, per se, is quite easy; the problem is to do it
selectively. A surgeon who excises the heart during an appendectomy loses the patient and gets sued for malpractice. So it is
with the protection engineer if service interruptions result from
opening an excessive number of breakers to clear a fault. Many
relays sense a given fault, but only those protecting the faulted
zone should initiate circuit-breaker opening.
While a multitude of highly specialized and sophisticated
relays have evolved, for our purposes we need only at this point
categorize their underlying principles. To sense a fault (or
perhaps locate it), all electrical relays use one or more of the
following.
1) Level detection: abnormally high current, low frequency,
voltage, for example.
2) Magnitude comparison: the magnitude of voltage compared to line current provides a basis for distance measurement,
439
{:}--
I-El
SHUNT REACTOR
-1E
CAPACITOR
AUTO TRANSFORMER
TWO-WINDING TRANSFORMER
POTENTIAL TRANSFORMER
paramount problem.
GENERAL APPROACH
Analog-to-Digital Conversion
For the station in Fig. 1 the computer (or peripheral gear)
must convert to digital form the instantaneous value of 129 ac
quantities (35 sets of currents, 8 sets of voltages). Each is
sampled every 0.5 ms. With such a fast rate the station control
cabling must be well shielded to minimize both magnetic and
capacitive coupling. However, this does not relieve the program
completely from avoiding tripping should a surge introduce a
substantial error into a sample.
Rated phase to ground voltage is 69 V rms, rated current is
5 A rms. Accurate distance measurement by the program should
occur down to 5 V peak; two successive voltage samples then
could be less than 0.5 V when the wave is crossing zero. A-D
conversion error should not exceed 3 percent of the correct value
(not full scale) over the range of 0.5-100 V, instantaneous.
Similarly, the current conversion error should not exceed:
3 percent for 0.5-100 A, instantaneous;
10 percent for 0.2-250 A, instantaneous.
Certain readings will be part of the same computation, so
care must be exercised as to the order in which all these quantities are sampled. Otherwise, the program must take the time
to correct for the nonsimultaneity of related readings.
The objective of 4-ms tripping of severe faults fixes the
sampling interval of 0.5 ms. A minimum of six postfault samples
is needed to trip a transmission line fault, so the maximum
speed is about 3 ms for this zone.
Data Storage
440
Prefault Routines
TABLE I
ROUTINE LISTINGS
Designation
Described
in Appendix
AD
BDF
BF
CPD
CPDSI
EX
FIL
FLOC
MA
OBK
OPX
OPY
ORTT
PDFC
QD
SDF
TOP
TDF
TRIP
VFD
WS
Zi
ZlGS
ZIPHS
Z2GS
Z2PHS
Z2T
Z3G
Z3PH
Z3GS
Z3PHS
Z3T
XIII
XI
XV
VIII
XII
XXI
XXII
XXI
Fig. 7
XI
XIV
II, X
XVI
XVIII
V, VII
V
v
V
XVIII
XVI
XVI
V, VII
V
XIX
Description
Analog-to-digital conversion of instantaneous value of ac voltages and currents
Bus differential
Breaker failure
Line current peak determination
Differential current peak determination
Executive
Voltage filtering
Line-fault locator
Memory action for Z3PH
Open breaker keying
Operation logic (90-270)
Operation logic (0-180)
Overreaching transfer trip
Current comparison
Quadrant determination
Station overall differential
Turnoff of fault programs
Transformer differential
Breaker tripping
Voltage fault detector
Waveshape analysis
Zone 1 distance logic, phase and ground
Zone 1 ground-distance calculations
Zone 1 phase-distance calculations
Zone 2 ground-distance calculations
Zone 2 phase-distance calculations
Zone 2 time-delay trip
Zone 3 ground-distance logic
Zone 3 phase-distance logic
Zone 3 ground-distance calculations
Zone&3 phase-distance calculations
Zone 3 time-delay trip
INTERRUPT
EACH
SAMPLE
EX
EVERY
64TH SAMPL
TD
EX
441
TRIP
MA
CPDIL
(b)
ORTT
Z2T (4*)
Z3GIPH
BDF/TDF
VFD
UPON
TRIP
ENTERED
BY:
(C)
Z3G/PH
BR
BR
BR
3
3
(C)
VFD
CTR. > 0
Z3G/PH
Z3G/PH (*)
Z2T (4#)
INTERRUPT
NONE
(e
VFD
TRIP
Z3PH & CPDI L
5
7
(SA)
(#)
( f)
16BMS
(4)
TOP
(5)
16)
(8)
(4*
(3)
BDF/TDF
(9
(6)
FLOC
Z3GJPH
SDF
5)
F
(8)
I5
n)
4* ) (8 -10I
BIDFrrDF (# ) (5e 9
#
I
(SR)
IZ1 )
Z3G/PH
-- -
__
Z3GPH(#)
W
Z
INI
Zi
BDFITOF
l~~~~~~~~
Z3G/PH
Z3G/PH
Z3G/PH
* ---PATIULR ON
LE ZGEND
r8
I#)
0
BDFfTOF
Z3GfPH I4*)
PARTICULA
ZONEl
I#4)
Z3G/PH
ZI
0--ALLLMES ON SAME
I8S 5)
I8 -7)
(8)
(9)
410)'
(BR)
(4*)
I8S -3)k>
(3)
4)
BDFJTDF
)9
Z3G/PH
(a)
BDF/TDF
(9)
Z3G/PH
II
ZI (4*)
3)
IS 314)1
OR (5
OR (7 -M 0))
I
Zi
(53)'
(3)
G
(4*) (9
8
B
.gskRS,2&4 LINEIK)
DELAY
BR
___
_3
(9)
(2-n)
10
5~
___
NOTES
PARTICULAR BREAKERS RELATED TO FAULTED ZONE
VOLTAGE LEVEL AS
LINE CHOSEN BY FLOC
BID
(C)
6
8
8
8
9
9
11
SA
NON-FAULT
BDF/TDF (*#)
TRIP
TDF
FLOC
UPON
6VFD
Z3G/PH (#*)
Z3G/PH (*
BDF
|SDF
NO
FAULT
FOUND
NONE1
S5
CPC (#)
VFD (2)
Z3G/PH
CTR.
>0
()5
FLOC
Ws
Z3G/PH
UPON
FLOC
OPER.
SDF
(a)
SDF
OPER.
3
3
3
3
Z3T (4#)
OBK
TOP
BF
PDF
CTR.
>0
UPON
VFD
OPER.
I
2
2
2
2
3
VFD
CPDSI
ZI (D*)
UPON
PREFAULT
ROUTINES
(8
3)I
-10O
(BR)
BR
VFD
FLOCOR Z3 CTR.
VFD
VFD OR EX
SA
PARTICULAR ZONE
442
12
ZONE 3
TIUVIA
'm.
ZONE 2
od
TA
.---
(b)
(a)
STA. R
STA. S
General
Faults cause aberrations in the currents and voltages. At
fault inception a sudden rise or drop in voltage can occur.
Current and voltage peaks can change in magnitude and/or
phase position with respect to prefault conditions. Of course,
these aberrations may also result from switching or load changes.
Therefore, the fault-detection logic must not be so sensitive
that nonfault programs are being interrupted too much for
"false alarms."
VFD looks only at one set of voltages for each kV level as
determined by potential selection logic referred to in Appendix
XXII. It ignores currents to minimize prefault duty since
there are many more currents than voltages (11 sets of line
currents, 3 sets of voltages in Fig. 1). Furthermore, voltages
can change magnitude instantaneously while currents cannot.
So the use of voltage offers faster fault detection.
Ground Faults
After each sample, VFD first calculates the residual voltage
3Vo (see Appendix I for definition) and the magnitude of its
first difference 3Vo'. If the latter exceeds 1/64 per unit, detection occurs. For most ground faults detection will occur on
sample 0 (i.e., the first sample after fault inception). The
voltage changes abruptly per Fig. 5(a) unless the fault occurs
when 3Vo is at angle 0, as shown in Fig. 5(b). With zero initiation there will be sufficient 3Vo', provided the 3Vo peak exceeds
0.09 pu. As described in Appendix II, if 3Vo' is too small due to
a distant fault, the phase-fault logic (to be discussed below) will
detect it.
This logic is performed once per sample for each voltage level.
Phase Faults
The phase logic looks for a peak by sensing a change in the
sign of the first difference (difference of the present and previous
sample value). If VAG, VBG, VCG, VAB, VBC, or VCA is 1/64 pu
less than the previous peak of the same voltage, VFD operates.
-2
~~~~~~PEAK
-I -.
00
(a)
(b)
0~~~~~~~~~
_I
-I
(c)
(d)
Fig. 6. An apparent voltage peak may occur at sample 0. (a) Detection at sample 0. (b) Detection at sample 7. (c) Detection must
await next peak. (d) Detection at sample 1.
Three-phase fault:
10 percent minimum voltage
collapse less than 10 percent
Phase-phase or two-phase-ground:
near fault
distant fault
0
0-3
0-2
0-11
443
It can be seen, then, that VFD detects all nearby (and hence
severe) faults by the time of sample 3, allowing time for the
locator programs SDF and Z3/Z1 to set up tripping by sample 7
(i.e., 4 ms after fault). Note that the locator programs start at
sample 0 rather than the latest sample (assuming the present
sample is 9 or less). This minimizes the possibility of waiting
for more samples to complete the necessary number of passes.
A.
OPERATING
N~~~~~~~ZN
DIFFERENTIAL ZONES
46
4~~~~~~~~4
z
w
Fig 7.Pretdfeeta.caatrsi
U-
Iso/a
sAI.CVo
neededI
.25
C
PF)
444
IEEE
1969
+7
I-CY. -
(b)
(a)
0
I
CY.
(C)
WS Waveshape Analysis
Substantial differential current may result from either magnetizing inrush [3], [8], [such as in Fig. 8 (a) or (b)] or from dc
CT saturation [10] [such as in Fig. 8 (c)] during external faults.
Successive peaks of either inrush current in Fig. 8 fail to occur
at about 8-ms intervals. In Fig. 8 (a) the peaks are too far apart;
in (b) the opposite is true. To distinguish between inrush and
legitimate fault currents, WS requires 15 to 20 samples between
peaks, or 7.5 to 10 ms.
The current in Fig. 8(c) meets the requirement of 15 to 20
samples between peaks during the first four cycles (if we define
point P as a peak). Accordingly, WS also imposes the requirement that a peak be 75 to 125 percent of the previous peak
and, for good measure, of opposite sign. WS then calls BDF (#)
or TDF (# ) if the peak requirements are met.
When BDF or TDF bid for WS, the bid is suspended until
CPDSI detects a differential current peak.
General
Fig. 4 shows the distance routine coverage. Zone 1 reaches
85 to 90 percent from station P toward station R, tripping
with no intentional time delay. Zone 2 reaches beyond station
R to insure detection of an end-zone fault such as at Fl. Since
zone 2 also unavoidably sees adjacent-line faults such as F2,
its tripping must be delayed by time T2 to allow breaker 4
relaying to clear F2. Zone 3 reaches still further to:
1) initiate T3 delayed back-up tripping of adjacent-line faults
should the primary relaying fail,
2) assist in locating which line is faulted,
3) provide instantaneous tripping of end-zone faults, working
in a pilot scheme, in conjunction with the breaker 3 zone 3 and
a transfer-trip channel.
All three zones are directional, operating only for faults in
the protected line or beyond station R.
These routines use the compensated voltage principle [6], [7]
where the protected line currents modify the station voltages.
Five angle comparisons are made: three for single-phaseground faults (one per phase), phase-to-phase unit for two-phase
faults, and three-phase unit for three-phase faults. In the analog
relays, compensators energized on their primary with the line
current induce a secondary voltage, which is a replica of the
actual transmission-line impedance drop from the relay to a preset balance point. For example, the phase-to-phase unit of the
phase distance relay develops a compensated phasor voltage
Vzy:
(1)
where Zc is the positive-sequence line impedance to the desired balance point (e.g., 90 percent of the line for the zone 1
relay)
Another compensator similarly modifies VAB to yield Vxy.
if Vxy leads Vzy by 0 to 1800 the relay trips. The routine then
must develop these compensated voltages and make a phase
angle comparison.
Voltage Compensation
Equation (1) can be written for a zone 1 reach in instantaneous
values as
(3)
445
TABLE III
QUADRANT DETERMINATION
VREF
Quadrant
First difference
Second difference
I
+
-
II
-
III
+
OPERATE ZONE
IV
+
+
'OPERATE ZONE
(a)
QD Quadrant Determination
In quadrants I and IV the first derivative of the sine function
is positive; while the second derivative is negative in quadrant I
and positive in IV. Thus, the combination of the sign of the
first and second differences defines the quadrant, as specified
by Table III.
(b)
VREF
VoP
>900
<
(b)
(a)
VREF
(c)
Computing Sequence
Zone 3 (Z3G and Z3PH) is called first since it covers a larger
than do zones 1 and 2. When one of its operation counters
is first incremented, zone 3 advances the priority of itself and
zone 1 (Z1G, ZlPH) of the same line. EX makes no priority
distinction between the phase (PH) and ground(G) routines
for two reasons: first, some of the computations are common to
G and PH, and secondly, Z3PH may operate for a phase-togrotnd fault sooner than Z3G, even though ZIPH will not
operate, while ZiG will.
area
Time-Delay Trip
When any zone 3 operation counter reaches two it enters a
T2 delayed bid for zone 2 (Z2T). If zone 3 remains operated
until T2 has expired, Z2T is called. After about the first 16 ms
of the fault, zone 3 computes only at 16-ms intervals, allowing
time for nonfault routines to run. When Z2T is called if Z2GS
or Z2PHS fails to operate, Z2T enters a (T3-T2) delayed bid
for Z3T. If Z3G or Z3PH continues to operate at 16-ms intervals
until (T3-T2) elapses, Z3T trips the line breaker (s).
446
1969
3 operation counter reaches 2, ORTT keys the transmitter as- routines to run in preference to nonfault routines. So the
sociated with that line. This signal (30-300 kHz) may be coupled program returns to prefault status except when these delayed
to a phase wire of the protected line or sent over a telephone or routines run.
microwave channel. Receipt of this information indicates that
the remote-end relays see fault-power flow into the protected Routtnes Not Included
line. If the local zone 3 also operates, ORTT immediately trips,
The areas covered in the above sections are representative,
provided a coordinating time delay has elapsed since zone 3 first but incomplete. Additional routines could include
operated. This delay equals the channel drop-out time plus 4 ms.
1) out-of-step tripping of line breakers,
OBK provides continuous transmitter keying when the line
2) automatic reclosing,
is disconnected from the station, so the remote zone 3 can
3) synchronism verification across an open breaker,
trip immediately, should the other end be closed on a faulted
4) automatic synchronization of breaker closing,
line. OBK also uses its breaker position information to advise
5) negative-sequence directional overcurrent protection of the
the potential-selection logic when the line again connects to the shunt reactors (SR in Fig. 1),
station. This logic prevents the program from using isolated
6) voltage unbalance detection and overcurrent protection of
line potential for relaying of the other lines on the same kV -the shunt capacitor bank (CP2 in Fig. 1).
level. Transfer to an alternative source occurs when the breaker
CONCLUSIONS
trip circuit is energized either by the computer or by nonautomatic means.
This study has disclosed no reason to bar eventual application
of the digital computer to perform the complete substation
Line Current Peak Change Supervision
protective relaying function. Many existing relaying techniques,
Z3PH contains logic to prevent line-breaker tripping due to such as digital logic, level detection, and timing, match identical
potential-circuit trouble with the power system normal. If, for stored-program computer operations. Others with no direct
example, a potential fuse blows, VFD erroneously detects a digital counterpart have been developed here. Key examples are
1) angle comparison (cylinder-unit equivalent),
fault. Even though FLOC fails to find a fault, EX will call
2) percentage-differential characteristic,
Z3PH. Unless Z3PH detects a 0.1 pu change in a current peak
3) magnetizing inrush restraint,
magnitude compared to the previous peak or the time interval
4)
line-drop compensation,
between these peaks is abnormal, tripping is blocked.
5) filtering of transients.
This paper also has demonstrated how to adapt present-day
MA Mlemory Action
protection techniques to fit the time-sharing, sampled data comFor a nearby AB or three-phase fault, where VAB drops to puter world. For example, fault-detecting and-locating logic has
nearly zero, MA provides the distance routines with prefault been devised to take control away from nonfault routines and
quadrant determinations. EX calls MA each sample to make a to direct the program to the routines most likely to effect fast
QD unless VAB and its first difference both drop below 0.003 breaker tripping. The following salient advantages to the use of
pu. MA stores the last 33 QD values. When the voltage is too a computer are foreseen:
low for a reliable QD, the program uses the stored QD value
1) faster breaker tripping, with security against undesired
determined 33 samples earlier (or a multiple thereof); at 60 Hz operations comparable to existing relays (the logic here provides
this is almost a multiple of one cycle (33 X 0.5 = 16.5 ms), 4-ms nominal trip time),
so the stored QD values tend to match those which would have
2) greater dependability, since the hardware is in frequent
been determined had VAB continued normally. The sampling use as contrasted to the protracted idleness of existing hardware,
period inaccuracy is tolerable, since the computations are not
3) more economical; cost can be shared with nonprotective
close to a balance for a nearby fault. However, eventually the functions for data acquisition and control,
drift would become excessive. Accordingly, access to these stored
4) readily adaptable to use with digital current and potential
values is blocked after 32 ms.
transducers.
While long-range optimism seems justified, many roadblocks
OTHER PROTECTION
promise to deter the development of an economical, serviceable
BF Breaker Failure
system. Some of the likely key problem areas are the following.
1) Programming effort: based upon an experienced real-time
When any routine initiates breaker tripping, it also enters a
programmer
being able to write and debug 10 program state64-ms delayed bid for BF (# ). After 64 ms, BF checks all
phases of each breaker to see that current has ceased flowing ments per day, many man years must be expended.
2) Input requiremenits: the instantaneous value of each ac
(less than 0.05 pu). After 8 samples indicate no current is
current
and voltage must be converted to digital form every
flowing, BF (# ) turns off. If current is still present, the routine
ms
0.5
(129
quantities in Fig. 1).
operates a multicontact lockout relay, which latches in the
3) Computer speed: the program imposes massive speed
operate position, tripping back-up breakers and blocking rerequirements. For example, six passes of ZlGS should be perclosing until the lockout relay is manually reset.
formed in 0.3 ms, requiring a total of 54 multiply, 162 add, and
6 left-shift operations. Also storage of converted ac inputs
TOP Turnoff
should not entail more than 0.2 ms of main-frame time (129
TOP resets flags and counters, allows VFD to run again, and inputs in Fig. 1).
restructures some of the priorities (see Table II). It will not
4) Security: the computer will find the substation environrun if any higher priority routine is suspended or on time-delay ment most inhospitable. Much programming and analog-tobid, with the exception of Z3G/PH and BF delayed bids. The digital conversion hardware effort will be needed to keep unlatter routines continue too long to allow low priority fault desired breaker tripping to tolerable levels.
447
APPENDIX I
DEFINITIONS
CT
CTR
d
DECR
Differential
Current transformer
Counter
ZOM/ZOL 1, a stored constant for each line
Decrement
Summation of all currents flowing into a procurrent
tective zone (e.g., bus 1 in Fig. 1)
First
Present sample value minus the previous sample
difference value
h
Sampling interval, seconds
H1
I4/h, a stored constant for each line
H2
L/h, a stored constant for each line
H3
L3/h, a stored constant for each line
Ha1
Lo,/h, a stored constant for each line
Lo2/h, a stored constant for each line
H02
H03
LQ3/h, a stored constant for each line
Phase currents, instantaneous
'A, 'B, IC
Phase currents, phasor
IA, IB, 1C
IA', etc.
Phase A first difference (prime always connotes
first difference)
Phase A second difference (double prime always
IA", etc.
connotes second difference)
Breaker current (any phase), instantaneous
IBK
Delta currents (difference of two phase currents),
ID
instantaneous
Transmission line currents, instantaneous
IL
Zero-sequence current, instantaneous
Io
Parallel line zero-sequence current, instantaneous
IOM
Value of latest current peak
IP
Value of previous current peak
IP-1
Ip (ss)
Value of steady-state peak
INCR
Increment
K
Sample designation [e.g., (IA)K]
L1, L4, L4
Positive-sequence line inductance of zone 1, zone
2, and zone 3, respectively
Lo,, L02, L03 Zero-sequence line inductance of zone 1, zone 2,
and zone 3, respectively
pu
Per unit 69 V rms; 5 A rms for line protection;
approximately 5 A for differential protection, but
adjusted for different CT ratios and for power
transformer voltage ratio
PT
Potential-transformer or capacitance device
R1, R2, R3 Positive-sequence line resistance of zone 1, zone 2,
and zone 3, respectively
Residual
3IU, instantaneous
current
of zone 1, zone 2,
and zone 3, respectively
Routine ab- See Table I
breviations
Routine States:
Running In operation
Bidding
Awaiting completion of higher priority routines
to
run
Time
delay
Suspended
Executive
turnoff
Turned off
Second
Present first difference minus the previous samples'
difference first difference
Zero-sequence voltage, instantaneous
Vo
Residual voltage = VAG + VBG + VCG (instan3Vo
taneous)
VAG, VBG, Phase-to-ground voltages, instantaneous
VCG
ZOL
ZoM
Zone
(# )
Single-Phase-Ground Faults
The residual voltage 3Vo and its first difference 3Vo' are
defined in Appendix I. Fig. 5 shows that the first difference at
sample 0 will be larger if the fault occurs when 3Vo is at other
than 00, due to the jump in voltage at inception. Fig. 5(b)
represents the most unfavorable situation with sample 0 at 10
from inception and sample 1 at 20.80. For this case the magnitudes and first differences for a unit of 3Vo at 60 Hz are:
Sample
3Vo
3Vo'
-1
0
0.174
0.355
0.524
0
0.17
0.18
0.17
0
1
2
to yield
13V0' max
0.188
0.564
448
E-O. 5 MS-)
0D
(a)
APPENDIX III
TIME TO CURRENT-TRANSFORMER SATURATION
I
R
T
(b)
Fig. 12. Voltage conditions for BC faults. (a) Distant fault.
(b) Nearby fault.
Three-Phase Faults
Phase faults are detected by a drop in peak magnitude of
(5)
449
PE
z
I
Ul
S
0
S
S
49
C
0
MI
!i4
APPENDIX V
DISTANCE ROUTINE CALCULATIONS
The following equations define the machine operations required for distance relaying. The currents and voltages are
instantaneous pu values of ac quantities. See Appendix I for
definition of terms. Appendix VII gives a numerical example.
(IADA')K =
(UADA')K =
(UBDA')K =
(IBDA')K =
(23)
(24)
(25)
(26)
(27)
(28)
Equations (6)- (8) provide polarizing voltage for the three-phase
mho unit [9] of ZlPHS, Z2PHS, and Z3PHS. Equations (9)(28) develop the first differences IADA', IBDA, ICDA'. and IOA'
required to compensate the bus voltages for the inductive line
drop. See Appendix IV for the basis for (16), (23), (24), (27),
and (28).
ANGLE - DEGREES
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)
(21)
(22)
(29)
(30)
(31)
(32)
Z3GS
Appendix VI describes how the magnitude comparison of [6]
is converted to an angle comparison. The compensated zerosequence voltage is
(33)
(34)
(35)
(36)
(37)
(38)
(VS3')K =
(VT3')K =
(VR3")K =
(VSS")K =
(VS3)K - (VS3)K-1
(39)
(VTS)K
(40)
(VTS)KR1
(VT3")K =
(VR3')K - (VR3')K-1
(VS3')K - (VS3')KR1
(VTS')K - (VT3')K_1
(VXG3' )K
(VXG3)K
(41)
(42)
(43)
(VXG3)K-1
(44)
(45)
(46)
(47)
(48)
(49)
4 D'
Call OPX to compare TR3 and Vxa3; Vs3 an-d VyG3; VT3 aild
VZG3.
(X+Y)
Z3PHIS
VXY3 =
(VXY3')K =
(VXY3")K =
(VzY3)K =
(VzY3')K =
(VzY3")K =
(50)
(51)
VXG3- VYG3
(VXY3)K- (VXY3)K-1
(VzY3')K- (VzY3')K-1.
Call OPX to compare VxY3 and VAB (reference).
Call OPY to compare Vxy3 and Vzy3.
(53)
(54)
(55)
(56)
(57)
VOTI-
(58)
(59)
ZIGS
These operations are identical to those of Z3GS, except that
here the equations contain terms with the subscript 1 rather
than 3. For example, (33) becomes
(60)
ZlPHS
These operations are identical to Z3PHS, except for the use
of subscript 1 rather than 3.
Operations Common to Z2PHS and Z2GS (Zone 2)
(61)
IADA H2 - VOT2
(62)
VI CG-
ICDR2 -
IBD1AH2 VoT2
ICDA H2 - VOT2'
(63)
(64)
VZG2
Z2GS
(65)
(67)
0co = dloM
(68)
(70)
CL -
(b)
(a)
(ICO')K + (IC)')K+l]
if ICo is in quadrant I or III. (71)
(COAA')K = (1/2)[(ICO')K + (ICO')K-1]
if Ico is in quadranit II or IV. (72)
Note fromn the alternative solution for V0T2 in (67) that
(ICOA')K
= (1/2)[
[6].
Z2PHS
These operations are identical to Z3PHS, except for the use
of subscript 2 rather than 3.
APPENDIX VI
2b = 1800 - 2c
(75)
2 (b + c) = 1800
(b + c)
(66)
(69)
where
A
(x-Y) -9
-
(VzY3)K- (VzY3)K_1
ICDA'H1 -
(X+Y)
-I
,-
(52)
(VXY3')K- (VXY3s)K_1
VZG3 - VYG3
VZG1 = V0 - ICDR1 -
(.X-Y)
,-"/
900.
(76)
(77)
so
(b+c)< (a+d)
(a + d) = 1800- (b + c)
(78)
(79)
451
so
Zos
138KV
Il-n)ZOL
ZIL
ZL
0--~iZO
n
ZoU
A-G
1000/5
--
(bA+c) <900.
techniques):
+ 1200/I
VXN
(VAG
VO)
Y = VWO
IMPEDANCES IN OHMS AT 138kV, 60 Hz:
0.05
Z OS- .02
nZlL - 1.6
j 5
Z IS
nZOL- 7.5
(80)
10
Z IU
Z OU
lnIZIL -
22.5
(X + Y)
0.10
0.13 +
0.5 +
j 11.5
Vo- VOT3
(81)
(82)
TVXN + Vwo
(83)
j 10.7
(X + Y)
j 7.5
j 67.5
IADR3 - IADA H3
(X-Y)= (X+Y)-2Y
PER UNIT QUANTITIES FOR FAULT AT X:
VAG
V0G
0.812 | -0.90
0.976 1-117.70
M982 1117.60
VCG -
(84)
(X
.6
.2- ~ /
VAG
=
\
50
2I0 3!0 40
710
2-
.8-
_BVG
8,0vAGP
A GLE
O- -0
1.2-
(IA + 1.334Y
/
/
.0- .8-
. -
I0
IB-.068 )
-.068)
~~~~~~~~~(IC
2T FAULT
4- - POINT
SAMPLING
1
0
2
Fig. 16. Inlpult quantities.
-I
Y)
- 2Vwo
APPENDIX VII
.8
VXG3
VR3.
Operation should occur when VR3 leads VXG3 by 90 to 2700.
IAISS) . 1.86I-85.80
IBI(SS . 0.092 91.9
ICss) . 0.092 91.9P
VAGFP- 1.00 t
General
EXAMPLE
Zone 3
Table IV performs the Z3GS computationis based on Appendixes V and VI with computer operations beginning with the 31o
calculations. Current first differences are averaged per Appendix
IV to obtain IOA', IADA and IBDA = ICDAI. At sample 4, Z3GS
increments the phase A operation counter based on OPX
logic described earlier. At this point Z3GS advances ZIGS
and ZIPH priority to 3. On the next pass (5), VR3 moves to
quadrant II, but VXG3 is delayed in moving to quadrant IV
until sample 6. Accordingly OPX does not increment the phase
A operation counter at sample 5 since VR3 and VXG3 are caught
452
~~~~~~~~~~~~~IEEE
TRANSACTIONS
ON
POWER
APPARATUS
SYSTEMS,
AND
APRIL
TABLE IV
FAULT SAMPLE,-Z3GS
FAULTED VALUES
SAMPLE-1
AT INCEPTION
QUANTITY1T
ANGLE
MIAGNIT.
~35.00
0.573
VAG
VBG.
ANGLE
VCG
IBaSS
I1C (SS)
MAGNIT.
39.10
-0.996
155.0
0.423
-(
)0
-85.0
ANGLE
I-0.962
131.9
131.9
0.068
-40.90
137.7
137.7
0.066
IA+ 1.334
0.574
-71.9
163.4
0.374
-1.334
0
0
I8-0.06S
-0.068
3VO
____
SAMPLE 4
QUANTITY
96.50
-0.853
0.099
-50.3
185.0
-0.906
0.062
0.062
0.138
-0.006
148.5
0.408
0.408
0.426
-0.020
148.5
-18.4
159.3
0.032
170.1
170.1
159.3
-0.052
-0.095
0.009
MAGNIT.
ANGLEj MAGNIT.
ANGLE
MAGN IT
109.70
-7.1
228.2
0.764
-0.121
-0.732
120.50
3.7
239.0
0.700
0.063
-0.842
14.5
249.8
ANGLE
131.30
35.6
1.063
46.4
1.348
180.9
-0.001
-0.019
-0.035
- 0.035
-0.051
213.3
-0.851
224.1
224.1
-0.064
-0.001
148
-0.069
-0.069
202.5
202.5
213.3
180.9
191.7
191.7
IB-0.068
IC0.068
-L.0.C19
.842.1
-0.08
____
] 0.088
___-0.002
MAGNIT.
0.781
IA+ .-334
SAMPLE 8
24.8
I-0.246
-0.036
-0.09
0.450
~3.2
0.792
-0.620
-0.267
0.747
-0.036
140
206.6
MAGNIT.
0.016
0.016
1.088
-0.052
0.062
217.4
-0.469
-0.440
77.3O
0.744
-39.5
195.8
- 7.6
-0.020
-0.084
SAMPLE 7
ANGLE
-0.751
-0.086
-0.587
0.012
SAMPLE 6
MAGNIT.
0.104
0.811
-28.7
Is(ss)
iciss)
98.90
-17.9
0.670
ANGLE
SAMPLE 3
0.802
-0.299
-0.596
88.10
VSS)
ANGLE
MAGNIT.
-61.1
174.2
-29.2
-0.072
SAMPLE 5
MAGNIT.I
ANGLE
VAG
VBG
VCG
55.70
SAMPLE 2
-0.927
0.281
-1.196
______0.072
ANGLE
-0.006
~~~0
3______
SAMPLE I
MAGNIT
44.90
0.509
-77.7
157.6
-46.8
0
0
SAMPLE 0
0.810
0.244
-0.922
-0.054
2.417
Z8
-0.119 1-0.132
-0.119 1-0.132
-0.103
[-0.103
SAMPLE
QUANTITY
310
0.126
0.386
0.875
0.984
1.300
1.608
1.909
2.179
2.424
I0
0.042
0.129
0.067
X
0.225
0.096
0.009
0.328
0.103
0.007
0.433
0.636
0.100
-0.003
0.726
0.090
-0.010
0.808
3if
If
0.105
0.002
if
0.536
0.103
-0.002
X
0.297
0.09
0.522
0.100
0.760
0.201
0.225
0.024
10o
X
x
IAlo
'O QD
3:6A
X
0.096
TAD
41)A
________AD_____________
XAD
-101103 ._&241'IO
-EIbA H03 1 1.491I0A
VoT3 = -kR03- IOA H03
X
X
QD
ILJ)A
~x
X
X
X
x
X
IADR3s- .04B rA D
11ADA H3 3I81~DA
-0.0481AD - 3.831 ~DA -VOT3
VX03 VAG -0.0411AD -3.83I"'ADA VOT3
XsD 'TCD
X
X
x
X
x
X
115V'DX
-0.848
"SD -l'CD
-0.149
-0.101
X
0.104
1.005
0.102
0.0905
1.479
0.066
X
1.878
0.238
0.013
0.245
0.007
0.243
-0.002
if
1.248
0.231
-0.006
1.691
0.212
-0.019
0.187
I:
0.242
-0.104
0.237
-0.129
0.222
-0.154
0.200
-0.175
X
-0.195
-1.057
-1.111
-1.149
-1.228
-1.194
-1.171
-1.298
-1.300
-1.091
-1.245
-0.988
-1.163
x
X
-0.025
-0.816
-1.952
-0.036
-0.888
-0.060
-0.906
-0.071
-0.846
-0.061
-0.766
-2.268
-2.162
-1.205
-2.152
-1.360
-0.048
-0.926
-2.272
-1.461
-1.466
-1.306
-2.010
-1.310
x
X
-0.261
-0.112
-0.011
-0.380
-0.119
-0.007
-0.502
-0.122
-0.003
-0.624
-0.122
0
3ii
-0.116 -0.120
0.024
0.016
0.444
0.480
N
N
-0.118 -0.110
0.030 j0.036
0.421
0.464
-0.768
-0.834
-1.303
1-1.033 1-1.274
-'SD R3 - 0.048'SD
:'B3DA H3*- -83.6'-BDA
X
X
X
X
31
-0.106
0.013
0.406
-5D
x
X
x
X
-0.692
-1.443
-0.778
Jj1
ifE
R3 -I'SDA H3 -VOT
VYG3=VBG %DR3 X~IBA H3 -VOT3
VZG3 = VCr -'CD R3 1'-CDA H 3 - VOT3
0.232
-0.079
if
0.213
-0.054
QD
I'BDA -'CDA
0.062
-0.018
-0.026
E
'CD
-1.385
0.739
-0.15
0.007
-0.845
-0.106
0.009
-0.840
-0.095
0.011
NE
-0.100
0.041
0.383
x
X
X
-0.816 1-0.788
-1.115 1-0.909
-0.739
-0.676
X
X
-1.412 1-1.520
-1.581
1969
453
QUANTITY
3vW03
-2VW03
TABLE IV (Cont'd)
SAMPLE
3
-4038
-3.993
3.286
2.692
0.978
-&779
2.519
1.159
.843
1.133
1.506
1.486
1.389
x
x
X
0.181
0.290
-0.022
X
-3429
V R3
V S3
V T3
V"lR3
SIGN
VS3
Vi"T3
SIGN
SIVN
V'XG3
V R3
V S3
QD
QD
V T3
QD
V YG;3
VIZG3
-3.567
2.662
1.196
1.154
1.068
1.43
1.032
1.702
0.072
0.256
1.547
1.250
-0.035
0.158
-0.072
-0.168
1.231
1.41
IC
3:EIT
X
X
x
-0.152
0.047
-0.101
0.083
-0.241
-Q255
x
-3.827
2.552
-0.042
0.096
2.378
0.797
-0.06
0.059
-0.235
_
-0_21
-
IE
131
-0.005
0.186
-0.138
i
-
3I
0.068
0.088
0.206
0.233
-0.061
-0.108
VIIXG3
SIGN
Vi'ZG3
SIGN
SIGN
x
x
x
x
+
+
+
+
+
+
VXG3 QD
HE
]I
3m
3mE
INCR.
Vu1YG3
VYG3
-1
jWv:
+
+
QD
VZG3 QD
(3NCR.
INCR.
X
0 ADVANCE
ZONE PRIORITY
I
KEY
() TRANSMITTER
H
impedance
base voltage/base current
69.3/5 = 13.85 ohms
base
2VW03
Lrelay = Zpri/21r(Rc/Rv)(1/ZB) PU
(60.0/377) X 0.167(1/13.85)
0.00192
pu
3.83 pu.
H3 multiplies times IADA', the averaged first difference of positiveplus-negative-sequence inductive line-drop compensation.
Fig. 17 shows the phasor relations for Z3GS at sample 2.
Input voltages VAG, VBG, VCG are compensated by the line
currents to produce VXG3, VyG3, and VZG3, respectively; these
4.54,
APPARATUfS
TABLE V
FAIULT EXAMPLE-ZIGS
SAMPLE NUMBER
QUANTITY
-10 Ro
VOTI
'ADA
HI
0.020
-"3BDA HI
I"BDA
- 1.63
-0.020 IBD - 1.63
VYGI
VBG - 0.020
VCG - 0.020
VZGI
-
3VWoi
2VWO
VRI
VSI
VXGI
VXGI
VyGI
VTI VZGI
VI RI
PH. C OPX
V'51
I'BDA-VOTI
ID- 1.63 IBDA- VOTI
ICD - 1.63 "CDA VOT
+ VYGI + VZGI
2VWoI
VT,I
V" RI SIGN
VTI
VXGI
____ V IYGI
QD
__
V'tZGI
QD
VYG
QD
0.200
-0.112
OPERATION
-0.349
-0.785
0i45
-1.065
1.755
1.170
-1.676
1.006
1.116
0.960
0.2
6.061
0522
0.662
0.365
0.033
-0.007
0.196
0.166
0.140
-0.128
-0.160
-0.174
-0048
_
I1
-0.07
3E
I
31
0.006
-0.160
-0.120
-0.033
0.130
-0.194
+
0.194
0.`66
mlI
t X IIL
Zone
-0.333
-0.454
-0.648
0.016
0.179
-0.346
x
x
-0.156
0.100
-0.212
-0.920
0.012
0.386
-0038
and
0.192
X
X
X
x
X
x
-0.030
-0.362
-0.528
-0.964
1.015
COUNTER
-0.025
-0.386
-0.583
0.196
-1.754
1.170
0.062
-0.065
-0.463
-0162
-0.814
t160
0.513
0055
-0.498
-0.155
6.OLO
.82
0.2
VZGI QOD
-Q0.966
-1.657
1.104
X
X
SIGN
VXGI
-0.3Q
-Q.4
-0.591
-0.379
x
x
-0.020
-0.324
-1.044
-1.507
1.004
0.920
-0.507
-0.561
189
-0.293
-0.347
-0.044
-0.015
-0Q378
-_ 914
-0.122
-0.82
-0.084
0.005
QO8
Q073
x
X
V'ZGI
PH.
-Q0.521
X
x
QD
QODX
VSI
-0.471
-0.010
X
X
SIGN
SIGN
VRI
-0.033
-0.488
OPERATI2O3N4
6ONE
V" T
-0.023
-0.040
2VWOI
2VWOi
V"ISI
-0.448
= -Io
-IAD RI
-.102 Io
--4.88 I"OA
HO1
- IVOA
._
ImL
z_
ME
31
INCR.
+
XINCR.n
x
x
APPENDIX VIII
FIL VOLTAGE FILTERING FOR DISTANCE ROUTINES
QD quadrant determination utilizes the sign of the first andl
second differences rather than the sign of the quantity and its
first difference in order to minimize the effect of transients below
the system frequency. The filtering routine described here attempts to eliminate the influence of higher frequencies such as
in Fig. 18, where a transient is superimposed on voltage VAB.
Note that the filtered values are plotted.
When a quadrant changes to other than the succeeding one,
FIL is called. In Table VI a jump occurs between samples 2 and 3.
If sample 3QD had been quadrant I, FIL would not have been
called. The procedure is as follows.
1) When a quadrant jump occurs at sample K, average the
voltages of samples K 3, K 2, and K
1, and replace the
(K
2) voltage with this average (e.g., sample 1 average of
-0.49 replaces -0.76).
2) Average voltages K
2 (original), K
1, and K, replacing the (K 1) voltage with this average.
3) Perform new QD for sample K
1.
4) Average voltages K 1 (original), K, and K + 1, replacing the K voltage with this average.
5) Perform new QD for sample K.
6) Using averaged K voltage, make QD for sample K + 1.
Note in the above that the "original" value may mean a
filtered value from a previous pass as is the case at sample 5 in
-
455
TABLE VI
FILTERING ROUTINE FOR FIG. 18
QOUANTITY
SAMPLE
}-1
V'AB
Vil AB
D
-076 1
-0.38
-0.77
-0.44
+0.38
-0.39
x
x
_ _
(K-2)
(K-1)
(K)
-0.32
X
x
(K-3)
-054
VAB
FILTERING RTN:
VAB
ABB
A_____B
_ _
-0.49
-0.64
-0.51
-0.39
-0.63
-0.2
-0.15
+0.13
+0.12
-0.24
+0.34
IV
Ir
JE
IV
IK-3\
(K-2)
(K-1)
-0.5
-0.44
-0.42
-0._
+0.07
+0.02
+0.07|
mI
+
V
ABB;
tTI
QD
(K-3
(K-2)
(K-1)
(K)
-0.40
+0.0Q
VAB
FINAL QD
2K)
____-0.48
____
__
FILTERING RTN:
:VAS
5
_
VAB
FILTERING RTN:
-0.17
|_
QD
:i
__
||||
iv
iv
FAULT
2
TIME
w
-I
SAMPLE
456
1969
APPENDIX IX
CPD LiNE CURRENT PEAK DETERMINATION
CPD processes each line-breaker current each sample. First
it computes the first difference. Upon a sign change of the first
difference the value of the sample is recorded as the peak along
with the time position. Two peaks are stored so that Z3PH
may compare the pre- and postfault values to confirm that
VFD saw a fault.
APPENDIX XI
SDF STATION OVERALL DIFFERENTIAL
Refer to Fig. 20. Block A sums the line currents. To handle
the phase shift in wye-delta power transformers, delta currents
or phase currents less zero-sequence current must be used. If in
Fig. 1 the 500 kV and 230 kV phases are identified A-B-C,
while the 66 kV phases are identified 1-2-3, the three line current
summations would involve
APPENDIX X
VFD VOLTAGE FAULT DETECTOR
VFD processes each voltage sample as shown in Fig. 19.
Block A looks for a significant jump in 3Vo since the last sample.
Block A makes one such check per voltage level (three in Fig. 1)
unless a fault is indicated sooner. Then block B looks for a sign
change or a zero value in the first difference of the phase to
ground voltage (one check for each of three voltages on each
voltage level, unless a fault is detected sooner). Block E stores
the measured value as the peak. If this value is at least 1/64 pu
less than the previous peak, a fault is detected.
457
APPENDIX XIII
BDF BUs DIFFERENTIAL
See Fig. 22, which applies for one phase. Block A sums the
phase currents connecting to the bus and looks for a magnitude
exceeding 0.25 pu. If the differential current is too small, block B
checks for a first difference magnitude exceeding 0.05 pu. A yes
from either A or B permits H and J to perform a percentagedifferential check per Fig. 7. If J detects a fault, block K increments operation counter CG. If the counter now reads more than
45S8
1, block L calls TRIP and BF. All breakers surrounding the bus
are tripped. BF watches to see that all are successfully tripped.
If block L answers no, indicating this is the first successful
pass, M advances the priority of this BDF(# ) phase. If the
WS flag is set (P) showing that the waveshape is satisfactory,
the program returns to EX for a bid check. If P answers no,
and the routine is processing one of the first four samples (R),
no WS call is needed, since the CTs have not yet saturated.
If block R answers yes, S clears counter Cl while T suspends
BDF(#) and enters a suspended bid for WS. When CPDSI
detects the next peak of the associated differential current it will
allow a WS call. If WS finds a satisfactory waveshape it in turn
removes the BDF (#) suspension. BDF(#) starts over, since
S cleared counter Cl.
Tripping requires two successive positive passes, since block
C clears operation counter Cl on each negative pass. Blocks
D and E provide a means to reduce a BDF (#) priority (F) if
no fault is found after three negative passes. If time permits,
another try is made; block G requires another WS call on a
subsequent BDF (#) call.
APPENDIX XIV
TDF TRANSFORMIER DIFFERENTIAL
Phase
Delta Side
Wye Side
IA
IB
IC
I1-I2
12-13
I3-It
459
XV
WS WAVESHAPE ANAIYSIS
When CPDSI senses a peak, WS checks to see if it occurred
15-20 samples since the last peak, is opposite in sign, and 75
to 125 percent of the magnitude of the last peak. If so, the
BDF (# ) or TDF (# ) bid suspension is removed. If the waveshape is not satisfactory,, WS enters a suspended bid for its
program, so it can check the next peak.
460
needed. If none exceeds 0.06 pu, Z3PH performs its angle comparisons in Fig. 26. Each time block C gets the opportunity it
checks all new samples since the last check; if necessary, it will
wait for the maximum first difference of 0.188 Ip(ss), corresponding to a sensitivity of Ip (ss) (0.06/0.188) = 0.3 A.
After block C in Fig. 25 detects a fault, block D clears all
five priority counters, and E increments the appropriate operation counter. If this counter reads one (F), block S in Fig. 26
advances this line's zone 3 priority and bids for zones 1 and 3.
If any operation counter reads 2 or more (F of Fig. 25) and
this is the first pass (G), block H enters a bid for Z2T to provide
for tripping in time T2 (see Fig. 4) if not cleared earlier. Block
J enters a bid for ORTT (# ), which provides for fast tripping
if the transfer-trip channel is operative.
If Z3G fails to find a fault, Z3PH (Fig. 26) angle comparisons
(A and U) are made. If either check indicates a fault and B, C,
D, E, or F answers yes, blocks J through S function as described
for Z3G. If block B answers no and if FLOC has not found a
fault (C), block D looks at all postfault samples of all three
phases. If block D fails, E checks for a difference in the magnitude of the last two peaks of all three phases. Also block F answers yes if the peaks on each phase are not 15 to 20 samples
APPENDIX XVII
Zi ZONE 1 DISTANCE
Z3G/Z3PH (#) bids for Z1. The computations are similar to
those for zone 3 (see Appendix V and VII). While Fig. 27
shows only two angle comparisons (A and F) for simplicity,
there are three ground and two phase checks, any one of which
connects to block M. When any of the five zone 1 operation
counters reads 2 or more (counter range is 0-3), tripping occurs
(R), a time-delay BF bid is entered (S), and the transfer-trip
transmitter is keyed (T), in case zone 3 did not have time to do
it. Blocks U and L withdraw zone 3 (# ) and zone 1 (# ) bids.
Each time a definite operation occurs, block E or J increments
its priority counter; when block K finds all five counters reading
3 or more, it withdraws the Zl (#) bid.
apart.
Block V in Fig. 26 provides the logic ample time to find a
fault and then permits block W to reduce the zone 3 priority
on this line. Blocks W, X, and Y reset various items which might
have been initiated on earlier passes.
Block Z provides for a 2-ms delay in removing transfer-trip
keying to allow for continuous keying even though the angle
comparisons fail to yield a fault indication on every sample for
an internal fault.
APPENDIX XVIII
Z2T ZONE 2 TIME-DELAY TRIP
461
APPENDIX XIX
APPENDIX XXI
(S).
MA MEMORY ACTION
Z3PH enters an ORTT bid. Fig. 30 shows this logic for pilot
tripping in conjunction with a transfer-trip channel from the
remote station (s). Block B initiates transmitter keying to allow
fast remote breaker tripping. This block also removes a transmitter turnoff bid which may have been entered by block Z
of Fig. 26 on a previous unsuccessful bid. The channel-delay
counter (C and D) allows time for zone 3 and channel drop out
if the direction of fault-power flow reverses due to sequential
breaker operation on a faulted parallel line. The counter setting
in D should be at least channel drop-out time plus 4 ms, allowing
for the 2-ms delay in block Z, Fig. 26, and for computation of
postreversal samples.
After the coordinating delay (D), if a transfer-trip signal is
being received (E), the line breaker (s) is tripped (F), and a
time-delay BF bid is entered (G). If the fault is not in the
immediate line section (between stations P and R in Fig. 4),
no signal will be received from the remote station(s), so block
H removes local transmitter keying. Block J then provides for
temporary turnoff of the zone 3 routines to allow time for nonfault programs to run. When EX calls Z3G/PH (#) and TOP
462
Discussion
I. F. Morrison and B. J. Mann (University of New South Wales,
Kensington, Australia): The application of computers to the field
of substation protection has been discussed for several years now
[13], [14]. We would like to compliment Mr. Rockefeller for his most
interesting contribution to this field. We were particularly impressed
with the comprehensive and detailed nature of the software organization he proposed. We were however a little unclear as to whether the
system he outlines is actually in the course of implementation or
simply a feasibility study.
We would be most interested in the technical details of the conm,puter (or computers) that the author has (or intends) to use for
this work and especially details of the peripheral equipment. It is
apparent, as he suggests, that the proposed system would impose
massive speed requirements on any computers used. It is thus
desirable to extract from the system all the information required by
the sampling of a minimum number of quantities. With this in mind,
has the author investigated the suitability of sampling sequence or
other derived relaying quantities in place of, or in conjunietion with,
the line currents or voltages?
We feel reasonably well qualified to comment on Mr. Rockefeller's
work since we have been studying the application of digital computers to substation functions at the School of Electrical Engineering,
University of New South Wales, for some little time now [15]. As a
result of this work, we would agree with the author's more general
conclusions and add that further salient advantages of the proposed
system lie in the provision of greater flexibility, ease of updating, and
compactness than present methods; this due to the software or
''programmable" aspects of the system.
The author's adaption of zone-packaged techniques [6] to digital
format is very interesting. Our own line of investigation has been
directed toward the actual calculation of the impedaTnce [16] rather
than its classification in a particular zone. This we felt wouild enable
greater flexibility in the choice of simulated relay characteristics.
The calculation has been achieved at satisfactory speed by the
utilization of table-lookup procedures. One of the problems encountered has been the existence of the transient dc component in
the current signal. Since this is not readily removed by hardware
techniques, we wonder if the author has experienced any diffic-ulty
with this.
(In a lighter vein, we note that the 50-Hz system in Australia
gives us a few ms per cycle advantage.)
463
REFERENCES
1968.
K. R. Shah (University of Missouri, Columbia, Mo.): Mr. Rockefeller must be congratulated for introducing a new concept of
protecting transmission line by using stored-program digital computers. This idea of replacing relays by digital computers (on a timesharing basis) gained momentum when it was realized that some of
the power failures which have occurred due to the malfunction of
protective relays could have been prevented if on-line digital computers were employed for the protection of power systems. The
function of relays in initiating the tripping of a circuit breaker can
be performed faster by employing computers which constantly
monitor live lines. In this regard the author mentions that digital
computers can initiate circuit-breaker tripping within four ms for
severe faults, but that this figure does not take into account the
processing rate of digital computers. So if we presume that the
464
supplier.
REFERENCES