stated that some form of internal corona effects of harmful corona, a routine test measurement with an oscillatory circuit is

measurement should be made at or above which measures the corona itself does not more sensitive than RIV measurements
operating voltage as a complement to the seem necessary. with the NEMA circuit. It may be further
power-factor and capacitance evaluation. Agreement with Mr. Mathers' statement stipulated that either of these methods will
Certainly there is a need for such a test if on the importance of leak-testing is cer- detect small amounts of corona more
disassembly examinations reveal deterior- tainly in order. It is fundamental that a efficiently than power-factor or capacitance
ation which a corona test has uncovered and modern bushing which is not properly measurements. The problem in routine
which power-factor and capacitance tests sealed stands little chance of survival in testing of oil-permeable insulation, however,
did not reveal. Experience to date at the service. is not to measure corona but to determine
author's company does not verify this need Data from disassembly examinations whether or not the dielectric has been dam-
since bushings exhibiting significant corona indicate that power-factor and capacitance aged by the high-potential tests. If this is
also have a noticeable change in power measurements are extremely sensitive to the proper question to ask then power-factor
factor and capacitance when these measure- test damage of an oil-permeable dielectric. and capacitance measurement at 110%
ments are made at 110% of operating volt- Although this is contrary to Mr. Vogel's of operating voltage must again be sub-
age. Since the power-factor and capaci- statement, it is doubtful that a real differ- mitted as the best indication of service
tance measurements definitely reveal the ence in point of view exists. Corona reliability.

Magnetizing Inrush Phenomena in The residual flux is the flux in the core
prior to energization of the transformer.
The amount of residual is determined by
Transrormer Banks the hysteresis loop of the core excitation
curve at the instant that the magnetizing
current disappears during the previous
de-energization. In Fig. 1 the residual
W. K. SONNEMANN C. L. WAGNER G. D. ROCKEFELLER flux is shown in the positive direction and
FELLOW AIEE MEMBER AIEE ASSOCIATE MEMBER AIEE
is presumed to be a maximum based on
the premise that the exciting current has
been extinguished by circuit-breaker op-
T PROVIDE a basis for the design of half-wavre shape commonly associated eration at a normal current zero following
a transformer differential relay utiliz- with m Lagnetizing inrush. Before at-
a positive half-cycle.
ing waveshape for discrimination between temptingK to explain the reasons for these
magnetizing inrush and fault currents, a various vwaveshapes and fix the minimum In a given transformer, the first peak of
study was undertaken to determine the expected harmonic percentages, single- the total flux is a function of the residual
most suitable proportioning of harmonic phase inirush theory will be reviewed. flux, the magnitude of the exciting volt-
restraint to fundamental-frequency op- age, and the instantaneous value of the
erating current as a means of preventing Single-IPhase Inrush energizing voltage at the time of closure.
relay operation during a magnetizing in- In Fig. 1 maximum transient flux is pro-
rush.' As a necessary corollary of this Magneetizing inrush occurs in a trans- duced by energizing the transformer at the
study, it was desired to determine logical former t( o a greater or lesser degree when- zero point on the voltage wave, choosing
reasons for the wide variety of waveforms ever the polarity and magnitude of the the particular zero point when the voltage
encountered. This paper presents the residual iflux do not agree with the polarity is increasing in the positive direction to go
salient features of this study. and magnitude of the instantaneous value with the assumed positive residual flux.
Magnetizing inrush theory, when en- of steady{-state flux which would normally With negative residual, closing the circuit
ergizing a single-phase transformer, is be requirred for the particular point on the 180 degrees later would produce maximum
well known.2-7 However, little has been voltage wave at which the circuit is transient flux of opposite polarity. If the
published concerning inrushes in 3-phase closed. A maximum ilrush may be ex- breaker closes 90 degrees later at point X
connected banks. Therefore, in addition pected vwhenever the residual flux is a of Fig. 1, the instantaneous value of the
.to reviewing the phenomenon of single- maximuxm of one polarity and the nor- steady-state flux is zero, and the transitent
phase inrush, it was necessary to develop mally r equired instantaneous value of flux, therefore, is equal to the residual
theories applying to 3-phase systems. steady-si tate flux is a maximum of the flux. If the transformer is energized at
other po larity. Fig. 1 shows the fluxes point Y of Fig. 1, where the steady-state
Inrush-Current Harmonic Analysis for such a maximum inrush condition,
wherein Ithe total flux is the actual flux for Paper 58-789, recommended by the AIRE Relays
and Transformers Committees and approved by the
Harmonic analysis of a variety of the first cycle, and wherein, for simplifica- AIEE Technical Operations Department for
selected waves from oscillograms of 3- tion, no decrement or decaying action is presentation at the AIRE Summer General Meet-
ing and Air Transportation Conference, Buffalo,
phase inrush currents quickly disclosed shown. N. Y., June 22-27, 1958. Manuscript submitted
that the second harmonic was the pre- The tcDtal flux of Fig. 1 may be viewed as March 14, 1958; made available for printing April
7, 1958.
dominant harmonic, and that in some consistinig of two components: W. K. SONN3MANN, C. L. WAoNnR, and G. D. Rocca-
cases the third harmonic was of a negligi- 1. Steaedy-state flux: the normal flux F9LLER are with Westinghouse Electric Corporation,
ble amount. Table I shows the results requiremiients corresponding to the steady- Mr. Wagner in East Pittsburgh, Pa., and Mr.
Sonnemann and Mr. Rockefeller in Newark, N. J.
of some of these analyses. In the table, state ma, Lgnitude of the exciting voltage.
The authors are grateful to M. W. Whitaker and
the peak value of fundamental component 2. Tran sient flux: the flux required to be B. Gentry of the Kentucky Utilities Company for
of each wave analyzed was considered to added to the steady-state flux to make the their enthusiastic co-operation in the matter of
to the residual flux at the
be 100% for comparative purposes. total nut ofx equal
circuit closure. This transient
obtaining oscillograms of magnetizing inrush cur-
rents in the field. They also wish to acknowledge
Note that the waves illustrated in istanalog ous to the d-c transient component the help received from E. C. Wentz of the Trans-
former Division, Westinghouse Electric Cor-
Table I are not all of the conventional of asymi metrical fault current. poration.

884A Sonnemann, Wagner, Rockefeller-Magnetizing Inrush Phenomen Oc-rOBBR 1958

 Table 1. Harmonic Content oF Representative Inrush Currents

REMARKS

SINGLE PHASE SHAPE-240- BASE-25MVA 13.6/66KV-ENERGIZED

FROM 13.6KV SIDE-CYCLE * 2.

SAME AS CASE A EXCEPT 13 CYCLES LATER.

OSCILLATORY SHAPE-AUTO-TRANSFORMER WITH DELTA TERTIARY-

TRACE 2, FIRST CYCLE, OF FIGURE 6.

SHOULDER ON BACK PORTION OF WAVE- AUTO-TRANSFORMER

WITH DELTA TERTIARY.

OUTPUT OF DELTA CONNECTED CURRENT TRANSFORMERS-

TRACE 6, FIRST CYCLE, OF FIGURE 10.

WYE CURRENT, FIRST CYCLE, NO CT SATURATION- NO CT

BURDEN EXCEPT OSCILLOGRAPH ELEMENT.

SAME AS CASE F EXCEPT 30 CYCLES LATER.

SAME CURRENT AS CASE G EXCEPT DIFFERENT SET OF CURRENT

TRANSFORMERS, BURDENED BY DIFFERENTIAL RELAYS.

flux value equals the residual flux value, eliminated and the inrush current be- saturation density is assumed to be at
no transient flux is present. Under this comes an offset sine wave with the nega- 140%0 of rated peak flux. This value is
condition, there would be no magnetizing tive portion of the current cut off. Note used in this paper. Fig. 3(B) shows the
inrush current. that as the inrush peak increases, the current produced with this value of
In Fig. 2, a simplified excitation curve size of these legs becomes comparatively saturation density and with a 90%0
has been assumed as two straight lines, smaller and the actual shape of the in- residual flux, assuming that the breaker is
from 0 to S, and from S to P. The inrush rush current closely approaches that of a closed at the angle to produce the maxi-
current I for the condition of maximum cutoff sine wave. mum inrush current, and assuming no
transient flux was determined graphically If the residual flux could ever be equal damping or decrement in the first cycle.
by entering the excitation curve OSP with to saturation density, and neglecting Note that the width of this inrush-current
the instantaneous flux values. For ex- decrement, the inrush current would be a base is 240 degrees.
ample, at 90 degrees on the flux wave 0, complete sine wave, fully offset as shown Harmonic analysis of the current wave
on the right-hand portion of the figure, in Fig. 3(A). Since this wave would be in Fig. 3(B) yields the following values:
a value of 4. is determined. Entering the identical to a fully offset fault current fundamental 100%
without harmonics, no means for dis-
=

saturation curve with this value yields the direct current = 74.6%
value of exciting current Ix. This is then crimination between fault and inrush cur- second harmonic - 17.1%
plotted vertically at 90 degrees on the rents would be available. third harmonic=7.4%
Practically, the residual flux will always peak= 187%
right-hand portion of the figure to yield
one point on the current wave 1. Other be less than saturation density. The This wave was found to have the mini-
points are determined similarly. If the shape of the hysteresis loop of modem mum obtainable amount of second har-
actual excitation curve is used, as shown transformer steel will result in a maximum monic for a single-phase inrush for any
by the hystersis loop, legs will be produced residual flux of not more than 90% of residual flux between zero and 90%,
near zero on the current wave, as shown rated peak flux, if rated voltage was pre- using the method of Fig. 2 for obtaining
by the dashed portion of the wave. viously interrupted. This 90% value of the inrush currents.
These legs result from the bend in the residual flux is assumed to be the maxi- The method of determining inrush cur-
excitation curve near the saturation point. mum that will be encountered. rent in Fig. 2 assumes that the source
By assuming that the excitation curve is In order to fix the shape of the most inductance is zero and that the resistance
two straight lines, intersecting on the severe inrush to be expected, the point of of the source and the transformer winding
flux axis at a point S, called "saturation saturation density must also be fixed. is zero. This represents the extreme limit
density," the inrush-current legs are Based on modern transformer steels, for severity of inrush which, of course, is

OCTOBER 1958 Sonnemann, Wagner, Rockefeller-Magnetizing Inrush Phenomena 885

 second harmonic as a result of resistance.
However, the damping effect can be re-
garded as a safety factor in preventing
differential relay tripping during an
inrush.
The foregoing theory and conclusions
concerning a single-phase inrush are now 140%
expanded and modified to include 3-phase
connected banks.

Inrush in 3-Phase Transformers (A)

INRUSH CURRENT
The inrush current into a 3-phase bank .--s (NO DAMPING)
Fig. 1. Voltage, Aux, and current during a is complicated by the fact that there may xz
X~Z TOTAL
magnetizing inrush. Transformer enegized be an inrush current into more than one _L
FLUX WITH DAMPING
at zero on the voltage wave LLi.
phase, and these separate inrush currents 0

may be affected by the electric connec- .A ^dw 1----

*__v lo
140% r 24 F-
SATURATION %
DENSITY
. Jk M

never reached. The assumption of zero tions and/or magnetic coupling between 90%
source inductance and source resistance is phases. In addition, for relaying, the _ -

most closely approached when the kva of effect of delta-connected current trans-
the transformer is small as compared with formers must be considered. In the fol- (B)
the kva of the power system from which it lowing, the possibilities of inrush occurring Fig. 3. Effect of residual flux on inruh-cufrent
is energized. The effect of source in- in more than one phase is discussed along shape
ductance is to reduce the magnitude of the with the effect of an electric connection
inrush current through reducing the as represented by a delta-connected wind- A-Assuming residual equal to saturation
ing. density
magnitude of the excitation voltage. The B-Assuming 90%o residual
damping effect of previously neglected
resistance in the source and in the trans- RESIDUAL FLUXES
former winding is illustrated by the As mentioned previously, the amount the load current to form the total cur-
dashed inrush current in Fig. 3(B). of residual flux left in the core of a trans- rent, it follows that the exciting current
This is intended to be a qualitative illus- former depends upon the point in the cycle may be interrupted at almost any point
tration. Quantitatively, the damping at which the exciting current was inter- in its cycle, depending upon its phase
effect is frequently not nearly so much as rupted at the time of bank de-energiza- relationship to the total current, which,
illustrated. The time constant for the tion. If the exciting current has just it is presuimed, will be interrupted at or
decay of the inrush current is a function passed through a positive half-cycle and is near its normal current zero. Again,
of the total resistance of the source plus interrupted at its normal current zero, it is most likely that one phase will be
the transformer winding. Since the without forcing, then the core will be left interrupted before the other two because
resistance damping reduces the width of with positive residual flux of maxi- of the phase difference between the total
the base of the inrush current, adding to mum anmount. Whether or not a circuit currents. With a star-delta-connected
the "dead time" in the cycle when there is breaker will force an unnatural current bank, interrupting the first phase on the
no appreciable current, the per-cent zero when interrupting only exciting cur- star side does not extinguish the excita-
second-harmonic contact will be increased rent is a question which is not debated tion for the first-phase core, because of an
over the amount of second harmonic ob- here. Such a question becomes more or electric connection on the delta side.
tained by neglecting the damping. No less academic if load current was inter- Again, if a star-delta bank is de-energized
attempt is made in this paper to evaluate rupted at the time of bank de-energiza- from the star side, when the second phase
quantitatively the amount of increase in tion. Since the exciting current adds to is interrupted, the final remaining phase
will furnish excitation to the first two
phases in series through the delta-con-
nected windings at 50% of normal value
with a change in phase angle. When
these contingencies are taken into con-
sideration, it follows that the amount of
residual to be expected in the 3-phase
T
cores is rather unpredictable. With all
this in mind, it will be presumed that the
residual flux may be up to m um value
of 90% in any of the three cores, and of
either polarity.
With the foregoing assumption of
maximum residual flux of either polarity
TIME
Fig. 2. Derivation in all 3-phase cores, there will be times
t 180° 270 360 of magnetizing in- when all three phases will experience an
rush current from inrush when the bank is re-energized.
v excitation charac- This condition will be considered as it
terstic affects differential relaying. However,

0806 Sonnemann, Wagner, Rockefeller-Magnetizing Inrush Phenomena OCTOBER 1958

 Fig. 4 (leFt). Instantmneous
current difections for transfor-
mef a experiencing an inrush

Fig. 5 (right). Circuit used in

Reld test
currents may show up in all three phases
with only one phase experiencing ain- symbol to avoid conflict with a subsequent traces for I., Ib, and 1, represent star cur-
rush. The reason for this is the electric figure in this paper.) rents, inasmuch as the oscillograph ele-
connection between phases, and this Under the conditions as described, an ments were inserted in the circuit back of
phenomenon, for the purpose of this oscillographic record of the currents in the point at which the delta connection
paper, is called "helping effect." This phases g, &, and G would sbow phase & and was made up. The delta currents, I.-Ib,
particular condition will be dealt with phase C currents of opposite polarity to etc., were taken as indicated, as well as a
first. phase g. That is, while the wave trace trace for the current in the tertiary wind-
HELPMG EFFECT for phase a is in the positive region, the ing of the transformer bank.
wave traces for phases & and c would be in Fig. 6 is a reproduction of the oscillo-
Fig. 4 shows a star-delta7connected the negative region. If the connection to
bank consisting of three single-phase graphic records for the three star currents
ground were opened, it follows that the and the tertiary current, where the ter-
transformers labeled a, &, and C. It is magnitudes of the phase b and phase
presumed that the circuit connected to minology agrees with Fig. 5. The helping
currents would each be equal to one half effect is plainly evident in the traces for
terminals A, B, and Cis open and that the the magnitude of the phase a current.
transformer has just been energized from the three star currents I., I,, and 4, In-
With the ground connection complete as spection of these traces indicates that the
its star-connected side at terminals a, b, shown, the magnitudes of the phase & and
and c. It is presumed that transformer a inrush current started in phase a first.
phase c currents will be something less For the first portion of the cycle, phases &
is experiencing an inrush. The current than half the magnitude of the phase a
arrows represent instantaneous directions, and c were helping phase a as indicated by
current, depending upon the amount of the wave trace from the starting time up
on the basis that transformers & and c current, 3Io flowing into ground. If an
are helping transformer a through the to the peak P on the trace for Ib and up to
oscidlogram shows phenomena in line with the point X on the wave trace for I.. At
medium of the delta winding. The direc- the foregoing with respect to magnitude
tion arrows for the phase a windings are point X on I., note that the phase c trans-
and polarity, then it is to be presumed former begins to experience its own inrush,
shown in the same relative direction. that only one phase experienced an inrush,
That is, the currents are represented as as the current begins to increase rapidly in
provided that the oscillographic record magnitude. This presents a circuit prob-
flowing in at the polarity marks of the was taken from star-connected current
two windings such that their effect is addi- lem to the transformer in phase &, in that
transformers on the star side of the bank, it is now called on to help supply the de-
tive. It does not make any difference to and provided that the bank was energized
the core of the transformer whether it gets from the star side.
its exciting ampere-turns from the pri- In Fig. 4, the condition of transformers
mary winding or the secondary winding, & and c helping transformer a has been
or both. In this case, transformer 0, for. shown. Similarly, another diagram could
the duration of the inrush, needs an excess be drawn showing how transformers a and
amount of exciting current, and an & can help transformer c when trans-
ailiary source of supply is available IO
former c needs it. This would require a
through its secondary winding energized different set of direction arrows repre-
from the delta-connected windings of senting instantaneous current directions.
transformers b and C. Note that the The helping effect thus described can
direction arrows for the currents flowing switch from phase to phase for the con- IC
in the windings of transformers & and c are dition of multiple-phase inrush.
in opposite direction. This is because
these two transformers are acting as trans- MULTIPLE-PHASE INRUSH
formers in the usual fashion wherein cur-
rent flowing in at polarity mark on one The circuit used to obtain oscillographic
winding is neutralized by current flowing data from a transformer bank, as installed
out of polarity on the other winding. on a utility company system, is shown in ID
Thus, the current -ID is a helping-effect Fig. 5. Inrush currents were taken from
current which may be conceived as the 69-kv side, with the 154-kv disconnect
originating in transformers & and C as a switches open, as indicated in the diagram.
source and being used in transformer G as Wave traces were obtained for seven cur-
a load. (The minus sign is used with this rent values, as indicated. The wave Fig. 6. Oscillograms from Reld test

OCrOBER 1958 Sonnemann, Wagner, Rockefeller-Magnetizing Inrush Phenomena 8807

 made to predict rather accurately how Zs a
-oA
much shoulder to expect or how much
oscillatory current to expect in the re- a t-IDOB
spective phase and delta currents. -oEb Z5 b
It is first presumed that the total
amount of magnetizing current required
for the core of each of the single-phase Zos c b
Fig. 7. Equivalent diagram for phase a transformers will be a known quantity.
transformer These currents are designated as I,,. Zos'IC
, ~~~~~~~c
I,,,1, and Im,. Fig. 7, which is the equi-
valent circuit for the phase a transformer, I G1_G o4-
mands of both the phase a and phase b shoNvs I,,, as the current flowing through 310
transformers. Since these two demands the magnetizing branch of the trans-
are in opposite direction, the helping cur- former. Since I,.a is a function of the
Fig. B. A star-delta-connected bank with
rent from phase & transformer flows in the source impedance on star side
flux in the core, as described previously
direction of the maximum instantaneous for the single-phase inrush theory, it is
demand. From point P to point Q the also a function of the integral of the volt-
phase a transformer requirements are 8. For the moment, and subject to later
age applied to the transformer. A rigor- discussion, let it be conceived that the
greater than phase c; therefore, the in- ous solution for this magnetizing current,
stantaneous value of lb remains with a analysis is restricted to the fundamental
therefore, requires a knowledge of the component only, in which case, all doubts
negative polarity but a decreasing magni- voltage drops in the system. A good
tude. At point Q the requirements for are resolved concerning what values of
approximation, however, can be obtained impedances to use.
phase a and phase c transformers are equal by assuming an infinite system connected
in magnitude but opposite in polarity; The first three equations of Appendix
to the transformer and using the method I are true, regardless of the frequency.
helping-current flow in phase & is therefore shown in Fig. 2 to determine I.., I,, and
zero. From point Q to point R, the re- There are:
Imc. A further discussion of this approxi-
quirements of phase c transformer exceed mation follows later in the paper. a -Ima I'D (1)
that of phase a. Phase b current there-
fore increases with a positive polarity to Fig. 8 shows the circuit used for the Ib-Imb ID (2)
help supply this demand. At the peak of analysis. The generated voltages at the
ultimate source are shown as Ea, Eb, and IC-imc ID (3)
current R coincidental with point Y on
the trace for I,, the inrush for transformer E,. The total system impedance up to Equation 5, shown in the following, is
the transformer input terminals a, b, and also true regardless of the frequency,
a has temporarily subsided, and a and b c is shown as Z. and ZO,. The impedance
help phase c until the peak of current in provided that 3Io is defined as the current
phase i subsides to zero. The whole Z, is intended to represent the usual posi- flowing in the neutral connection or
tive- and negative-sequence impedance ground.
process is repeated in the next half-cycle. of the system, while the impedance ZOJ is
The flip-flop action of the helping effect is int.ended to represent the zero-sequence Ia+Ib+Ic 3Io (5)
also indicated by the wave trace for the impedance of the system. Three single-
tertiary current ID. In equation 9 the magnetizing current
The wave traces for Ia and I, are typical phase transformers are shown as a, ~, and for phase b transformer is arbitrarily set
of those with a "shoulder," and the wave c. The delta terminals A, B, and C are equal to zero for the first part of this
traces for lb and ID are typical of the
disconnected from the rest of the system analysis because first equations are to be
so that no current can flow out of these developed which show the helping effect
oscillatory wave form, both of which have terminals. Note that current can flow when only two transformers, a and C, are
been mentioned earlier in connection with within the delta winding, as shown by ID.
Table I. undergoing inrush. The derivation
Note that the reference direction arrow makes use of the knowledge that the sum
Up to this point the helping effect has for ID is shown as opposite to the direction of the three voltages around the delta
been used as a means of logic to explain of the direction arrows for the phase circuit from A to B to C and back to A
why the waveforms for inrush current into currents in the star winding. This is in must equal zero. Also, the three gen-
a 3-phase bank look like they do. linie with the usual concept of trans- erated voltages Ea, Eb, and E, must equal
With the waveforms explained qualita- former action, where current flowing in zero based on an assumed balanced condi-
tively on a logic basis, it now becomes at the polarity of one winding is neutral- tion, as shown in equation 12.
desirable to examine the circuitry to- ized by current flowing out of polarity in
gether with a mathematical analysis of the other winding. In this respect, this EAB+EBC+ECA=O (12)
the circuitry to determine if an approxi- figure differs from Fig. 4. The reason for Ea+Eb+Ec=O I
mate quantitative analysis is possible. this difference is that it facilitates the In normal circuit work voltage EAB
CIRcuT ANALYSIS writing of general equations which will would be determined by taking the gen-
allow for the possibility of any two trans- erated voltage Ea and subtracting out the
In Appendix I, certain formulas are formers helping the third, regardless of impedance drops up to terminals A and B.
developed that define the 3-phase inrush which one is involved. This would involve the term IGZTR. In
currents and the delta current in terms of Inasmuch as magnetizing inrush cur- such an operation, ZTR would represent
the total magnetizing current for each rent of the half-wave shape consists of a the total leakage impedance of the trans-
core and the transformer and system im- d-c component, a fundamental com- former, and the operation would further
pedances. In the following, the deriva- ponent, and a multitude of harmonics, a presume that the primary current equalled
tion of these formulas is discussed to- question is likely to arise concerning the the secondary current, neglecting mag-
gether with the assumptions that can be values of impedances to be used in Fig. netizing current. In this case, it is de-
888 Sonnemann, ITagner, Rockefeller-Magnetizing Inrush Phenomena OC rOBERR1958
sired to make allowance for the fact that pedance is zero (for an ultimate limit), 10 degrees apart, the waves for Ia, Ib,
ID will not equal the primary current then equation 27 becomes and 1, were plotted as shown. These
I., if transformer a is experiencing a mag- waves show a striking similarity in form,
netizing inrush. For convenience, then, Ia =6Ima- limc (28) relative position, and relative magnitude
6 6
each transformer is conceived to be repre- to those illustrated from the actual field
sented by its equivalent diagram as shown Under these conditions, equation 28 in- test in Fig. 6, even though no attempt
in Fig. 7, where the total leakage im- dicates that the actual phase current in was made to determine these particular
pedance of the transformer is split into phase a is five sixths of the total magnetiz- constants from the circuit used in the field
two parts, Zp and Z&s. In Fig. 7, Zp ing current for phase a, minus one sixth of test.
represents the primary leakage im- the total magnetizing current for phase c. A statement was made earlier that it
pedance, and Z88 represents the secondary The other one sixth of the magnetizing would first be conceived that the deriva-
leakage impedance. The magnetizing current for phase G must of necessity flow tions in Appendix I, and as discussed in
branch is represented by XM across which in phases b and c. Considering the other the foregoing, would be considered as
appears the excitation voltage E,, and extreme, when the zero-sequence source being applicable to the fundamental com-
through which flows the exciting current impedance approaches infinity as a limit, ponent only. However, the equations
Ima=Ia-ID. With the transformers rep- equation 27 becomes: used were fundamental to circuitry with-
resented by their equivalent diagrams, 2 1 out regard to frequency. It can well be
equations 13 through 15 are written to Ia 3Ima-I3 Imc
- (29) conceived that the coefficients of I,,m and
yield voltages EAR, EBC, and ECA. Im, in equation 27, for example, could
Equation 13 is This indicates that under the foregoing change somewhat as the frequency is
conditions, one third of the magnetizing changed, and certainly the reactance por-
EAR =Ea IaZs -o(Zos-Zs) - current for phase a transformer must flow tion of these terms would have very little
IaZp-IDZSS (13) in phases b and c. This particular limit effect so far as the d-c component of the
In the consideration of transformer can be easily justified by inspecting Fig. 8 magnetizing inrush is concerned. How-
equivalent diagrams, it is the usual prac- and visualizing that a switch is open in ever, it is believed that in so far as a
tice to assume that the primary imped- the neutral circuit. comparison between the harmonics and
ance is equal to the secondary impedance. Equation 30 yields an oscillatory-type the fundamental is concerned, there is
This has been done, as shown in equation wave whenever the magnetizing currents probably very little change as a function
26, in order to simplify operations. Im and I,. are of opposite polarity and of frequency. This statement is made
In Appendix I, use is made of the fore- are shifted in time phase with respect to based in part on the previously mentioned
going fundamental equations to derive the each other. Equation 31 is quite similar strong similarity between Fig. 9 and
following three equations: to equation 27, the difference being that actual test results as illustrated by Fig. 6.
the coefficients for the two magnetizing
II 5ZBB+2Zo8 \ Zs$+ Zos currents, I, and Imc, are interchanged. RELAY CURRENTS
\6Zss+3Zos J \6Z88+o3Zs / In Appendix I the derivation of the The currents considered so far have
(27) general equations 37 through 39 continues been the phase currents on the star side of
for the condition of all three phases ex- the bank with the inrush taking place on
Ib ID =(Ima+Imc) (6Z8+3ZO) (30) periencing an inrush. that side. In differential relaying, how-
In order to test the usefulness of equa- ever, it is almost universal practice to con-
(5Z88+Zos\ Ima Z88+ZOJ tions 27, 30, and 31, a sample derivation nect the current transformers in delta on
6Z,gS+3ZO,/ \6Zss+3Zo° was made as shown in Fig. 9. Two mag- the star side of a star-delta bank. (This
(31) netizing currents, ,,. and lInc, were as- connection eliminates zero-sequence cur-
sumed of equal magnitude, each with a rent from the relay, and yields the correct
It should be remembered that the fore- 180-degree base, and with Imc shifted 60 phase angles to go with star-connected
going equations are predicted upon an degrees from the point where it would
initial assumption that the magnetizing have been a mirror image of I,,,a. Two
inrush current in phase Q transformer is such magnetizing currents may be derived
zero. In examining equation 27 for limits, by the methods discussed in connection
note that if the zero-sequence source im- with Fig. 2 on the basis of the following
assumptions: 1. that the switch is closed TRACE
when phase a voltage is zero and increas- LNO.(4)
1. ing, 2. that phase a core has 40% residual
with positive polarity, and 3. that phase
C core has 90% residual of negative polar-
.0.5- I ity.
An arbitrary assumption was made - (5)
/lb that the zero-sequence source impedance - (6)
U ~ ~~~
I ~% DEGREES
was equal to ten times Z8s. This assump-
- 90 2820 p60 tion, when used in equations 27, 30, and
L( 0.5-
I IC I 31 yielded the following:
Ia =0.695Ia - 0.3051mc (55)
I = 0.305Ima-0.305Imc (56)
1.01
IC t.695h1ese
i 0.305tio,a (5a7) Fig. 10. Delta curmnts coincidental with
Fig. 9. Calculated inrush-current waveEorms Using these equations, and using points Fig. 6

OCTOBER 1958 Sonnemann, Wagner, Rockefeller Magnetizing Inrush Phenomena 009

 amount of magnetizing inrush involved in Fig. 12. Com- IRAImoaImb
phase b. If this is so, then trace 4 in bination of 180-
Fig. 10 for IRA is simply the total mag- degree base and .Ma imb7
netizing inrush current in phase a. On 240-degree bas X
this basis, its half-wave shape is logical, Inrush currents \
and there are no shoulders discerible.
Similarly, trace 5, RBB, is the negative of DEGREES
the magnetizing current in phase c. Note 0 120 240 :Si
that it has almost the same peak magni-
z 60 120 180 240 300 360 tude as scaled from the film and that it is -240e
TIME-DEGREES
displaced in time phase from the top
\SATURATION DENS IT Y/ trace, which is logical. The bottom
trace I,RC is the negative of the sum
of the top two traces, and this is PHASB RELATIONSHIPS OF HARMONMCS
consistent with its broader base. It is Tlhe near agreement in figures of 16.5%
also apparent that the wave trace for IRC second harmonic stated for Fig. 12 and the
-3.0-- should have a double shoulder to a slight
degree. This would be more obvious if 17.1% second harmonic previously given
Fig. 11. Limiting Inrush condition the oscillograph had been running faster for a wave having a 240-degree base, asi's
to spread out the time scale. (The
shown by I.. in Fig. 12, requires discus-
double-shoulder effect is seen in Fig. 12, sion. A separate analysis for a half-
current transformers on the delta side of wave shape having a 180-degree base, as
yet to be discussed.) shown for -I,, in Fig. 12, indicates a
the bank.) The actual relay currents are
derived in Appendix II, and they may be MOST SEVERE CASE FOR THB RELAY second-harmonic content of 42% of the
as shown in equations 46 through 48, or fundamental. Offhand, one would think
as shown in equations 49 through 51. Since Appendix II has established that that, with 42% second harmonic in
any one relay current is the difference in -I,, and 17.1% second harmonic in I,..,
IR.A Ia - Ia = Ima -Imb (46) the total magnetizing current of two Iu would contain more than 16.5%
phases, it becomes possible to establish second harmonic.
IRlb -Ic'mI - Inmc (47) the. most severe case from a relay stand- Analyses made of half-wave shapes
IRC Ica4Imc-Ima (48) point for any one phase. For this pur- having bases of more than and less than
RAAMIa'cIa -ImC (49) pose, equation 46 for phase a relay is 180 degrees, as well as 180 degrees, lead to
chosen. the following conclusions. The funda-
IBBI' bIaX=Imb -Ima (50) In Fig. 11, a flux plot is shown for the mental-frequency component comes to a
IRC' Ic Ib =mcZnc- I (51) maximUM possible inrush condition as it positive peak coincidental with the posi-
affects phase a. Residual fluxes of 90% tive peak of the total magnetizing current.
From inspection of these six equations, have been assumed for all three phases, The second-harmonic component also
it is seen that positive for phase a and negative for comes to a positive peak at the same time.
IRA - I'BB' (52) phases b and c. With the bank energized With this in mind, note that the funda-
at the zero voltage point on phase a volt- mental components of I,,. and -Im_ in
Ihus, -ht0(53)
- age, while phase a voltage is increasing in Fig. 12 will add 60 degrees apart, giving a
'RC =-IRA (54) the positive direction, a maximum inrush fundamental value for the combined wave
is experienced in phase a accompanied by which is not much less than the arithmetic
From a consideration of the last three a substantial inrush in phases b and c. sum of the two separate fundamental
equations, it is seen that, for the purposes The flux for phase a extends 1.5 per unit components. However, since the peaks
of studying waveform, there is no prefer- beyond saturation density, and the flux of Ina, and -1,, are 60 degrees apart on
ence between equations 46 through 48 in phases b and c extends 1 per unit be- the time scale for the fundamental, the
as compared with equations 49 through yond saturation density. In accordance peaks of the second-harmonic component
51. with the method of Fig. 2, this yields mag- are 120 degrees apart on their own time
A significant feature of these equations netizing currents for phases a and b with scale. Adding the two second harmonics
is that they show that the helping effect peaks in the ratio of 1.5 to 1.0, and with 120 degrees apart on their own time scale
is no longer present. The current flowing. bases of 240 degrees and 180 degrees, leads to a new value of second-harmonic
to any one relay is simply the difference respectively. These are shown in Fig. component which is less than the larger of
between the total magnetizing currents 12. Fig. 12 also shows the relay current the two components which were added.
of two transformers, as typified by equa- IRA= Ima-Im. The authors conclude that there is no
tion 46. For the relay in phase a, re- In Fig. 12, a double-shoulder effect is particular significance to attach to the fact
ceiving rBu, it does not matter whether seen wherein the shoulders are less pro- that the figures are so nearly the same for
phase c transformer is experiencing an in- nounced tham those previously examined. the two waveshapes.
rush or not. Also, the amount of the Nevertheless, they are there. It is easy to reach a false conclusion con-
helping effect is of no consequence, inas- An analysis of the waveform for 13A of cerning the appearance of third-harmonic
much as it disappeared in the delta relay Fig. 12 yields a second-harmonic com- currents in the current delivered to the
currents. With this in mind, it is of ponent equal to 16.5% of the funda- relay from delta-connected current trans-
interest to study Fig. 10 and compare it mental component. This is practically formers. After the transformer bank has
with previously discussed Fig. 6. the same as the 17.1% second harmonic reached steady-state condition, it may be
In the previous discussion of Fig. 6, it reported for a wave of the single-phase presumed that the magnetizing currents in
was stated that there was no material shape having a 240-degree base. the three phases are equal in magnitude

890 Sonnemann, Wagner, Rockefeller-Magnetizing Inrush Phenomena OCTOB:R 1958

and 120 degrees apart on a fundamental- A.- r a rents, the predominant harmonic is the
ima4,
frequency basis. It may then be pre- IA second. 2.The minimum possible second-
sumed that, with like third-harmonic harmonic component which could be ob-
a

components in each magnetizing current, ob tained is in the order of 16 to 17% of the

this third-harmonic component would be IB fundamental. 3. This low percentage will
subtracted out from the relay currents by b
actually never be attained, inasmuch as
means of the delta connection. This is be- l oc the values of 16 to 17% second harmonic
cause, with the fundamental components IC are predicted upon the assumption that
120 degrees apart, the third-harmonic there is no decrement, as well as on other
components would be 360 degrees apart. -L pessimstic assumptions. 4. In a large
During the inrush condition, however, a number of measured waves, the lowest
third harmonic may be present in the re- Fig. 13. Inrush currents on delta tide of star- second-harmonic component found was
lay current. In Fig. 12 there is a third- delta bank 23% of the fundamental value. 5. The
harmonic current in the total relay cur- third-harmonic current component is not
rent IRA* This is because there is a reliable as a means of restraining a differ-
third-harmonic component in Ima and energize differential relaying. The re- ential relay during the inrush period.
none whatever in the current -I.b. In lays will thus receive the phase currents,
this connection it was noted in the anal- IA, IB, 'c, of Fig. 13. Inspection of Fig.
13 yields the following equations.
yses of various half-wave shapes that a
half sine wave with a 180-degree base has
Appendix 1. Magnetizing
IA Inma-Imc (58) Inrush Helping Effect
no odd harmonics whatever, including
the third. With bases greater than 180 IB Imb -Ima (59)
degrees, the third harmonic comes to a IcImc'-m (60) With reference to Figs. 7 and 8, the
negative peak when the fundamental following applies
component is at a positive peak. Con- In these equations, the magnetizing cur- la Ima=ID (1)
versely, with bases below 180 degrees, rents ',,m, I,0, and Imc are, of course, the
the third-harmonic component comes to total magnetizing currents involved for 1b-I'mbID (2)
a positive peak when the fundamen- the three respective phases. This is true IC imc ID (3)
tal component is also at a positive peak. because there is no helping effect possible
from the star-connected side with this Adding equations 1 through 3 gives
Thus, it is seen that there is a radical
phase shift in the phase position of the side on open circuit, as it would be if the la +lb +IC (IM4a+Imb+IMC) = 3ID
-
(4)
third-harmonic component, depending bank is first energized from the delta
side. I4+Ib+Ic =31o (5)
upon the width of the base. When a
severe inrush starts, with a base of more There is no difference in form between 3Io-(Ima+Imb +Imc) 3ID (6)
than 180 degrees, the third-harmonic equations 58 through 60 and equations 49 1
component will have one polarity with through 51, developed in Appendix II. ID =IO --3 (Ima+Imi+ImC) (7)
respect to the fundamental as measured Therefore, it is concluded that the inrush
at the peak of the fundamental. As the problem, from a relay point of view, is the
same with star-connected current trans- IO='ID +13 Ima+Ima,+Imc) (8)
inrush subsides, the peak of the total
magnetizing current decreases and the formers measuring inrush currents into a
delta winding as it is with delta-connected Let
base shrinks. As the base shrinks
through 180 degrees, the third harmonic current transformers measuring inrush Im,, 0 (9)
disappears. As the base of the imrush currents into a star-connected winding. Then
current shrinks to less than 180 degrees, 1
the third harmonic reappears with re- Current-Transformer Saturation ID-Io-0(Ima+imc) (10)
versed polarity to what it had before. It 3
is thus seen that to assume that the delta Inrush tests were conducted using two
connection of the current transformers sets of current transformers, one of which IO =ID+3 (Ima+Imc) (11)
subtracts out all third-harmonic com- showed signs of saturation because of the
d-c component of the magnetizing inrush In Fig. 7
ponents from the relay is a false assump-
tion. At the same time, it is seen that if current while the other did not. Har- EABs+EBC+ECA A -

an attempt were made to design a har- monic analysis of identical cycles of the (12)
two traces did not disklose any case where
Ea+Ea+Ec- 0
monic-restraint transformer differential
relay utilizing third-harmonic component current-transformer saturation reduced EAB Ea -IaZs -IO(Zo8-Z,) -IaZp -IDZSS
the second-harmonic percentage below the (13)
only, it would be liable to fail on inrush
currents whenever one phase only started value obtained from the trace where cur- EBEC=Eb-IZ - Io(Zos-Z) -IbZP -IDZBB
out with an inrush having a base of 180 rent-transformer saturation was not evi- (14)
degrees or more. dent. It is logical that this should be so,
based on the general theory of operation E04 =EC -IcZ8 -Io(Zos -Zs) -IcZpI-IDZg.
of current transformers, which is not (15)
Inrush Currents on Delta Side of discussed in detail here. Adding equations 18 through 15 gives the
Star-Delta Bank following
Conclusions 0 =0-3IoZs-31oZo+3IoZ -3IoZp-
The current transformers on the delta
side of a star-delta-connected bank usually In general summary, the authors con- 31DZs,s (16)
will be connected im star when used to clude that: (1). In magnetizing inrush cur- -IOZOS-IOZP-IDZBB O (17)

OCTOBBR 1958 Sonnemann, Wagner, Rockefeller-Magnetizing Inrush Phenomena 891

 I ZP+ZO*)( 18) lb-ID= (Im_+_Im)Z_+ZO_ )
-
Appendix II. Relay Currents
or As in Appendix I and with reference to
Using equations 3, 23, and 26 gives Fig. 8, the following applies

ZP+ZOlD
19) IC-Imc=-I(Ima+Imc)( zs+Z7) IaeIma+ID (40)
Ib =Imb+ID (41)
Substitute equation 19 in equation 10
Icimc(i 6Zss+3Zos) IC=ImC+JD (42)
ID -ID (Z +Z
\Zp+Zoe 3 (Ima+Imc) (2 0) With delta-connected current trans-
/Ia Zss+Zot formers, the currents to the relay are
ID[1+(Z Z )I (Ima+Imc) (2 1) IRA =Ia-b (43)
( 5Zss+2Zo, _/ Z__+ZO_\
_
Ic-Imc t6Z48+3Zo8 / \6Zss+3Zo,/ JI -Ib(-4IC4)
ID((zp+z08t+z88\
1
(212)
3 (Ima+Imc)
Z IRtC -la (45)
Zp+Z0.
Now assume that Use of equations 40 through 42 im equa-
zp+zO tions 43 through 45 cancels out the term
1- I( m +mc)(Za (
) 2) ImbHO (32) ID and yields
Substitute equation 23 in equation 1 Substitute equation 19 in equation 7 IRA Ia Ib -Ima -mIn (46)

Ia Ima+ID =Ima-13 (Ima+Imc) X ID= -ID (ZSS )-(Im+IMb+ImC) IRB Ib-Icl Imb-ImC (47)
(33) fCt=Ic-Ia=Im=cIIma (48j
If the delta connection is made up the
/zp+zos+zss IDzI
'1+ +z08 1 other way, there results
[ 3\PZP+ZO5ZB) (4 3Ima±Imb+Imc)
_

= Ia-IC'Ima
(34) IRA mc
-
(49
laIa1 I ZP+ZOS
1 Zp+Zos, Substitute Z88 for Zp (equation 26), then I'RB' Ib-Ia Imb-Ima (50)
Inc IPZO IRC' =l b-=Ib = Ime
3 \Zp+Zo+Zss/
ID
/2Z883+Zos 1
(S+Ima+lm+ImC)
(51)

'ma
I4, [3ZP+3Zos+3ZSB-Zp-Zo]
__
Zs's fZOs 3 By inspection, it is seen that
3(ZP+ZOS +Zss) IRA = -RB' (52)
ID=_ 3(ZS +Z
3\2Zss +Zo. (I,imi+Imb+Imc) (35) IRB = 'IRC (53)
Substitute equation 35 in equation 1
2ZP+2Zos+3Zss_ IRC 'IRA' (54)
I
3(Zp +Zo,+Zs)J la-ID+ima =I71,A Z z X
ReFerences
Inc [3(Zp+Zos+Zs) (25) (Ima+Imb+Imc) (36) 1. A TNtAnssORunR DIFFNDRBNT[UL R*LAY WITH
SBSCOND HAxomNc RBST?RANT, R. L. Sharp, W.
Let Ia =Ima (l Z8S+ZO)_ E. Glassburn. AIEE Transactions, vol. 77, pt.
III, 1958 (Paper no. 55-818).
zp = ZRs (26) 2. TRANS0SSxN AND DISTRIBUXON RzVBZBNCx
ZS+ZO BooK. Westinghouse nlectric Corporation, B.
Then (Imb+IMC) Pittsburgh, Pa., 1950, pp. 126-28.
6Z88g+3Zo5 / 3. PRINCIPL3S op ALTBzNATINO-CUR*DNT MA-
Ia Ima
= 5ZS8+2Zo,
( 6Z,8-+3Zo, / Im (,Zz+Z0.
6Z8s+3Zosl I
/5Z88+2ZO\ / Z88+ZOa
6zss+3Zo8J \6zss+3Zo8J
CmsNBRY (book), Ralph R. Lawrence, Renry B.
Richards. McGraw-Hill Book Company, Inc.,
New York, N. Y.. fourth edition, 1953, pp. 27-31.
(27) 4. DiB WUCHSBLSTROMT3CNNIK, BAsw 2: DIn
(Imb+Imc) (37) TRANFORAETORNK (book), la Cour, Faye-HaneaL
With the limit Zes -O 0, the following results Edwards Brothers, Inc., Ann Arbor, Mich., 1944,
Similarly pp. 143-51.
5 1 / 5Zgs+2Zo8 5. TsNsFORamc MAONUTZrNO INRUSH CUmNT,
Ia 66 Ima -6 Imc (28) _/ Z8s+Zo, ,
'b =Imb \ 6Zs+3Zo0 6Z8S+3Zo8 x T. R. Specht. AIER Transactions, vol. 70, pt. I,
1951, pp. 323-28.
With the limit Zos , the following results (IUma+I,.c) (38) 6. TRANSFORMBR MAONRT[Z[NO INtUSH CUR-
NT AND INFLU3NCR ON SYSTEM OPnRATrrot,
2 1 L. F. Blume, G. Camlli, S. B. Farham, H. A.
Ia Ima-Imc (29) Peterson. Ibid., vol. 63, 1944, pp. 388-75.
3 3 Ic-Imc (6Zss+3Zo)s(6Z +3 ) X 7. TRANsnoaxu ENOINBBDINO (book), L. F.
Blume et al. John Wiley & Sbns, Inc., New York,
Using equations 2, 9, 23, and 26 gives Ima+Imb) (.39) N. Y., 1938, pp. 23-36.

892 Sonneimann, Wagner, Rockefeller-Magnetizing Inrush Phenomena OCTOBBR 1958