IEEE Transactions on Power Apparatus and Systems, Vol. PAS-IOO, No.

7 July 1981 3195

PERFORMANCE OF THIRD HARMONIC GROUND FAULT

PROTECTION SCHEMES FOR GENERATOR STATOR WINDINGS
R. L. Schlake G. W. Buckley G. McPherson
Westinghouse Electric Corp. Detroit Edison University of Missouri-Rolla
East Pittsburgh, Pa. Detroit, Michigan Rolla, Missouri
Member Member Senior Member

ABSTRACT
The paper shows how the normally generated third-harmonic voltage can should not be removed from service. To allow for these small ground currents,
be used to protect the lower ten to twenty percent of generator stator windings trip settings for over-voltage ground relays are generally set to detect neutral
against ground faults. A method of determining the applicability of the scheme displacement voltages in excess of 5- 10% ofthe phase-to-neutral voltage. The
to a given machine is described which takes into account the design of the potential hazard of an undetected ground fault in the lower portion of the
generator and its externally connected apparatus. The effects of finite resis- winding, where the neutral bus displacement voltage is below the ground relay
tence ground faults are also investigated and other advantages of the scheme, trip setting, therefore arises. Moreover, a fault at the neutral end of a
such as protection of the neutral, are also highlighted. generator phase winding will effectively short circuit the neutral grounding
transformer. Under these conditions the conventional ground fault protection
INTRODUCTION scheme will be totally inoperable, with potentially catastrophic results.
The reliability requirements placed on modem power systems demand If a ground fault occurs and remains undetected because of its location or
increasingly complete protection of the generator. Because the major causes otherwise, the probability of a second fault occurring is much greater. This
of stator winding ground faults are electromechanical stresses and local second fault may result from insulation deterioration caused by transient
insulation deterioration, the locations of ground faults on the stator cannot be over-voltages due to erratic, low current, unstable arcing at the first fault
predicted a priori. Thus the entire stator winding (neutral end included) must point. This second fault may yield currents of devastating magnitudes.
be considered when designing schemes for protection against ground faults. The present paper describes an investigation into the feasibility of utilizing
The conventional unit type generator has the neutral stabilized by a resis- the third harmonic (180 Hz) emf; induced in the stator phase windings of all
tance loaded distribution type transformer, see Figure 1. The load resistor is large generators, to give ground fault protection in the lower 10-15% of the
sized to limit the in-phase component of line-to-ground fault current to a value winding. In conjunction with the conventional 60 Hz scheme, which relies on
a resistance loaded distribution transformer to limit ground fault currents, this
not less than the total charging currents of the capacitances-to-ground of the
can be designed to give ground fault protection over 100% of the stator phase
generator winding and the associated isolated phase bus, transformers and winding. In addition to the above, the paper also highlights a number of other
leads. This value of neutral effective resistance has evolved over the years and
results in moderate ground fault currents, generally under fifteen amperes for operational advantages of the scheme and investigates its range of validity. A
a solid phase-to-ground fault at the generator terminals. Currents of this
further major advantage of the scheme is its simplicity and low cost. The relay
can also be easily retrofitted on existing installations.
magnitude generally do not cause serious damage to the core steel. However,
while it is unnecessary to use high speed detection for ground faults,
good industry practice dictates prompt tripping of the generator with the DISCUSSION
minimum delay practical.
Typical Unit Connected System
A three-phase circuit model of a typical unit connected generating station is
-UNIT AUXILIARY shown in Figure 1. Normally the potential transformer (P.T.) neutral is di-
TRANSFORMER
rectly grounded. However, some power companies connect the neutral point
of the P.T. primary winding to the generator neutral, although this practice is
depreciated by an IEEE Committee Report1.
When providing protection for the generator stator winding, it is important
to consider the winding neutral for several reasons. During high potential
testing of the stator winding it is necessary to attach safety grounds to the
generator neutral. Other conditions occasionally require disconnecting the
generator neutral and leaving the neutral open. There are many recorded
cases of these grounds not being removed when the generator is returned to
service or alternatively, the generator neutral to grounding transformer con-
nection is not made. This is a potentially dangerous situation, both for the
machine, and also for plant personnel. In addition, the presence of the cable
which connects the generator neutral bus and the potential transformer prim-
ary neutral increases the exposure of this neutral to ground faults. If a ground
occurs in this cable, the neutral grounding transformer is short circuited.
The third harmonic relay will detect all of these conditions.
7The Problem
The relay scheme consists of two relays in parallel with the load resistor, as
shown in Figure 1. A regular over-voltage relay, operating at the fundamental
generator frequency of 60 Hz, protects the upper 90%o to 95% of the stator
winding as explained earlier. The lower l1o (neutral end) of the winding is
For a single ground fault near the neutral end of the winding, there will be protected by a relay which detects a reduction in the third harmonic, or
proportionately less voltage available to drive current to ground, resulting in a 180 Hz, component of the neutral bus voltage following a fault.
lower fault current and a lower neutral bus displacement voltage. At the lower No 180 Hz component of line-to-line voltage appears in either delta or wye
limit, a fault on the neutral bus will result in no fault current or displacement connected machines. However, most generators are connected in wye to
voltage at all. eliminate 180 Hz circulating currents which can cause undesirable heating,
Small 60 Hz ground currents may flow under normal conditions, possibly but a phase to ground 180 Hz voltage does exist. This voltage causes 180 Hz
due to generator winding imbalances or to faults on the secondaries of the current to flow, through the various leakage capacitances to ground, returning
generator potential transformers. Under these conditions the generator to the generator neutral bus through the neutral grounding transformer. This
current will cause a 180 Hz voltage drop across the load resistor. If a ground
fault occurs in-the lower end of the generator winding or on the neutral bus
(see Figure 1), the grounding transformer will be effectively short circuited
and this 180 Hz current will bypass the ground transformer. This will cause a
reduction in the 180 Hz load resistor voltage. Thus a ground fault in the lower
10% of the generator winding may be detected by a marked reduction in the
180 Hz neutral bus displacement voltage.
In developing a method for determining the applicability of this relay
scheme for a typical unit-connected generator, the standard 60 Hz scheme of
81 WM 028-0 A paper recommended and approved by the Figure 1 is assumed appropriate for the upper 90 to 95% of the winding. The
IEEE Power System Relaying Committee of the IEEE applicability of the package thus depends on the proper operation of the
Power Engineering Society for presentation at the 180 Hz relay. There are two problems associated with the operation of the
IEEE PES Winter Meeting, Atlanta, Georgia, February 180 Hz relay.
1-6, 1981. Manuscript submitted June 13, 1980; made 180 Hz Neutral Bus Voltage: The applicability of the protection scheme
available for printing October 29, 1980. depends on the magnitude of the 180 Hz neutral bus voltage under all ex-

1981 IEEE
 3196
pected operating and fault conditions. The unfaulted 180 Hz neutral bus The general circuit model for one phase of the generating station system is
voltage must be within the relay's operating range. In addition, for a fault in shown in Figure 2. In the circuit models developed the following assumptions
the lower 5 to 10lo of the winding, the resulting reduction in the 180 Hz neutral are made:
bus voltage must be sufficient to cause the 180 Hz relay to trip and indicate the 1. All distributed windings, bus bars and cables are represented by Pi-line
presence of the fault. equivalents.
A method had to be developed to calculate the 180 Hz neutral bus voltage 2. The generator neutral bus grounding transformer has been approximated by
under all expected load and fault conditions. These estimates could then be the load resistor reflected to the primary of the transformer.
used to determine the resulting 180 Hz neutral bus displacement voltage range
for a given machine. 3. The driving voltages, Ei, E2, E3, in Figure 2, are the emfs induced in the
coils of the generator stator winding for a 3 coil machine. Additional coils
Effect of the P. T. Connections: The second objective of this project was to would lead to further subdivision in the same manner.
determine the ef ect of the P.T. connection described previously on the 4. No 180 Hz current flows in the line and the delta primary winding of the
applicability of the relay package. The performance of the 180 Hz relay generator step-up transformer, and the h.v. grid appears as an open circuit to
depends on the generated 180 Hz current returning to the generator neutral the third harmonic. This renders the equivalent impedance of the unit and
through the neutral bus grounding transformer. This P.T. connection provides auxiliary transformers equal to the phase-to-ground capacitance of the delta
a second current path in parallel with the neutral bus grounding transformer. winding only.
This current path might bleed enough of the 180 Hz current around the 5. For simplicity, an equivalent 180 Hz impedance, Zeq, will be substituted for
grounding transformer to limit the range of applicability of the relay package. the circuit elements external to the machine terminals. These components are
Thus a method for determining the 180 Hz currents flowing through the P.T. the generator isolated phase bus, the station auxiliary bus and the unit and
primary windings was also desired. station auxiliary transformers.
The 180 Hz neutral bus displacement voltage is directly proportional to the 6. Since a Pi-line equivalent has been made for the P.T. windings, one-half of
180 Hz current flowing through the grounding transformer primary winding. the capacitance to ground of each phase of the P.T. appears in parallel with the
Thus both of the problems associated with the operation of the 180 Hz relay grounding transformer. An equivalent impedance, Zgnd, is then made for these
can be solved by determining the distribution of the 180 Hz currents in the parallel circuit elements.
third harmonic, generating station impedance network. 7. Interturn capacitances between adjoining stator slots are negligible due to
Method of Solution the electrical shielding properties of the core steel. The winding
As part of t e present work, a simple, but extremely versatile computer betweencapacitance-to-ground in a given slot outweighs the interturn capacitance
program named GTHAP was written to fully model the machine on a coil by to-width ratio the conductors in that slot. Assuming a typical conductor height-
of 3:1, the total ground capacitance will be at least 26 times the
coil basis, together with its externally connected apparatus such as the iso- interturn capacitance
lated phase bus, main generator step-up transformer, etc. This program is turn capacitive reactance between two windings in the same slot. Thus the inter-
used to estimate the 180 Hz circulating ground current for any given machine ground and can safely be neglected. will be 26 times larger than the impedance-to-
based on either a calculated or measured third harmonic voltage value. Thus
the applicability ofthe third harmonic scheme, and its range of protection for a A further reason for neglecting the interturn capacitance is that very little
given machine, can be determined. voltage is available to drive leakage current through it. Since the three third
Since the third harmonic voltage acts as a zero sequence component, no harmonic phase voltages are in phase, the highest possible voltage between
third harmonic currents flow in the low voltage delta windings of the unit and turns of different phases is equal to the third harmonic phase voltage. This is
auxiliary transformers. This means no 180 Hz voltage appears on the wye not a large voltage. Also, a practical winding will be short pitched, resulting in
connected, high voltage side of the unit transformer. Thus the main power only one or two slots where the potential difference exceeds 80%o of the third
system appears as an open circuit in the third harmonic current network. harmonic phase voltage. Thus, there is generally less voltage available to
Using the assumed method of connecting the potential transformer (P.T.) drive interturn leakage currents than there is to drive leakage currents to
primary neutral to the generator neutral, see Figure 1, there are two parallel ground.
paths external to the generator through which 180 Hz current is driven. One of Mutual inductances under zero sequence excitation are non-existent. By
these paths is through the various machine capacitances to ground, returning definition, zero sequence impedance is measured by applying the same vol-
to the generator through the neutral grounding transformer. The second path tage and current to all phases of a machine. Thus, each of the phase windings
is trhough the Y-Y connected P.T. primary windings and back to the generator will develop an MMF in opposition to the others. No flux will link any two
neutral. However, initial calculations and subsequent simulations with windings and only leakage flux will flow. Thus, only leakage inductances are
GTHAP have shown that the 180 Hz series impedance of these P.T. windings involved in the zero sequence impedance ofthe unfaulted generator windings.
is very large. Thus the effect of this second current path on the relay perfor- These (zero sequence) impedances should be used in modelling the generator
mance is almost undetectable and it can safely be neglected in evaluation of stator winding, although their overall effect is small.
the relay scheme Under fault conditions, one phase is partially shorted to ground. This fault
Large generators often contain several parallel paths in each phase winding, will therefore cause a smaller opposing MMF in that phase. Such an imba-
so the precise path for the third harmonic current depends on the type of lance in the phase MMF's will cause a mutual flux to flow and the zero
generator connection. It also depends on whether the machine winding is sequence impedance will no longer involve only the leakage impedance. This
faulted or unfaulted. The precise equivalent circuits for each case are shown imbalance will depend on the magnitude ofthe current in the faulted section of
in Figures 2 through 5 and described later in this section. the winding. However, the fault current path includes the grounding trans-
former which appears as a very high impedance. This will keep the fault
General Outline: Each of the circuit models yields a set of mesh current current small and the MMF imbalance will be negligible. Therefore, three
equations and GTHAP was developed to solve any of these sets of equations times the normal 60 Hz zero sequence impedance should be used for the
for any fault point and fault resistance. 180 Hz winding impedance under fault conditions.
In carrying out a solution, the 180 Hz emfs and the various 180 Hz system
impedance values are calculated first. The 180 Hz system impedances are Cases to be Considered
then used to fill the appropriate impedance matrix. A complex matrix inver-
sion subprogram is then called to determine the resulting system admittance Case 1: Un aulted Systems: The circuit model for the unfaulted machine is
matrix. The system voltage vector and admittance matrix are then multiplied shown inPFigure 3. When the machine is unfaulted, all three phases are
to calculate the desired currents. identical. As explained earlier, the three single phase third harmonic induced
emf s are equal and in phase.
180 Hz Circuit Models: The unfaulted system is of primary importance in POTENTIAL TRANSFORMER
determining the normal 180 Hz current distribution. Under faulted condi-
tions, two types of circuit models are relevant. A faulted, single path perphase
model must be used when there is only one current path per phase. When
there are two or more parallel current paths per phase, another circuit model is
needed. For these reasons three cases must be simulated.
rI I m
.r---r

PHASE A
,q
- -mA
KEY
Ct,. HALF COIL TO GROUND CAPACITANCE
X. COIL ZERO SEQUENCE REACTANCE
-

FIGURE 2 SINGLE PHASE CIRCUIT MODEL GENERATOR

To minimize the computer core and time required for the unfaulted case, a
KEY-K W!TH_3ABRIl COI S_ER PHASE_ - _
-
single phase equivalent circuit can, and has been developed for the three
I * GENERATOR POTENTIAL TRANSFORMER I - GENERATOR AND GROUNDING TRANSFORMER
m- ISOLATED PHASE BUS I5. UNIT AUXILIARY TRANSFORMER
phase network. To adjust for the effects of coupling to the other phases, the
S-UNIT STEP-UP TRANSFORMER
grounding impedance in the single phase equivalent is three times the actual
grounding impedance, Zgnd.
 3197
Case 2: Faulted system with a Single Current Path per Phase: The circuit Where q = number of slots/pole/phase for the fundamental.
model for the faulted case with one current path per phase is shown in Figure To illustrate the manner in which the partial coil voltages, eal and ea2, shown
4. This system is not symmetrical, so all three phases must be included in the earlier in Figures 3-5, are obtained, Figure 6 shows a case for a fault part ofthe
vector-matrix voltage equation. way into the first coil. It should be noted that each coil emf, including those in
PT WINDING the faulted coil, is used with its correct in-phase and quadrature components
to determine the correct ground path and leakage current. Similar construc-
PT. WINDING tions are used for other fault points and winding configurations.
T4
-_I
PA- TH 1
PATH T PATH 2
IrIECD'
-r=EO>t,

PHASE A
la) FUNDAMENTAL: UNFAULTED
KEY PATH I PATH i
VX-* COIL ZERO SEOUENCE REACTANCE BELOW FAULT
PHASE B Xou * COIL ZERO SEOUENCE REACTANCE ABOVE FAULT Sy=a
4 Cw AS FIGURE 3
PATH 2
zoq [ Rf a FAULT RESISTANCE

FIGURE 4. FAULTED SYSTEM EQUIVALENT CIRCUIT 3 COIL PER

PHASE MACHINE, SINGLE PATH WINDING

Only one fault is considered at a time. This fault is placed in Phase A as

shown in Figure 4 and the fault resistance denoted Rf. This phase is then
modelled by a voltage source and Pi-line coil equivalent below the fault. The
remainder of the faulted coil above the fault and all succeeding coils are
modelled individually. The winding capacitance and series impedance are PATH 2 PATH 2
distributed according to the proportion of the winding faulted. The 180 Hz
voltage induced in each section of the winding is discussed later. See Case 3 (b) 3RD HARMONIC: UNFAULTED (c) 3RD HARMONIC: ST COIL FAULTED
below for an explanation of the coil voltages eal and ea2. FIGURE 6. FUNDAMENTAL AND 3RD-HARMONIC COIL VOLTAGE PHASOR DIAGRAMS
2 POLE, 2 PARALLEL PATH, 36 SLOT MACHINE

Case 3: Faulted System with Two Parallel Current Paths per Phase: The
circuit model for the faulted case with two parallel current paths per phase is Effect of Load and Power Factor on Generated Third Harmonic Voltage
shown in Figure 5.
POTENTIAL TRANSFORMER The current paths for the 180 Hz (third harmonic) stator currents, discussed
earlier, present very high impedances. Therefore, the 180 Hz stator currents
are very small in relation to the 60 Hz load currents. These currents produce a
very small- 180 Hz armature reaction stator MMF, resulting in a negligible
effect on the 180 Hz line-to-neutral terminal voltage.
A major source of third harmonic line-to-neutral voltage is -the effect of the
rotor ampere-turn distribution. The fact that the field winding coils are in
distinct slots around the rotor surface and also that the pole face may be
asymmetrical, gives rise to a stepped rotor MMF wave. This stepped MMF
space wave may be Fourier analyzed to yield many odd harmonic compo-
nents. The effect of this source of third harmonic air-gap flux depends on the
rotor winding configuration, the field current and the core steel saturation
level, which in turn determines the reluctance of the magnetic circuit in the
machine. The portions of the third harmonic MMF caused by the rotor turn
and airgap permeance distributions are space stationary with respect to the
rotor direct axis (the center of the pole face).
A further source of third harmonic voltage is the third harmonic component
PHASE A (FAULTED) of the air-gap flux wave due to saturation of the rotor and stator teeth. Little
saturation occurs in the stator yoke (the area of iron behind the stator teeth).
FIGURE 5. FAULTED SYSTEM EQUIVALENT CIRCUIT 3 COIL PER This is because the stator yoke is sized mainly from mechanical considera-
PHASE MACHINE 2 PARALLEL PATH PHASE WINDING
tions. Hence, this component of the third harmonic flux magnitude in the
KEY-CIRCUIT ELEMENTS AS FI6URE 3 a 4 machine depends mainly on the saturation level in the rotor and stator teeth
and the rotor core.
Again, only one fault is considered at a time. This fault is placed in Phase A
as shown. Phases B and C are modelled as in Case 2. The voltage sources in
the unfaulted current path are the normal 180 Hz induced voltages, Ei, E2, E3,
etc. The sum of the voltages eat and e a2 in the faulted path is equal to the coil
voltage, the individual voltages being obtained from the ratio of unfaulted to
faulted coil length as in Case 2. 300C x _/
If the generator winding contains two parallel current paths per phase, the
-

faulted and unfaulted paths in Phase A must be modelled separately. The OAPU.
OA4PU -
capacitance per path will be one-half of the total capacitance per phase. The 250 -
series impedance per path will be twice the total series impedance per phase.
Again, simple circuit theory is used to solve for the current flowing through
the grounding transformer. 200
A faulted machine with more than two parallel paths per phase may be
modelled as a two path machine by approximating the unfaulted paths by one
equivalent winding. The faulted path must then be modelled separately.
Calculation of 180 Hz Faulted Phase emf s
In order to calculate the individual third harmonic coil emf's for a particular o' - - 7- __-
- ___
machine, the number of slots, poles and parallel winding paths must be 10 _ _
known. Either a measured or calculated 3rd-harmonic rms line to neutral 5~~~~~~. OD 10PIW
voltage must also be known. A conventional coil voltage phasor diagram for _ - - ...
the particular machine can then be constructed as shown in Figure 6. The 0.5 0.75 1.0 Pul(MW)
relationship between the nIh harmonic phase-to-neutral voltage, 'Er, (mea-
sured or calculated) and the nib harmonic coil voltage, 2Ec, is given by Uf F THIRD HARMONIC VOLTAGE EXPRESSED AS A PERCENTAGE
OF NO LOAD VALUE.
q5Ec = 'E,/HKd (1) PU =PER UNIT POWER OUTPUT.

where 2Kd, is the n1" harmonic distribution factor. For a three phase winding:
FIGURE 7 VARIATION OF THIRD HARMONIC VOLTAGE WITH
'Kd= sin (nr/6) I sin (nir/6q) (2) REAL AND REACTIVE POWER. ( REF. 2)
 3198
With a fixed terminal voltage, the flux density in the air-gap will remain Reliability of the Unfaulted System Model
nearly fixed. This results in a nearly constant level of saturation in the rotor
core and stator teeth. However, as the load angle increases, the angle between The second investigation was an attempt to verify the reliability of the
the rotor d-axis and the peak of the air-gap flux wave increases. Thus more of unfaulted system model of Figure 3.
the air-gap flux passes through the rotor teeth instead of the pole face. This Measurements of the neutral grounding transformer and potential trans-
renders the rotor teeth saturation component of the 180 Hz flux wave depen- former 180 Hz secondary voltages were made at various output power levels
dent on the power factor of the generator load. on the chosen test unit.
This rotor tooth saturation component is not space stationary with respect The 180 Hz neutral grounding transformer secondary voltage was mea-
to the pole face. Rather, for a symmetrical pole face machine, it can be sured under generator start-up conditions. The voltage was recorded for 75
expected to be space stationary with respect to the position of peak air-gap hours during which time the machine was off line five times. The maximum
flux, and is therefore load dependent. power output attained was 400 MW, which was 0.5 p.u. on the generator MW
The combination of these effects renders the third harmonic terminal vol- base.
tage dependent on the excitation level, the load and the power factor at the On another occasion, the 180 Hz generator terminal line-to-neutral voltage
generator terminals. The curves of Figure 7, taken from Reference 2, illus- and the generator output power were recorded, beginning in the evening and
trates the behavior of the third harmonic terminal voltage for a typical running until the morning. This experiment was run overnight to record the
machine under varying load conditions. The exact relationships governing the 180 Hz terminal voltage behavior at reduced load.
interactions of the sources of third harmonic terminal-to-neutral voltage, Knowing the 180 Hz voltage on the secondary of the neutral grounding
discussed above, are extremely complicated, and beyond the scope of this transformer and the load resistance, the transformer's 180 Hz secondary
present paper. current was calculated at various generator output levels. The unfaulted
system simulation was then computed using the 180 Hz generator terminal-
to-neutral voltage measured at each power level.
The tested generator's load resistor is rated at 0.588 ohms when a maximum
EXPERIMENTAL VERIFICATION OF THE MODEL current of 307 amperes is flowing through it. This current corresponds to a
solid phase-to-ground fault at the machine terminals. The resistor was mea-
In order to verify the reliability of the computer program developed for this sured at 0.498 ohms, prior to installation, with no current flowing through it.
project, three experimental investigations were made. To determine the effect of variations in this resistance on the error encoun-
tered in calculating the grounding transformer current, two computer simula-
Model Compatibility tions were run. The load resistors were set equal to the limiting cases of 0.498
ohms and 0.588 ohms.
The first of these was an attempt to determine the compatibility of all three The comparisons of the estimated and measured 180 Hz neutral grounding
models, i.e., whether or not they predict the same grounding transformer transformer primary currents are shown in Tables 2 and 3. Table 2 assumes
current when no fault exists. First, the program was run for the unfaulted case the grounding transformer load resistance to be 0.498 ohms, while Table 3
using impedance data from a typical machine on the Detroit Edison system. assumes it to be 0.588 ohms.
The unfaulted system was then approximated by running the faulted system of As indicated in Tables 2 and 3 the error encountered in predicting the
Cases 2 and 3 with 0.01% of the winding faulted through a nine megohm 180 Hz grounding transformer current was approximately 20 to 30%. There
resistance. The results of this investigation are shown in Table 1. These three are several possible sources ofthe error but most relate to either experimental
models gave the same results with negligible error. or data errors. Fortunately errors of this magnitude can be tolerated in making
a determination of the applicability of this protection scheme to a particular
TABLE 1 - COMPARISON OF PREDICTED GROUNDING generator. It is believed that the errors arise from faulty data as to winding-
TRANSFORMER CURRENTS UNDER to-ground capacitances, and to the somewhat uncontrolled conditions under
UNFAULTED CONDITIONS which third harmonic voltages were measured.
In Tables 2 and 3 the effect ofthe grounding transformer load resistor on the
180 Hz 1 Grounding 180 Hz grounding transformer current magnitude is evident. However, the
Line-to- Transformer Transformer percent error is not greatly affected since the load resistance appears in the
Neutral Fault Fault Load Primary % calculation of both the measured and estimated currents.
Case Volts Position Resistance Resistance Current Error
Predicted Reduction of the 180 Hz Grounding Transformer Currents Under
1 730 Unfaulted 0.498Q1 0.138A - Various Fault Conditions
2 730 0.01% 9M(1 0.498Q1 0.138A 0.0%o
3 730 0.01% 9MQ1 0.498Q1 0.137A 0.7% The third area of investigation is an example of the method used to deter-
mine the applicability of the relay package to a specific generator. This
The single phase unfaulted model is effectively equal to the three phase method entails determining the expected reduction of the 180 Hz neutral
faulted system shown in Figure 5, with 0.01% of the winding faulted through a grounding transformer current under various fault conditions.
nine megohm resistance. Since the three phase faulted circuit with such afault The 180 Hz terminal line-to-neutral voltage is assumed to be 730V in this
agrees with the single phase, Case 1 circuit, the validity of the single phase experiment (730V was an actual measured voltage, see Table 3). In an actual
model is proven. application of this program, the smallest expected voltage would be used since

TABLE 2- ERROR IN PREDICTING THE 180 Hz GROUNDING TRANSFORMER CURRENT

Assuming the Load Resistor = 0.498 Ohms
Grounding Transformer
Computer
Load Terminal-to- Secondary Primary Estimated Error
Neutral 180 Hz Load 180 Hz 180 Hz 180 Hz 180 Hz
MW MVAr r.m.s. Voltage Resistor Volts Current Current Current mag. %
607 59 730V 0.498(1 4.6V 9.24A 0.11IA 0.138A 0.027A 20
404 30 605V 0.498(1 3.3V 6.63A 0.080A 0. 114A 0.034A 30
224 6 437V 0.498(1 2.4V 4.82A 0.058A 0.083A 0.025A 30
234 -2 440V 0.498(1 2.5V 5.02A 0.060A 0.083A 0.023A 28
380 25 590V 0.498(1 3.2V 6.43A 0.077A 0.111A 0.034A 30

TABLE 3- ERROR IN PREDICTING THE 180 Hz GROUNDING TRANSFORMER CURRENT

Assuming the Load Resistor = 0.588 Ohms
Grounding Transformer
Computer
Load Terminal-to- Secondary Primary Estimated Error
Neutral 180 Hz Load 180 Hz 180 Hz 180 Hz 180 Hz
MW MVAr r.m.s. Voltage Resistor Volts Current Current Current mag. %
607 59 730V 0.588(1 4.6V 7.82A .094A .1 19A 0.025A 21
404 30 605V 0.588(1 3.3V 5.61A .068A .099A 0.031A 31
224 6 437V 0.588(1 2.4V 4.08A .049A .071A 0.022A 31
234 -2 440V 0.588(1 2.5V 4.25A .051A .072A 0.021A 29
380 25 590V 0.588(1 3.2V 5.44A .066A .096A 0.030A 31
 3199
this would produce the smallest magnitude of unfaulted grounding trans- TABLE 4 - DETERMINATION OF RANGE OF PROTECTION
former current and the smallest reduction in this current following a fault. This
experiment considers both single path per phase and a two parallel path per
phase winding machines. Driving Fault Fault 13 % Reduction
First, the unfaulted case was run to determine the expected 180 Hz Voltage Resistance Ql Position % Amps in Current
grounding transformer current with a given 180 Hz terminal line-to-neutral
voltage level. Simulations were then run with the faulted system models of 730 0 10 0.033 76
Case 2 and Case 3. Six different fault resistances were placed at each of six 730 0 20 0.065 53
fault locations in the lower 10% of the winding. The cases were all run with the 730 0 30 0.095 31
grounding transformer load resistor equal to the resistor's value measured 730 0 40 0.123 11
prior to installation. The resulting 180 Hz grounding transformer current 730 0 50 0.150 -9
magnitudes and percent reductions for the Case 3 model are shown in Figure 730 0 60 0.169 -22
8. The results for a Case 2 model agreed so closely with those for the Case 3 730 0 70 0.187 -36
model that Figure 8 can be used for both cases. Figure 8 shows that the scheme 730 0 80 0.199 -44
is most sensitive for fault resistances in the range 0 to 1000Q1.
Unfaulted base case. V3 = 730, 13 = 0.138 amps
FIGURE 8. PREDICTED REDUCTION IN 3RD HARMONIC GROUNDING TRANGFORMER
TION AND RESAISTANCE
z

SEI
80-
-t: CONCLUSIONS
z loon
,, 60-
t DECREASING
The following conclusions can be reached as a result of this investigation.
100 FAULT RESISTANCE 1. Calculations indicate that the additional loop provided by connection of the
z- P.T. neutral to the generator neutral does not bleed sufficient current away
0
20 from the ground path for the effectiveness of the scheme to be impaired.
looon
UNFAULTED CURRENT VALUES 2. Good coverage, typically 20% of the winding, can be obtained with the third
t'I:

, 40- CASE 2: 0.138 AMPERES

harmonic relay package.
CASE 3: 0.138 AMPERES 3. Preliminary studies given in Figure 8 show that the most significant reduc-
tions in third harmonic grounding transformer current are produced for
I,6
fault resistances in the range 0 to 1000 ohms. This applies for all fault
2 4 8 10 positions between 0.01 and 10.0% of the stator winding.
FAULT POSITION 1% OF WINDING FROM NEUTRAL END
4. The combined relay package can provide 100% stator ground protection if
Percentage of Winding Protected sufficient third harmonic voltage is generated. However, it should be
pointed out that, even though 100o of the generator winding is protected,
The only questions which now remain to be answered are, firstly, what no backup protection is provided.
percentage of the stator winding can be effectively protected by the third 5. As shown in Figure 7, because the third harmonic voltage passes through a
harmonic scheme?, and secondly, how can we determine this? minimum-value at low loads, as the generator is brought up to full load, the
Under normal circumstance with no faults present, a certain amount of 180 Hz relay may indicate an invalid alarm if the ratio of minimum to
third harmonic voltage will be measurable at the neutral bus. If the neutral bus maximum third harmonic voltage is less than the relay setting. For this
suffers a fault, then this voltage will fall to zero because the neutral grounding reason, if this ratio is not known, the 180 Hz should only be used to alarm
transformer is short-circuited. At the other-extreme, if a zero resistance short and the normal 60 Hz relay used to trip the machine in the usual way.
circuit is applied to the machine terminals, then the level of third-harmonic
current flowing through the grounding transformer will increase because the
current can flow through this fault as opposed to flowing through the leakage
capacitances of the circuit. This increased current will raise the neutral bus
voltage above the unfaulted level. Between these two extremes, there exists a
position on the winding at which, if a zero resistance ground fault were to ACKNOWLEDGEMENTS
occur there, there would be no reduction in the third harmonic bus voltage,
thus the third harmonic relay would not detect the fault. It is obviously The authors are indebted to their respective organizations for the provision
important that this position is above the lower point protected by the conven- of support and facilities in the course of the present work. In particular, the
tional 60 Hz relay. authors would like to thank Mr. Tom Roberts and other members of the Relay
The above questions can be answered very simply by simulating a fault at Division, Electrical Systems Dept. of Detroit Edison for their help and en-
progressively different points along the stator winding. Zero resistance faults couragement. The authors would also like to thank Mr. Lon W. Montgomery
in the range 10 to 80% in steps of 10% were simulated and the results are of the Generator Stator Systems Dept. at Westinghouse Electric Corporation
shown in Table 4. As we move out from the neutral bus, zero resistance faults for his consistent help, and support during the course of the present work. The
will result in a reduction in the third harmonic neutral bus voltage, until some authors are grateful to the authors of Reference 2 for permission to include
position is reached where an increase is noticed. A finite resistance short Figure 7, taken from their work, in this paper.
circuit will alter this position somewhat and it is possible to construct a family
of curves from an extension of Table 4 which would depict the variation of
third harmonic neutral voltage as a function of fault location with various
resistance values as a parameter.
On a practical note however, because we are attempting to protect the
lower portion of the stator winding, the likelihood of a high or medium
resistance short caused by insulation cracking, surface tracking or discharges REFERENCES
is small because the available voltage is small. It is much more likely that such
a fault would be as the result of mechanical damage such as fretting caused by 1. "Potential Transformer Application on Unit-Connected Generator", IEEE
relative motion of the stator winding and the stator core and therefore be of Power System Relaying Committee Report, IEEE Trans., Vol. PAS-91,
very low resistance. According to Table 4, theoretically, a zero resistance 1972, pp 24-28.
ground fault at the 45% position would be undetectable on this particular 2. Ilar, M. et al,"Total Generator Ground Fault Protection", Protective
machine. Practically though, the relay would typically be set to detect a 50% Relaying Conference, Georgia Institute of Technology, May 3-4, 1979.
reduction in neutral bus displacement voltage3, then Table 4 shows that the 3. Stien, M. and Linders, J.R., "Ground Fault Protection of Complete
180 Hz scheme will protect at least 20% of the winding giving good overlap Generator Windings", Fourth Annual Western Protective Relaying Con-
with the conventional 60 Hz scheme. ference, October 18-20, 1977.
3200
Discussion
K. H. Engelhardt (British Columbia Hydro and Power Authority, Van-
couver, Canada): The authors have done an excellent and thorough job
in analyzing, on the basis of generator design data, the third-harmonic
voltage and current distribution for normal conditions and for stator
ground faults at various winding locations, and applying a 100%
ground fault detection system on these principles. V3N
The utilization of third-harmonic voltages for the detection of SEC
ground faults at or near the neutral end of the stator (being a "blind"
zone to conventional 60 Hz voltage relays) poses some practical pro-
blems and constraints to relay application engineers. These are briefly
described in the following.
1. Determination of Third Harmonic Voltages
Details of generator winding design, and the kind of computer pro-
gram described in the paper are generally not available to utility A) Simiplified Schematic
relay engineers. On the other hand, instrumentation for measuring
the third harmonic voltages under various unit output conditions is
relatively easy to come by. Having such measured voltages obtained
at the neutral end and the bus end- of a generator, it is then possible 3V3E
to determine the approximate 180-Hz voltage reduction as a func- ..SEC.

tion of fault distance from the neutral. This need not be accurate, as /
long as it can be established that the 180-Hz detection zone extends -
beyond the 60-Hz blind spot near the neutral. (A
2. Application of 180-Hz Undervoltage Relay z
The typical relay setting for detection of 50Qo reduction in neutral-> V3NSEC Y
end 180-Hz voltage as mentioned in the paper, may be very difficult
to accomplish in practice with the wide operational variations of this NORMA coND. GROUND FAULT Alr NEUTRAl..
voltage sometimes found. Figure 1 is a striking example. B) Third-Harmoni
ic Secondary Voltage Distribution

V3,N
Fig. 2 Principle of Third-Harmonic Voltage Magnitude Compargtor

As the third harmonic voltages at the neutral end and at the terminal
end will shift their relative magnitudes for faults in either of the end
zones (in respective opposite directions of course), such relays will
see ground faults in both these zones, as shown in Figure 3.
NEUTRAL END TERMINAL END

S//////////A
/////WINDING COVERAGE
COMFARATOR
10 HZ

WINDING COVERAGE OF ALL BUT

NEUTRAL END 5 TO 6% BY 60-H
VOLTAGE RELAY

Il
I
AMOUNT OF OVERLAP
VARIES WITH GENERATOR
OUTPUT

Fig. 3 Stator Winding Coverage Provided by 180 Hz Comparator and 60 Hz Over-

voltage Relay
The blind spot of the 180-Hz relay has a different location than the
blind spot of the 60 Hz relay. Thus, a considerable portion of the
OUTPUT P winding will be covered by both relays, thereby providing some
0.5 I p.u. degree of mutual backup. The calculations described in the paper,
and/or third-harmonic voltage field measprements at both ends of
Fig. 1 Example of Measured Neutral 180-Hz Voltages
the generator stator, can be used as basis fqr the application of the
180-Hz voltage comparator.
Here the 180-Hz voltage is so low under certain generator output Manuscript received February 20, 1981.
conditions that the secure application (i.e. without possible false
operations) of such a scheme is barely practical, or if so, then only
by supervising it with judiciously set minimum current or
MW/MVAR detectors. Such supervision would of course limit the R. J. Marttila (Ontario Hydro, Toronto, Canada): The authors have
unit's operating range within which neutral end ground faults are presented an interesting paper on the performance of a third harmonic
detectable. ground fault relay for generating units.
3. 180-Hz Voltage Comparator Method The relay principle discussed by the authors is based on detecting a 50
An alternative detection method is offered by the fact that for stator per cent reduction in the third harmonic voltage at the neutral of the
ground faults, a 180-Hz voltage reduction at one end is accompanied generator. Could the authors elaborate on the criteria used in
by an increase at the opposite end, except for the 180-Hz "blind" distinguishing between a 50 per cent reduction in the third harmonic
area (given at the 45% position in the example of Table 4 of the voltage caused by a ground fault and that caused by normal variations
paper). The discussor has applied a third-harmonic voltage com- in
parison relay based on this principle, a simplified schematic of which a) real power output
is. given in Figure 2, on a number of generators. b) reactive power output?
 3201
In the case of a), Figure 7 of the paper indicates that reduction of Modolf Stien (ASEA AB, Vasteras, Sweden) and John R. Linders
rnore than 50 per cent in the third harmonic source voltage can occur as Sarasota, FL): This is an interesting paper which in a clear and straight
the unit loading is reduced from the maximum to the minimum. Also, forward way presents the basic facts about the Third Harmonic Voltage
in the case of b), our measurements indicate that a reduction exceeding Scheme for complete stator ground protection. As noted by the
50 per cent can occur in the neutral third harmonic voltage in response references, third harmonic voltages, measured on the neutral of the
to a normal change in the reactive power output. machine, have been providing this protection for some years. This
To have an accurate measure of the third harmonic voltage reduc- paper extends the understanding of third harmonics, including, as
tion, the 60-Hz component needs to be filtered out. What rejection shown in Figure 6, their use in the proteciton of parallel winding
ratio is required of the filter to limit the 60-Hz component to an accep- machines.
table level under fault conditions, to allow the necessary measurements The correlation between the computer program and the field data
to be made on the third harmonic component? What is the minimum would be more meaningful if all of the component capacitances were
unfaulted value of third harmonic voltage required to make the given. For example, it is not evident from Table 4 that the refinement of
necessary measurements? using the third harmonic phase displacements of Figure 6 and PI sec-
Alternate approaches to that used by the authors have been discussed tions of Figures 2-5 add to the accuracy of the calculations. Also what
in the references cited below. Both of these approaches require as input percent of rated L-N voltage is the 730 V of third harmonic?
the neutral voltage and the residual terminal voltage and depend on the The third harmonic load profile of Figure 7 should be used with cau-
ratio of the third harmonic components of these voltages to be relatively tion since it applies to a specific machine. The paper notes that several
constant over the full operating range of an unfaulted machine. For a sources of third harmonics are developed in the generator. With minor
ground on the stator at or near the neutral, the relationship between the changes in machine parameters, these sources may come into- phase
two voltages changes sufficiently from the quiescent to allow detection reinforcement of opposition at significantly different (relay and reac-
of the ground condition. Have the authors considered and what opinion tive) loading conditions. Thus Figure 7 should not be taken as typical of
may they have of this type of an approach? We have several relays of a given machine without confirming tests or calculations. From the ap-
the type described in reference [1] in service to give an alarm with out- plication point of view it would have been of interest if the authors had
put from the relay. Some relays have been in service for more than three included some information on how the 3rd Harmonic Voltage of the
years. So far, no alarms have been recorded. We would like to point generator can be estimated from basic design data of the machine.
out, however, that according to our measurements, the radio of the two In Tables 2 and 3, on what is the estimated error based? Does the
voltages used by the relay is not as constant as would be expected under computer program calculate the third harmonic voltage (Column 3) or
normal operating conditions of some machines and could, therefore, is this the starting point of the program? In the reference to Figure 8,
result in nuisance alarms from this relay. We are currently investigating has any work been done to represent the high resistance fault as a non-
alternate implementation techniques of the "two-voltage" approach in linear resistance which is most likely in a real situation?
an effort to accommodate the possible variations in the ratio of the The philosophy reflected in the final conclusion that the third har-
voltages, as well as provide sensitivity for detection of grounds with the monic relay should only alarm because it may falsely indicate a stator
relatively small no-load third harmonic voltages. ground fault is not favored by most operating utilities. If a given relay is
REFERENCES not secure it would probably not be used except under very unusual
cases. An essential reason for only alarming on third harmonic is that
[1] K. H. Engelhardt. "A Composite Ground Fault Detection Scheme the condition is not immediately critical (as contrasted to a differential
for High Resistance Grounded Generator Stators". Presented at relay operation, for example). A procedural shutdown is thus quite ac-
the meeting of Canadian Electrical Association. September 1973. ceptable and surely less hazardous to the system. But the "alarm only"
Quebec, Canada. concept should not be used to transfer the responsibility from the pro-
[2] L. Pizmindi. "Stator Earth Leakage for Large Generators".
-
tection engineer to a plant operator. False signals from third harmonic
IEEE Transactions on Power Apparatus and Systems. Vol. relays cen be avoided by use of suitable arming relays or otherwise
PAS-94. No. 4. July/August 1975. pp. 1436-1439. adapting the system automatically to the prevailing non-fault condi-
tions.
Manuscript received February 20, 1981.
R. L. Schlake, G. W. Buckley, and G. McPherson: The authors are
grateful for the kind comments made and the interest shown by each of
M. S. Baldwin (Westinghouse Electric Corporation, East Pittsburgh, the discussors.
PA) and W. A. Elmore (Westinghouse Electric Corporation, Coral So far as making measurements of the third-harmonic voltage at the
Springs, FL): neutral bus are concerned, Mr. Englehardt is correct in stating that such
1. This paper affords considerable insight into the modeling and measurements are relatively easy to make and, provided that the more
analysis of third harmonic effects in generators. Much of the dif- difflcult values of third harmnonic terminal voltage can be obtained,
ficulty described by the authors in setting a relay that will be both some estimate of the 180Hz voltage reduction as a function of fault
secure and sensitive can be overcome by utilizing a relay that com- distance from the neutral can be obtained. It is of course vital to suc-
pares the third harmonic voltage, neutral-to-ground, with that at cessful operation of this relay scheme that a significant reduction in the
the machine terminals (as derived from a wye-ground broken-delta 180Hz voltage occurs for fault positions past the blind point of the con-
voltage transformer connection). These voltages "track" one ventional 6OHz relay. As shown in the paper, the fault resistance can
another (rise and fall together) reasonably well as loading and/or have a significant effect on the amount of winding protected. Thus cau-
excitation is changed. A fault at one end (either end) of the winding tion needs to be exercised when using these simpler methods of evalua-
will cause the third harmonic voltage to drop at the faulted end and tion.
rise at the other. The relay using this concept complements the 60 Mr. Englehardt makes an important point when he speaks of the
hertz relay conventionally used for generator ground detection and variation of the third harmonic voltage with real and reactive power
overlaps its coverage for a large part of the machine winding. A output from the generator. A user unfamiliar with this variation could
much larger part of the winding can be covered by a relay using this potentially design a scheme which would generate an unnecessary se-
concept than by one responding to third harmonic alone. quence of nuisance trips or alarms. There is a definite need for further
2. Does Detroit Edison use the "alternate" vt connection with the vt measurements of the variation of this voltage on typical units, because
primary neutral connected to the generator neutral rather than relatively few published records of such measurements exist. There is
ground? If so, why? also a need for a simplified method of calculation to help steer potential
3. Page 1 Middle of right hand column "During high potential-it is
-
users of this scheme around possible pitfalls. We are pleased to see the
necessary to attach safety grounds to the generator neutral." Ac- additional measured results supplied by Mr. Englehardt, but still advice
tually, to hi-pot the generator winding it is necessary to isolate the they be interpreted with caution because of the small number of obser-
neutral from ground. Following the hi-pot testing the entire win- vation points in his results.
ding should be grounded to drain off any charge. However, it is When considering the practicality of using a relay setting of 50%.
agreed that the absence of third harmonic should detect the condi- reduction in neutral end voltage, one has to remember the length of
tions stated in the paper. time that a given unit is to operate at reduced load. In reality, most
modern steam turbine generators operate at or close to full load for the
Manuscript received February 20, 1981. major part of their operating life and only pass through these lower
 3202
loads during unit loading or unloading. We therefore do not consider In order to emphasize the dangers of this alternate connection, it may
this point a major problem area, provided steps are taken to defeat false be of interest for you to know that a Detroit Edison machine did in fact
alarms at low loads. suffer a fault where the cable connecting the vt primary neutral to the
The relay scheme considered by the authors is only one of several generator neutral failed. The fault originated with failure of a single leaf
possible alternatives. Our primary purpose in studying the particular of a multi-leaved flexible connector on the center phase of the isolated
scheme we did was that it was to be applied at two new generating sta- phase bus. The failed leaf touched to ground, initially causing a single
tions on the Detroit Edison system. The 180Hz Comparison Method is a line to ground fault at a relatively low current. The fault arc traversed
admirable way of achieving the same result and our calculations and the length of the isolated phase bus (forced air cooled type) and blew
computer programs can also be used to assist in the application of this out the de-ion grids at the transformer end of the bus, creating a three
alternate method. phase short-circuit. Fortunately, the hv breakers had already tripped
In response to Mr. Marttila's questions, we have considered three the unit from the system but sufficient flux still remained in the machine
possible solutions to the problem of distinguishing between genuine for the fault to cause considerable damage. During the single line to
reductions in third harmonic neutral voltage caused by winding faults ground phase of the fault, the generator neutral is raised to line to
and those resulting from changes in load. Firstly, the 50% reduction ground potential, as is the cable connecting the two neutrals. This cable
was selected based on a combination of suggestions made in the paper failed and flashed over to ground, effectively short-circuiting the
by M. Stien and J. R. Linders (see Reference #3 of the paper) and the grounding transformer. Fortunately, the cable withstood line potential
need to obtain a good range of protection. This 50% figure is not un- long enough for the normal protection to have time to operate. Had this
changeable and can be altered to reflect the particular users needs and not been the case, damage which was already severe, could conceivably
conditions. have been worse. We learned three major things from this fault; firstly,
Ideally, the voltage at which the relay is set to trip should be not more the quality of the cable installation (if it is to be used) is very important;
than 80% of the minimum value of voltage through which the machine secondly, all multi-leaved flexible connectors in isolated phase buses are
passes during normal operation. This would completely avoid spurious to be taped to prevent possible grounds on failure; and thirdly, all de-
trips or alarms due to load changes. However, this has a penalty. Based ion grids are to be bolted in place.
upon a limited number of published test results, plus our own The authors are, of course, very much aware of the procedure for
measurements, the 180Hz neutral bus voltage does not fall lower than high potential testing and we appreciate the time taken by Mr. Baldwin
about 60/o of the open-circuit value, and this figure is readily available and Mr. Elmore to improve on our poor wording.
from the generator manufacturer. If we were to select a relay setting of In answer to the questions by Mr. Stein and Mr. Linders, we believe it
80%o of this value, say 50% of open-circuit value, using a typical ratio is important to correctly include the phase of the third harmonic voltage
of full load to open-circuit 180Hz voltages of 2: 1, this means the relay is if accurate results are to be obtained. We see no benefit in further
set to detect a 75% reduction in 180Hz voltage when at full load. simplifying what is already a fairly simple precedure and it did not occur
Reference to our Table 4 and Figure 8 shows that, for the machine to us to leave this out because unnecessarily large errors would other-
studied in our paper, we could only protect 7-10qo of the lower part of wise have resulted. The error in the coil voltages that would occur if we
the winding for fault resistances up to IOOY. This range of protection assumed no phase shift in the coil voltages, can be obtained from the in-
barely covers the blind spot of the conventional 60Hz relay and is not verse of the distribution factor, and is about 55% in the 6
felt to be satisfactory. The measurements necessary to determine the slot/pole/phase case studied. We must here apologize for the omission
minimum voltage must obviously come from field tests during commis- of the term q in equation (2) which should read
sioning of the unit. nKd=sin(nu/6)/qsin(nir/6q). 730 volts is equal to 2.8% of the rated
An alternative method would to use an arming relay to arm the device line voltage of 26kV.
only for loads greater than, say 40 to 50%o full load and to set the relay Mr. Stien and Mr. Linders are quite correct in stating that Figure #7
at 50%o of full load and to set the relay at 50%o of full load or near full of our paper should be used with caution. It is explicitly stated in the
load value. paper that Figure #7 contains only typical results and is certainly not ap-
To aid our understanding of the use of this relay, we are very serious- plicable to all machines. So far, we have not perfected a simple method
ly considering adding a requirement to future generator specifications which can be relied upon to give acceptable accuracy for a wide range of
that the manufacturer supply curves depicting the variation of 180Hz machine designs. However, initial work in this area has been done and
terminal voltage with changes in real and reactive power. Using the kind we hope to complete this in the close future.
of modelling described, the range of protection could be definitively The errors quoted in Tables 2 and 3 are derived from the differences
calculated. in the values in columns 7 and 8 of these same tables, i.e. it is the
We are surprised to hear that machines on which Mr. Marttila has percentage difference in the measured and computed results. At the pre-
made measurements have changes in neutral bus 180Hz voltage of more sent time, the computer program uses measured values of third har-
than 50%o for practical changes in reactive power only (i.e. at constant monic as input. When the previously mentioned mathematical techni-
real power). We strongly encourage Mr. Martilla to publish these que is perfected, calculated values could also be used.
measured results at the first convenient opportunity. They may aid the We do not feel it is necessary to represent any nonlinearity in the fault
industry significantly. resistance because we are considering faults close to the neutral which,
With regard to filter requirements, the unit used is a six (6) pole active because of the distribution of potential across the winding, are likely to
filter with a roll off of 36 dB with a corner frequency set between 120 be low resistance faults caused by mechanical damage as opposed to
and 150Hz octave. Thus the filter attenuates the 60Hz feed through by carbonizing of the insulation by slot or internal discharges. Hence, even
about 40dB at the point of connection of the 180Hz relay. if a severe nonlinearity existed, say a resistance change of 5:1, this value
In answer to the questions by Mr. Elmore and Mr. Baldwin, Detroit is still likely to be small compared to roughly 4000Q in the grounding
Edison did use the "alternate" vt primary neutral connection until transformer circuit plus the winding capacitive reactance.
about five years ago when design of our latest power plant started. Our We are indebted to Mr. Stein and Mr. Linders for clearly explaining a
reasons for using the alternate connection were primarily because it benefit of choosing an alarm only condition for the delay. We were
eliminated the need to coordinate the setting of the 60Hz ground relay aware of this and it was in fact incorporated as one of the reasons for
and the fusing of the vt secondary winding. Our latest practice though is recommending an alarm only condition to our Company. This recom-
to directly ground the vt primary neutral. We were always aware of the mendation has been neither accepted nor rejected at this time.
potential dangers of this alternate connection but did not actually The authors would once again like to thank the discussors for their
change our practice until publication of the IEEE Committee Report interest and stimulating questions.
listed as Reference #1 in the paper.
Manuscript received April 2, 1981.