You are on page 1of 8

which would exist if there were no satura-

Underexcited Operation o1 tion. An equivalent synchronous react-


ance equal to the reciprocal of the short-
Turbogenerators circuit ratio has been assumed in order to
approximate the saturation effect. Re-
sults, when using this approximation,
C. G. ADAMS J. B. McCLURE have been found to closely agree with
ASSOCIATE AIEE MEMBER AIEE those obtained by other more detailed
methods described in reference 1, par-
ticularly for the condition of pullout at
PERATION of turbogenerators at 2. For most practical system conditions, low excitation.
high-power factors and even extend- operation as indicated before retains ade-
quate steady-state stability margins.
ing into the leading region is increasing. Representation of System by an
This trend is due in some cases to the 3. Voltage regulator control with an auto-
matic lower limit rationally permits a Equivalent Reactance
location of large steam-electric stations greater utilization of leading reactive capa-
remote from the load and the transmis- bility than does manual control of excitation. Representation of a system by an
sion of their outputs at high voltage. In equivalent reactance is fairly well known.
other cases, it is due to the application of Simplified Analysis of a It is simple and produces practical re-
corrective reactive kilovolt-amperes in Typical System sults. B3ased on generator rating, a stiff
the subtransmission and distribution system may have an equivalent reactance
areas of the system, or the use of high- Figure 1 shows a one-line diagram of in the order of 15 per cent. Systems
voltage cable in the transmission system. the generator or station, under considera- involving long power transmission may
These conditions are a natural result of tion, supplying load to its system. approach a value of 40 per cent.
system growth and economic pressure to Figure 2 shows the general relations be- A fairly good approximation to the
more effectively use the generation and tween the system, machine terminal, and equivalent system reactance may be ob-
transmission equipment in the supply of machine excitation voltages. Their tained in the following manner. In most
electrical energy. specific relation for various conditions cases the short circuit kilovolt-amperes
This paper reviews underexcited opera- will be discussed later. The following for circuit breaker application is known at
tion of turbogenerators from the stand- conditions have been assumed. the generator bus. Reduce this kilovolt-
point of limits imposed by system sta- 1. The generator is rated 0.8 power factor amperes by the amount contributed by
hility and the application of lower excita- 0 shorircuit ratio (SCR) Maxmum the generator in question. Divide twice
tion limits on voltage regulators. Re- turbine output equals rated generator kilo- the rated generator kilovolt-amperes by
sults are shown for a typical turbogenera- volt-amperes which is 1.0 per unit. the reduced short circuit kilovolt-amperes.
tor operating on an average system. A 2. Theequivalentsynchronousreactanceof The result is approximately the system
method for readily estimating the stability the generator is equal to the reciprocal of equivalent reactance in per unit. A more
limits for other machine and system short-circuit ratio. accurate method is to set up the system
characteristics is included in the Appen- 3. The system is represented by an equiva- on a calculating board and insert a syn-
dix. lent reactance Xe. chronous reactance of all generators
It is hoped that in this way reassur- 4. Resistance in the circuit is neglected. equal to the reciprocal of their short cir-
ance will be given to those who find high- 5. Since most turbogenerators have a cuit ratios. Find the reactance looking
lpower factor operation of generators cylindrical rotor, pullout power is assumed into the system at the entry point of the
desirable in planning their normal system to occur when there are 90 degrees between generator being studied. The latter
the generator excitation voltage and the method is more complicated but more
expansions. equivalent system voltage.
Conclusions ~~~~~These assumptions simplify the deter- Papera48-81 rcommlteendded by the AIEE power
mination of operating characteristics technical program committee for presentation at
1. Moentroeeaosmnfcue maufactued
Modernturboeneratrs hl no
whle fetn hhe acuracy
ral affcting
o greaty cuay the
January winter general meeting,
AIEE 26-30,
i948. Pittsb.urgh.
Manuscript Pa.,
submitted
by the company with which the writers are Of the results over a wide range of system November i0, 1947; made available for printing
associated may be operated successfully in conitins December 4, 1947.
the range of 60 per cent underexcited reac-cntos.c. G. ADAMS and J. B. MCCLURE3 are both with the
tive kilovolt-amperes at zero kilowatt load central station engineering division of the Generat
to 30 per cent underexcited reactive kilovolt- Effect of Generator Saturation Electric Company, Schenectady, N. Y.
amperes at rated generator kilovolt-amperes. The authors wish to acknowledge the particularly

thereby made available to supply most sys- Freao


nicrae
ata
steady-state pulllout power over that
h eibflurghoeseioE. ad.Hunstley, H. F. s
MCloug
and HI. D. Taylor of the General Electric CCompany.
tem demands.

1948, VOLUME 67 Adams, McClure- Underexcited Operation of Turbogeneraitors 521


4/(0rXi@aJ-08~ FORCNSTAF
GENERATOR OR turbogenerators manufactured by the power factor for several constant kilowatt
STATION
Xd X.±STATINXsX* company with which the writers are loads. The power increment between the
associated operate successfully in the sug- curves and the dashed line at the kilowatt
Ed tEt T gested area for voltage regulator lower loading for which the curve is drawn is the
I' limit setting without excessive tempera- power that may be picked up under the
,/l GO
GROUND
tures and retain adequate stability mar- condition of no change in excitation or
gins. It should be recognized that for system voltage from their initial values.
Figure 1. One line diagram of system hydrogen-cooled machines, operation with These power increments have been plotted
increased hydrogen pressure permits a against power factor in Figure 5. The
accurate. Since steady-state pullout higher kilovolt-ampere output under effect of changing generator terminal
powers are not affected greatly by small comparable operating conditions. voltage on allowable power increment is
variations in equivalent system reactance, shown in Figure 6. Pullout power as
not much effort can be justified in obtain- Manual Control Operation function of initial kilowatt load for dif-
ing an extremely accurate system rep- ferent initial power factors is shown in
resentation. Figures 4 through 9 illustrate the fac- Figure 7. The curves of Figure 8 illus-
tors influencing pullout power when the trate the effect of system reactance on
Generator Operating Limits generator excitation is manually con- pullout power. The effect of generator
trolled. Referring to Figure 2, the short-circuit ratio on pullout power is
Figure 3 is a plot of reactive kilovolt- generator excitation voltage Ed and the shown in Figure 9. Figure 10 shows the
amperes and kilowatt output for the system voltage ES are determined so that variation in pullout power produced by a
assumed generator operating at its rated with the initial loading there is rated volt- change in system equivalent reactance for
terminal voltage. Positive reactive kilo- age at the generator terminals, except for a low short-circuit ratio generator.
volt-amperes is supplied by an over- Figure 6 for which the initial generator The question of primary importance in
excited generator. The circle ABCE is voltage is varied. Holding Ed and ES at the system operation is the ability of its
the well known limit due to rated line their initial values, the maximum pullout component parts to endure the normally
armature current. The curve IB is the power is assumed to occur when these encountered emergencies. Generators
limit due to rated field current. Area voltages are at 90 degrees. Figures 4 will be required to quickly pick up load in
CGKL is suggested for the setting of the through 9 show the pullout power that case a generator is tripped off; in case a
operating range of lower excitation limits can be developed with the excitation and major tie-line opens; or if a high rate of
being applied on modern voltage regula- system voltages unchanged from their system load increase is encountered.
tors. Stability limits for two values of initial values for different initial loads, For most system conditions, it can readily
equivalent system reactance are shown by initial power factors, system equivalent be determined what load increases may be
curves JE and DF. (See Appendix I for reactances, generator short-circuit ratios, expected on the several generators as a
curve derivation.) and initial generator terminal voltages. result of any emergency. This deter-
The tendency for heating in the end Figure 4 shows the pullout power versus mination should take into account any
structure of the armatures of turbogenera-
tors with rated armature current and low Figure 3. Rated o.e PE CYLINDRICAL ROTOR GENERATOR
field current was recognized as long ago as voltage generator SHORT CIRCUIT RATIO OF GENERATOR 0.9 z

1920. Up to that time, all machines circle diagram EQUIVALENT Xd IISCR.


MAX. TURBINE OUTPUT :1.111
r

GENERATOR KVA.
were built with magnetic materials in the 0.8 power factor +.A0A
end structures, which contributed to heat- cylindrical rotor
ing in these portions of the machine. The generator z
operation in the underexcited region of Short circuit ratio Of J
+0.8 _ _ 8_\
_ _
machines built prior to about 1925 should generator = 0.9 ~_\
be approached advisedly and preferably Equivlent Xd = +0.6 -RATED FIELD CURRENT
should be checked by temperature meas- 1/TCR=1.111LMIT
Maximum turbine
urement under actual operating condi- output = generator +0.4
tions. As a result of design progress, kilovolti i RATED PF.
experience and tests indicate that modern
X +0,2 _ ,;4 - d -1-25-
Figure 2. System vector diagram i
0H
X / KILC0.2WATTS 0.4
PER UNIT \_ 08_
L / CONSTANT FIELD CURRENT LOCUS 0.6 0 RATED ARMATURE
90 AT PULLOUT GCURRENT LIMIT
w -
0.2 AREA OF OPERATION
d PULLOUT POWER d EAd s /Y OF VOLTAGE REGULATOR

/ CONSTANT\ Fy ;4/tT \1\/ { ~~~~~~~~~~~~~RATED


KVA
4 C;URRENT
LOCUS \ qa~~~~~
-0.6 /_ .5D i DASHED CURVES ARE

//I\ < -IX CONSTANT TERMiINAL ] t


VOLTAGE LOCUS -1.0 l -X* 0
E6
~~~~~~~~~STEADY STATE STABILITY LIMIT

522 AdaLms, McClure -Underexcited Operation of Turbogenerators AJEE TRANSACTIONS


-01.8
X'e : 0.20
0.80
a. o.so5CR. 0.90s SCR 0oa
i
wa: d .1 5CR..1
0 u

0.~~~~~~~~~~~~~~~~~~.0
4-1.6

KW. L OA D I0.00

__<
0.8 0.9 ~~~.0 0.9

GEN. KVA. J~~~~~~~Fgue5(aoe.Liiigvau fpoe nceetfo tbliyvr


0.80
______
0.90
______ LEAD
~~~~.00 0.008
LEAD POWER
______ FACTOR
u nta la oe atrfrdifrn nta od
OFLINITIALGLOAD.LAG
POWER FACTOR ~~LOAG .

at low generator terminal voltage is


limitations in boiler capacity; speed It is also apparent from Figure 5that with emphasized by the reduction in allowable
governor characteristics; and the action an initial 20 per cent kilowatt load at 0.9 power increment as the generator voltage
of any load-limiting devices which may be leading power factor, a 50 per cent load decreases. At 0.95 leading power factor
employed in the control of the prime- increase can be accepted by the generator and 0.8-kw load, the power increment at
mover input. In this way an estimate before instability is encountered. 90 per cent generator voltage is one-half
can be made of excitations which should In practical operation, the4_LEA generator its value
POWER FA TRO LAO3
rated generator voltage.
at NIILL A

be maintained for machines on manual voltage will probably be less than rated In Figure 7 the difference between the
control in order that they may pick up when the machine is required to absorb straight line at 45degrees with the horizon-
their expected share of load without loss of large amounts of leading reactive kilovolt- tal axis and the curves may be interpreted
synchronism. amperes. Figure 6 shows the effect at any kilowatt loading as the permissible
For a typical machine, Figure 4 shows generator terminal voltage has on permis- power increment at that initial loading.
that a machine initially loaded to 80 per sible power increments for power factors. The increase in the power increment as
cent kilowatts at 0.9 power factor lead between 0.95 lag and 0.95 lead. The initial power factor goes from the leading
can pick up about 7 per cent in power be- desirability of voltage regulator control to the lagging region is evident. To
fore going out of step. It also shows that when operating in the leading region and accept the maximum turbine output, an
with 80 per cent initial kilowatt load at 0.6____________
1.0) power factor, the pullout power ex- SCR.,: 0.9,
ceeds the turbine capability. Margins in F xe: O Z ||0.2
permissible pwrincrement aesmall in > NTA s.LA:O[13
mintaned
thebehigher or mchies o
load regions manal
at 0.8 power0factor vlcrement forostabily
l bels0.9 la atd
w 0.4 1
I Fgr
0.9the. dife05c bewenh

If the generator load increase for the X _ .95 LAG.-


most severe system disturbance is known, l5 l v o p f _ v
the required power factor for the initial sus a ld pe t
kilowatt
F igure
load may be obtained from
5. The dashed curve AB on 0.2
o |
¢ F- -| 5 t
W --r l l r
t v i
Figure 5 iS drawn through the points 0. .9 LA
where the initial kilowatt load plus the L- a tage
permissible power increase iS equal to the generator_ _ad8k ldt INITIAeLOAD w PF.
maximum turbine capability. If opera- Figure 6. Limiting |- -
tionismairintned in the region to theiright value of power in- lote 90 per c e l-
of curve AB, the generator will not pull crement foractcabl rty ioh aor is vu
out within the turbine rating. This type versus generator ter-
of operation would be most conservative. minal voltage GENERATOR TERemINAL VOLTAGE PER UNIT

1948, VOLUME 67 Adams, McClure-Underexcited Operation of Turbogenerators 523


initial 40 per cent kilowatt load may be excitation on the generator, it appears 1.6 - _ I
carried at 0.8 lagging power factor and an from Figure 10 that the available genera- UNITY PF. INITIAL LOAD
initial 75 per cent kilowatt load at 1.0 tor pullout power is about twice the maxi- l l 0 .20 °°_
x

power factor. mum turbine output for X, from 20 to 40


In Figures 8 and 9 the permissible per cent corresponding to the average
power increment is again the difference system under normal conditions. It is 1.2
between the curves and the straight 45 realized that these torques are seldom re- t MAX. TURBINE OUTPUT
degree line. The change in pullout power quired or seldom exist and to that extent
for an initial 1.0 power factor load result- are somewhat fictitious, but at the same X 0.9

ing from doubling the equivalent system time it does give an indication of the mar- LI SCR: 0.9_//
reactance is shown in Figure 8. Figure 9 gins which are available through the auto- e -f---
shows the effect of decreasing generator matic control of excitation, which are
short-circuit ratio from 0.9 to 0.6. Quali- available to meet emergency system condi- J S DA
tatively from Figures 8 and 9, the con- tions.
elusion may be drawn that generator pull- Some voltage regulators permit the 0.4
out power is not greatly affected by small INCREMENT
changes in its short circuit ratio or in sys- SCR. a 0.6
tem stiffness. 1.6 UNITY PF. INITIAL LOAD
SCR. 10.9
Voltage Regulator Operation Xd__Xd _ I_
_-_
0 0.2 0.4 0.6 0.8 1.0
When the excitation of a machine is t
under the control of a voltage regulator, 1.2 _INITIAL KW. LOAD PER. UNIT
adjusted to maintain rated voltage at the n
TRNE OTU lue99. Pullout
Figure power
oe versus initial kilowatt
Z.,r_/E.KYA. load Forulu essIiilklwt
-
machine terminals, the generator will ,, . -_ --
_ - - dKierent short-circuit ratios
stably accept the maximum turbine out- X x 0.2
put in any practical case. The curves in s voltage regulator, and machine stand-
Figure 10 assume that the excitation sys- Jpit
tem has a ceiling voltage 50 per cent in
excess of the required slip ring voltage at X_x 0.4. Transient Stability
the rating of the main generator. This is
equivalent to having an exciter of 300- 04 Since the voltage behind transient react-
volt ceiling and rated generator slip ring PERMISSI BLE POWER ance is not affected greatly by power fac-
volts of 200 which is representative of INITAL 0.8 KW LOAD 1.0 PF. tor at rated terminal voltage, transient
present day practice. Terminal voltage X 0.4 stability limits are almost unaffected by
of the generator is assumed to be 1.0; initial loading conditions. Voltage regu-
the exciter is at ceiling; 90 degrees exists 0o lators with usual rates of response and
between machine excitation voltage and 0 0.2 0.4 0.6 0.8 1.0 ceilings affect these limits by a rather
equivalent system voltage. For ceiling lNITIAL Kw. LOAD PER.UNIT minor amount. Of course, transient syn-
Figure 8. Pullout power versus initial kilo- chronizing torques are high as compared
watt load for different external reactance to the steady-state stability limits and
1.6 - 0.20 e of importance in determining whether
SCR.: 0.90 synchronism will be lost after a system
Xd R1.111 adjustment of the'lower limit of excitation shortcircuithasbeencleared.
MAX. TUJRBINE $ |,- as a function of turbine kilowatt load or of For a turbogenerator on an average
.- 1.2
OUTPUT = GEN. KVA.>N
.1 the generator kilovar output. Area system operation at unit; or in the leading
1 CGKL on Figure 3 suggests a range for power factor region will have a small
PR:0.B LAG -1 .041efc pntanin iis With ihmn
a: PZ L |\| the setting of these limits which will in effect upon transient limits. man-
i.0 0 *_,:
PF
5 most cases leave a considerable margin ual control of excitation, generators should
between system requirements and ma- be operated with sufficient excitation so
w
0.80.8 /T <OP.BLEAD
/ | chineincapability.
I ;//
tion this Manualthecontrol opera-
region
allows generatorsto that they can
increment after stably
the transient phaseload
accept their is
0 _ _ _ _ |_ |
accept a reasonable load increase before
pullingoutofstep; however, itshouldbe
over. Vo'tage regulators
excitation rap dly enough during the
can adjust
_J pullI | realized that Figure 3 represents condi- transient perioe, so that any load may be
|J ;/ .\ tions for rated generator terminal voltage, assumed up to the generator rating.
0.4 | ASPERMISSIBLE POWER | and that allowable power increments are Voltage regulator control permits opera-
|/ INCREMENT reduced as the generator voltage is de- tion closer to the steady-state stability
PE PNfLEADIONAG 04K8|
creased. Voltage regulator control ration- .limit.
/l 0 | | ~~~~ally permits operation closer to the steady-
___ ___ __l_l _ state stability limit and possibly beyond Nomenclature
0 0.2 0.4 0.6 0.8 1.0 those shown by operation in the "dy- UntKa=uiklol-mprssth
INITIAL KW. LOAD PER UNIT namic" stability region. However, opera- generKatori ralvoted kilvot-mpre
pee
Figure 7. Pullout power versus initial load at tion below the line GC on Figure 3 should which is equal to the maximum rated
different power factors be approached advisedly from the system, turbine output

524 Adams, McClure-Underexc'ited Operation of Turbogenterators ATEE TRANSACTIONS


2.5 GCEILING EXGITATION 5Eds
3/2 EXCITATION ATMACHINE RATING QEgE4J(
p+jQ =- e 2 (11)
Et 1.0 Xd Xd
IMAGINARY AXIS For constant values of Ed and Et, equation
11 is a circle with its center on the negative
2.0 \ -_/_\ reactive axis at Et2/Xd and with a radius of
EtEd/Xd. Applying to Figure 3, the center
of circle IB is on the negative reactive axis
____ jo;_A____ at Ei5/Xd=1/l-ll=0.9, and its radius is
\ 129- V/sn EtEd/Xd= (1)(1.89) /1.111 = 1.70. Circles
1.5 \ ____
. for other constant field currents than rated
may be easily drawn by keeping the same
SCR. s0.6 center and varying the length of the radius
t___ ____ ____ 1.667 in proportion to the field current.

o°0. IC'° 1 i__ __ \_


t REAL AXIS
o// Steady-State Stability Limit
MAX. TURBINE OUTPUT The limiting pullout power is taken to be
____ Figure 11. Circle in vector notation that power flow existing when Ed and E8 are
at 90 degrees and E = 1.0. For this case
taking Es along the real axis,
0.5 generator supplies positive lagging reactive
power to its connected systemt Ed=EdE)a =Edejr/2 =jEd (12)
P+jQ=EtI* (I) Ed=EJO (13)
Referring to Figure 2, the generator ter- I=Ed Et iEd-EtE (14)
0 0.4 . - 1. minal voltage in vector notation is jXd iXd
0 0.4 0.6 1.2 1. 6 2.0
Xo IN PER UNIT Et=EEtE (2) 1 -jEd-EtE-J Ed-jEtE'# (15)
Figure 10. Pullout power versus external and the generator armature current is -JXd Xd
reactance similarly FEd-jEtE-J
I =IEi(+)
The conjugate of the current is
(3) P+1Q=EtI =
L Xd ] (16)
P =real power in per unit E Et'
Q =reactive kilovolt-amperes in per unit p _ = (17'
Ed=generator excitation voltage which is 1* (4) +Q -Xd Xd
proportional to generator field cur- The vector power is found by substituting From the geometry of Figure 12
rent. Unit excitation voltage pro- equations 4 and 2 in equation (1).
duces the field current required at [ Xe Ed
rated no load voltage on the genera- P+jQ=E, f(0+0) (5) sil 0 Xd+XeE(18)
tor saturation curve
Et = generator terminal voltage in per unit of or or
rated generator voltage P+jQ =ErJe-G (6)
ES =equivalent system voltage in per unit of
rated generator voltage Equation 6 for the condition that E =1l.0 Edt=6fotsintonE)Xd+Xe
E L Xe jsi E (9
SCRa=egenerator short circuit ratio is the and I= 1.0 (rated voltage at the generator
ratio of field current at no load rated terminals and rated current in armature), Eliminating Ed from equation 17 by using
voltage on generator saturation curve represents a circle with its center at the equation 19
to field current that produces rated origin with a radius of EII = 1.0. This is Xd+Xe E2
circle AB1CDE on Figure 3. When the cur- (20)
Jo

generator armature current with P+jQ =-2LI2sin E -j]-


armature short circuited rent lags the terminal voltage, the imaginary L XeXdJ Xd
Xd = equivalent generator synchronous part of equation 6 is positive since the angle
reactanice in per unit which is assumed 0 is negative. This corresponds to over- By usIng the relation developed in Appendix
equal to reciprocal of generator short excited operation. II, equation 20 iS:
circuit ratio X+Xe 1
Xe =equivalent system reactance in per unit Rated Field Current Limit P+iQ =EtIL 2XeXd J
of generator base ohms. Base ohms The maximum field current is determined
of generator equals rated generator by the field current required when the ma- jE 2 d±e 1 (21)
line to neutral voltage divided by chine is carrying rated kilovolt-ampere at L2XeXd XdJ
rated generator armature current rated power factor. Referring to Figure 2 Equation 21 is the equation of a circle with
I =generator armature current in per unit of and determining the vector power in the its center on the reactive axis at
rated armature current
6, ,B, a = angles defined by Figure 2 same manner: E2 Xd+ Xe 1 1
P+jQ+EtI* (7) tL2XeXd Xd

I- Xd = EdEja-EtEJ1
Ed-Et and a radius of
Appendix I jX (8) E[d.
Figure 3 shows approximate reactive and I* EdiaEi (9) 2eXd]_
real power limits on the basis of rated volt- -iXd In Figure 3, the center of circlewDF is at
age at the machine terminals. These limits rE J ~ i 11102
may be readily estimated in the following P+jQ =Ete'rI d IXl (10) (1)2[(.1+.)_1 j=2.045

on the positive kilovolt-ampere axis


Rated Armature Current Limit
RatedArmatureCurrent Multiplying out equation 10 has a radiusreactive
~~~~~~~~~~and of
Th vetrpwritie neuto See Nomenclature for definitions of symbols. (12(11102 294
for the convention that an overexcited *ojuasters adaentotr od.aetyeidiae 2(0.2) (1.1111)

194S8, VOLUME 67 Adams, McClu4re-Unzderexcited Operation of Turbogenerators 525


Appendix 11 Appendix 11 Ed J IIXd Eimaginary axis and radius of andThe
a.

down the
center

of the circle may be shifted up


imaginary axis by the simple addition of
Referring to Figure 11, a circle is de- \ another vector.
scribed by a vector when its scalar magni-
tude remains constant and its angle com- References
pletes 360 degrees. A circle passing through R
the origin with radius a and center on the 1. POWER SYSTEM STABILITY (book), S. B. Crary.
imaginary axis is represented by equation John Wiley and Sons, New York, N. Y., 1945.
22. E\ x 2. STABIL*ITY CHARACTERISTICS OF TUJRBINE GEN-
1/ I ERATORS, C. Concordia, S. B. Crary, J. M. Lyons.
j=aS(20
r =aE( r/2)+j*a (22)
(22)
AIEE TRANSACTIONS, volume 57, December 1938,
pages 732-41.
From the geometry of Figure 12, the scalar
magnitude of r is that shown in equation 23. Figure 12. Vector diagram at pullout 3. STEADY STATE STABILITY
AFFECTED OF SYNCHRONOUS MA-
BY VOLTAGE REGULATOR
CHINES AS
TIONS, volume 63, May 1944, pages AIEE
CHARACTERISTICS, C. Concordia. TRANSAC-
r =2a sin (23) notation eq
vector notatton equation 24 shows two 215-20.
equivalent forms for a circle with its center 4. MODERN EXCITATION SYSTEMS FOR LARGE
The angle subtended at A by the arc OB is at a on the SYNCHRONOUS MACHINES, J. B. McClure, S. I.
26, where 6 is the angle that the radius vector Whittlesey, M. E. Hartman. AIEE TRANSAC-
makes with the real axis. Therefore in r =2a sin OEJ=aiEJ2 /2)+ja (24) TIONS, volume 65, 1946, pages 939-45.

Discussion nonmagnetic end plate starting


through the rotor to its
is used point. If a
a consider-
only by flow of compound
adjacent to the hot fingers.out of insulation
able portion of this flux passes through the To determine the extent of this heating,
end plate into the end laminations of the tests were conducted on a machine which
M. D. Ross (Westinghouse Electric Corpora- core, producing losses in these parts. had identical characteristics as the one which
tion, East Plttsburgh, Pa.): The authors are Where a magnetic end plate is provided, this failed. Thermocouples were inserted in the
to be congratulated on giving a very interest-
ing treatment of stability limits in the under-
flux passes around the stator in the end plate
in a thin skin on the surface of the solid
fingers to obtain actual temperatures. The
data in Table I, giving the test results, show
excited region of operation. The capability iron. It is our practice to use alloy steel the extent of this heating.
curves included in the AIEE standards for which will reduce the losses in the magnetic Not only was the temperature in the
the larger machines were proposed-to the end plate to a reasonable value. In this clamping fingers higher than that of the
committee by McClure and have beRen a case practically no flux enters the lamina- other parts of the machine, but the rate of
valuable addition to thestandards in explain- tions back of the end plate. rise was also greater. The finger tempera-
ing the load limits of the turbine generator. We tested a 25,000 kva hydrogen cooled, ture change followed the field current change
Figure 3 of the present paper expands that 3,600 rpm generator about a year ago at a by a matter of seconds.
diagram to the underexcited region. number of armature current values at zero
In discussing the limitations of the genera- power factor underexcited, and found the
tor so far as heating of the end zones is con- temperature rise of the stator end plates J. W. Batchelor (Westinghouse Electric
cerned the authors, in the second paragraph with zero field and 90 per cent armature cur- Corporation, East Pittsburgh, Pa.): The
under the heading "Generator Operating rent was below 60 degrees centigrade as generator capability curve illustrated in
Limits" would infer that improvements in compared with an American Standards Figure 3 of the paper for which the authors
the design included the use of non-magnetic Association temperature limit for such parts have presented an admirable derivation is
materials for the stator end plates or clamp- of 70 degree centigrade rise. While this quite useful to the designer and the applica-
ing flanges and perhaps for the retaining temperature rise will not necessarily be tion engineer. It is suggested that a deriva-
rings of the rotor. It might be of interest to duplicated in all sizes of machines, we have tion with a different approach might bring
state the experience of the company with to date found no difficulty in building the out some features which would make the
which I am associated in which we have largest machines with end plates and retain- information as useful to others as it has
always used magnetic materials for stator ing rings of magnetic materials. proved to us.
end plates and rotor retaining rings and This derivation is based upon the concept
continue to do so today. of an unsaturated machine, as suggested by
For a given load at leading power factor Matthew S. Schneider (The Cincinnati the authors, in which the no-load field cur-
the flux in the end zones is largely governed Gas and Electric Company, Cincinnati, rent is equal to that of amachinewith satura-
by the long air gap between the rotor and the Ohio): The authors offer a word of caution tion. Thus, the reciprocal of the short cir-
stator end plates in the end of the stator in regard to the operation of machines made cuit ratio equals the equivalent synchronous
core. This flux, which rotates with its prior to 1925 at low excitation. An investi- reactance. If the excitation for the syn-
rotor, passes from one pole of the rotor to gation of the insulation failure in one of these chronous reactance is designated as 1.0, the
the end of the stator, passes circumferen- older machines led to the conclusion that no-load excitation equals the short circuit
tially around the stator, out the opposite side extensive finger heating was the cause. ratio. These two values of excitation may
of the stator to the opposite pole and The magnetic structure mentioned con- then be added at the rated power factor
sists of armature lamination clamping fingers angle to determine rated full load excitation.
of magnetic material held in place by a ring Therefore, it is necessary to know power
Table 1. Data on Tests to Determine Extent of the same material. The fingers are so factor, short-circuit ratio and system react-
of Armature Lamination Finger Heating situated in the machine that they are in the ance at rated kilovolt-ampere, to deter-
flux area of the main field retaining ring, mine the per unit generator capability curve
The amount of flux through the fingers is a as shown in Figure 1 of the discussion. The
Xps ~~~Temperature Rise, function of the saturation of the ring due to curve A CF with center at H and radius equal

bO
B
ZQ
Xo
W
--.
Arms-
.--
Hottest
d-c field current. The greater the field cur-
rent the lower will be the flux in the fingers
to 1.0 represents the full load temperature
limit of the armature. Full load field current
t, ;, X ~ture Field Finger and the less the heating effect produced by may be determined by adding the field cur-
rents for no-load and synchronous reactance
iron losses. The maximum heating occurs
-42.3.. 470. .0.97 Lag.. ..33.7... 50.5... 59.5 where the a-c flux density is greatest, which at the rated power factor angle. This is the
42__._8_5__._9_Lea_______.____.___.__ in the machine that failed is at the center of triangle GHB and since GB is the full load,
Note: Armature temperature by embedded detec- the phase belts. field current, the curve BJ with center at G
tor. Field temperature by change in resistance. This local heating effect will not be ob- represents the maximum loading in the
Machine rated 44,705-kva 0.85-power factor, 13,200 served in readings of the usual embedded region where field temperature is the limit.
volts ' temperature detectors and will be apparent The steady-state stability limit is an arc

.526 Adams, McChsr-Underexcited Operation of Turbogenerators ALEE TRANSACTIONS


starting at G with center along the vertical El El.O E2 a concentrated load area. Th3 other ex-
axis, and with radius equal to SCR/2+ 2 treme of a remote station supplying power
1/2Xe, where X. is the system reactance on 1. l X2 over a high-voltage transmission link is dis-
the machine kilovolt-ampere base. The Z cussed later. Referring to the equivalent
expression for the radius may be obtained C0.6 r--- circuit in Figure 2 of the discussion, all
by considering the condition for pullout at
machine voltage and power factor equal to
z
: |
xz=O,4, Z=0.4+j0
X0.25, z-os+;o
(b)
(c)
generation except unit (1), the machine
being studied, is combined into an equivalent
1.0, and by solving similar right triangles. ______ 1 (d) generator as shown at (2). This assump-
The expression derived by the authors may z /) tion is valid if the load is almost evenly
be reduced to this form. In the case of an divided among machines of similar charac-
infinite bus, the circle has an infinite radius, teristics. The system load is combined also
and the limit reduces to the line GE. >_ 0.4. into a single equivalent load represented by a
Once the capability curve has been estab- z constant impedance Z. An additional step
lished for full load conditions, it may be / X2=0.25, Z=O.63+ j0.31 of paralleling Z and X2gives the equivalent
used for other operating conditions, by keep- 0 x2=0.2, z=0x reactance Xe of the system. In this circuit
ing the center for armature conditions at H 0 the magnitude of the system load and the
and for field conditions at G. For example, W relative loading of connected generating
with a hydrogen cooled turbine generator capacity are not entirely concealed. Upon
nominally rated at 0.5 pound per square 00.24 specifying the terminal voltage and the
inch gauge gas pressure, but operating at 15 Z initial loading and power factor of unit (1),
pounds per square inch gauge the stator the loading of the equivalent generator and
current may be increased 15 per cent and the 3 the excitation voltages El and E2 can be
field current increased 8 per cent. Thus, a X02 FOR ALL CURVES determined. The pull-out power at con-
the radius HC is increased 15 per cent, and Z INITIAL KW LOADING OF stant excitation is obtained from the equa-
radius GB is increased 8 per cent. The UNIT 1)-0.5 PU tion:
intersection of these two circles is on an I_I
extension of the line HB. It should be 0.90 0.95 1.0 0.95 0.90 Mx p -12 s i a EiE2
noted that the per unit scale for field current
is not necessarily the same scale as is used
LEAD
POWER. FACTOR OF INITIAL LO.
LAG M i Zn2l
for kilovolt-amperes since one per unit field
current is determined by the short circuit
Figure 2. Limiting power increment for
generator (1) as a function of initial power fac-
Zni =magnitude of the driving
where:
point impedance viewed from El
ratio. tor at initial kilowatt loading of 0.5 per unit. Z,,=magnitude of the driving point imped-
All impedance and power quantities in per ance viewed from El
J. B. Ward (Purdue University, Lafayette, unit on machine (1) kilovolt-ampere rating =XI+X2Z/(X2+Z)
Ind.): This timely paper will be very useful
to engineers concerned with system operat- clusion that these do not seriously affect Z12 = magnitude of the transfer impedance
ing and planning problems. The curves in accuracy over a wide range of system condi- from E1 to E2
the paper clearly show the influence of vari- tions is not immediately obvious. At least
ous factors upon operating limits at low the range of applicability is not clearly =XI+X2+X1X2/Z
excitation. It is the purpose of this dis- evident in the assumptions themselves, and
cussion to examine more critically the simpli- a planning engineer might wonder, first of 3
fied method of analysis used to derive these all,if hisparticular system and operating con- jo.2 .2
curves. The authors list the assumptions on ditions would fall within the range referred o.46
which the method is based, but the con- to by the authors. In addition, rational 2 O
E.4.6 Z".4+jO.2
decisions on safe operating margins which
1.0A are based on stability calculations require tZ 0.6
some knowledge of limitations and potential =
errors inherent in the method used. w
,0zFI£LD -
In connection with the third assumption X ___ I __ I
vi
|\ of the list, the method of representing the
RAD. GB F| system by an equivalent reactance for sta-
Z
FROM METHOD OF
(a)(a)
0
_ _ _ !_ bility calculations is fairly well known, but ADAMS a McCLURE
. ] perhaps 0 /t is not as well understood. Basi- .4/
cally it is an application of Thevenin's /
__
_ ____ >45s' \ _ theorem, but unfortunately with the pres- IJ.
/ \ ence of speed governors, magnetic satura- KW LOAD ON 0O-0.8
tion, and other influences, power systems do Z 0.3
o not submit to rigorous application of linear L
S
network theory. With most of the react- W KW LOADON( 1.2
/ ANGLE ance residing in the machine to be tested for C) 2 -______
_/RTDP0UF_8 ANGLE
RATED Z 2 _
d__ . {c pull-out limits and a relatively small l
I-H
+ 0.2 0.4 0.6 0 amount in the equivalent system it seems
KILOWATTS-PER UNIT logical to assume that system conditions 3
___ O2I
__ I [ t T 1would not seriously affect results. Never- oL 0.
Al
r r RMT I-IT theless, it is desirable to know how much ) Xe"0.2 FOR ALL CURVES
ollR.* HG/l and in what direction the pull-out limits i- INITIAL KW LOADING OF
041/:
>| / might vary with system conditions. This E UNIT (1)" 0.5 PU l
| c:> i /o | / variation can be demonstrated to a certain - C 90o 0.95 .0 0.95 0.90O
I t^J ;/ < l / extent by numerical examples based on LEAD LAG
____ /____ [ equivalent circuits which represent a POWER FACTOR OF INITIAL LOAD
z I0 </ A slightly less complete reduction of the sys- ON UNIT (I)
< | §/ / ~tern than the single equivalent reactance. FigUre 3. Limiting POWer inCrement fOr
J~ ~~ ____I __I _l As an illustration, the following examples Snr{r()a UC;nO nta Oe
1/1 6 ~STEADY STATE are presented, giving some results for the gnrtr()a ucino nta oe
- E TABILITY LIMIT Irelatively stiff equivalent reactance of 0.2 factor at initial kilowatt loading of 0.5 per
___ oI IVIA.Px. per unit which is discussed in detail in the unit on (1) and with constant kilowatt loading
F ~~~~~~~~~paper
as an average condition. on (2). All impedance and power quantities
Figure 1. Generator rated voltage capability Consider first a metropolitan-type system in per unit on machine (1) kilovolt ampere
curve in which the generating capacity is close to rating

.1948, VOLUME 67 Adams, McClure- Underexcited Operation of Thrbogenrtors :527


all=900-e0l reactance was substituted for inductive tion and what information is needed to de-
reactance to avoid the inherent resistance of tect its presence?
il= impedance angle of Znl the inductive circuits. After adjustment to Under the section Voltage Regulator
the desired initial conditions, machine (1) Operation the following sentence appears:
The curves of Figure 2 of the discussion was advanced to the pull-out limit, with the
show some
(f)
results for an,
kilowat
O initial kilowatt pweravncdtthof -ouandt
power oup (2) and exctaionvotages
output of (2) excitation voltages "When the excitation of a machine is under the con-
loading on machine
loadngmacine
o 1) of 5perunit.
o 0.5per nit.The held constant. In this equivalent circuit
trol of a voltage regulator, adjusted to maintain
rated voltage at the machine terminals, the gen-
values assumed for this machine are: Rated there are many possible combinations of erator will stably accept the maximum turbine
power factor 0.8, SCR 0.9, Xi = 1/SCR, initial reactive power loadings on machines output in any practical case."
initial terminal voltage 1.0 per unit. The (2) and (3) for a given loading on unit (1). I understand this to mean that a generator
various combinations of X2 and Z shown For purposes of comparison the reactive under the control of a voltage regulator can-
with the four curves are chosen to give a power was shifted from (1) to (2) as the not become unstable when operating at rated
value of 0.2 per unit for Xe in each case. power factor of (1) was varied, in order to voltage. (1 per unit.) Suppose the genera-
All quantities are based on the kilovolt- keep the initial line loading and load bus tor were operating at a voltage of say 0.95 or
ampere rating of machine (1). voltage constant. In both curves the 1.05 per unit, would this have any effect on
Curve (a) is obtained by the simplified power factor of (2) was adjusted to approxi- its stability characteristics?
method presented in the paper. For curve mately unity at the 0.9 lagging power fac-
(b) the load is 2.5 per unit, unity power fac- tor condition on unit (1).
tor, resulting in an initial kilowatt loading The kilowatt loading on (2) of 1.2 per unit
of 2.0 per unit on the equivalent machine. represents half-load on the basis that the C. G. Adams and J. B. McClure: The
It X2 represents generators with the same connected machines have the same SCR authors welcome the supporting data and
SCR as unit (1), this corresponds to nearly as unit (1). The maximum power incre- calculations submitted in several of the dis-
full load on the connected capacity. Curve ments are considerably less at laggi ig power cussions. In general, they serve to justify
(c) represents a different system in which the factors than those obtained by the sirnpli fl-J some of the assumptions and emphasize
0.5+jO load impedance corresponds to 2.0 method, but with a smalier discrepan-v limitations on older turbogenerators in the
per unit system load and results in a 1.5 per appearing at leading power factors. As the unlderexcited region.
unit loading on the equivalent machine. load on (2) is increased beyond this valuc the Ross has presented valuable data regard-
This is a loading of approximately one third maximtn-i power incremens for (1) 11 ing zero power factor operation of modern
the connected capacity on the same basis. decrease -.apidly since th system i t <- generators which his company manufac-
Curve (d) is for the same value of X2 as proachiilg .-he steady-stat stiljt. . tures. With regard to detailed generator
curve (c) but with a system load at 0.9 For 0.8 per unit kilowatt lcad on (2), a less design to avoid serious stator end heating,
power factor lagging. Note that all three severe condition, the curve conformn more there are several design approaches. As
curves lie above (a), indicating approxi- closely with that of the simplified nmethod, well as the one outlined by Ross, the use of
mately the extent by which the simplified particularly in the leading power factor nonmagnetic end materials and step backs
method described by Adams and McClure region. in the stator core is successful.
gives conservative results for these system For the conditions considered before, the Schneider's data clearly show that serious
conditions. single equivalent reactance representation heating can develop at leading power fac-
Next consider a system of the type illus- yields sufficiently accurate results for most tors in older machines. Our tests also show
trated in Figure 3 of this discussion. The practical purposes at unity and leading that the finger temperatures rapidly follow
machines represented by (l) and (2) supply power factors, which is the area of primary changes in loading. When underexcited
power to a remote loadl center (Z) over a interest in the present paper. The simpli- operation is contemplated on older machines
transmission system, and the generators at fled method probably should be applied in service, field tests provide the best data on
and beyond the load ceniter are combined with caution to initial loadings in the lagging which to base future operation.
into an equivalent machine (3),. Unit (1) region when long, heavily loaded transmis- Ward's discussion supports the assump-
represents the machine to be tested for pull- sion systems are involved. The applica- tion of a generator equivalent reactance
out limits and (2) the combined effect of bility to other operating conditions and equal to the reciprocal of its short circuit
nearby machines. If unit (1) is a 50-mega- types of systems can be checked in a similar ratio, especially at leading power factors.
watt, 0.8 power factor machine, for example, manner. Obviously the purpose of this assumption is
the line reactance of 0.2 per unit corresponds to simplify calculations while retaining
to approximately 80 miles of double-circuit reasonable accuracy. In actual operation,
132-kv line with transformer banks at each W. H. Osterle (West Penn Power Company, borderline cases should be approached only
end. For simplicity resistance is neglected Pittsburgh, Pa.): As I understand this when using automatic control of excitation.
except in the load impedance. This system, paper, the authors have taken the mathe- The use of modern voltage regulators
viewed from the terminals of unit (1), again matical equation Ed Es/Xd + Xe and for dif - eliminates the significance of incremental
reduces to an equivalent reactance Xe of ferent loads and various power factors with power limits established for manual control
0.2 per unit. Ratings of unit (1) are the different values of external impedance (Xe) of excitation.
same as in Figure 2. calculated the pull-out power, using Xd, the We refer Osterle to data presented by
Ctirve (a), obtained by the method of the reciprocal of the short-circuit ratio, and Et, other discussers and to Figure 6 in the paper
paper, is included for comparison. The terminal voltage of the machine, as unity. for the effect of operating at other than rated
other curves show the maximum power Have the authors any experimental data or voltage. Voltage regulator control will in-
increments which can be picked up by unit results of any actual cases of generator static crease system stability for steady-state load
(1) from an initial kilowatt loading of 0.5 instability that can be plotted on any of these changes.
per unit with two different constant kilowatt curves to see how near actual cases compare? Batchelor presents another method of
loadings on (2). These results were ob- Will the authors elaborate a little on the arriving at Figure 3. In the use of this
tained by means of an a-c network calcula- subject of "end structure heating with low figure, the assumption of rated generator
tor. In the calculator set-up capacitive field current?" How serious is this condi- voltage should be borne in mind.

528 Adams, McClure -Underexscied Operation of Turbogenerators AJEE TRANSACTIONS

You might also like