Você está na página 1de 10

rI based on system readings or previous

quiva ent o.
E .| .
wircuits t
£or
n
Power-Flow studies, is reproduced in the stub loads
and generators at the interconnection
Studies
tuies terminals. In system planning studies

involving proposed future transmission


development or location of new generating
J. B. WARD stations, the system changes are fre-
ASSOCIATE AIEE quently so radical that it is not possible to
predict, with sufficient accuracy, the load-
ings on the stub loads and generators used
PRESENT-DAY extensive intercon- commonly-used methods of network sim- to represent multiple ties to an adjoining
nection between power systems com- plification, and presents additional system.
plicates the analysis of steady-state net- methods for devising simple equivalent
-,, M~~~~Aethod B. The system to be simplified
. >. , ., , .,
work performance in the study of plan- - circuits which reproduce with reasonable
ning and operating problems. This com-
. .
accuracy the terminal performance of the l~~~~~~~~~~~s
replaced by an equivalent circuit in
which the identity of the tie-line ter-
plication of analysis is increasing con- actual system under conditions prevailing minals and generator busses is retained,
tinually with system expansion and addi- in the usual tvpe of power-flow study. Bv I
and the line network iS represented by
tional interconnections. Network ana- these new methods an extensive power transfer impedances between these van-
lvzer studies of power-flow conditions on network,' containing numerous loads and ous busses. Equivalent load impedances
- . . , . . .
large systems frequentlv require pre- generators, is reduced to an equivalent appear as transfer impedances from each
liminary simplification and reduction of circuit, as viewed from several selected retained bus to neutral. The process of
the network, because of limitations im- terminals, requiring few network analyzer evolving this type of equivalent circuit
posed by the number of analvzer circuits circuits and sources for its representation. from a network analyzer set-up of the
and generator units available. Changes in generation or loads within
In studies for determining the distri- the reduced system can be reproduced syse ispovtine h for d the p
bution of power flow over the lines of a readilv in the equivalent circuit. Al- pnew methods
o and of explaining certain
system under various emergenev operat- though the methods are described, for steps in obtaining the new type of equiva-
ing conditions, or with proposed future the most part, in terms of the necessary lent circuit.
developments, interconnections mav have tests and measurements on an a-c network A 1-line diagram of the system to be
an important effect, particularly if ,two analyzer set-up
. .
of the system to be re-
^ . . . . ~~~~~~~~~replacedbyin an equivalent c
circuit iSi rep-
or more ties exist forming parallel paths duced, derivation of the equivalent cir resented Figure 1. This diagram
through the adjoining system. The total cuit analytically or by use of a d-c board shows the loads, generators, and inter-
interchange of power between separate is feasible. Based on constant-current connection points, but the internal line
companies usually is established bv in- representation of system loads and genera- connections are omitted. It is assumed
terchange contracts, but in order to de- tion, the new methods will accomplish in
termine the proper division of this power many instances a greater simplification
than previously-used equivalent circuits.
that be
must symp ined
te simplified in Figestudy1
before a detailed
flow over several tie lines under various of the adjoining network can be carried
operating conditions, it^ mav be necessary Limitations of this treatment of loads and oout. If I t
* - ' to load- tie-line tertminals, generator
to set up the adjoining svstem or an generation restrict
I
application ~~~~~~~~~~~~~~~busses,
and neutral bus are to be retained,
equivalent circuit representation of it. flow studies, but in studies of this type the branches of the resulting equivalent
The problem of network reduction in acetal acurc
acceptable accuracy ca athebanheafihnedutngeqla
can be attained, circuit, shown in Figure 2, can be ob-
power-flow studies is not related solely to tained as follows.
inter-company ties. Some individual Present Practice Disconnect the generators and repre-
svstems have grown to such an extent that sent loads as constant impedances cor-
the complete network of major lines, Present practice in reducing the size responding to the load real and reactive
stations, and substations exceeds the of a network for analyzer power-floxr power at normal bus voltage. Connect an
capacity of the largest existing network studies is confined largely to two methods. analyzer source from one of the terminals
analyzers. Even on smaller systems, Method A. The network to be studied
. .. .. . . ~~~Paper 49-96, recommended by the AIEE system
engineering committee and approved by the AWEE
simplification of the network by- means in detail is terminated at interconnecting
of equivalent circuits can be employed points, and the adjacent system is rep- pIEwitrogeram cmmittete for presentation at
tehnia
advantageously, to speed up the progress resented by aload or generator attached January 3i-February 4, i949. Manuscript snb-
of studies when interest is focused on a at each of these terminals. Connections mitedctober 4, i948; aeaalbefrpitn
certain section. through the reduced system are omitted J.BWADSwihtecolofEcrcaEn-
This paper outlines briefly the existing entirely, and an estimated tie line flow, neering at Purdue University, West Lafayette, Ind.

194I9, VOLUME 68 Word-Equivalent Circuits for Power-Flow Studies 373


a branches is equal to n(n - 1)/2, where n is 3. Angular displacements between bus
the total number of terminals retained. voltages in a system usually are not large.
b 2 TIES TO The procedure just described is equiva- Initial angles between various load and
NETWORK } ADJOINING generator busses can be taken into ac-
lent to applying star-mesh transforma-
SYSTEM
C
A 30J tions to remove successivelv all busses count accurately, and then the assump-
except generator busses, tie-line busses, tion can be made that effects of subse-
LOADS and neutral bus. The approximation of quent variations in these angles can be
loads as constant impedances is required. neglected.
NEUTRAL 4. Synchronous condensers, static ca-
Figure 1. Typical system to be replaced by New Equivalent-Circuit Representa- pacitors, and line charging capacitance
an equivalent circuit tion can be treated as separate sources of re-
active power or combined with nearby
to be retained, terminal I for example, to The basis for the new equivalent circuit loads and generation.
a new junction established at x, as shown representation presented herein is the 5. Although the derivation of the
in Figure 3. Connect zero-impedance treatment of generation and loads, in the equivalent circuit is based on the con-
metering circuits from junction x to all network being reduced, as constant-cur- stant-current approach, it is assumed that
other terminals which are retained, in- rent quantities. By means of this ap-
cluding the neutral bus. Adjust tEe proach, the internal generator busses as _,
a r-
A
source at terminal 1 to an arbitrary value well as load busses in the reduced system I
of voltage and read the current in each can be eliminated, and the identity of
metering circuit. The impedance of tle only the tie-line terminals retained. The b 2-
branch between terminals 1 and 2, in the effects of suppressed generation and load
equivalent circuit of Figure 2, is obtained are reproduced by an equivalent gen- c _ - 3
by dividing the source voltage by tlhe erator or load at each tie-line terminal, | - -' -. -J
current in the metering circuit coiinected and the actual network of line connections
to terminal 2. It should be emphasized is replaced by an equivalent mesh network
th all metering
a t
circuits are closed and interconnecting these terminals. The 75 ,
carrying current simultaneously during equivalent circuit of this type for the
this test. Similarly, the impedance of the system of Figure 1 is shown in Figure 4 Figthed 2B. Equtivalenstircuit obtfained
equivalent branch between terminal 1 In principle the methods are rigorous for a
and neutral is the voltage at terminal I hypothetical network in which all loads
divided by the current in the metering and sources are fixed impressed currents. the equivalent soturces and loads, once
circuit connected to neutral. The analysis of such a system is outlined determined, can be held at a fixed power
This process is repeated with the source in Appendix I, and, in the following schedule, for convenience in manipulation
connected successively to each retained paragraphs, the methods are adapted to during studies.
terminal. At each step, all other ter- the practical determination of simplified 6. Initially, transformer turn ratios
minals are connected to x through meter- equivalent circuits for sections of power- are assumed to be the same as ratios of
ing circuits. Thus, two measurements system networks. Some of the important the nominal voltages of the sections con-
are obtained for each transfer impedance. assumptions and approximations involved nected to the respective windings.
The branches from the various generator in this adaptation are listed briefly: Methods of handling transformer tap
and interconnection busses to neutral can 1. It is assumed that system loads settings which result in other than nomi-
be considered as equivalent loads. How- can be considered as constant-current nal ratios are discussed in a later section.
ever, in the equivalent circuit derived in quantities. Usually loads are treated as 7. It is assumed that, when a section
this way, the distinction between actual constant impedances in analytical solu- of a power system is to be replaced by an
loads and line losses is lost. tions of power-flow problems, and as con- equivalent circuit, detailed information
The equivalent circuit, Figure 2, is stant-power loads in network analyzer on the performance of this section, with
connected at the proper points to the de- studies. Differences in the three view- various line outages, for example, is not
tailed set-up of the adjacent network. points become of major importance onlv of immediate interest. Rather, the effect
The desired internal generating schedule under extreme voltage variations, as may of this section during detailed study of
in the reduced system is established by occur during short circuits, for example. the adjoining network is desired.
adjusting sources connected to the gen- 2. Generators are usually held to a
erator busses of the equivalent circulit. If fixed power and voltage schedule, except Determination of the Equivalent
a different condition of loads is required for certain regulating machines. Reac- Line Network
in the reduced network, it is necessary to tive power generation may vary appre-
measure a new set of transfer impedances ciably, but, as a rule, large blocks of After the section of the system to be
from the various terminals to the neutral reactive power cannot be transmitted reduced and the interconnecting ter-
bus. The transfer impedances, repre- economically over long distances. Fur- minals are decided upon, this section is
senting I;ne connections between gen- thermore, a large proportion of system set up in detail on a network analyzer,
erator and interconnection busses, are studies are made at peak-load conditions, with all loads, generators, and capacitors
modified only slightly by the presence of with generators operating at or near their open. The transfer impedances between
load impedances. maximum excitation. Hence, limited the terminals to be retained are measured
Certain branches of the resulting variation of power, reactive power, and in the manner described under method B
equivalent circuit are sometimes of such voltage corresponds essentially to con- in "Present Practice," with the exception
high impedance that they can be omitted stant-current loading, in magnitude at that, in the present method, only the
entirely, but in general the number of least, on generators. points of interconnection are retained.

374 Ward-Equivalent Circu4its for Power-Flow Studies AIEE TRANSACTIONS


Equivalent Generators and Loads: and with the source current, if the ratio tion of these equivalent source loadings is
Method I r/x of the major lines is reasonably uni- indicated in equations 1
forn. If appreciable angular displace- I1 =Dall+Db1I+Dc1Ic+
Regardless of the number of internal ment appears, these currents or distribu- 12=Da2ia+Db2Ib+DC2I+ +(1)
busses suppressed in the reduced network,
an equivalent generator and load can be
tion factors can be measured as complex
quantities, with the source current as ref- and so on.
a +D2sD
on
.

The equivalent per unit


(1)

established at each terminal of the equiva- erence, by means of the analyzer vector- source currents can be converted back
lent circuit, corresponding to a given metering equipment. into power and reactive loadings, using
schedule of generation and loads at the 3. In order to establish a convenient known or estimated tie-line terminal
actual stations. This can be accom- notation for the distribution factors, let voltages.
plished through the medium of a set of the internal busses be indicated by letters, 5. The system generation and load
distribution factors, determined as de- and let the points of interconnection can be reflected separately to the tie-line
scribed in the following paragraphs. which are retained in measuring the terminals in this manner, to appear as an
This description is given in terms of equivalent line network be indicated by equivalent generator and an equivalent
generation. The loads can be considered numbers. Use the first letters of the load at each terminal, or the two can be
as negative generation and can be treated alphabet to indentify generator busses, combinedinto.anethloadornet generation
in exactly the same way. and assign the remaining letters to load at each terminal. If the currents
Iae ub,
1. In a detailed set-up of the network
being reduced, with load and generator
busses. If a load or generator appears at
tion. on the right hand side of equa-
one of the retained busses, assign both a tions 1 include all busses, with the proper
circuits disconnected, connect each in- suitable letter and a number to this bus algebraic signs to distinguish between
terconnection bus to the analyzer neutral When station a is energized by impressing generation and load, these equations give
bus through a metering circuit. In the unit current, denote the currents in the equivalent net current sources. The
system of Figure 1, for example, connect metering circuits connected to terminals distribution factors, once determined, per-
points 1, 2, and 3 to the neutral bus. 1, 2, 3, . ., by the symbols Dai, Da2,Da", mit representation of any desired schedule
2. Connect an analyzer source be- . , respectively. Similarly, when unit of generation and load.
tween one generator bus, such as ter- current is impressed at station b, label 6. In reducing a large system contain-
minal a in Figure 1, and the analyzer the corresponding distribution factors ing many loads, this method may become
neutral bus. Adjust this source so that as: Dbl, Db2, Dbs, . . .
quite lengthy and cumbersome. An
its output current is equal to one ampere 4. The distribution factors are used approximate apportionment of the total
or unit current. Measure the currents in to obtain an equivalent generation at system load at the terminals of the
the several metering circuits connecting each terminal of the equivalent line net- equivalent circuit can be carried out more
the terminals to neutral. work, from known or specified power and rapidly as follows,
These metering-circuit current readings reactive generation and bus voltages at In the detailed set-up of the network
are the distribution factors associated generators in the reduced system. Ac- being reduced, disconnect all generators,
cording to the usual practice in analyzer but represent all loads in the usual fashion
ANALYZER studies, voltages are expressed in per cent as constant impedances to neutral, ad-
LSOURCE
SOUE
orsperunit of a nominal base voltage, and justed to consume the correct scheduled
power
___0 quantities are
powr quntiies recorded either
re rcored etlie inin real and reactive power at normnal bus
b 22. terms of actual svstem value or in per voltage. Connect all the terminals re-
*NETWORK unit. tained in the equivalent line network
c 3 .
Express the powcr and reactive loading
. through metering circuits to one terminal
. a single analyzer source, which
of each station as a complex number in of in turn
_~~ ~ ~~~~e per uni of th
unit of chsnpwr1ae
tTe base.d
NEUTRAL power s-
s S connected
c

minal. Adjust
to neutral
this
at the other ter-
source to deliver unit
' |
| ~~~~suming that normal angular displace-
suming tha noZ,/,z,,,,z,rmal,,,7,/,,/ o
angular current to the system. Read the currents
.-,,METERING CIRCUITS--_ ments between bus voltages are known, in the metering circuits and label these
either from previous studies or from a
i _JUNCTION (X) J_
__
- special analyzer test made for this pur- circuits to terminals 1, 2,
express each bus voltage in per unit tivelv. These are treated as total-load
, respec-

Figure 3. Network analyzer circuit connec- pose,


as a vector quantity, with one of the re- ttion t
tions for measuring transfer impedances taed terminals as reference. For each factors.
bus, the per unit current loading, corre- The equivalent load at terinal 1 is
with the bus which is energized. If one sponding to the scheduled power and
of the interconnection terminals happens reactive loading, is obtained by dividing } IL

to be also one of the generator or load the per unit vector power by the conjugate EQOUvACLENT
busses, it is not necessary to carry out the of the corresponding per unit bus voltage. OR LOADS
test for this bus. Unit-current input to If the bus angles are not known, the use, { ~T2
this bus all would return to neutral either of estimated angles or of only\ :
through the metering circuit connected to voltage magnitude may give sufficiently\\
the same bus. In this case, the one dis- accurate results. I3
tribution factor is unity; the others are Represent the resulting per unit current
zero. Writh this exception, distribution loadings of stations a, b, c,...,by
factors are measured with unit current Ia Ib, h, ,respectively. Denote by
impressed successively at each bus. Il, '2, Ih,...,the equivalent generator
Usually, the metering circuit currents per-unit-current loadings at terminals 1, Figure 4. New simplified equivalent circuit
will be nearly in phase with each other 2, 3,...,respectively. The determina- for the system of Figure 1

1949, VOLUME 68 Ward-Equi7valent Circuits for Power-Flow Studies 375


taken as the product of di and the total Equivalent Generators and Loads: X
system load current. The equivalent Method II C
load at terminal 2 is the product of d2 and
the total system load. The equivalent The method outlined here resolves
load at each of the other tenninals is directly the equivalent net generation or z3\
calculated in a similar fashion. net load, as the case may be, at each ter-
In the event that large loads, that is, minal, and can be employed if the system
low-impedance load circuits, appear at to be reduced is set up in detail, together I3
some of the terminals being retained, the with the adjoining network with which it
accuracy of the equivalent loads deter- is interconnected. It may not be neces- 3
mined in this way can be improved bv sary to represent the entire adjoining Figure 6. Typical equivalent circuit which is
modifying the procedure slightly. Dis- network if a portion adjacent to the to be reduced further to one requiring a single
connect large loads at any of the terminals interconnecting ties is sufficient to es- source
before measuring the distribution factors. tablish normal power flow and voltage at
Apply the resulting factors to the total the points of interconnection. Some
load, reduced by the total amount of load particular power-flow condition, here sources in Figure 5(C. In realitv, these
disconnected in the tests. Then, at each termed normal flow, serves as the basis
of the terminals where load was removed, for determining the equivalent net loads work, but it is not necessary to determine
increase the resulting partial equivalent and generators. their distribution in the various branches.
load by the amount which was discon- 1. First, establish the normal-flow 4. When these tests are made, it is
nected from that terminal during the condition on the analyzer set-up and desirable to establish two or three addi-
measurements. measure carefully the voltage and angle tional conditions in the detailed portion of
7. Another method for reducing all at each interconnection bus. Measure the set-up which represent extreme de-
loads simultaneously to equivalent loads also the normal power and reactive flow partures from normal flow, involving line
at the terminals suggests itself if the over the interconnecting ties. outages or other emergency operating
system loads are viewed as constant- 2. Set up the equivalent line network, situations. The equivalent loads and
impedance quantities. with only tie-line terminals retained, and generators, derived in the same way from
Set up in detail the network to he re- connect an analyzer source to each ter- these extreme conditions, will serve as an
duced, leaving open all generator circuits, minal. Adjust these sources to reproduce indication of the range of applicability of
but representing loads by the proper im- the normal-flow values of voltage and the normal-flow values, and also will
pedance in the usual manner. Measure angle at each terminal, and record the serve as a guide for adjusting the equiva-
the transfer impedance, between the magnitude and direction of the resulting lent reactive generation.
several retained interconnection points power and reactive loading of each source. a. If it is desired to separate the
and neutral bus, as described under Since loads are not represented in this equivalent net generation or net load at
method B, and use these as the equivalent test, the loadings of the sources consist of each terminal into the component parts
loads. If desired, the load real and reac- real and reactive power circulated be- contributed by the load and bv the gen-
tivre power can becomputed, using known tween sources by the normal-flow ter- eration in the reduced network, distribu-
or estimated terminal voltages, and then minal voltage and angle differences. tion factors can be measured for each
these loadings can be combined at each 3. The measured circulating flow at generator bus as described under method
terminal.
loadings.
with the equivalent generator
each terminalinterchange
normnal-flow
can be combined with the
at the same ter-
I. For the particular generating schedule
within the reduced network, which was
minal, to determine the equivalent net used in the normal-flow condition, the
load or generator which should be placed equivalent generation can be determined
50 VLBI, so , at each terminal of the equivalent circuit. at each terminal, and then the value of
OfW NRK (0) TO ETORK This process of combination is described each equivalent load can be separated
TEDUCE
REDUCEDD 2 .1.. BESET UP
IN DETAIL A 4
(5)
best by example.
Figure 5(A) shows the normal-flow
from the corresponding net generation
or net load.
v3Le3 0 \vY3Le3 5 interchange and terminal voltages in a
-(4) (9) system with three ties between the two Reducing the Number of Equivalent
A. NORMAL FLOW B. CIRCULATING FLOW sections. Figure 5(B) gives the flow Generators
circulated between the terminals by the
40 i4i'" 50 same normal-flow terminal voltage con- In cases where normal power inter-
((5) (10) ditions. Figure 5(C) illustrates how the change is from the reduced section to the
equivalent load or generator at each rest of the system, the equivalent circuit
o16
(110)(5)
20+ t
terminal deduced5(A)
given in isFigures the information
fromand 5(B). The probablyofwill
several the require a net Ifgeneration
terminals. at
a sufficient
5 p fo'9"1 equivalent net load or generator at a number of power source units is not avail-
t03) ()given terminal contains the real and able on the analyzer to set up the equiva-
AND SOURCESreactive power components required to
C. NETLOAOS lent circuit together with the adjacent
Figure 5. Numerical example illustrating the attain a balance with the normal-flow detailed system, the method given here
determination of net equivalent loads and interchange and with the normal cir- can be used to combine the generators
generators by' method 11 culating flow at that terminal. For con- appearing at several terminals of the
Direction and mdgnitude of real and reactive -venience in calculating the net load or equivalent circuit into a single source.
power flows are shown. Reactive power is generator at each terminal, the normal The reduction to a single source intro-
enclosed in parentheses circulating flowr is indicated by dotted duces additional line circuits in the final

376 Ward-Equivalent Circuits for Power-Flow Studies AJEE TRANSACTIONS


equivalent network, but the shortage of terminal current sources could be com- tion to supply losses can be estimated, in
analyzer sources often is more severe than bined at the star point without changing order to facilitate establishing the initial
that of line units. terminal performance, and the constant terminal conditions of the study. Then,
The procedure is described in terms of factor k/l does not disturb this condition. in the course of the study, the single
an example involving a 3-terminal equiva- The final equivalent circuit with a single source is adjusted to produce the same
lent circuit, shown in Figure 6, with net source is shown in Figure 9. total power and reactive input to the
generation appearing at each terminal. If the source currents, representing terminals as calculated for the original
It can be extended, by analogy, to any either equivalent generation or net gen- equivalent network from which the
number of terminals. In the following, all eration, have relatively large phase dis- single-source circuit was derived.
quantities are retained in literal form. A placements with respect to each other, the If a change of internal generating
numerical example is given in Appendix procedure just outlined may produce schedule in the reduced system is to be
II. negative resistance terms in the branches simulated, a new single-source network
1. Figure 6 represents the equivalent of the final equivalent circuit. It may be must be computed, unless it can be as-
circuit only, with connections to the ad- possible to avoid negative resistance by sumed that all generation which has been
joining detailed system not shown. The changing the angle 4. If this is done, one shifted to the star point changes loading
equivalent generator loadings in per unit additional branch Z12'will appear between in the same proportion.
current, complex or vector values, are terminals 1 and 2.
indicated as Ii, 12, I3at terminals 1, 2, and Another way to avoid negative re- Transformer Taps
3 respectively. Several intermediate sistance is to divide each original source In network analyzer studies, convenient
steps are necessary in deriving the single- current into two parts, one a quadrature base voltages are chosen for the various
source circuit. Consider a star network, } which can be represented by a
component ~~~~~~~~~~~transmission and subtransmission levels,
Figure 7(A), connecting three current capacitor at each terminal, and chosen and the analyzer voltage measurements
sources equal to those shown in Figure 6. so that the remaining parts of all source
Arbitrarily assign a value of (1.O+jO) to currents are in phase. This can be stated of these atkina
usually nominal values. ofperentage
termVery frequently,
the impedance of the branch connected to v of power quantities.
more clearlv in terms
~~~~~~~transformers represented in a study bave
terminal 1. If the other branches, Z2' and Assume that the terminal voltages in the actual turn ratios deviating from the
Z3 are chosen ' to satisfy the relation present example are nearlv in phase. r o t
1.-1.O=12 Z2'=13.Z3', see equation 4,
> ~
Then, if the three generator loadings are
~~~~~~~~~~~~~~~
ratios of the arbitrary nominal vralues
Appendix I, then the three source cur- at widely different power factors, negative gnect by the tagnsfor ers Althoug
rents can be added together and impressed resistance may appear. To avoid this, all nominal tae transformais douno
at the common junction of the star net- of the real power generation is transferred appear in the analyzer set-up, any such
work without disturbing the terminal con- from each terminal, at some convenient deatin fro nnal ratio are rere
ditions. uniform power factor for all sources in- sented at points of transformation by
2. Convert this star network to an volved, to the star point; the balance of adjustable autotransformers provided for
equivalent mesh network, equation 2 of the reactive power is generated locally at the purpose. The occumrrence
of off-
Appendix I, as indicated in Figure 7(B). each terminal by means of capacitors. If nominal turn ratios within a system
Comparing the impedance values of this an equivalent source consumes reactive should be considered in devising an
mesh network with those of Figure 6, power, a shunt reactor can be used at equlent
considere sysing
an
multiply the impedances of Figure 7(B) that terminal instead of a capacitor. these ratios may have an important
by a complex constant, /kO, which is This method of reducing the equivalent effect onntria
efc terminal voltages and the dis-
otgsadteds
chosen so that one branch of the resulting p e sources to a single source is based
power
asoureto tas sue isubased tribution of reactive power. Since the
network is equal to the corresponding on the assumption that. the equivalent possible situations in regard to off-
branch of Figure 6; at the same timecthe
other branches (those of Figure 7(B)
gend kown urrents. In terms of nominal transformer tap settings are so
numerous, a general treatment is not
multiplied by k/I) are greater in magni- power quantities the single source
power tarnt mustgnee tuelocated laddi- fered to assistSeveral
attempted. suggestions are of-
tude than the corresponding branches of
Figure 6. In this example, assume that
.tional I R andInt12X losses incurred in the. approximately, the effects of transformer
at sar must gnere th e in reproducing, at least
flow of current from the single source to
the relation Z12'-k/4=Z32 satisfies this the terminals of the equivalent circuit. turn ratios.
condition. This effec
This effect appears higher sstar-point
appars as a hg A residual, off-nominal turn ratio, en-
countered in tracing from one tie line
3. Next, calculate values of imped- voltage and as a shift in its angular posi- terminal to another through the system
ances Z23" and Z31, which, after being tion ahead of the terminal voltages. The being reduced, can be represented in the
paralleled with Z23'.k/4 and Z31'.k/4 re- star-point voltage and additional genera- e c b c
spectively, will be equal to the corre- ecquivalentautotransf.orer
circuit by connectingy an
analyzer in series with
sponding branches Z23 and Z31 of the oria n o Fw one of the terminals. n If terminals are
original network, Figure 6. The network'
, < \ ~~~~~~~~~~~~~~retained,
and a different residual ftrturn
of Figure 8 then will have the same trans- \1.0 rai aper bewe2ahpi
fer impedances between terminals as that 2 2 mias ths2a erpeetdb
of Figure 6. / Z~ 2 - autotransformers at any (nt-1)
(i(n1)
4. The mesh network composed of ,3 Z23 terminals. This assumes that the product
branche Zii
Z-'k// and3 Z3/i
Figure 8, is equivalent to the original star A. AU ULARY STAR B. EQUIVALENT MESH
of turn ratios around any closed ioops in
the reduced system is unity. If this is not
network of Figure 7(A), with each branch Figure 7. Auxiliary star network and its the case, it is possible to encounter differ-
multiplied by k/4;. This star originally equivalent mesh, used in the derivation of the ent residual ratios in tracing through
was established in such a way that the single-source circuit different paths between two givren ter-

1949, VOLUME 68 Ward-Equzivalent Circuits for Power-Flow Studies 377


minals. Normally these differences will from the equivalent circuit, against the
be small, or the paths invrolved will be of / 9 <> difference between generation and load
high impedance and an average ratio can plus estimated losses in the reduced sys-
be used. < tem. The equivalent loads may be
A simple test on a network analyzer can shaded, if necessary, to establish the
be used to determine the net turn ratio , Z: 2 correct total power interchange.
between retained terminals in a complex Z31 k. Z r t
network. A voltage of 100 per cent is Conclusions
impressed at one terminal on an analyzer LCV
set-up of the complete line netwvork, with In steady-state power-flow studies, the
capacitors, loads, and generators open, t o
treatment of system loads and generation
andanwith autotransformer aduse
all auornsomr
wit al adjste tot as constant-current quantities gives a
represent normal transformer tap set- 3 sufcinl relsi rpeetaint
/ ~~~~sufficientlv realistic representation to
tings. The voltages measured at the w t
other terminals can be used to compute otherFigureter-ininals be used
network
8. Mesh can equivalent
to to computewarrant taking advanitage
plification in analysis
of anv sim-
which may result.
the net turn ratios between terminals. Figure 6
Transfer-impedance~~~~~~~~
Transfer-impedance measurementsmesurent This assumption
satisfactory reproduction of actual give
fails, however, to
char-a

shouldemfresiuade wthr ratas setwt norml the equivalent loads are held at constant acteristics under certain conditions, no-
tamned terminals are small or zero. If a values of real and reactive power. tably short circuits and power swings.
Assuming that generators in the re- A general statement on accuracy is not
relatively large turn ratio exists between
one terminal and the rest, it mav be duced system are operated at a scheduled attempted, because the importance of
to rede tpower output and fixed terminal voltage, various assumptions and the degree of
mossibleasurement bydaltis ringpossible
t apsetings the real power loading of the equivalent approximation involved depend on the
ofmransoremersbyalteorine taserminal
the i generators is held constant. The reactive nature of the system being reduced. Trial
quetranston.Then,at thea teratio in-
if tun loading of the actual generators, par- on several types of power networks, and
tuedtion. her,ies wthe thrnratitermi
troduced in series with that terminal of of ticularly those near the tie-line terminals in actual system studies, has yielded good
the equivalent circuit, the transfer-im- might vary during studies of different results. The equivalent circuits derived
pedance representation will be aroxi- conditions in the adjoining system. This by these methods are capable, in the
mately correct. introduces a serious deviation from the cases tried, of reproducing with accept-
The sum of the distribution factors basic assumption of constant-current able accuracy the terminal performance
sources, and gives rise to the problem of of the reduced network with various
'described under method I may not be
adjusting the reactive generation in the emergency operating conditions repre-
ecual to unity in the presence of off-
equivalent circuit. For the purpose of sented in the adjoining system.
simplifying this
factors as read wi.th normal tap settings choose, as pointsproblem,
it is desirable to Wlhen an equivalent circuit is used to
retained in the equia-
choeaspitrtindnteeqv- reproduce power flows over inter-com-
should be applied to the source and load lent network, busses where, in normal
shouldent. applyiedt mthesourceIandold pany ties, accuracy requirements are not
currinents. vapplying mhoads Idtden- system operation, the voltage is held at a extremely rigid. One company, in the
termines, thes quialernatiloas
erators, residual aetwend ter-'
turn ratios between
gen- definite, known value. Then, only the
power component of equivalent loads and
process of making a planning study of
proposed future developments, usually
minals should be represented during the
deternination of normal-voltage circti- generation need be determnined. Equiva- has incomplete information on future
plans of its neighbors, yet an approximate
lating flow tlent
reactive generation is adjusted to hold
network voltage. It may be helpful to calculate equivalent representation of neighboring
the equivalent generation at each ter- systems may be necessary in the study.
minal corresponding to the scheduled Joint studies of interconnection problems
Using the Equivalent Circuit real power loading of all suppressed gen- offer opportunities to obtain equivalent
erators and the maximum reactive power circuits of the systems involved as viewed
After the desired equivalent circuit has capability of all generators and other from the inter-company boundaries. The
been derived, it is set up, connected at sources of reactive power. This informa- methods described herein are well suited
the proper points to the detailed repre- tion can be used as a guide to avoid ob- to this type of equivalent-circuit applica-
sentation of the adjoining system, and the taining an unreasonable amount of reac- tion.
desired tests are carried out on this de- tive support from the equivalent circuit Additional approximations can be em-
tailed portion. Some additional sugges- when some particularly severe emergency ployed to simplify determination of the
tions are given here on using the equiva- condition is being studied on the ad- equivalent circuit if the study at hand
lent circuit during these tests. joining detailed network. allows liberal accuracy limits. For ex-
If the information on equivalent loads Since total tie-line power interchange ample, the distribution factors in method
and generation has been derived in terms between the equivalent circuit and the I might be applied directly to kva load-
of per unit current, the corresponding adjoining network depends on the dif- ings, neglecting voltage magnitudes and
loadings in terms of real and reactive ference between total equivalent genera- angles. A d-c board could be used to
power can be calculated, using known or tion and equivalent load, small errors in determine the distribution factors, if
estimated voltages at the tie line ter- these quantities may assume major pro- transformer taps are neglected and uni-
minals. portions in reproducing a relatively small form impedance angles assumed.
If it is necessary to estimate these power transfer. If inter-company ties In general, the new equivalent circuit
voltages initially, the actual conditions are involved, the total transfer usually is representation occupies an intermediate
are established by trial on the first specified. In other cases, it is well to position relative to methods A and B
analyzer study. During subsequent tests, check the initial measured transfer to or mentioned earlier in the paper. It is a

378 Ward Equivalent Circuits for Power-Flowv Studies AJEE TRANSACTIONS


nmiore complete representation tharl
metl:od A, but it may lack some of the
I 4-cthe ori,inal impressed current to be trans-
ferred to erminal k
flexibility and accuracy of method B,
I

tLrough suppression of internal gen I + 13)


/I++ 12+ ~~~~~~~1k/Z(+z+
(ZkZ Z +
erating stations. It is intended to supple-
ment existing methods; in practice, all This method of eliminating junctions con-
three methods might be used simulta- Z \ / taining impressed current accomplishes the
neously to advantage in some studies. k/6 Za 2 same result as a reduction formula in matrix
or tensor formn discussed by Kron.' In the
case where n is equal to two, equation 5
reduces to the familiar relation for replacing
Appendix 1. Analytical Meth- a tap load on a line section by
loads at the two ends of the line.
equivalent
ods and Basis for the Simplified In carrying out successive applications of
Equivalenit Circuit /
Equivalent Circuit23
this reduction, it is desirable to identify the
original impressed currents by letters or
symbols, and to retain them as literal rather
The principles employed in methods I and 3 than numerical expressions, keeping a record
II of the paper apply rigorously to an ideal- of the numerical coefficients of these cur-
ized network in which loads and generators Figure 9. Final single-source circuit rents as junctions are eliminated. When
maintain constant-current loadings, that is, the process is completed, that is, when all
currents fixed in magnitude and relative except the tie-line termninals have been
phiase position regardless of bus voltage and theii the process repeated to eliminate eliminated, all of the original sources and
variationis. Although these conditions are another junction. In this way, all busses loads will have been shifted to the retained
not realized in practice in a power-system e terminals with, in general, a portion of each
network, suchschanin iealzednetorkls
netvror, idealized network is con-
on- except the tie-line
nated, and, in will teriiiinalsallcansource
the process,
be elimi-
and apernatactrmalUonolc-
appearing at each terminal. Upon collect-
sidered here for the purpose of illustrating load currents have been shifted to ing coefficients of like terms in the expression
the basis for practical applications. An to appear as equivalent im-
these terminals to for the total equivalent impressed current
analytical method iS described for eliminat- thessedterminals. app as equivalence im accumulated at each tie-line terminal, the
ing all busses, except the tie-line terminals pressed currents. As long as equivalence is distribution factors of equations 1 will be
in the section of the idealized network which maintained at each step, the final circuit, obtained. Any desired schedule of source
is being reduced, and its equivalence to such as that shown in Figure 4 for the system and load currents in the original network
methodsmethds Iand III iS
I and isshow. Ths method
shown. This mehod of Figure 1, will reproduce the performance anlodcretinhergnlntwk
can be converted into equivalent currents
can be adapted to the analytical determina-
of the reduced network as viewed from the
impressed at the tie-line terminals by
tion of a power-system equivalent circuit if adjoining network. Hence, it is only euations 1
necessary to show that this process IS valid in eqaios1
neither an a-c nor a d-c board is available for
this purpose.
eliminating the common junction of a
eral n-point star connection of impedances
genw Either metbod I or method II of the paper
can be applied to the idealized constant-
Referring to Figure 1 as an example, with a fixed current impressed at this current network to obtain the same result as
assume that thle generators and loads satisfy unction that yielded by the analytical method just
the specified constant-currenit character- The general star-mesh transformation is described. In the case of method II, the
istics. Assume that connections exist to an well-known, and is included here merely for normal-flow interchange and the normal
adjoining network at the three points as convenience and consistent notation. Let cirlculating flow would be expressed in
indicated, and that it is desired to replace Zi, Z2, Zn represent the impedances terms of current rather than power quanti-
the system of Fi-,ure 1 with an equivalent tieZsrprsntte.medne
0
circuit. Assume further that the neutral connected from the common junction to
terminals 1, 2, reectiel of n The equivalence of these different meth-
bus is common to both the section to be re-
duced and the adjoining section, and that
point star network. tpvym,
flInener
and k denote any two of the n terminals.
an n ods can be demonstrated by the prin-
ciple of superposition. Any voltage differ-
the adjoining network contains at least one Then represents the impedane be- ences existing between the tie-line terminals
point of fixed or controlled voltage to neu- tween terminals ej and k in the equivalent can be considered as arising from currents
tral. mesh network and its value is given by circulating through the line network of the
Under these conditions, the distinction reduced system, with the source of these
between load and generation in Figure 1 1 1 1 1 circulating currents residing in the adjoining
is merely a matter of relative phase position Zi7c =Z,Zk - +_ +..
+ + ) (2) system. If the tie-line terminal voltages
of the current. Both the generators and 1 2 3 Zn are equal, there can be no circulating cur-
loads can be treated alike as known, fixed, An impressed current (representing either rents; in the equivalent circuit, the im-
impressed currents. By Kirchhoff's law, a source or a load) existing at the common pressed current at each terminal then will be
the vector sum of the tie-line currents must junction of an n-point star can be divided equal to the tie-line flow at that terminal.
be equal to the vector sum of the loads and into n parts and relocated at the n remote In other words, the equivalent impressed
sources (impressed currents). The equiva- terminals of the star. If the total impressed currents are equal to the corresponding tie-
lent circuit must reproduce the division of current remains the same and the voltage line currents in the actual network, when
this total tie-line interchange among the the star the circulating currents are suppressed. In
several ties in the same manner as the actual are not disturbed, the performance of the method I, the circulating currents are sup-
network of Figure 1, when changes are being network beyond the terminals of the star pressed by connecting all tie-line terminals
studied in the adjoining network. Within will be the same. In general terms, let I, together. The resulting natural distribu-
the section to be reduced, any bus or junc- represent the portion of the original im- tion of each internal impressed current in
tion can be eliminated by a star-mesh pressed current, I, which is shifted to termi- the tie lines is measured separately, and then
transformation of the line impedances nal k. The following relations must be equations 1 accomplish the proper super-
terminating at the junction. If an im- satisfied to preserve equivalence of the position of the effects of all internal im-
pressed current exists at the Junction being modified network pressed currents.
eliminated, it can be removed, prior to the In method II, any arbitrary "normal"'
star-mesh transformation, by replacing it I =I1+Ii+I3+ .,..+ (3) operating condition of the complete net-work
with separate currents at the neighboring is used as a basis. The circulating-current
junctions in such proportions that perform- JlZl =I.2Z2=IZ3~.. . =InZn (4) components of the resulting tie-line flows
ance of the network beyond these adjacent are determined by impressing the "normal'"
terminals is not disturbed. New branches, By combining the relations given in equa- voltage differences at the terminals of the
appearing between neighboring junctions as tions 3 and 4, the values of the new im- equivralent line network, and then the equiva-
a result of the star-mesh transformation, pressed currents at the terminals can be lent impressed currents are obtained by
can be combined in parallel with circuits determined. In general terms, this leads to separating the circulating components fro.n
already existing between these juncetions, the following expression for the portion of the "Inormal'> tie-line loadings.

1949, VOLUME 68 Ward-Equivalent Circuits for Power-Flow Studies 3,79


Appendix 11. Sample Calculation degrees anid 5.71/0 degrees respectively,
of Single-Source Equivalent
Single-SourceEquivalent
19 ty from the relation 0.6-1.0/0 degrees=1.7.
Z2'=0.105 Z2'; thebranchesof theequiva-
Circuit (2.28-j0.76) lent delta, Figure 7(B) are: Z12'=1.42,
\v9< /t ~ Z2Z8.06, and Z25'= 22.86. By inspectionl,
~~~~~~
It is assumed that the equivalent circuit
<
.r * r r to replace Xz /
the ratioZ80u6
ratio larger in mgniuecthan
Z12,/Z12' isiS largern
~~~~~~~~~~~~~~~~~tile magnitude than
of Figure 6 has been derived a the corresponding ratios of the other
power-system network, and that an addi- branches, hence the value of kla is taken as
tional reduction in the number of sources is 151/74\ 10.3274° (41.4/74 degrehes 74 degres) =29.2/74
necessary before this circuit can be used in degrees.
the detailed study of an extensive system, Whdegrees. imp
to which it is interconnected at terminals 1, mut/lWhen the edances in Figure 7(B) are
2, and 3. The numerical values associated (o.6I minals 1 and 2 in the resulting network is the
with the equivalent circuit are:
Impedances in per cent on 100,000 kva / ,
(-jOa16) menas the corresponding branch in the
same stecrepndn rnhi h
base i p ck base original equivalent circuit of Figure 6.
Additional branches of the proper impe-
Z12= 1±+j40=41.4/74.6 degrees dance can be connected between other pairs
of terminals, see Figure 8, to make the re-
Z23= 6 +j24 = 24.8/76 degrees
Zn =6+j24 8=24.8/736_degrees (-JO,087)
jO.087
sulting network entirely equivalent to that
Z31 = 34.8 +j 1I8 = 123 /73.6 degrees Figure 10. Single-source equivalent circuit of Figure 6. This requires, in the present
Source currents in per unit (referred to derived in the numerical example example: (Z=273-k/74 /(Z23a+k/m-Z23')m =
Z23')
voltage at terminal 1) Values of current are in per unit impedances Z Z23=27.7/74 degrees, and, similarly
I = 0.57 -jO.l9 =0.607-18.5 degrees areSince the network of Figure 7(B), with all
12 = 1.61 -j.70 = 1.76/ -23.5 degrees branches multiplied by k/l, is contained in
Figure 8, this portion may be replaced bv
13_A12
I3 = 0.10-jO. 12 = 0.156/15/-50.2
50.2 degrees component
aa quadratureand
capacitor,
cobie at
to be generated
the remaining
a
portion to by
new jucto
be
esalse
tthe sstar from wvhich it was derived, Figure
It is assumed further that the impedance combtned establwishoe
at a new junction 7(A), with all impedances multiplied by k/l.
angles are sufficiently uniform to use scalar moved to the new junction is chosen so that The sum of the modified source currents
impedance with a constant impedance angle the phase angles of all three currents are the can be impressed at the common junction of
of 74 degrees, and that normally the inter- same. These modified source currentst this star, and the residual quadrature com-n
connection terminal voltages are displaced
only a few degrees. These latter assump-
used in subsequent calculations,
are ponents of source currents can be impressed
at the terminials to obtain the final equiva-
tions are not required by the method, but I =0.57 -JO. 19 = 0.6/- 18.5 degrees lent circuit, requiring a single power source,
they simplify calculations in cases where 12 = 1.61 -jO.54 = 1.77-18.5 degrees as shown in Figure 10.
they can be made. -

In order to avoid negative resistance and I3 =0.10-jO.033 =0.105/-18.5 degrees Reference


to permit scalar computations, the original
source currents are modified by separating In the auxiliary star network of Figure 1. TENSOR ANALYSIS OF NETWORKS (book),
each of the currents 12 and 13 into two parts: 7(A) the values of Z2' and Z3' are 0.353/0 GabrielKron. John Wiley and Sons, Inc., N3e7

Discussion Figure 1 of the discussion, corresponding to


the mesh equivalent of Figure 2 of Ward's
dance between each pair of jumpers, with the
other jumpers separated electrically.
R. D. Goodrich, Jr. (United States Bureau of paper. The series of readings obtained gives
Reclamation, Denver, Colo.): Ward's in- It can be proved mathematically that the values of Za+ Zb, Za+ Z,... Z2+ Z3. Evi-
troduction of the constant-current-type two equivalent circuits are interchangeable. dently, for the 7-point network shown in
equivalent circuit is a distinct contribution However, the star equivalent has the ad- Figure 1 of the discussion, 21 such equations
to the technique of a-c network-analyzer vantages that: can be written, but, since there are only 7
operation. It provides a quick and simple unknowns, 14 of the equations are redundant.
method for representing the essential feature a. It requires less impedance units to set up. Although it has certain limitations, the
of a secondary network, namely, its ability b. It permits the elimination of an impedance leg, method of least squares often is considered
to transfer power from one point of inter- if no connection is to be made to it. to yield the most reasonable values for the
connection with the main network to c It allows, in some cases, theinalteration of an
line reactance
unknowns under these conditions. The
impedance to simulate a change
another. generator.reactance, or total system load. least squares procedure is as follows: first,
If power flow at the points of interconnec- multiply each of the original observation
tion with the main network is the only pur- The star equivalent has the disadvantage equations by its coefficient of the first un-
pose for which the equivalent circuit is re- that the evaluation of Za, Zb, and so on known, Z5, and add the products. Since
quired, it would appear that the equivalent requires the solution of a set of simultaneous the coefficient of Za is one in six of the equa-
generators in Figure 5(C) could be adjusted equations, one equation for each unknown tions, and zero in the rest, six equations will
by trial rather easily, and thus the labor of impedance. However, the symmetry of the be added together to yield a normal equation
computing the fictitious power outputs of coefficients of the unknowns simplifies the of the form
these units could be avoided. If the labor involved.
equivalent generator at one point of inter- To determine the Z's, a reasonable power 6Za+Zb+Zc +ZG+Zs +Z2+Z3 =M
connection is omitted, there should be only flow first is established in the secondary net- Next, multiply each observation equation
one combination of adjustments of the re- work on the network analyzer. Then the by its coefficient of Zz, and add the products.
maiming generators which will yield the de- secondary network iS disconnected from the Th'eodnra qainwl,teeoe
sired values of interchange power. The main network and from thze generator units, be
significance of individual generator powers and a meterable jumper is connected to each
in the equivalent circuit will, however, be point to be retained, including the ground Za+6Zb±ZC+ZG+Zl+Z2+Z3 = M2
lost, bus. In some cases, it is permissible to bus
Where a more complete equivalent circuit together all of the points of generation in the This process is repeated until seven nor-
is required, as in stability studies, for ex- secondary network, so as to retain a single mal equations have been formed; their
ample, it has been found convenient to use equivalent generator. The final step of solution will yield the impedances desired.
the star-type equivalent circuit, shown in analyzer operation is to measure the impe- The resulting equivalent circuit will be

380 Ward-Equivalent Circuits for Power-Flow Studies AJEE TRANSACTIONS


a e >s O network analyzers, for the type of equiva- generators usually oscillate together as a
lents introduced by this paper. It is a unit.
H2 Z welcome contribution and should prove use- For such cases, equivalents are meas-
b -2i- 2 ful to most utilities that have calculating ured by means of method B, the generators
boards. Although the Bonneville Power in the network being represented by their
t3 $ 3Administration's a-c analyzer has 18 board transient impedances. The end result is an
C X SZG 3, generators, the Northwest Power Pool con- equivalent which contains a single equiva-
1 tains more interconnected systems than can lent generator. However, the equivalent
be set up in detail; thus, we require equiva- rarely satisfies the initial steady-state
lents. conditions at the retained terminals and is
Figure 1. Star type of equivalent circuit In many problems on our board, these not rigorous. It should be possible, by
equivalents contain too many impedance means of an extension of Thevenin's theo-
units, since there are generators in the net- rem to n terminals, or by some other de-
the
t original
i systemv
t both for steady state works we desire to reduce. Ward's method vice, to determine an equivalent
for supplies an ingenious method for reducing would be much more satisfactory than which
those
the number of impedances needed and used at present in swing studies. If any
allows for changes in loads and generation. system engineers or network specialists
C. E. Parks (Public Service Company of In the majority of our studies, generators know of such a method, it would be appreci-
Indiana, Inc., Indianapolis, Ind.): The and loads are maintained at constant power ated greatly if they would publish a paper
author is to be congratulated for his novel output, and usually the reactive power on the subject.
approach to the subject of the setting up of changes very little so that it isn't necessary
equivalent circuits for power-flow studies to be able to alter these. However, to be
on an a-c network analyzer. able to reduce the number of generator R. J. Tallent (Ebasco Services, Inc., New
Most electric utility companies are now sources is of much value, and the author York, N. Y.): Ward has developed equiva-
employing the a-c network analyzer in the should be complimented on his neat scheme lent-circuit methods which are quite an
solution of present, and future, system for reducing the number of equivalent improvement over the old conventional
operation and development problemns. A generators. methods. Power systems have grown so
large majority of the systems now being We have attacked the problem of equiva- large, and interconnections have become so
analyzed on the calculating board are inter- lents and solved it in another way. It is common, that it has become quite difficult to
connected with at least one other company. based on the fact that generators can be re- condense, by the conventional equivalent
Where these interconnections become nu- placed by negative impedances in analytical methods, some systems sufficiently to get
merous, the problem of representing the solutions, assuming constant impedances. them on the board. It has been a problem
adjacent system's transmission line set-up, To do this on an a-c analyzer, we have de- to set up the critical sections in enough de-
taking into account that system's possible veloped an electronic negative impedance tail so that line loadings and voltages may
development in the future becomes a very capable of use with the analyzer. This be observed, and yet leave enough board
important factor in the study under con- means that a system to be reduced can have generators and impedance elements to
sideration. If the adjacent system is set up its generators replaced by the negative im- accurately reproduce the adjoining system.
in detail, it usually requires that a joint pedances, and transfer impedances may be By using Mr. Ward's method, it is possible
study be conducted. While such joint measured between the network terminals, to reduce considerably the number of ele-
studies are very desirable, they consume the as in method B described in the paper. ments in the adjoining-system equivalent.
time and effort of the operating engineers of Should any of these impedances contain The method in the paper described under
both companies involved. Such an arrange- negative resistances, one of our negative method I, paragraphs 1 through 5, would
ment will result usually in abbreviating the impedances can be used to represent it. In give a flexible set of results suitable for read-
system set-up for both systems inivolved, this fashion, it is possible to eliminate the ily representing a variety of load and gen-
because of the limitations for board facil- use of distribution currents, which, as Ward eration conditions, but, as noted, the work
ities, and further generally will limit the has pointed out, can be quite "lengthy and involved in reducing a large system with
number of studies that can be conducted in a cumbersome. " Incidentally, these nega- many loads would become lengthy and time
prescribed period. tive resistors can be used as board generators consuming.
The author of this paper, since he is in conjunction with the ordinary board gen- The methods described under method II,
engaged actively in power-flow studies of the erators, as well as provide a representation paragraphs 1 through 5, are an improvement
numerous systems which employ the Purdue of negative resistor elements. However, on the above insofar as reducing labor and
network analyzer, has appreciated this even with negative resistors, it is helpful to time, but have the disadvantage of requiring
problem, and in his excellent paper, has have a method for obtaining single-source both systems to be set up simultaneously.
presented a very sound solution of the equivalent circuits in parts of systems re- This requirement nullifies one big advantage
problem. mote from the central portion being in- of the method, since shortage of analyzer
The methods that he suggests have been vestigated. In this connection, negative elements is frequently the motivating reason
employed in power-flow studies made by impedances would simplify the procedure for using equivalents. Also, as recognized,
the Public Service Company of Indiana, given by Ward to secure a single-source results are not as flexible, since they pertain
Inc. It has been found, by actual practice, equivalent. In those cases where negative only to a given generation and load condi-
that the results obtained by using the equiva- resistances appear, they easily can be repre- tion.
lent circuits established by these methods sented by the electronic units, thus obviat- The method described in paragraph 6
has proved sufficiently accurate in the solu- ing the need for mathematical stratagems, under method I sounds promising and prac-
tion of the study being conducted. Other such as "changing the anglek," to eliminate tical. However, as described, the results
methods of representing the adjoining sys- negative resistances. would not include the effect of varying
tems network have been tried out by our The author's pertinent comments on generation in the neighboring system.
company, but it has been found that the the effect of transformer taps in obtaining This would limit its usefulness to a degree.
method presented by the author in this pa- equivalents should be helpful to those who If this method could be extended simply to
per produces by far the most accurate re- have had to contend with interconnected sys- include generation effects, it would find
sults. tems. Fewv, if any, articles have mentioned frequent application, since it would be very
Again, may I congratulate the author in this subject, and the author deserves praise easy to carry out, requiring only one system
his fundamental thinking in the develop- for bringing it to the attention of system set up at atime. Results would be accurate
ment of the ideas presented in this paper, engineers, enough for a large variety of studies, pre-
and express the hope that other operating Now that at least two methods exist for suming, of course, the studies deal with like
companies will make use of this time- and obtaining satisfactory steady-state power- load conditions and the same system.
labor-saving tool in their future studies. flow equivalents, it is desirable to obtain one Perhaps, by applying the superposition
which would be satisfactory for power-swing principle, it may be possible to extend this
studies. In those cases where the gener- method to include generation effects without
Glenn W. Bills (Bonneville Power Admin- ators are in an area electrically remote from either increasing the work and time appreci-
istration, Portland, Oreg.): There is a the network upon which are inflicted shocks, ably or causing need for major set-up re-
definite need, by those engineers who use a-c such as line faults and line clearances, the connection, such as disconnecting loads.
1949, VOLUME 68 Ward-Equivalent Circuits for Power-Flow Studies 381
This might be done in the manner describedl The current of each source is equal to the 1 of the papler, this would appear as a star
below. short-circuit current of the corresponding connection of six legs, wita the common
1. Proceed as described in paragraph 6 terminal. junction at the seventh terninal and with
and obtain load distribution factors for the mutual coupling between eacih pair of the six
interconnection points. legs, thus providing the required minimum
2. Next alternately and singularly con- J. B. Ward: Tallent points out a disad- of 21 impedances. Goodrich presents an
nect, by a metering lead in parallel with the vantage of method II in the necessity of interesting method for seeking the "best
leads to the interconnection points, the setting up simultaneously the network to be fit" of impedances in an approximate star
various generating buses to the board gener- reduced together with the section to be network without mutual coupling. It is
ator. Adjust loads and record flows in each studied in detail. On the surface, if such a noted that the general "normial" equations
metering lead for each generating bus. set-up were possible, it might seem that the for a given number of points of entry need
3. By application of the superposition equivalent circuit was not required. The be solved only once, since the coefficients are
principle, the distribution factor for each author had in mind a particular situation in fixed by the number of terminals. Then
generating plant to the interconnection proposing method II, namely, the occa- this solution may be applied to any problem
points may be obtained by subtracting re- sional necessity of interconnected operating involving the same number of terminals. A
sults of step 1 from step 2, and dividing by companies making joint studies wherein simple relationship between self and mutual
the flow in the lead to generating bus. skeleton representations of the major trans- constants and the "through" impedances
4. The net load at each interconnection mission networks are used to permit setting used by Goodrich is developed in Starr's
point may be computed by following equa- up simultaneously the several systems and paper.
tion: studying interconnection problems of mu- The reduction of a network containing
tual interest. While such a set-up is avail- sources as described by Bills treats both
Load, =Di (Load) -Dia(MWa) - able on the analyzer, it is a simple matter to generation and loads as constant impedances,
Dib(MllAlTb) * * obtain equivalent circuits by method II. while, in the present paper, these quantities
Then these equivalent circuits will be useful are considered primarily as impressed cur-
where in subsequent detailed studies made by the rents. The successful use of the two differ-
D= distribution factor for loads as obtained in step separate companies on their individual sys- ent approximations of power-system load
1 for interconnection point 1 tems. and generator characteristics seems to indi-
Load = total system substation load In connection with Tallent's comments on cate that, for the normal power-flow type of
Dia, Di = distribution factor for power plants a and the approximate apportionment of load in study, there is considerable freedom in de-
b to the interconnection point 1 paragraph 6 under method I, it was intended signing equivalent circuits, based on com-
Loadi =net load or generation at interconnection that the equivalent generation be obtained binations of the two viewpoints. If it is not
point 1 by separately measured distribution factors, necessary to deal with changes in loads, their
with loads disconnected, as described in representation as constant impedances
5. Adding these equivalent loads or preceding paragraphs 1 to 5. Tallent's dis- lends itself more readily to wholesale reduc-
generations to the interconnection points, cussion outlines an interesting application tion to equivalent loads at the tie-line
and connecting these points by the transfer of superposition, leading to essentially the terminals.
impedance between them, will give a close same result with loads left connected. Kimbark's discussion brings out the close
equivalent. The star-type equivalent circuit suggested relationship between Norton's theorem and
by Goodrich is not interchangeable with the Thevenin's theorem. Although most
mesh circuit, Figure 2 of the paper, in the power-system networks do not conform ex-
Edward W. Kimbark (Northwestern Uni- sense of exact general equivalence. For a actly with the idealized network character-
versity, Evanston, Ill.): The representa- general n-terminal circuit, a minimum of istics on which these theorems are based, it
tion of a multiterminal active network by a n(n 1)/2 links is required in an equiva- is felt that there could be more widespread
mesh of impedances, with a constant-current lent circuit.1 The links of the circuit in and effective application of these principles
source at each terminal, as described by Figure 2 of the paper, expressed as admit- in power network analysis. Figure 2 of the
Ward, is really an extension of Norton's tances, are the D coefficients of Starr's discussion is submitted to supplement Kim-
theorem to a network of more than two paper. The C coefficients of Starr's paper, bark's pertinent comments on extension of
terminals. which are commonly known as "self" and these theorems to more than two terminals.
Thebvenin's theorem, in which a 2-terminal "mutual" constants,2 may be arranged in Two equivalent circuits are shown for a 4-
active n-etwork is replaced by an equivalent an exact equivalent circuit, in the form of a terminal active network. Terminal n is
circuit consisting of a constant-voltage star network with mutual coupling between taken as the reference terminal, correspond-
source and an impedance in series, is known branches. For the sample system of Figure ing to neutral in power networks. Note
and used now by many power engineers. that the star-with-mutual circuit mentioned
Norton's theorem, the dual of Thevenin's previously is shown in the generalized
theorem, is not so well known. According .Thevenin equivalent circuit. This form
to Norton's theorem, a 2-terminal active net- _ I yields a close analogy with the familiar 2-
work can be replaced by the parallel com- I z12
terminal case.
bination of an impedance and a constant- In the generalized Norton's equivalent,
current source. The impedance is the same 1
0
2 the links are transfer impedances, and the
as that of Thevenin's theorem. The cur- Z23 Z13 current sources are equal to the short-circuit
rent is equal to the short-circuit current of \ . r 3 currents obtained by connecting terminals 1,
the original network. Z,, Z2n5Z35 2, and 3 simultaneously to n. The methods
Another fact not generally known is that - -10 n of the paper lead to a circuit of this type, and
Thevenin's and Norton's theorems can be - - the impedances to neutral will appear if
extended to networks having more than two A. NORTON'S EQUIVALENT loads are retained as constant impedances.
terminals. Instead of one impedance The voltage sources in the generalized
branch, there is now a branch between each Thevenin equivalent are the open-circuit
pair of terminals. These impedances can be r .voltages of the original network, measured
found by reduction of the original network Zj Z;-2Z|~-2j3) with reference to terminal n. It should be
after short-circuiting each constant-voltage ©
@ j -I emrphasized that the self and mutual impe-
source and open-circuiting each constant- | |Z23 (Z22-2l2-Z23)l dances appearing in Figure 2( B)of the discus-
current source in that netwrork. Thdvenin's 2 sion are not the same as the transfer impe-
equivalent circuit has a constant-voltage l I: (z-z4-z) dances in Figure 2(A) of the discussion.
source in series with each terminal, except E3 3
the reference terminal, equal to the open- REFERENCES
circuit voltage between the corresponding -s~~ QIAETCRUISI .M tr.AE
terminal and the reference terminal. Nor- L_________ --_______-_- TRANSACTIONS, volume 51, 1932, pages 287-298.
ton's equivalent circuit has a constant- B. THEVENIN'S EQUIVALENT2.ECTIATRNMSONNDITIBIN
current source shunted between the refer- REFERENCE BOOR. Westinghouse Electric Corpo-
ence terminlal and each other terminal. Figure 2. Generalized equivalent circuits ration (East Pittsburgh, Pa.), 1943, pages 87--96.

382 Ward-Equivalent Circuits for Power-Flow Studies ATEE TRANSACSTION.S

Você também pode gostar