You are on page 1of 10

294 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. IA-15, NO.

3, MAY/JUNE 1979

Current Transformer Burden and Saturation


LOUIE J. POWELL, JR., SENIOR MEMBER, IEEE

Abstract-A tutorial review of the steady-state and transient behavior Ip Ip/N R+iwL Is
of current transformers used with power system relays and meters is
presented. ANSI accuracy standards are discussed and consideration is
given to the significance of burden and saturation in the application of N:1 | Zb
protective relays.

INTRODUCTION
IN RECENT YEARS, as power systems have grown both in
complexity and in the magnitude of available fault current, Fig. 1. Equivalent circuit for current transformer.
engineers have become increasingly aware that current trans-
formers are not ideal devices and that the error introduced by
current transformer (CT) saturation can contribute to incorrect Coupling between the primary and secondary windings of a
transformer is determined by the flux patterns of the two
operation of system protective relays. This awareness has led
to concern on the part of users who, in an attempt to assure windings. If the windings are suspended in air, the flux set up
in the source winding extends to infinity in all directions with
proper system operation, have written stringent performance the result that only a small fraction of the available flux links
specifications. This in turn has caused anxiety among manu- the second winding. This arrangement is referred to as an air-
facturers who are faced with the challenge of meeting these
core transformer, and when applied to CT applications is
specifications while at the same time maintaining and hopefully
reducing costs. sometimes called a "linear coupler" because the relationship
Current transformer performance under dynamic system between secondary and primary current is fixed by the geo-
conditions is an extremely complex, nonlinear phenomenon metry of the windings and the number of turns.
which in general cannot be accurately determined using analy- Because only a small fraction of the available flux in an
tical methods. Shortcut procedures exist which result in an air-core transformer actually links the windings, this arrange-
approximation to CT behavior which is sometimes satisfactory. ment is relatively inefficient. The coupling between windings
can be drastically improved by putting them on an iron core
This paper will review the qualitative aspects of saturation and
explore some of the limitations of the quasi-linear approxima- which acts to confine the flux and maximize the flux linking
tions. After a brief discussion of the American National the two windings. It is the iron core that produces the non-
Standards Institute (ANSI) specifications for CT performance, linearity and inherent error of practical current transformers.
consideration will be given to the significance of burden and Fig. 1 illustrates the equivalent circuit for a current trans-
saturation in applying protective relays. former with a ratio of N: 1 which can be used to approximate
the nonlinear behavior of the CT [7]. This circuit comprises
CURRENT TRANSFORMER FUNDAMENTALS five major elements. An ideal transformer is included to ac-
Except for mechanical differences necessitated by the way count for the nominal turns ratio of the CT. Current trans-
they are used, current transformers are like any other kind of formers manufactured for use in the United States are generally
transformers; they are four-terminal (two-port) passive devices rated for SA continuous current in the secondary winding, and
in which the output is a function of the input. In the simplest the ratio is expressed on that basis. The ratio of the ideal trans-
model of a current transformer, the relationship between the former in this equivalent circuit is the ratio of rated primary
output quantity and the input quantity is a fixed number current to 5A rated secondary current. Series resistance is
called the turns ratio. The input and output quantities can be included to account for the resistance of the windings. In
either current or voltage, although for current transformers it wound type CT's, the resistance R of Fig. 1 is the sum of the
is conventional to talk in terms of primary and secondary resistance in the secondary windings, and the resistance in the
currents. In an ideal, or fixed ratio transformer, the principle primary windings modified by the square of the turns ratio
of volt-ampere conservation applies; i.e., ampere turns input N. In the case of bar type current transformers which have no
equals ampere turns output. actual primary winding, R is only the resistance of the second-
ary windings. jwL is the leakage reactance of the windings.
Paper IPSD 77-22, approved by Power Systems Protection Com- The significance of this term will be discussed later. As will
mittee of the IEEE Industry Applications Society for presentation at become evident shortly, the behavior of a current transformer
the 1977 Industrial and Commercial Power Systems Conference, Pitts- is due in part to the characteristics of the CT itself and in part
burth, PA, May 9-12. Manuscript released for publication February 28, to the impedance with which the CT is loaded. This impedance
1979.
The author is with Contractor Equipment Business Division, General ZB is referred to as burden and in the equivalent circuit of Fig.
Electric Company, Schenectady, NY 12345. 1 is considered to be a complex impedance including the entire

0093-9994/79/0500-0294$00.75 i 1979 IEEE


POWELL, JR.: CURRENT TRANSFORMER BURDEN AND SATURATION 295

T - - -

IC~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

t
I-
z
0
U)
Ia
0

.001 I .01 0.1 10 30 50 70


RMS SECONDARY AWERES EXCITrNG CURRENT DEGREES PHASE ANGLE

Curve Ratio Winding Resistance

A 100/5 0.06 Q2
B 600/5 0.33 Q
C 1200/5 0.67 Q

Fig. 2. Typical CT secondary excitation characteristics.

external secondary circuit-wire, relays, meters, etc. Zc is a The left portion of Fig. 2 is the familiar secondary excitation
complex nonlinear impedance used to represent the effects of characteristic of a current transformer which relates the core
the iron core of the current transformer. exciting current Ie to the secondary excitation voltage, Vs. It
The term "fully distributed winding" is often used in is important to note from the secondary excitation curve that
application literature to describe a prerequisite condition for the core always requires exciting current. This means that the
current transformers. This refers to a winding which is evenly secondary current produced by a CT can never be a true
distributed around the circumference of the toroidal core as replica of the primary current. That is Is Ip/N. Another way
opposed to one wound on only a portion of the core. The of saying this is that the phenomenon called saturation is
large current transformers used for ground fault protection on actually continuous; there is no point above which the CT is
low-voltage systems are examples of CT's which do not have saturated and below which ideal operation can be expected. It
fully distributed windings: one segment of the core is remov- is true that above certain voltages (corresponding to the knee
able to allow retrofit installation, and the segment does not point of the secondary excitation curve) the exciting current
contain its share of the total winding. drawn by the core increases far more rapidly than does the
Winding distribution is significant in that the leakage react- secondary excitation voltage, and thus the ratio error of the
ance of a fully distributed winding is negligibly small and is current transformer rapidly becomes more severe.
normally considered to be zero. This means that application It is also important to note that, while the secondary excita-
calculations can proceed using only winding resistance. This is tion curve appears to be two straight line segments joined by a
of practical significance in that the leakage reactance of a non- small nonlinearity about the knee point, the coordinates of
distributed CT is generally not known. If the reactance can be this characteristic are expressed in logarithmic terms, which
determined (for example by laboratory measurement) a non- means that the relationship between the secondary excitation
distributed CT can be used in any application. voltage and exciting current is truly nonlinear. Thus, the
Note that Zc is an artifice used to model an observed physi- artificial Zc shown in Fig. 1 is not a constant impedance; not
cal phenomenon. The iron core of a CT demonstrates a behav- only is the magnitude of this impedance variable, but the angle
ior which can be analyzed by means of a shunt impedance. undergoes nonlinear change as well. In the analysis and exam-
296 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. IA-15, NO. 3, MAY/JUNE 1979

ples given in this. paper, this artificial impedance will be treated


as being purely resistive, and the real and imaginary compo-
nents of winding impedance and secondary circuit impedance
will be combined by scaler arithmetic (rather than vector
algebra) to form a single immittance. This method results in a
"worst case" approximation. Actual secondary excitation
characteristics are often accompanied by phase angle curves
(the right portion of Fig. 2) which can be used to restore the
complex nature of the artificial core impedance Z,, by expres-
sing the angular shift between exciting current and excitation
voltage. The use of the phase angle curve along with complex
manipulation of impedances will give greater precision to the
approximation calculations.
Therefore, the equivalent circuit of Fig. 1 taken together
with the secondary excitation characteristic of Fig. 2 can be
used to analyze the performance of a current transformer
feeding a specified secondary burden. This analysis will fall
into one of two categories-either determining the primary -
'
current corresponding to a given secondary current or deter-
mining the secondary current corresponding to a given primary Fig. 3. Hysteresis curve for iron core, illustrating relationship between
core flux and exciting current.
current. The former is relatively straightforward and can be
accomplished with little effort, while the latter is more diffi-
cult because the nonlinear relationship between secondary steel that the reluctance of the core increases with magnetic
voltage and exciting current requires an iterative solution. density. The saturation effect described in the secondary
It should be noted that the "rms secondary voltage" given excitation curve of Fig. 2 is actually another way of showing
on the secondary excitation curve of Fig. 2 is the voltage across that as the level of excitation (VJ,) increases, the flux in the
the output terminals of the CT and not the V8 of Fig. 1. This iron core increases producing an increase in reluctance, thereby
slight difference simplifies some calculations, although numeri- requiring more exciting current. Also, as the reluctance of the
cally it is not greatly significant. core increases compared to the reluctance of air, more flux
Outdoor oil circuit breakers are often fitted with current spills out of the core and attempts to link the two windings
transformers with tapped windings. The manufacturer generally via the less efficient air path.
provides the secondary excitation curve for the full ratio, and Actually, the secondary excitation characteristic as shown
it is frequently necessary to apply relays and meters on a in Fig. 2 is a simplification of the far more involved B-H
lower tap. It is quite simple to produce an excitation curve on hysteresis curve of magnetic materials which relates core flux
the lower tap if it is recalled that all windings on the same core density (B) and magnetizing force (H). For the purpose of
will have the same exciting ampere turns at a given flux density. this paper, flux density B can be expressed in volts per turn
Thus, the knee point on the excitation curve for the full ratio and magnetizing force H in ampere turns. Then, flux density is
(N turns) can be converted to the knee point on the curve for proprotional to the voltage impressed on the secondary of the
a tap (Nt turns) by multiplying the secondary voltage by Nt/N current transformer V8 and the magnetizing force is propor-
and the secondary current by N/Nt. The new curve could then tional to a current Ie which in the model appears to flow
be drawn parallel to the old curve. through the core impedance Z,
As Fig. 2 indicates, the performance of current transformers A significant point to note about the B-H curve is that it is
is better when the CT ratios are higher. This is because, with not a straight line. As indicated in Fig. 3 this means that with
constant volts per turn of flux density, higher voltages can be sinusoidal excitation, and hence sinusoidal flux, the magneti-
developed because more turns are available. zing force and hence exciting current are necessarily nonsinu-
It should be noted that most manufacturers of tapped soidal [7]. Referring to Fig. 1, if Ie is nonsinusoidal, and if
bushing-type CT's in the United States design them to have Ip/N is sinusoidal, Is must be nonsinusoidal. Even more
fully distributed windings on each standard tap as well as on significant is the subtle distinction this implies between
the full winding. the hysteresis curve of Fig. 3 and the secondary excitation
It is of course true that this approximate analysis is based characteristic of Fig. 2. The secondary excitation curve relates
on a number of simplifying assumptions. To gain a better the rms values of exciting current and excitation voltage, while
understanding of the saturation phenomenon it is necessary to the hysteresis curve relates the instantaneous values of core
examine these assumptions individually. It should be recalled flux density and magnetizing force.
that the iron core of a transformer is necessary to improve the In general, short circuit currents are offset to some degree.
efficiency of the device by assuring that most of the flux The dc component of an offset fault current introduces a dc
created in the source winding links the load winding. In this flux in the core of a CT about which the ac flux oscillates.
sense, then, the iron core forms a "conduit" for flux linking Thus, offset fault current requires more excitation current than
the two windings. It is an essential characteristic of iron or the secondary excitation characteristic would indicate, and as
POWELL, JR.: CURRENT TRANSFORMER BURDEN AND SATURATION 297

a result, the output of a current transformer in the presence of


offset fault currents is even less than an rms approximation 20.
would lead one to expect. Of course, once the dc transient
15
decays, the rms characteristic of the secondary excitation curve
can be used to approximate CT performance. The time con-
stant of the dc transient in CT secondary current is a function
of system and CT circuit parameters; dc saturation has received
a lot of attention recently as it has been observed that system
X/R ratios have been climbing. In especially severe cases, :c time
special CT's with gapped cores have been used to reduce the
secondary circuit time constant and assure that current is LJ
available for relaying sooner after the fault occurs. CLL
The worst case for saturation occurs when the core of the
current transformer contains flux prior to the start of the
primary current wave. So-called "premagnetization" (or
"'remanence' in recent literature) comes about as a direct
result of interrupting the current circuit when CT flux is not
zero [3] . This can happen during severe offset faults or through Fig. 4. Output current of 50/5 Type JCB-O CT; primary current: 1200
careless use of dc in ringing-out CT circuits or in determining A. RMS; secondary burden: one PJC 1 lA relay.
CT polarity. The effect of remanence is to increase the time
required for the current transformer to arrive at the rms burdens is sufficient. "Worst casing" the solution by this
operating characteristic represented in the secondary excitation method gives more conservative results because it magnifies CT
curve. inaccuracies.
Before going further, it would be well to summarize the Fig. 4 has been included to illustrate these points. A com-
points which have been made or illustrated so far. monly found CT in industrial plants is the 50/5 current trans-
1) Saturation is a continuous phenomenon. While there is a former with a large window used to measure ground fault
point above which saturation becomes far more severe, current current. This CT is often coupled with an instantaneous over-
transformers always require exciting current and therefore current relay. Fig. 4 is the actual waveform of the output of a
always include some ratio error. 50/5 ratio General Electric type JCB-0 CT loaded into a
2) The output of a current transformer is always less than PJCllA relay with a 0.5A pickup calibration. The primary
the ideal output. Ip/N, because there is always exciting current. current for this test was 1200A, and with a nominal ratio of
3) Actual saturation includes an hysteresis or backlash effect 10:1 the secondary current should be 120A (170A crest). In-
which cannot be analyzed quantitatively with simple mathe- stead, the crest of the nonsinusoidal wave is 16A, and the rms
matics. magnitude is 8.75A.
4) The secondary or output current from a CT has a dis-
torted wave form because exciting current always is nonsinu- CT ACCURACY STANDARDS
soidal. Industry standards have been written which give a method
5) Because the output of a CT is a distorted sinusoid, pre- of classifying the accuracy of current transformers. ANSI
cise calculation of CT performance is not possible. C57.13-1968 makes a distinction between CT's destined for
6) DC saturation is a special case arising from asymmetrical metering applications alone and those which will be used with
fault currents and/or a premagnetized CT core. DC saturation protective relays [1]. The reason for this division is that CT's
is especially important because in severe cases it can reduce the used for metering alone must be accurate in the lower range of
output of a CT to almost zero for the first few cycles of the currents up to full load (usually 5A secondary) while accuracy
fault. The duration of this condition is determined by the in relaying circuits is of importance at higher currents. Also,
system X/R ratio. metering CT's may be used to determine energy and demand
In addition to these basic principles, an approximate analysis levels for billing purposes where slight errors, which might be
procedure has been suggested which combines the equivalent tolerable in relaying applications, can have a significant finan-
circuit of Fig. 1 with the simplified secondary excitation cial impact on either the seller or consumer of electric power.
characteristic of Fig. 2. This analysis involves four critical The ANSI accuracy classification system makes use of a set
assumptions which must be recognized if the accuracy of the of standard secondary burdens and a set of accuracy classes.
analysis is to be correctly assessed. 1) Hysteresis is neglected. The standard burdens for metering CT's are given in Table I.
A single line version of the B-H curve, called the secondary Table II lists the standard accuracy classes. In this table, the
excitation characteristic, can be used to represent the nonlinear correction factor is both the ratio correction factor and the
relationship between V8 and I, 2) The output of the current transformer correction factor. The former is the factor by
transformer will be assumned to be sinusoidal. This will permit which the nominal current transformer ratio must be multi-
calculation of rms quantities using simple mathematics. 3) DC plied to account for error, while the latter is the factor by
saturation will be neglected, and only ac effects will be calcu- which a wattmeter reading must be multiplied to correct for
lated. 4) Arithemetic addition of winding impedances and both ratio error and phase angle error in the nonlinear CT. The
298 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. IA-15, NO. 3, MAY/JUNE 1979

TABLE I An extremely good CT might be described under this pro-


ANSI STANDARD BURDENS FOR CURRENT TRANSFORMERS cedure as C800 or T800 while a poor transformer could be
(60 HZ) rated CO0 or T0. The C indicates that the CT accuracy can be
For 5 Ampere calculated while a T prefix means that the CT has appreciable
Standard Burden Volt- Power Secondaries
Designation Amperes Factor R(Q) L(mhy) Z(Q) leakage flux in the core and the accuracy must be determined
B-0.1 2.5 0.9 0.09 0.116 0.1
by test.
B-0.2 5.0 0.9 0.18 0.232 0.2 As useful as the ANSI classification standard may be there
B-0.5 12.5 0.9 0.45 0.580 0.5
B-1 25.0 0.5 0.50 3.3 1.0 are important limitations both to the user and the manufac-
B-2 50.0 0.5 1.0 4.6 2.0 turer. First, while it gives the user information about the per-
B-4 100.0 0.5 2.0 9.2 4.0
B-8 200.0 0.5 4.0 18.4 8.0 formance of CT's at certain discrete points, this method does
not give enough information to permit calculation of approxi-
mate performance for specific burdens at specific current levels.
TABLE II Secondly, it gives the user a way to specify to the manufacturer
ANSI STANDARD ACCURACY FOR METERING CT'S
the degree of ratio error tolerable in a CT for a specific relaying
Accuracy Limits on Correction Factor application, but it does not provide a mechanism for specifying
Class 100% Rated Current 10% Rated Current
1.2 0.988 - 1.012 0.976 - 1.024
the allowable phase angle and harmonic distortion errors, both
0.6 0.994 - 1.006 0.988 - 1.012 of which can be critical. These problems are most important to
0.3 0.997 - 1.003 0.994 - 1.006 the relay engineer whose needs can rarely be satisfied with
anything short of secondary excitation curves.
TABLE III SIGNIFICANCE OF CT SATURATION ON
ANSI STANDARD BURDEN CAPABILITIES OF RELAYING CT'S
PROTECTIVE RELAY APPLICATIONS
Secondary Voltage Ohmic Burden
As has been indicated, the analytical procedure of deter-
10
20
0.1
0.2
mining the performance of a current transformer can be
50 0.5 laborious and will not produce anything more than an approxi-
100 1.0
200 2.0 mation to actual CT behavior. It is frequently necessary to
400
800
4.0
8.0
employ this procedure in applying relays, and often it is
necessary that special techniques be employed to minimize
apparent relay error. It is appropriate that a portion of this
limits of transformer correction factor must be met with the paper explore the significance of CT performance on relay
additional requirement that the power factor of the load applications to put these things in perspective.
being measured must be between 60 and 100 percent. The In general, it is desirable that relays be presented with pure
correction factors listed in Table II apply at 1.0 power factor; sinusoidal currents which are exact replicas of the primary
ANSI C57.13-1968 gives adjustment factors to correct to currents they are intended to measure. In actual practice
other load power factors. distorted waveforms and erroneous current magnitudes often
Used together, Tables I and II can be used to describe the must be accepted. Protective relays receive this less-than-ideal
accuracy of any metering CT. For example an extremely good input and make a decision whether or not to operate. Based on
CT might be rated 0.3 B-0.1, 0.3 B-0.2, 0.3 B-0.5, 0.3 B-1, 0.3 erroneous inputs then, relays may fail to see an actual fault, or
B-2, 0.3 B-4, 0.3 B-8 meaning that it meets accuracy class 0.3 they may not respond as quickly as intended. In some instances
for all standard burdens. At the other end of the scale, a poor relays may operate on fault conditions for which they are not
CT might be rated 1.3 B-0.1. That the accuracy is specified intended to trip. Any of these misoperations can have serious
only at the 0.1 burden means that the error on all other bur- consequences, and it is important that the role of saturation be
dens exceeds the minimum classification (1.2) and no other understood and anticipated at the time the relays are applied
rating can be given. and set so that problems due to CT performance can be
Because accuracy in relaying CT's is of importance at higher minimized.
magnitudes of current, the American National Standards In- The most commonly used relay in industry is the induction
stitute has devised a method of specifying accuracy where it disk time overcurrent relay. This application is subject to mis-
counts most. For relaying CT's, accuracy is described by operation of two types. First, as saturation occurs the output
specifying the secondary voltage the CT can develop at twenty of the CT is reduced relative to the input. Thus, a relay set by
times rated secondary current without exceeding 10 percent test to pick up at 5A could be connected to a 100/5 CT
ratio error. No restriction is placed on phase angle error. On a (apparent sensitivity of 100/5 X 5 = 1 OOA of primary current)
typical CT made in the United States this amounts to specifying and, because of saturation, the combination might require
that the accuracy is 10 percent or better at 100A of secondary 120A of fault current to operate. In this example, the relay
current, and the standard further requires that the error must may fail to see a low level fault just above its pickup because
not exceed 10 percent at secondary current levels below IOOA of errors in the CT.
as well. An electromechanical induction disk relay has the following
For 5A CT's the ANSI standard describes seven secondary torque equation
voltages which can therefore be related to ohmic burdens, as
listed in Table III. Tn = K1 (I2)-k2 (1)
POWELL, JR.: CURRENT TRANSFORMER BURDEN AND SATURATION 299

TABLE IV
FOURIER ANALYSIS OF CURRENT WAVEFORM OF FIG. 4
Harmonic Crest Magnitude Phase Angle
1 7.329 43.420
3 3.537 -12.45° 10
5 2.163 -46.34°
7 1.506 -76. 70
9 1.069 -104.33° 0
11 0.746 -135.790 z
13 0.499 -165.920 0
15 0.306 159.980
17 0.157 112.990
19 0.104 19.56°
1.0

where
I- TEST DATA
Tn net torque on the disk,
k1 design constant, "PILISHED CURVE
k2 restraining torque due to the control spring, -0.1i
I rms operating current.
It can be shown (by test) that the pickup sensitivity of such a
relay decreases drastically at frequencies greater than the third
harmonic. Since operating torque and pickup sensitivity are
,.I. __
I ........_
related, it may be postulated that the net operating torque 5 50 500
AMPERES
with a distorted operating current is the sum of the operating
Fig. S. Comparison of published relay time-current characteristics
torques due distorted current wave, i.e., with actual operating time on 50/5 CT (number 2 time dial setting).

Tn- (kIn2) -k2. (2) practical CT's. The higher the available fault current, the
greater the secondary voltage the CT will have to produce for
Note that the constant k in this equation is not the same as k1 a given burden. Since the performance of window type CT's
in (1) above since the design constant includes the effects of typically improves as the nominal ratio increases, this suggests
the frequency-sensitive impedances of the relay core, coil, and one approach to reducing the impact of saturation on the
disk circuits. application of relays. Because small feeders on stiff systems are
A Fourier analysis of the badly distorted current wave of much more subject to these phenomena than large feeders,
Fig. 4 yields components listed in Table IV. For these first ten using feeders as large as possible (with appropriate CT ratios)
harmonics, the rms magnitude of current is 8.75A. Note that will minimize the concern for relay error due to saturation.
the odd harmonics predominate as should be expected since Actually, this philosophy is true for economic reasons as well.
the waveform is periodic. The phase angle is with reference to Small feeders generally require less stiffness than larger feeders,
the distorted current wave. and it makes little economic sense to provide system capacity
Based on this example it should be apparent that a relay and switchgear to handle greater stiffness than is actually
connected to a badly saturated CT will produce less operating needed.
torque than a relay connected to a CT which saturates little if Of course there will always be situations in which a IOOA
at all, with the same apparent rms multiple of relay pickup in a feeder is needed on a 1000 MVA bus. In such cases careful
case. Furthermore, it might be speculated that if the analytical consideration should be given to the selection of CT's and to
procedure described early in this paper were used to approxi- the third method of controlling saturation error reduction of
mate the rms output of the saturating CT, the result would burden. Judicious choice of relays- can often reduce the total
still produce an erroneously high estimate of operating torque burden on CT's significantly. For example, there is little need
because the waveform distortion and its effect on torque could for a 0.5A relay setting on 10015 CT's if the bus stiffness is
not be correctly accounted for. That these speculations are 1000 MVA at 13.8 kV. The 5A setting is one tenth as sensitive
true can be shown by comparing the time-current characteristic in absolute terms yet still provides ample sensitivity for the
of an induction disk relay under ideal test conditions with that available fault level. Furthermore, the burden of a 5A relay is
obtained when it is connected to a CT known to saturate badly 0.01 times the burden of a 0.5A relay!
(Fig. 5). DC saturation, either due to offset current or remanence in
The point, therefore, is that the torque produced by an the CT, is not especially a problem with time overcurrent relays
electromechanical time overcurrent relay is not sufficient to because the dc effect decays very rapidly and is usually over
give the operating time predicted by the published relay curves when time delay relays are expected to operate. However,
for the prevailing conditions of CT ratio, fault current magni- instantaneous relays may respond incorrectly to this effect.
tude, and relay setting. The actual operating time will be slower. Consider the circuit of Fig. 6. Suppose a three-phase fault
There are three principle factors contributing to the under- occurs such that phase A CT is severely saturated. The resulting
reaching and timing error problems of overcurrent relays on phase A CT output will be less than the outputs of the B and
300 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. IA-15, NO. 3, MAY/JUNE 1979

A aB DC

1 Residual

Fig. 6. Typical feeder relay connections. Fig. 7. Typical bus differential connection.

C-phase transformers even though the actual fault current has


the same magnitude in each of the three phases, and an instan- 1 2 3t t 4 1. Oi
taneous relay on phase A may not see enough current initially 50a. 50a 50a 2600n | ia
to operate. At the same time the unequal secondary currents
will sum to form a "fake" residual current which the residual (a)
relay may incorrectly see as a ground fault, especially if the
residual relay is instantaneously operating. For this reason the
use of an instantaneous residual relay should be carefully
considered. 1 2 3 4{ {
A problem frequently encountered by relay engineers is 50a.
"O |Ei 50a. j5Oa. iSa
| 32600n
obtaining sufficient residual time overcurrent relay sensitivity
for the level of grounding employed in the system. When CT (b)
ratios are selected on the basis of circuit capacity the need for Fig. 8. Equivalent circuits for bus differential connection shown in
ratios of 2000/5 or even greater often arises. Using 0.5A relays Fig. 7. (a) Fault on circuit 4, CT4 fully saturated, (b) Fault on bus,
no CT saturation.
the minimum apparent ground fault sensitivity is 200A which,
in the case of 400A grounding, is generally not sufficient. One
common solution is to use 0.1A relays to achieve an apparent incorrect relay operation can be prevented by including through
sensitivity of 40A. However, the burden of a 0.1A relay is 25 restraint in the design of the differential relays, and secure
times as great as that of a 0.5A relay. With type IAC53 relays instantaneous operation can be obtained. There are, however,
the voltage at pickup across the 0.1 A relay coil is approximately two instances requiring further consideration.
five times the voltage at pickup across the coil of a 0.5A relay. One of the most interesting and secure bus differential
Thus, depending on the CT ratio and saturation characteristic, relaying schemes does not include any provision for through
it may be that using a 0.1 A relay will not give greater sensitivity current restraint, and in fact the basic principle behind its
than using a 0.5A relay. The best solution in this case, of operation is that current transformers saturate. Suppose each
course, is to use a 5015 donut CT to measure ground fault circuit of Fig. 7 is capable of supplying 10 000A of current to
current directly. a bus fault (total of 40 COOA). Then, a fault on circuit 4 just
Incidentally, the calculations required to determine the outside the relay zone causes 30 000A to flow in breaker 4.
approximate sensitivity of a residual relay are particularly Assume that the CT on breaker 4 saturates so badly that it can
complicated. It is necessary to account not only for saturation produce no output at all (it is "soaked"), but each of the
in the current transformer in the faulted phase but also for the remaining CT's can correctly duplicate 10 000A of primary
exciting current required by the CT's on the two healthy current. The equivalent circuit of Fig. 8(a) shows that if CT-4
phases. produces no output all (30 000/200 = 150A of secondary
It might be noted that the familiar "ground sensor" consis- current flows as exciting current) the CT is effectively short
ting of a 50/5 CT and an instantaneous overcurrent relay is an circuited. However, this means that the CT magnetizing branch
example of how saturation can occur by design. In this case, also short circuits the relay coil, and the only impedance pre-
the secondary excitation characteristic is such that, no matter sent is the parallel combination of relay coil resistance and CT
how much primary current is available, the secondary current winding resistance. If the relay coil resistance is large (say
and hence secondary voltage are limited to fairly low values 2600 Q2) it will be swamped by the very small CT winding
which will not exceed the modest short time voltage and resistance-on the order of 1 Q. Thus, assuming that no other
current capabilities of the relay. At the same time, the relay CT saturation takes place, the voltage across the relay is
sees enough current to operate reliably. 3(50) X 1 = 150 V.
A related problem is the effect of unequal saturation of However, if the fault is on the bus, each CT sees a primary
CT's connected to a differential relay. If there is an error due current of 10 OOOA, and by the assumptions above, none satu-
to saturation in any CT in the differential circuit of Fig. 7 rate. Hence, each produces an output of 50A which sum and
POWELL, JR.: CURRENT TRANSFORMER BURDEN AND SATURATION 301

flow through the 2600 Q relay coil causing a drop of 520 000 V Positive Torque
(Contact Closing)
(Fig. 8(b). There is obviously a great distinction between an
internal fault (520 kV on the relay) and an external fault (150
V). Of course these assumptions are clearly extreme and both Negative Torque
voltages would actually be lower than indicated here. Further-
more, a nonlinear resistance is wired across the relay coil so
that the actual voltage under bus fault conditions will be
limited to a reasonable value. The basic tendency of this circuit
to produce extremely high voltages for internal faults, however,
has lead to its usefulness as the familiar high impedance voltage
differential relay for bus protection.
Often a differential circuit will be built up using a low
impedance time overcurrent relay with whatever CT's are Fig. 9. Torque angle characteristic of directional overcurrent relay.
available. Examples of this are retrofit bus differential connec-
tions using extremely inverse relays and ground differential written in terms of the polarizing voltage instead of the current
connections which require the use of an auxiliary CT to match in the polarizing coil
the ratio of a CT on the neutral of a generator or transformer.
In such cases the desire for sensitivity competes directly with Tn = klIoEp cos ( -t)-k2 (4)
the need to desensitize the relay to avoid incorrect operation
on the error current due to unequal saturation of the CT's. where 0 is now the angle by which Io lags Ep. Fig. 9 is the
Generally such schemes can be made relatively secure by using familiar torque angle characteristic which shows the region in
enough time delay to ride through dc saturation and by careful which this directional relay can develop torque in the contact
application calculations. Note that even though a feeder does closing direction.
not supply fault current its CT may require exciting current. In theory, fault current lags system voltage by the imped-
For this reason, the calculations associated with nonstandard ance angle of the system components. The directional relay is
differential schemes can become extremely messy. It is also designed with a maximum torque angle r such that for normal
necessary that particular attention be given to the burden fault currents the quantity (0 - r) is small, thereby maximizing
amplification effect of auxiliary CT's in these crude differential
contact closing torque klIoEp cos (0 - r). The type JBCG
schemes; in the case of the ground differential scheme the directional ground overcurrent relay has a torque angle of 600,
auxiliary CT ratio required to correctly balance the differential
for example.
sometimes becomes so large that the burden reflected into the It can be shown that an induction cup produces torque only
residual of the phase CT is intolerable. In such cases other when Io and Ep are of the same frequency [5].
means of protection, such as directional overcurrent relays, Consider now a resistance grounded system which is peculiar
must be employed. in the sense that because ground fault current magnitude is
Another relay often found in industrial installations is the
determined primarily by resistance, the fault current is very
directional overcurrent relay. In its modern form this relay close to being in phase with system voltage. For example, on a
consists of an induction disk time overcurrent relay torque 500 MVA, 13.8 kV system with 1200A resistance grounding
controlled via a wound shading coil by an instantaneous induc- the fault current would lag by about two degrees. A directional
tion cup directional unit. This latter device is similar to an relay in such a case would probably be polarized from a set of
induction disk relay except for lower inertia (for faster opera-
broken delta potential transformers which would undergo
tion) and the fact that it is equipped with two coils. One coillittle or no saturation and thus would produce an Ep relatively
is wired to receive the current the relay is intended to measure
free of harmonics.
and the other is connected to a reference current or a reference Suppose now that in an attempt to get sensitivity, the
potential. The relay torque equation is operating coil of the directional overcurrent relay was con-
nected to a 50/5 donut CT so that the operating current has a
T, = klIoIp cos (6 - t) -k2 (3) waveshape similar to that shown in Fig. 4. Only the funda-
where mental frequency component of this current could interact
Tn net torque on the cup, with Ep to produce torque in the directional unit. Referring to
k1 design constant, the table, the first harmonic has an angle relative to the dis-
torted wave of 43.420, and since the distorted current wave
Io operating (measured) current, by 20, the fundamental frequency position of
Ip polarizing current-either from a reference CT or driven operatingvoltage
lags the
current leads Ep by 43.42-2 = 41.420. Therefore, if
by a reference potential transformer (PT),
0 angle between Io and Ip, the relay is a JBCG with a torque angle of 600, the quantity
(0 - r) is (-41.42-60) or -101.42 . These angular relationships
T design constant called the "angle of maximum torque,"
are illustrated in Fig. 10. Note that the magnitudes are not to
k2 design constant of the restraining spring.
scale in this illustration. Since the cosine of a negative angle is
In most cases directional overcurrent relays are potential negative, the operating torque klIoEp cos (0 - r) is also nega-
polarized. For these instances the torque equation can be re- tive, or in the contact opening direction. Thus, because of CT
302 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. IA-IS, NO, 3, MAY/JUNE 1979

A 3) Wherever possible use standard relay schemes, employing


relays specifically designed for that application, instead of
"cobbled" arrangements which may contain hidden problems.
4) On differential schemes, carefully consider the location
of the relay with respect to the CT's. Sometimes locating the
relay at one end of the protected zone will minimize the effect
of lead burden in causing saturation.
5) Obtain secondary excitation curves from the CT manu-
Fig. 10. Phase angle relationships for ground directional relay example. facturer and, keeping in mind the unavoidable limitations of
Curve A -system voltage. Curve B-Fault current. Curve C-relay the procedure, calculate the performance of the relay and CT
current. Curve D-60 Hz component of relay current. (Magnitudes
not to scale.) combination to see if the performance is acceptable even
though severe saturation does occur. Manufacturers will often
assist in performing these calculations.
saturation, the directional element will not operate even though 6) If it is apparent that special CT's are unavoidable, specify
the fault current is in the correct tripping direction. It can also them using the procedure outlined in the ANSI standards.
be shown that if the fault current is in the reverse (nontripping) Generally this amounts to asking for CT's with a larger iron
direction, the directional element will be fooled into developing core cross section. Sometimes gapped CT's may be specified to
contact closing torque and incorrect tripping can result. moderate a problem with dc saturation.
Normally, the degree of saturation encountered in practical 7) Use two or more CT's in series to reduce the secondary
CT's will not be so severe as to cause the kind of incorrect voltage each has to develop [6].
operation illustrated here unless to begin with the relay appli-
cation is marginal. To be sure, this example illustrates that SUMMARY
directional ground overcurrent relays should never by operated Saturation in CT's is an area where qualitative understanding
from donut CT's which are intentionally designed to saturate abounds [4], but quantitative methods are limited in accuracy.
heavily. The point is, however, that in general directional relays The performance of specific protective relays on saturating
can suffer error in directionality because of saturation. CT's is fairly well understood and is a subject of considerable
It might be noted that if the polarizing source used with concern on the part of engineers in industry. This paper has
directional relays also saturates and produces a distorted wave, identified some of the major ways in which saturation can
the net torque will be equal to the sum of the torques produced affect the performance of relays and has suggested several steps
by like frequency polarizing and operating quantities. Because that can reduce the significance of CT saturation in planning
some frequencies may produce operating torques at the same an industrial power system.
time that other frequencies produce restraining torques, an
analysis of the relay performance is complicated indeed. In REFERENCES
practical terms this is only a consideration if the relay is [1] "Requirements for Instrument Transformers," American National
current polarized since potential sources saturate very little. Standards Institute, ANSI Standard C57.13-1968.
The foregoing discussion illustrates that protective relays [2] H. C. Barnes and A. J. McConnell, "Some utility ground relay
problems," AIEE Trans. vol. 74, p. 417, 1955.
can misoperate in a number of ways if severe CT saturation [3] R. G. Bruce and A. Wright "Remanent flux in current transformer
occurs. The important question, then, is how to avoid satura- cores," Proc. IEE, vol. 113, no. 5, p. 915, May 1966.
tion so severe that incorrect operation occurs? Several methods [4] IEEE Protective Relay Committee, "Transient response of current
transformers,"IEEEPubl. 76CH11304 PWR.
are readily apparent and have been suggested already, while [5] C. R. Mason Art and Science of Protective Relaying. New York:
other methods are not difficult to imagine. Wiley, 1956.
1) Use larger CT ratios. In a well designed system where [6] D. J. West, "Current transformer application guidelines," 1977
I&CPS Conf: Record.
feeder capacities are reasonably selected considering the avail- (71 A. Wright, Current Transformers. London: Chapman and Hall,
able system stiffness and where CT ratios are correctly chosen Ltd., 1968.
on the basis of circuit ratings the CT's will generally have
ratios which will minimize saturation problems.
2) Keep secondary CT burden low. Within the limits of
reason, use low-burden relays and instruments. Avoid extremely
sensitive overcurrent relays when alternate connections will Louie J. Powell, Jr. (S'67-M'69-SM'78) gradu-
provide equal or better sensitivity without the attendent ated from the Georgia Institute of Technology
saturation problems. Do not place all single phase metering with a Bachelor of Electrical Engineering degree,
and relaying on the same phase-distribute the burden among cooperative plan. He has done graduate work in
the Power Systems Engineering Course spon-
all three phases. Use shorter CT leads, or if this is not feasible, sored jointly by General Electric Company and
larger wire sizes. When step-up auxiliary CT's must be used Union College.
While in college, he was employed by Rey-
carefully, consider the burden amplification effect. If several nolds, Smith, and Hills, Architects and Engi-
CT's are available on a breaker, consider relocating some neers where he worked on electric utility
devices to another CT. generating plant and transmission substation
IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. IA-15, NO. 3, MAY/JUNE 1979 303

designs. After graduation he joined the General Electric Company and dustrial power system relay protection at seminars throughout the
was initially assigned to the Chicago District Office where his principal United States, in Europe, Africa and Central America. In addition to
work was on electrical systems for steelmaking facilities. He then his work at General Electric he holds an appointment as Adjunct As-
transferred to the Industrial Power Systems Engineering Operation in sociate Professor at the Center for Electric Power Engineering, Renes-
Schenectady as an Application Engineer, working primarily in the selaer Polytechnic Institute, Troy, NY.
field of power systems analysis. He has made numerous electrical Mr. Powell is a senior member of the Industry Applications and
power systems studies for various industries, and his experience in- Power Engineering Societies of the IEEE and is Vice-Chairman of the
cludes participation in the development of an extensive power system Schenectady IAS Chapter. He is a Registered Professional Engineer in
serving an isolated portion of Indonesia. He has given talks on in- the State of New York.

Motor Protection Characteristics of Ambient Insensitive


Overload Devices
ARTHUR F. KOLB AND HOWARD W. THOM, MEMBER, IEEE

Abstract-The performance of thermal overload devices with respect overload relays. It is also desirable that the relays be single
to the effect of ambient sensitivity factors on motor winding insulation heater based such that the performance of one heater is inde-
temperatures is investigated. It is shown that ambient sensitivity charac- pendent of the others.
teristics must be defined in order to optimize output horsepower while
maintaining insulation temperatures within design parameters. Overload relays should also be capable of being coordinated
with other circuit components such as a motor circuit protector
INTRODUCTION (MCP) or fuse. This will assure that at high level faults, in
regions beyond the self protecting point, damage will not
T HERMAL overload relays are required to perform a occur, or if it does that it be restricted to specific levels. Fig. 1
multiplicity of electrical and mechanical functions to displays a heater coil in a size "0" combination starter prop-
well defined specifications. With an appropriately selected erly coordinated for either a one time service (OTS) fuse or
heater coil, they must function at a value of current I, yet an MCP. Although these characteristics may not be a require-
hold indefinitely at a second value of current (see Nomencla- ment of all applicable standards, they are required by most
ture) and implied by others [1].

Ih >KiIr (1) RELAY TYPES


Thermal overload relays may be classified by type as either
where K1 is less than 1, and trip within specified times at stored energy or self-energizing. Typical of the stored energy
2I4 and 6Ir. It is desirable that the relay be self protecting at system is the eutectic alloy thermal overload relay, while the
current values between 10Im and 13Im and have one second self-energizing system is considered to be represented by the
thermal capabilities at 1 51m where bimetallic overload relay. There are some significant differ-
ences between the two systems, and it is because of these
Im= K2Ir (2) differences that a comparison is appropriate.
The relays should be trip free, with memory capability, and Self-Energizing
may also have: a function indication, isolated normally open
(NO) or alarm contacts, and automatic reset and adjustability. The self-energizing or bimetallic relay is available in several
versions (see Fig. 2). Heaters may be integral, requiring re-
Some relays have a feature commonly referred to as ambient
compensation, more accurately defined as ambient sensitivity,
placement of the overload to meet a specific motor current, or
which is a characteristic implicitly common to all thermal they may be replaceable. The contact structures are usually
single pole double throw and have provisions for automatic
reset. They display some shock and vibration sensitivity at or
Paper IUSD 7841, approved by the Industrial Control Committee
of the IEEE Industry Applications Society for presentation at 1978 near the tripping current.
Industry Applications Society Annual Meeting, Toronto, ON, Canada, Generally, in a device of this type wherein the energized
October 1-5. Manuscript released for publication December 11, 1978. thermal elements cooperatively engage a single lever or trip
A. F. Kolb is with Standard Control Development, Cutler-Hammer bar to operate a common set of contacts, the heaters are not
Inc., Bowling Green, KY.
H. W. Thom is with Standard Power Control Products, Cutler- single rating based. Therefore, under single phase conditions,
Hammer Inc., Milwaukee, WI, 53216. they require a 22 percent current increase in the remaining
0093-9994/79/0500-0303$00.75 © 1979 IEEE

You might also like