Calhoun: The NPS Institutional Archive

Theses and Dissertations Thesis Collection

1952-05

The experimental determination of the performance

of a capacitor-excited induction generator with an

inductive reactance in series with the load

Goode, Richard William.

Massachusetts Institute of Technology

http://hdl.handle.net/10945/24719

Library

U. S. Naval Postgraduate Schoo]Monterey, California

raE EXPERIMENT-^ DETEPiAINATIQ^' OF THE PERFORMAMCE OF A

CAPACITCK-EXCITED IIDUCTiaM GEMFilUTOl WITH AN IIDUCTIVE REACTANCE

IN SERIES WITH TI-iE LOAD

by

Richard William Qoode, Lieutenant, U. S, Coast GuardB. S. , U, S, Coast Guard Academy, 1944

Henry Acker Hoffmann, Lieutenant Junior Grade, U, S. NavyB, S,, U, S, Naval Academy, 1947

Willard Franklyn Searle, Jr, , Lieutenant, li. S, NavyB. S«, U, S. Naval Academy, 1945

Submitted in Partial Fulfillment

of the Requirements for the

Degree of Naval Engineer

from the

Massachusetts Institute of Technology

1952

CAf^^CITCB-DCITHD INDUCTION C^MERAXm WITH m im^K^riw ^ !^..CTAI^C;E

m smiEs MUH im wmby

nd^hard WlXliiim Good®, ^i^>jimnMit, U, •.-• Coast Q^j^3:ii

Uimxy A£k%x- Hsffmann, Lieutenant Junior C^rade, U, ?. KtvyV5»iiiar4 Fr&nklyn S«arie, Jr., Lieul^rt^ntj U, 5. !^avy

^'uia'aXti.«&<ii in r«:uaial ;-yitxlimentof th« fleqytirtm^fits for th«

Ds^r-^e Oi U»ivai '-nglfHi^rfro® th«

I9t.2

a^cont developis#ntft in thft fit la ox pavaer capacitors hav®X«ad to r«n«w»4 Activity in theoretical &rui experi0ientalinvestigations oi: the &elf;-9xclt^c ifiductior* generator v»lthparticuiajt consideration being given to high-frec^ei^cyappiicaiioFia, The 9«r^«rclii'«g anil using shurst capacitorsto provide excitation current ha© not presented satisfactoryvoil*;^e rdgul^: tioA cJ'«ara€ terisiics. /-.h tne loaa posa^er

factor becomes more iagging^ the volxao© regulation b®c<Rs©e -

ir*cr«asin-^dY poor.

In an dtte£f^)£ to iitsprov^f ih^ voita^^- r«';^ui*2.i.iOi; ti*« ir«duccion^«n«rator h«9 been investigated with cots^ene^tion attained bytlm us« of additionai ca^iicit-anco iit series with the io«d.This operating technic^e causes po©sibi« ios* oi- excitation©vex* a c&xl€,lti JCMiq^ o.' pov^r a@a«iid ai i^iv^ginc-f power factors.Theoretical studies indicate that, if coiapeneation *s«re«ccomj:>ii$ii*»d by -.isiiivg iia indue ti-/^, ^'ath@r vhar^ capacitive,reactaiKe in eerie$ with the load, generation co^id be main-t-ainsd coiitinao-asiy Tor ali lo<id dssiaxsds at any iaggin^ p&msfactor*

Eji^erimentcttion in this theeis vgork verifies the«e facts.Furiijsr, volta-je- rejul^itiou can b® Lt^^roved by either d^s-

creating ©eries inductance or increafiin^ shtint c&pacitunce.O^'ssration -^ili Lt? Gontinaous itQm n© load to Jaaximujis powigroutput.

Ko excessive currents or voltages occur in the generatingeqaiptsfitnt or in th« iine as ^ r^^uit oi transients foiiowingfstults. Frequency variations^ howeve:i; exist after suddenload ch^nQss. Ails proiil^m, && '-^11 iis that Oi voit.4g«rec|ulation, may weii be handied by t-he aopiication of'automatic control ©q[uij>aent.

nvviThesis Supervisor: Alexander Kusko

, « .

Title: Associate Professor of Electrical Engineering

in \i«

I'v/f,.

Cambridge, A^assachu settsMay 16, 1952

Secretary of the FacultyMassachusetts Institute of TechnologyCambridge, ivlassachu setts

Dear Sir:

In accordance with the requirements for the Degree

of Naval Engineer, we submit herewith a thesis en-

titled "The Experimental Determination of the Per-

formance of a Capacitor-Excited Induction Generator

with an Inductive Reactance in Series with the Load"

Respectfully,

Richard W, GoodeLieutenant, U.S.C.G.

Henry A, HoffmannLieutenant j.g,, U«S,H,

VVillard F, Cearle, Jr*Lieutenant, U,S.N«

a.l:iO^L''1C»o2c sO

Y'^^l-

^IcHfemticH .A ytnm^

^:^n::'r\-^ :.i'lw-

The authors wish to express their appreciation

to Professor Alexander Kusko of M. !• T, and to

Dr. James B, Friauf of the Bureau of Ships, Depart-

ment of the Navy, for their advice and encouragement.

lomj^i^ti

TAPLg Qf QQNTPNTS

Page

I. INTRCDUCTION 1

II. PROCEDURE « . . . • 3

III. RESULTS AND DISCUSSION OF RESULTS 4

A, Steady State Characteristics 4

B. Transient Studies 18

IV. CONCLUSIONS 33

V. RECOMMENDATIONS 36

VI. APPENDIX 33

A. Detailed Procedure 39

B. Data b2

Co Bibliography 58

The advantages of the induction generator, stemming

from the squirrel-cage rotor construction, are known.

Power companies have instail<5d induction generators as

supplementary pov/er sources. In these installations,

magnetizing current is supplied by the synchronous machinery

of the system to iviiich it is connected. The inherent rugged-

ness, low maintenance, high speed possibilities, and low

cost of construction of an induction generator could be

enjoyed to a greater extent if the machine could operate

satisfactorily as an independent source. Independence can

be accoiTplished by the use of static capa-citors as a source

of magnetizing current. However, experimental and theore-

tical analyses included in the literature lead to the

conclusion that until satisfactory schemes of voltage

regulation are developed, the induction generator will

not be acceptable as an independent source of electrical

power.

Moreover, if the induction generator is compensated

with capacitance inserted in series in the line, power

generation will be lost for certain lagging power factor

loads. Inasmuch as improvement of the voltage regulation

may be accomplished by some form of series compounding,

it seems logical to investigate the effect of inductive

reactance on power continuity. Examination of the theory

reveals that if inductive reactance v/ere used, voltage

mix:.

>Err^ s?.''OnC'T.d')fif^ p^

isoi^i^aeie ^o a^^woe tn-

lOi^OB^ i^ewao i^Pl:

111 X&9«*

compensation will result and power generation is assured

for ail lagging power factor loads.

The purpose of this thesis is to exjT'erimentaily in-

vestigate the characteristics of a self-excited induction

generator using an inductive reactance in series with the

load, steady state perforrriance is obser\'ed for unity,

leading, and lagging power factors and for variations in

shunt capacitance and series inductance. Transient

behavior is studied under the conditions of short circuits,

suddenly applied loads, and suddenly removed loads.

-U4?9i IIlw noLtisntarriOO

lit p[ipgi;PVfi??

Prior to conducting the experimental v/oxic, the per-

fonnanco of the induction generating unit was analyzed

Velu.es of capacitive and inductive reactance v;cre estimated

for use in conducting the experimental v/ork. In the labora-

tory, the generator %ves operated at constant frequency.

Combinations of resistors, capacitors, and inductors were

used to provide the desired load magnitude and power factor.

Load voltage and current, generated voltage and current,

power factor, power input, and power generated v^ere recorded

to determine the steady state characteristics of the generat-

ing unit. The efficiency of the generator v/as determined for

all operating conditions.

An oscillograph was used to record the voltage and

current transients occurring under single*- and three-phase

short circuits, suddenly applied loads, and step unloading.

These transients were observed at bath the load and machine

terminals.

A detailed discussion of this procedure appears in the

Appendix, Included also is a brief reviev^f of the perfomiance

calculation technique en^Dloyed,

Superscripts refer to references as listed in theBibliography,

T!03

b9b-

?n.ttG-:;c:7C llB

Ait ^X^^^Yu 3%^%^ Qh^r,^ptgri^1;?-c^

The analytical determination of the performance of an

induction machine when driven above synchronous speed has

been covered by Dr. Friauf-^ and verified by Svvift^ for

both simple capacitor-excitation end for capacitor-excita-

tion with capacitance ccinpounding. Estas and Hussong^

extended this v/ork to parallel operation but without cori-

pounding. In all three of the above, it vms noted that

for certain lagging power factor loads the excitation of

the generator, and hence its voltage, was lost v/hile the

power v/as building up; and further, that this condition

became more severe as the lag angle increased, (This

phenomena is explained in the Details of Procedure in

the Appendix,

}

In order to verify the anticipated results that there

would be no lagging power factor excitation discontinuity

for an inductively compounded generator, values of X^ and

Xj_^ were chosen, using the method of Friauf-^ as explained

in the Appendix, to give a fairly high value of open-circuit

voltage and a value of short circuit reactance only slightly

belov/ the excitation-liniting condition. Thus continuous

excitation for the unity power factor condition v/as assured,

A value of 13,5 ohms for X^ and 10,25 for X|^ was used for

the family of curves shown in Figures 1, 2 and 3,

obni/oqfaoo xi9yilsoiSL

:.-,. -.> ...^ -.:^.?aK»

,„v A

vT

'

1

'

Q<

a; O(f)

-1 yLjOq

^ .1

(£>LlI *>"*

Q og< ~ „- L- jx"

< IS'^- ^ o ' oLlI

—1

1^

ff5 o „X

6 **^

\^,^" /W'^

3 -o o> I <^ ^

\^^^^^ ^\ p (_ < •« L o

\y. \ oo 2 ro ^x crN 3=3M / NX O Q •- \ / o\ " UJ

)s-\ \ y o\ —^ X

CE

\x-—^

CO

>^^o \ y^ X$^^^s ^^^^^

1^ // //)^^ ^

fo // / /yy^\. q: -^/

^ /3 y

//A^

/ \

\Li- CD 1

\ f^

\

:

'—

\

\\

d UJ '

^-1\ Ad CD CD CD CD

\ XI o

\ Z5/-z.

CD

OCD

Z ZL

CD CDCD CD

•z.

CDCD

1

r\ 1y / < < < < < Ll.

\ Q- E / _l _l —1 _l _l _l \

- V(7^00

d

00CD

d

ro roID ro

d dLO

d

DQ<

\i

.rz3

^i\tr 1 1 1 1 1

o o

2 q: UJ-. < Q:

V > 3

(3Nn 01 3Nn) SinoA-39VinoA avon

"

/

r ^1

Q<O

-

"

/ \ aC/1

^

/\ o UJ ->"-!

f \

-/

\ O Q i 1^

/ \ cr

QH 00

I¥ ix^ Ix'

1

/ o O

:l \<ITUl

UJ

oQ.

o

>

o

6

6

X

X

-

S^Ao \ z UJ I cj

z \ o > <

:<UJ_l

z

< vluQ

5

i -1z

)^

in _l \ \i^ QC \ \6 CD

200 ,

/ o

\ \v~ 2 5

<

6 /

CD !j \ \-'

LiJ

o_j

CO I

<-I ro

\>

\ zCL c3

1lO O^ 'S \ (D U

\>- ^/^ K \ O _J

- 1-/ V\\^

< GD

:1

3

•/V ^y^^^ \\\

ID (/)

=5\ / / >v. \\ cr

-A / / / > \ o

"\ // \^ ~-^-A\^ 1—

> cr

-z \ j1 / / ^ ^\

k,^ V \\ ^ -

- \ ///^^^^^ A, V\\s

-"^\ ////>

^^^ ""^^^,,\

^\\ -

\'

\luC^ vs^- \ \i^ .>^ \

^^^^^Wi^r 1 1

1

1

^^^"^f^

o a: o ^o UJ - orir^ > cr 3^ o < o

'^ ^[I

o 9 9^o ^ o>< O LL

(3Nn-0i-3Nn) sinoA - 39vinoA avon

2 4 6 8 10 12 14

LOAD CURRENT-AMPERES (LINE CURRENT TO LOAD)

GENERATOR CURRENT VS LOAD CURRENT FOR VARIOUSLOAD POWER FACTORS

FIGURE 3

It is shovni in these figures that a capacitor excited,

inductively compounded induction generator will hav© con-

tinuous excitation from open circuit to short circuit, and

hence continuous voltage and power characteristics, for ail

lagging load power factors, provided of course that the

excitation is continuous at unity power factor. It is

further shoivn in Figures 1, 2, and 3 that for leading powsr

factors a point may be reached at vrxich the magnetizing

inpedance is insufficient to give excitation current, and

consequently the voltage and power will collapse. Mote

that in Figures 1 and 3 the leading power factor curve is

discontinuous at the origin. This is the condition of no

excitation. The theory leads to the fact that if the para-

meters are such as to permit a sustained short circuit,

then generation should reoccur at some lower value of load

impedance. It v/as found that the generator was very unstable

in the region approaching loss of excitation and tended to

drift toward no excitation at load voltages of about 60,

It was also impossible to obtain re-excitation after passing

continuously through the region of no excitation. The same

condition was obtained, however, by O'/erspeeding the

generator, exciting at a higher freqijency and hence a value

of X^ below critical (see magnetising curve. Figure A-3)

and then slowing to 60 cp.s. On Figures 1 and 3 the path

of the curve for leading power factor between the point of

re-excitation and short circuit is estimated and shown

dotted.

-^-» f^vsfl iX^ ^o^s-x^an^ noid^Subnl b^bnuoqpioo Yi»vld^oi;bfil

9*0W .a«qi5il03 Xilw Tswoq ,

«X dVTua t<i^9a% :t9y?oc, gnibjt^i. 31! t C bnz X ^dio'eiH nl ^tsrf^

on ^o fioi:r gi slrfl .niglio »rii :^» tuounl^nooalb

-siaq 9ti^ Iti ... .... ad;)' o.t sb&Bi scioedS d/ft .n«i:|-s;fl0x«

• "" «9onBb«qwl

oi^llc^ldx* on bccfiwo^ d'flicb

erne- .. .-on

} Xndi^i^o woXsd ^X %•

— - -- ^......,,.. ...-. j.o od o:f pnlwoXs n»ri^ bus

*betd-ob

It is raost important to note here that the loss of

excitation occurred well after the power peak and not

during the power build-up as was the case described by

Swift for the capacitance-compounded generator. This

is, of course, because the critical value of magnetizing

reactance as seen by the machine terminals themselves

(inside X^. and Xv^) does not drop to the critical value

until well after the peak conditions. Ttiis is explained

in detail in the Appendix,

Figures 1 and 2 indicate the fact that the vol cage

regulation becomes very poor and the peak power available

drops off sharply as the power factor loads beconie increas-

ingly lagging.

Figure 3 brings out the fact that for unity or lagging

power factors the highest value of the generator current,

Ig, occurs at the no load condition, (it thus came as no

surprise that an easy way to cool off the test machine was

to load it doivn. ) It was for this reason that no attempt

was made at interpreting the results as per-unit values.

Conventional rating methods do not apply to a generator of

this type inasmuch as the ratio of the KVA rating of the

machine proper to the corresponding KVA rating of the output

terminals is so large. This is due to the unavoidable fact

that a large part of the generator current must go through

Xq, reducing Ij-^ relative to Ig, notwithstanding the phase

angle consideration.

lo

A cross curve of peak power is shovm in Figure 3, This

has no significance other than perhaps predicting the load

current and the I^ to Ig ratio at peak power for some other

power factors. It is worthy of note that as the pov>fer

factor increases from highly lagging to unity to leading,

the ratio of I^ to I„ increases. This fact of itself would

lead us to the conclusion that since more of the rated current

of the generator proper is getting to the load, the machine is

operating more efficiently at the higher power factors. Table I

shov/s the approximate value of the efficiency of each run at

the maximum power point. The values were arrived at by sub-

tracting the d-c drive motor losses from the d-c input power

and calculating the induction generator efficiency by using

the shaft input power and wattmeter readings at the load.

No account was taken of meter losses or the losses occurring

in the variac.

Efficiency at Maximum Power Conditionj;Qr V^i:l0M,$ l9^^, Poy/ffy, j-.a^1;p;r^

X^ = 13.::; \ = 10.2b

0.15 lag 16.6

0.33 33,6

0,53 40,5

0.68 40.5

0.80 b6.^

1,00 unity 63.6

O.Tj lead 68,0

JO iiif% i^#pa

J :j 'tjtto J» A

l^-^Cp ^-.'l? %o

The voltage waveforra for the generator unit v^as observed

on an oscilloscope connected across the load during all the

runs involved in Figure 1 and was found to be a pure sinusoid

at all loadings. An example for 0,8 lagging power factor,

at pesk load is shov/n in Figure 4. This confirms for the

compounded generator the results described by Hstes and

Hussong^ for the uncompounded esse. This may also be observed

on the oscilloc3rarn8 of transient behavior included in the

following section of this chapter.

In order to extenoL the results described above, a series

of runs, all at unity power factor but with various Xl and Xc,

have been plotted in Figures 5, 6, and 7, On each figure the

center curve (X^ == 10,25, X^ » 13. o) is a duolication of the

unity power factor curve of Figures 1, 2, and 3 respectively,

A decrease in the value of Xj relative to that of the

first set of figures v/hile holding X^- constant (sane open~

circuit voltage) gives perhaps the most important result

obtained. Figures 5 and 6 show that for a given power output

the voltage regulation frori no load to a given load is con-

siderably improved. For exan>ple, on Figure 5 at a 2,5 VJi

load, the improvement is from 18,5 percent to S,9 percent.

IVith an X^ of 3,7 the generator would undoubtedly be operated

at a higher power v/herc its regulation would drop off. It is

seen in Figure 5 that the voltage characteristic is cpi'-e

linear until about 3,5 KVV. It is further noted that the

peak power available is considerably greater as Xr^ is decreased.

.vL:^

iua.

, o&ol

098

FIGURE 4

Voltage Waveform

/5

o oc

(3Nin 01 3Nn) siiOA- 39vinoA avoi

f^

(3Nn 01 3Nn) siiOA-39vinoA avon

40

24

~1\

r-

;^OPEN CIRCUIT

"1\ \ r

INDUCTION GENERATOR -60 C.RS.

INDUCTIVELY COMPOUNDED

MIT MACH NO. 704-8 POLES

r, F N /X,_ V

J -\JUUU

/V^LOSING/y EXCITATION

,/^SHORT CIRCUIT,

J_ _L'0 2 4 6 8 10 12 14

LOAD CURRENT- AMPERES (LINE CURRENT TO LOAD)

GENERATOR CURRENT VS LOAD CURRENT FOR UNITY

LOAD POWER FACTOR WITH VARIOUS PARAMETERSFIGURE 7

16-

This all has been achieved by accepting a discontinuity

in the excitation at the Xj^ = 3,7 condition, Tlie value

of the short-circuit magnetizing reactance is such that

it is above the critical value as shown on the maqnetizing

curve, aeferring to Figure A--b(a), the short-circuit con-

dition is below the limitinq value of susceptance. As was

noted previously, the point of initial loss of excitation

is well beyond the peak power point, therefore, at a point

v/here operation is unlikely. dso, and of equal importance,

is the fact that lagging pov/er factor loads iwuld not cause

loss of excitation at any earlier relative value of power.

Figure 7 shows that the two curves at Xj. =* 3,7 \i^^m

no sustained short* circuit current, v^vereas the condition

of the runs in Figures 1, 2, and 3 permits a sustained

short-circuit of 9 amperes. This is borne out by the

discussion in the preceding paragraph and will be mentioned

at greater length in the section of this chapter on transients.

The results shown on Figure 7 indicate that for a given

generator rating the ratio of load current to generator

current is greater as the value of X, is decreased. This

of itself -.^ouid again lead otiq to suspect better operating

efficiency for the set with a lower Xj . At any rate, it

would be expected that the generator-set rating (at the

terminal beyond Xj ) would be closer to the required KVA

rating of the machine proper. Kote in Figure 7, as in

Figure 3, that the rating of the machine needs to be much

^\

-nor. ,• ; or.' •^niTTs'iS'" .t^vrya

:?nioq 6 J^ft ,»xo^#:i«/i* ,J'nir>-' -':^-"^.- '6«a ©rfcf bnoYsd Haw •!

©3U53 :'On bi oq ^alppL

©Y«if{ V.e - . - j(U idfl^J- swofia V aiL/plH

bsnoi^nera »g LUm bn;- df;c::r ci/^c r^'roc^^-ra arii nl nol3«L;:>tifa

.a^neiinsTcJ- no ^laiqs . dj-^ndl isieoip ^s

^aa 3iiJ' Tol ;:.-ii3xc.'m9

17-

higher than the rating of the generator set. The run at

the lower value of Xt was indeed found to be more efficient

as noted in Table II,

Efficiency at Maximum Povwr Conditions

X|^ X^

10.25 13.5 63,6

3.7 13.5 68.3

3.7 22.8 Ci^.G

The third curve on Figures b, 6, and 7, namely, Xr = 3.7

and Xj. = 22.8, was run as a verification for the inductively

compounded generator of the results predicted by Priauf

and checked experimentally by Swift*^. Tne open circuit

voltage is found to be much lower as the value of X^ approaches

the critical X^ as shovrn on the magnetizing curve. As a

conseqijence, the voltage regulation is very poor and the

peak power available drops off rapidly. Th^ set itself

operates efficiently on an output-over-input basis. There

is nonetheless a gross inefficiency in the use of the machine

size involved.

To briefly summarize the important steady state results

of the conditions investigated:

1. Continuous generation for all lagging power

factors is assured if excitation is contin-

uous for unity power factor.

YX»vJlj-dL'tM-

13.

2. In cases of leading or lagging power factor,

where discontinuity of excitation exists, the

discontinuity occurs after the peak power

condition,

3. The voltage regulation and efficiency drop

oft rapidly for the more lagging power factor

conditions,

4« Voltage generated is £ purs sine v/e-ive.

5, Reduction of X- , v/ith X^ held constant,

effectively improves voltage regulation

and peak power,

6, KVA rating of trie machine proper must be

larger than the KVA rating at the terminals

of the set and must be based on no-load

(open circuit) conditions which are the

most severe,

ryansie,

nt Studies

Sudden interruption of steady-state operating conditions

in an electrical system may cause abnormal transient values

of current and potential. Disastrous peak values of voltage

surge or current in-rush may instantaneously exist even though

safe, steady-state conditions prevail before and after the

change, .Uso, th© effects of abrupt faults, if not localized,

may cause an entire systen to become unstable. Generally, in

an electrical system being supplied with induction generator

power no intolerable transients occur, though the generator

may become unstable and cease generation of voltage^.

a. ) Dirge -rai; s5. Ch oy.t-Circuit - The transients at the

load texTninals and ct the generator terminals resulting from

a three-phase short-circuit are illustrated by the oscillo-

grams of Fiourc 3. The value used for shunt capacitance

in all transient studies was one -/j^iich provided an open-

circuit voltsge of SOT. volts. With the inductor selected,

a reasonable pov/er output of about 4,v8 KW v/3S obts';in<?d.

The steady-state frequency before application of the short

was 60 c,p*s. Thus, the initial portion of each trace rr.ay

be referred to as a time scale.

Currents and voltages of only one phase are shown

because balanced conditions existed. Figure 8-a shows

current and voltage at the load terminals; Figure 8-b,

the same at the generator terr^.inals, Tlie shorted load

was at unity pov/er factor.

Tlie load voltage, Br of 8-a, obviously must change

iiTsmediately to zero. There is an immediate peak of current,

,'-, of about tv/o and one-half times the initial current

peaks. The current wave fo-rrn is enclosed in an exponen-

tially decreasing envelope. In approximately one-half

second tlie short-circuit current is practically zero.

Tiis tine constant involved here is determined by

the impedance of the equivalent circuit as seen looking

back from the shorted terminals. Because four energy

storage elements are involved j there v«'ill be four roots

to the characteristic equation describing the impedance.

No decaying direct-current component of line current can

, J'nork'ii.i

:)l&i

ao.

rMTT-'ifnMrirrT "• .ra

W\AA

y y \

\ f\l\i\i\f\^^

-a. LINE TRANSIENTS

^^\f\/\/\/\/\/\A^^^/^^C\^^^'^^^

8-b, GENERATOR TRANSIENTS

c 196 ^f

L 9.8 m h.

I.M. MIT ' 704IND. MACH

TRANSIENTS FOLLOWING A THREE PHASE SHORT-CIRCUIT

FIGURE 8

be distingviished, "or only about 0.1 seconds did fhe

transitory line current have an effective value greater

than that of the initial line current.

After the load v/as shorted out, the serle-^ inductance

was effectively plc»ced in p-irallel with the shunt capacitance.

This combination anpe.-»rs to th© induction raachine as an

Inductive reactance of 5.1 ohms. Thus, excitation current

cannot be supplied ^nd the machine fails to •generate.

Referring to Figur® 8~b, it is seen that the stator

current, i , and voltage, e.,, both ao to sero without9 "

sj '

surging, and furthermore, that the response times are the

same as that for the short circuit current in the line*

It can be concluded that no daraage to this generating

unit will result because of a three-phase short-circuit.

However, it is significant to note that unless a short-

circuited branch can be re:rioved from the system, vdthin a

fraction of a second, all pow^r service will be lost.

It should be pointed out that In the trace of machine

current, i,, there aooears an Initial surge oi dlrect-

current, followed by an exponential decay to zero. As a

result, th^ entire exponential envelope of the sinusoldally-

varying component is offset above the zero level. Th^ offset

phenomena depends upon the instant of transient introduction*

The rate at v^iich the offset decreases is considerably greater

than the rat© at ''v^ilch the envelop* decays.

AS

r^Sui W^H v"t0.^1*»n«Td^

XlfrlOtibfti 9ftlt9t

i^al'';

brii

aa.

fe>,t )-, „§^afnX^-|?i\ft.^? gbp„yt;„~Q;iyff,\i\'i; - of the two, the

fault more commonly encountered is the single-phase, rather

than a three-phase short-circuit. The transient effects

of a single-phase short-circuit, as exhibited at the load

terminals and machine terminals, are illustrated in the

oscillograms of Figure 9-a and Figure 9-b, respectively.

It was necessary to show traces for only phase B and either

phase A or phase C, However, all three phases were trace-

recorded and, obviously, the conditions of phase A were

repeated in phase C. Actually, the pure resistance load

was adjusted so that voltages in all three phases were

closely balanced. Some unbalance is indicated in both

oscillograms because the sensitive elements of the oscillo-

graph did not permit fine adjustment of galvanometer sensi-

tivity.

As in the previous case, with voltage, v^^^, across

the shorted phase going instantly to zero, the currents,

ij^g and i|^c» ^^ ^^^ shorted lines initially surge and

thereafter exponentially decline to zero, Movi/, however,

the rate of decay is nearly one-half that for the three-

phase case. The maximum instantaneous current carried by

the shorted lines is about tvfice the steady-state current

peak. In the third line, the current never exceeds the

steady state value, since neither end of the shorting

cable is connected to phase B*

With voltage, v , equal to zero, the instantaneous

voltages, v^v^ and v^^^ , must remain equal in magnitude

and separated by a 180-degree phase angle, A "ghost"

s&

•-• X j ^. oa

9bu ••

^3.

9-a LINE TRANSIENTS

I V V V V;*y^/^^\\V/vV\MW^\VM\\^^

9-b GENERATOR TRANSIENTS

{^ • y ^'ginr^'

^g^— ^MT^

(Q 1 1 ^innr^

C = 196 ^f

L = 9.8 mh

,M.= MIT *704IND. MACH.

TRANSIENTS FOLLOWING A SINGLE PHASE SHORT-CIRCUIT

FIGURE 9

^4r

trace is apparent in the center of the upper oscillogram.

Thlg is attributed to a stray reflection from one of the

potential - recording galvanometers.

The oscillogram of Figure 9-b shows transients at the

machine terminals and is similar in pattern to the corres-

ponding oscillogram o£ Figure 8~b for the three-phase short

circuit. Two noteworthy differences are found, howeve.-%

The first is manifested by the doubled time of decay, in

agreement with conditions in the line. A possible

explanation of this may be found in considerimj the

transient voltages across the three shunt capacitors.

VUth the three phase short circuit, these capacitors are

each shunted through a series inductance. The voltage

across the capacitors diminishes rapidly to zero. On

the other hand, in the instance of single-phase shorting,

only one capacitor is shunted through series inductance.

It would be expected that the voltages across the other

two capacitors would have a longer transient period. As

long as these voltages are sustained, the induction machine

will receive excitation current. It follows that power

generation will not completely end as quickly as when

all three phases are shorted.

A second major difference between the oscillograms of

Figure 9-b and Figure S-b is evident in the trace for

current, i . , of the latter. It was pointed out previously

that the line currents, 1^^^ and i^^^, momentarily surged to

:tiorf« ©edfiq-««'Si11 e^^ " io jng-XDoXXiofiO on.tbncq

•TCB »to*l9&q«o «»8»fi;r ^jxwsuia i-ioiia iissi'iq 5*©ai-;j :iu^ iiilW

se»;tXov ariT .a^^naioufefli ««Xxa8 6 ri^uoid;^ b«:tfiuff« rioat

nO .oi«s oi vXbiqkJi aerlalniiaib siOiTXosqiJO aii^ aaoioa

^pnijaorfB sajsriq-aXgnie "io aonsj-anx ad^ nx ,bn6d -xadlo arfi

,oofts,1Dub;-ii gaxrioa fipuc:td;t b©jm;ds 8 J', ^olxosqao ©no Y-t«o

2»ri^o adi 880T0JB eai?6iXov ^rf.* isdi bdiJ-^aqx© ocl bXkJow tl

aA .boi^raq fn^k^nhi^ 'x§; : iiXuow 3*io:txDeq6a cmi

^(li^^^iR rtol^o: -aalB^aua aiB sagfcJ-Xov saari:}- ei pnoX

«c9vvo: lol .rl .Jnairsua rioXcJ-cdiaxo 9vx93s^ XXlw

ao :ijp 2s bna vXa^tsX^fnoo j-on XXxv/ aoXts^sanac

.b9J"Tori2 9T6 aaaftdq eoidd' Xla

^o ams^©oXXi33o sdi neawi^ad ddflaTte^lib -xotBcr bnooaa A

ic\ BO&ti 9iii ni ictsbiy^ aX (<-8 aiuielH bns d-9 aaugl^

V.r.'?J0*v-3ic- .n-0 bsJ-rrXoq ^«w .tl .•:!:*>.tt6l ad.t \o ,^^,.1 ,J-naT:5U3

higher-than-normal values. To accommodate the cuirrent surges

in the two phases, A and C, current is raost readily drawn

from phase B through the inter-connecting capacitors.

This causes a sudden compensating drop in machine terminal

current for the Q-phase, as is shown in Figure 9~b.

Povifer is eventually lost because, as explained in the

discussion of a three-phase short circuit, the reactance

seen by the induction machine is not sufficiently capaci-

tive to furnish the necessary excitation current.

S t }, J^p^MPt Jlff,n, MfftPy., § ,t^ ,t4M - To simulate a varying

load demand on this power generating unit, several trials

were made wherein the system load was suddenly increased

or decreased. This section investigates the transient

effects of suddenly starting an induction motor. Oscillo-

grams obtained include those of figures 10 and 11.

Figure 10 carried traces of line current, ir , and

voltage, v?^, and machine current, iq, and voltage, v ,

For the trial illustrated, there was a prior balanced load

on the generator of about 4.4 KW before starting the induc-

tion motor. The induction motor v/as started and run with a

eontinuous shaft load equivalent to about 700 watts. From

Figure 10-a it is seen that the line voltage initially

dipped in agreement with comments elsewhere concerning

voltage regulation. Once the inertia of the motor and

connected load had been overcome, the line voltage increased

to almost the original value. The line current during this

j^^^jjUa oi

Mil ''"•"lli"lillil'l"lMlll|l|ll|ll|l|llllMftl MIlllMIIIIIIIMIIIHIillMIIIIIMI

III! Ill IIJilllllllllllllillJllllllllllHlllilllllillllllllllllllllllHIlMli

'""" MM^|((||p|||(p)l!(lliii«'|IIIIIIHMIIIIIIII||ll|llllllllll|||

lO-a. LINE TRANSIENTS

lO-b. GENERATOR TRANSIENTS

\Y'^

\ *^'= =

C = 196/if.

L= 9. 8 mh.

I.M.= M IT *704IND. MACH

MOT = INDUCTION MOTORI HP. 220V 3(i

TRANSIENTS FOLLOWING INDUCTION MOTOR STARTINGWITH LOADED GENERATOR

FIGURE 10

2,7

^(itHHmHmmmmmmmmHHMmHHmmmHHmHmmmHmmH

fcilllliyiilllHlllllii, iinntn.ii.>i...u

ll-aLiNE TRANSIENTS-LIGHTLY LOADED INDUCTION MOTOR

'^^^l^nmmm^^

II -b LINE TRANSIENTS -HEAVILY LOADED INDUCTION MOTOR

C = I96^fL = 9 8mh.

I M = MIT ** 704IND MACH

MOT = INDUCTION MOTOR1 HP 230V 3<^

TRANSIENTS FOLLOWING INDUCTION MOTOR STARTING WITH

UNLOADED GENERATOR

FIGURE II

period, naturally, first increased abruptly and then

exponentially diminished to a steady-state value not

much higher than that value prior to the load change.

This v/as a relatively ligl'it load addition, though believed

to be not unlike a condition that is comnion on an expanded

scale in power distribution systems, Ko transients of any

consequence are observed. The length of the transient

period is governed principally by the characteristics of

the load.

There is noted, superimposed on the envelope of the

sinusoidal line current, i^^ , an oscillating component of

small amplitude and low frequency. This is thought to be

e feature introduced by the motor and its connected load.

At the machine terminals, current and voltage vary

as indicated by the traces of Figure 10-b. As can be seen,

slightly different load conditions prevailed in the run

during w^ich this oscillogram was recorded. The tendency

for terminal voltage to remain nearly constant is partly

due to the high degree of saturation at which the machine

v/as being operated. The stator current dip was caused by

a decrease in excitation current as the imaginary part of

the impedance presented to the machine teiT.iinals became

less negative while the motor was gathering speed*

The next step was to produce the oscillogram of Figure

11-a, showing the effects in the line of an induction motor

being started with an unloaded generating unit. The oscillo-

gram presented in Figure 11-b is for a similar condition

except that here the induction motor was heavily loaded.

b^bm<^^ a^ ..c noii^vo ?-:. ^-1- r(oiJx/-:-.o:) fe ©:^Unu Jon sJ oi

vnr. h"; -i.i'^.i;>'-^<rij o^i .^:\sy^yz .-rtiij'^li&^^.b T^oq nx oXsOf

o one ifj) 9' ii no

iq L^iiXdvo^' el boxiaq

.bfiol tri^t

odi lo aqoXsvns ea,i nc '•>*?-«>''-r-!:'^'^'a ,b*t"n ai etiedT

^o inenoqfiTO.0 eaiisi. -^ Isbloeunlt

,bsoI b9i5«nn©o «^1 bne ^o^cbi »ri:t yc^ bsouboid-nl s^lo^b*^ s

^IWfl? to e©-«): 'b3;J-fiolbni ••

iiij^ anj n^ ^jl^:;ve:Lq sn-

:

"'

.oi -aer^iiib YXirfeii«

yonobi-ti arrr ,b^t',o:>^1 -^ac iirii rlt>iifetf pal:cub

YXt^»q 81 w-ns^ '-^ov isnl«n^l :£ol

^0 j-^. : ^ri& cjs if»iTU3 rroliBilaxs nx tesd^o^b 6

a,,^,,

.;xdofilB •rfi o:J" b^d-n^s^Tq ^onftbeqmi erii

cw Toi-om ©rij allfiv/ 9vli»5en attX

©-xu^;.iL-: ...V. ........,.v-..^a... -.v. .; . j3ubonq oi ««w q»l8 :^x^n erlT

,j.^,j.o-, p,^_f>^;,bf^l HF io 9nlX 9Ai nl at^eYiQ ©n'.t ^Alwori* ,«-XX

«qXX.: tand5 b^baoXnu a& rtJ-lw bsiisi* gnisd

..,..:j..vJ X;^j.li.ii.2 -: "xolt ai d-XX ft'ia^il. nl b«JTi*a»Tq cfti©

;'&9n >=^6w a:o.tom nolJ-aubnl »ri:r aa^ri f^tiJi jq9dx«'./i.:ve

a5

In neither case do the line transients show instantaneous

values for current or voltage that might cause a generating

unit casualty. In the latter trial, the induction motor

load was approximately one-fourth the maximum pov^er output

obtainable with the combination of capacitance and inductance

used. As the jnotor-starting load is further increased, the

induction machine eventually will lose excitation and fail

to generate* As in the short circuit case, no dangerous

currents or voltages would be produced. In comparison, the

instantaneous line current resulting from starting an

induction motor with a synchronous generator may be six

times the peak current of the steady state^^,

d. ) Unloading - To complete the study of load varia-

tions on the induction generating unit, it was necessary to

consider the effects of abrupt load removals. For this

purpose the generating unit was loaded with parallel banks

of resistors and then unloaded by cutting out resistance in

three-phase sets. Figure 12-a illustrates both the line

and the machine electrical transients following a sudden

load reduction of 50 percent. The accompanying oscillogram

of Figure 12-b was obtained by dropping the remainder of the

load.

The traces of Figure 12-c were recorded when the entire

load was dropped in one step. In the first and third instances

the initial load was approximately 3 KW, In all cases the

load was symmetrical; thus, the effects in only one phase

are illustrated. Again, in all three cases, the value of

....... '

ff + rw 9i^

. ^ oi

»»o«*Xov TO %in9^'S»i9

-ioiv-xfrud^^ Sk*Q;iO'iii«:i\is B Ai'i -^•*

1^

ni

aoX

UpA ,b«j6T[+«uIXl •16

3a

y\AAr\V\^^\A^K.

A/VA'

fi „

'

•

' \Ay\,

l2-a. REMOVAL OF ONE HALFTHE LOAD

AV\A''^ /v'VVv^

.\^*v^^

12- b. REMOVAL OF REMAININGHALF THE LOAD

A ^ A A A A /^ A /*'

V V V V V V V V V

12-c. REMOVAL OF ENTIRE LOAD

C = I96 ^f

L = 27nnh

R^IM=MIT **704

IND, MACH.

TRANSIENTS FOLLOWING ABRUPT UNLOADING

FIGURE 12

shunt capacitance was the same as for all previous transient

investigations. However, the value of the series inductance

was increased somewhat in order to give poor voltage regu-

lation between the unloaded and the loaded conditions. By

doing this, it could be expected that voltage surge, if it

existed, would be demonstrated more clearly. An examination

of the three oscillograms shows that no voltage surge occurred.

The line voltage instantaneously jumps to a value nearly

equal to the new steady- state value. The remaining small

Increase in voltage is attained gradually in the next

quarter of a second. The generated voltage remains nearly

constant} the small change involved is completed within

5 cycles, Upon decreasing the load by one-half, the line

current (ij^ of Figure 12-a) exhibits a remarkable transient

effect. It appears that this transient is of an extremely

over-damped nature.

Unbalanced loading after initiation of the transient

produces a harmonic in the final stator current wave, i ,

of all three oscillograms. The voltage trace, however, is

consistantly a pure sinusoid. Here, as in a transformer,

the exciting component of current has varied irregularly

to preserve sinusoidal voltage output.

gj.l fy^.qM^PPY Y§f,j.#1iJ.Qh - In all of the oscillograms

recording the effects of sudden changes in load, it is noted

that frequency decreased with increasing load. This is due

to the change in slip accompanying power variations. As

power is increased from zero, the slip is approximately

.\£.

-uDsa •->] Av/emoe bstseionl sew

ji Ji ,3.0 uje &^siiov tBti:^ b»io«cpc« 9<i feluoo :M , aider «?niob

nol:fcni':fX^ n/' .vXiboIo ^tc?^ b*».t&Ttanom®h •d faXuow <,b»d-«ix«

YXaesn anlfim«^ see^Xov b^cJ-fi^seasg sni .onoo^a i to a;«:fTftup

nidtlw b6:teXqfi»o al b«vXovni •ens.iD XX&fiie a^ii lined-ano!)

enXX 9d:t ,UBd-dno yd b«oX tdd enlaseioeb rioqU .aaXoYO e

yXsnisiirxQ HA -10 31 ^tnaiens-SEt «M* #»di sifiaqqs ^I •.+39««

.diu4'fcn baqm&b-i^vo

j-aeiansTcr ad:^ io noX^&l:^lfll ir«*1fs e«ll3»oi bdonfiXednU

,1 ,9vsw ^ridiiuo XQd-^^s ii&nilt ^di ai ainomifi/l s asoubo-xq

8i ,:i9vwori ,d3«x^ ee«*^ov 9dT .aniAxpoXXXoao dead;^ XXfi lo

,a«£.n:oli«afi'x^ 6 ni as ••^•H •bioaunla %iuq s YXiruiidianoo

YXi6Xue«xxl belxBV eari c^nsiiuo %o ;fn»noqffl03 e«l^l»>'«> •'<*

.:^L;qj-uo d^aiiov X&bXoaunia ©vx«8S'iq o^

emsieoXXioao ©dJ io XXii nl - ilffillli^^f^V Yi^fi^Mg-t^Ll—L»a

b8>;rou ai ti. .bfioX cil 8#enfido rmbbk-e ^o sc^oelid ^di gnibrod^i

^[ ,b«oX ealaBeionX diXw b®aB®^3»b Y^^^^t^P®^^ ^fi'^^

a«w«q enlYn»qfiioi>3« qlXa nX «pnAdo 9d:t oi

.^„^. .. ., t r-JCp ''•" ' '.-t-»v '.;.-^^'; Mrt^c.-^ionc ax 'iswoqL -J . . ,;U. X'- i.'- • .'s c. .*. ^A.A«^ -• •

32-

proportional to power. The inertia of the generator rotor and

connected drive motor maintained essentially constant speed

through each load change. The increase in slip speed causes

an equal decrease in what is conventionally terraed synchronous

speed. With a fixed number of poles the result is a pro-

portional decrease in frequency.

Conclusions drawn from the discussion of experimental

results of transient investigation are included in Chapter IV.

:;Jt^«u^iij

33.

The work performed with the self-excited induction

generator with series inductance compounding leads to

the following conclusions:

1» The no-load characteristics of this generating

unit are determined by the exciting capacitance

and machine constants,

2. With load applied, the line voltage regulation

is improved as series inductance is reduced,

holding shunt capacitance constant,

3. '*Vith constant series inductance, the line

voltage regulation is iraproved by raising

the shunt capacitance,

4, Providing power service is continuous for the

range of unity power-factor loads from no-load

to short circuit, the generating unit will not

cease generation for any lagging power factor

load. The maximura power output may be in-

creased by either reducing series inductance

or increasing shunt capacitance.

5, Loss of power due to insufficient magnetizing

inductance does not occur before the generating

unit has delivered maxinium pov;er obtainable

with a given combination of inductance and

capacitance.

^^

.^yol.

jou IxXv/ ;?xnr v.i^j ;. .;Su^ -' - .^ -xfjue OJ'

34-.

6. Efficiency of the generating unit is poor

for partial loads due to the presence of

a large excitation current.

7, The transients following faults and sudden

load changes do not produce excessive

currents or potentials in the generating

unit or line. Sustained short-circuit

current is dependent upon the relative

values of series inductance and shunt

capacitance.

From an engineering standpoint, it seems that the design

of a practical induction generating unit based on present

knowledge will involve a compromise between voltage regu-

lation and power output on the one hand and stability and

efficiency on the other. Good voltage regulation may be

obtained by working the generator highly saturated; however,

excessive inefficiencies will result, Capacitive compen-

sation can produce desired voltage characteristics under

load, but may lead to excessive values of terminal voltage

and machine current if the line becomes shorted. Moreover,

the shunt capacitance must be a value low enough not to

produce excessive open circuit voltage; yet, high enough

to assb're generation continviity through a reasonable range

of lag'jing power factor loads,

Faults will not produce dangerous currents and potentials

if inductive compensation is employed. Inductive condensation

r!pi>*>. .

'^O^q

35

may lead to discontinuous voltage generation, but this will

not occur before reaching the point of maximum power.

Furthermore, and equally important, inductive values that

lead to tMs condition of instability also result in the

maximum pov^/er obtainable with a given machine size. In

the limit, series Inductance may be reduced to zero, i.e.,

no compensation, but this condition causes voltage regula-

tion to bscoraa rapidly worse as load power factor becomes

more lagging.

Inductive compensation may be adjusted to give con-

tinuous power generation for all lagging power factor loads.

However, in order to carry reasonable loads, the unit must

have shunt capacitors of large size^and will, therefore,

suffer from the above-mentioned disadvantages. Again,

voltage regulation is far from satisfactory. It is believed

that a comprciTiise design for a self-excited induction genera-

tor, statically compensated, will require inductive corpensa-

tion. The value of inductance v/ill be selected after

considering stability, povver output and voltaga regulation

for expected load demands.

^^

.,«^aoX '^^f'^n'^ f-mmr

Poor voltage regulations is one of the major disad-

vantages of tlie induction generator without compensation.

Static compensation improves voltage regulation, but

introduces other objectionable characteristics of operation.

Adjustable compensation may be a satisfactory solution. For

this purpose, capacitive compensation would net be acceptable

because smoothly-varying capacitors in power sizes are not,

at present, feasible. On the other hand, an adjustable

inductance could be continuously increased to give the

desired line voltage characteristic.

One scheme that appears promising involves use of an

automatically controlled inductance as series compensation.

As load increases, line voltage would be held nearly con-

stant by decreasing the inductance through a regulatory

system. The inductive reactance may be supplied by a

saturable core reactor, or a movable core inductance coil.

At this time, it is felt that sufficient knowledge of the

induction generator has been gained to permit a study of

self-adjusting voltage regulation from a servo-mechanisms

viewpoint.

The advantages of an induction generator will become

more apparent at higher frequencies. Size and cost of

capacitances could then be considerably reduced. Also,

the squirrel-caae rotor will permit relatively high rotor

speeds. For these reasons it is recommended that further

work be performed at 400 c.p.s.

9mC3£5vi Jix

' 'i*»TP(?f:^P' f^iCMW

37

Frequency ragulation should also be invescigated with

a view toward reducing the frequency transient foli owing

changes in load demand, For practical use in a turbo-

gejnerator set, che speed of the induction generator must

be varied with load deivtand to maintain conscant frequency.

The conventional fiy-Daii governor of the turbine must be

modified to provide for control of spring tension. This

control will be actuated by deviatxons from the desired

frequency, For parallel operation, load will be distributed,

as now, by adjusting the throttle opening. The combined

rotating inertia of the turbine and generator will make

transient frequency variations unavoidable, A study of

the engineering problems involved in frequency control

should be conducted by both analytical and experimental

means.

In conclusion, it is suggested that present disadvantages

of the induction generator may be reduced somewhat by i*e-

designing the electric and magnetic circuits of the induction

machine* Machine currer'.ts could be reduced by specially

formed rotor conductors. Operation in relatively saturated

regions may be tolerated if the volume of the iron in the

magnetic circuit could be appreciably reduced.

,ba»a*.

9d(?6Ji

n, ^?WQU

33

At PfciAa^p fm^m^

The msthod that Friauf developed for performance

calculations was modiiied to include the effect of in-

ductive compounding and then employed to predict values

for shunt capacitance ^tiA series inductance for laboratory

work. A qualitative description of this analytical work

follows.

Voltage build-up in an induction machine is initiated

by residual magnetism in the field poles and progresses to

a potential determined uniquely by the magnitude of shunt

capacitance. A no-load saturation curve was obtained

experimentally and appears in Figure A-3. In the lower

region of this curve, points were obtained by connecting

the induction machine through a variac to a 230-voit,

3-phase power supply that resulted in no power transfer

from line to stator tor each of a number of terminal

voltage values. Data in the high-voltege region was

obtained by operating the iiiduction machine under no load

aft a capacitor-excited generator for various values of

shunt capacitance. Terminal voltage and current v/ere

recorded and plotted to give the typical saturation cux^ve

of Figure A-3, Referring to the equivalent Circuit of

Figure A-2, it is seen that at no load conditions the

current from the terminals of the machine (marked t~t )

¥'^^^^^8^, g^.JIiSkl:

4^

2.30 V DC

nauRE Ai-oiRcuiT piAt^RAM FOR src^ot smre oPtHhriou

f gwwwu-Ty WV OC/t

^I 1 1^ "^

>^—'y2^OTC^—

f

^yznnuj^ ^

Si*^ Vc Z,i^

FiGURC AZ-EOUiy/AL€NT CtRCt^lT OF ONE PHASC

^^

is essentially equal in magnitude to the magnetizing branch

current, I , It will be noted that the slope of the straight

line of Figure A~3, indicated by OX ^ , represents a critical

value oi capacitive reactance, below which the generator will

fail to buxid up voltage, C^en-circuiu voitag© is fixed by

the intersection of the saturation curve with the straight

line, OX^, the slope of which is fixed by the value of shunt

capacitance.

Knowing the macnine constants, the stator irapedance drop

was calculated for various values of no-load excitation

current. By subtracting /ectorally these impedance drops

from the corresponding terminal voltages on a per phase basis,

values of air-gap voltage, Ej^ , were determined.

It was assumed tnat there was no resistance included in

the magnetizing branch. fhe sueceptance of the magnetizing

branch can then be expressed by the relations

.^ - ^ (1)

Magnetizing 6us,cepcance, Y^ , was calcuirited asid plotted

as a function of air-gap voltage, L]_. The result is the

susceptance curve of Figure A-4. The distance o-a along the

abscissa of Figure A-4 represents a minimum value of mag-

netizing susceptance below which no air-gap voltage ivill ue

generated. The existence of this minimum magnetizing

susceptance is a direct consequence of the critical value

of excitation reactance.

^-^

Xaollxay h .;.:!ii&>i»actJi , ,-..w vd bd3B3ibni: ^e-A •lueiH io ©all

Yd b^xii 8l dj^eii: ^«i

qoiD 9onsb«qfni itio&m •ricf cniwon;!

,8i86d steriq leq a no 3«p6Jio\ nnibnoqesTioo 9di raoai

jnoXiaX*i ®rij y<^ b#«a9iqx» »d nsri:f a£0 rlansid

o£oxoadfi

^i.^^.^^ ...,. ^^... „. .„-.,....- :'-^ ^-i .ho^j-si^nsg

45

Referring now to the equivalent circuit of Figure A-2,

Kirchoff's current equation applied to the point a, will

result In the expression:

I,^ - I^ + li- (2)

\^ich may be ejQ^ressed as:

^1 "^2 '^ ^1 '^^ "^ ^1 ^1*

Y2 is that admittance entirely to the left of points a - b;

Yj^ is that admittance entirely to the right of points a - b;

and Y is the admittance of the magnetizing branch.

Rearranging terms and employing the assumption of purely

reactive impedance in the magnetizing branch, the final

desired expression for magnetizing susceptance is.

= Im [ Y^^J- Im [Y2] (3)

The admittances H^ ^^^'^l ^^Y "^^ expressed as;

(4)

and ( 5

)

;;(^.);zLjC^»s/.

An admittance diagram after the manner of Friauf was

then prepared showing the locii of Yj^ and Y2 on a complex

plane. First, it was necessary to determine experimentally

the stator and rator constants of the machine as described

elsev/here. From equation (4) it is seen that for a given

^^

,£~A 9TL' ozio :fn©lBvlup ot ^IiisIbF.

. ionsid gnlsli^n^e jTi «fii io •»n)sJ^;tiffibs •rii^ di Y bns

YXeTLfq io noiSqmjeas. sriS ^nlYoXqrti« bat enrxaJ pnignaiiBa:^

Isnii drii ,ri3n6id gnisij-sagsm firi* ni «on&b«qmi svlc^ossi

i.Y(£) [sYJ ml - [^y] ml

(e)

xdiqmoo b no ^Y bnd ^Y %o liaoX 9flS pnXworf* baieq^iq nsriJ-

YXXfiinemlxaqxfi enlfin9;^«b oJ^ Yis«3»o»n , 2:i:lH -aaeXq

bediidaeb as enxriossi ddi lo eJ^ne^anoo i^jj^jl oob 10^0:^2 •rU

iisvip 6 101 ^^i.(;j a^Qe ax ^x (4^) noxlsLfps raoi'i .dia/IwaaX©

umachine, Y2 is a uniq\ie functi^an of slip, With slip «s

a parameter, the locus of Y- w-as plotted as indicated In

Figure A-Si(a),

It can be shovm tliat the locus of ac^aittanc® Yj_, with

magnitude of load impedance, Zt^, as a parafswter, is a

circle for a given machine vtf^»«n shunt capacitarKe, series

inductance^ Sktv^ load power factor angle are ail held constant.

The radius can be expressed byj

Hie contex of the circle- is locsucu ac the point having

an abscissa of s

/^, i- e. ^c X,,- X3

[same denominator as Eq, 6)(7)

and an ordinate of:

(S)(same denominator ^g £q. 6)

Such a circle locus for Yj_ is drciv/ii in Figure A-ti(a)

unity power f^actor. As the load iinpedance angle

is varied to give lagging power factors, a faraily of

circles is found for the resulting iocii of H-^* fne centers

oi" these circles trtov*^ to the left as the load power factor

decreases. Ui circles pass tlixough the ^exv-^ircuit point

and ths short-^circiiit point, because for losd impedance

-^^

'^'^^^r^^-^^f).1^rfy^-^ -v^C^-^- -.>^-/^)

(B)

55 .^^ _^^_. ^

id ^jHI' tti t^jftnlmensfe •mi..'

tX.S-a><.

^

oc ^X

InM

OC - OPEN- CIRCJiT Point

SC - SHORT ciACJir Po/sr

A'5 f*) - tWDoCTivf C0r\PinSf^TlON

KM-

A S(b) CAP'^cnivt COi^PtU^AT****

ReL-<]

IJCURZ^A^ _ /jpnarA/vcf loc (

4-7

equal to infinity or zero, power factor is meaningless.As brought out earlier, the open-circuit point is fixedby the value of shunt capacitance. The short-circuitpoint falls below tho open-circuit point and moves nearlyvertically up or down with increasing or decreasing seriesinductance respectively,

Fron equation (3) it is seen that rr.agnetizing suscep-tanoe. Y^. is equal to the vertical distance fror. the locusof Y2 to the locus of Y^. A straightforward mathematicalanalysis of the performance of the induction generatormay be commenced at this point. In brief, the methodinvolves graphically picking off values of magneti^ingsusceptance for selected values of admittance. Y^. Hien.use is made of the susceptance curve to determine air-gapvoltage. Having a value for air-gap voltage and a correspond-ing value of admittance, Y,, application of simple circuit«»lysis will lead to calculated values of load voltage.

\. and load current. I^. of the equivalent circuit.Figure A-2.

This method of performance calculation is found inthe notes of Dr. Friauf^ and experimental confirmationis included in the -.vori. by Swift^. For this thesis themethod served to establish orders of magnitude for valuesOf shunt capacitance and series inductance necessary forprospective laboratory investigations. It should be notedthat as the inductance is reduced, the magnetizing susceptance.

linsam b1 iojOb: lawoq ,u-:.«5 x^ w^^.^.u^^.x w^ *»^^.,.

jnioq Ji.ww:id.i-i^qc i^AJ viCi-:Kl tiii^'i iiUoc

Bqeei ©DfiBiot/bni

auooX eri:r laoal ©cmiicrsib iL;;i..':jev oi- o.: ^supa ax ,>j^,^.....:.

Xfioi:r*rr«»fflsfli biawioltniQi6ti» A -x^'** ^"^^"^ •'^^ ^^ ^^ "''

qeg-Tibfi •nimd^t^b oj- ftv^ciiO 9oci&ict90%us eni 20 ©be

-bnoqasTt-xoD 6 fane ^efciXov qfijrxls^o^ ©uXav 6 enivBH .o,fc.:iov

:fXL;3tXo «Xqml« lo nol:^solXqqA « ^Y .•Dflia:r;Hraba lo euX£V ©rti

,»g.stXov b«©X Ito «©vXsv b©:rftXtiDX60 o:^ bs«i XXxw alBxUoM

,ni}0'iio ii*®Xi5viup« .^^"5 ^^^^ **"» 'J^

.LOW ^iU a* ^iiiij'jioni 8X

. doqeoaq

Y^, may become less than the minimum value for sustaining

voltage generation, V/lien this condition prevails the

locus of Y-, has intersected the curve marked limiting

line of Figure A~5. The limiting line is constructed by

following the curvature of the locus of Yg ^"^ ^ vertical

distance above equal to the value of minimum magnetizing

susceptance*

It might be pertinent to point out here that if

admittance locii are similarly plotted for the case of

an induction generator with capacitive, rather than induc-

tive, compensation, two important changes occur. Locii

.are indicated qualitatively in Figure A-5(b} for this case.

Obviously, the locus of Y2 remains unchanged. However, the

short-circuit point is noiv found nearly vertically above

the open circuit point. Moreover, in opposition to the

previous case with inductive compensation, each family

curve representing the locus of Y^ for lagging power

factors moves to the right as the load impedance angle

increases. As illustrated in the sketch of Figure b-A(b),

this means that voltage generation would collapse for certain

lagging power factor loads.

In preparation for -che laboratory work, the equivalent

circuit parameters, Figure A~2, v/ere experimentally deter-

mined in accordance with A. I.E.E. standards. The values

of machine constants so obtained are listed in Table A~l,

4^.

Machine Cofistants for M.I,T. Induction Machine I\o, ^u4

r-, - 0,43 ohms

r = 0,174 ohms

X, =* 0,58 ohms

X,.- 0/w»8 ohms

x^ *= 7.79 ohms

A value of shunt capacitance v/as selected that would

give an open-circuit voltage not exceedin^j a safe upper

limit. A value of series inductance was then determined

which would locate the short-circuit point near, but above,

the limitincj line, w;ith these viilues of capacitance <xnd

inductance, the test ec^ipment as shown in the circuit

diagram, Figure A~l, v/as set up. The induction machine

was operated as an eight-pole machine and was driven with

a d-c motor. Speed was controlled by manually varying

field resistance, R^^ , to maintain oO c.p.s. output of

the generator as indicated by the frequency meter.

It was found necessary to make slight readjustments

of the values of inductance, L, , and capacitance, C-j^, as

previously determined to assure continuous voltage genera-

tion and to provide balanced conditions for the three phases.

Upon completion of each run, the reactance of these elements

was measured.

For the steady-state investigation with constant para-

meters, the generator was operated at power factors varying

it? i\&

it

p ^.«

•vtf.'. b • 1 f,JO It '*.CV-."

. - ^.ifi noli

^o

from .15 lagging to ,7b leading by adjusting the load

components R^' ^o' ^'^^^ ^'"'' ^^® data recorded is shoivn

in the Appendix (Runs I ~ VII) and is plotted in Figures

1, 2, and 3,

Further investigation vs/as concerned with the possibility

of improving voltage regulation by decreasing the value of

series inductance, L-j . First (Run VIII), the same value of

shunt capacitance, Cj_ , was used as in the previous test.

Then (Run IX), to demonstrate the effect of saturation,

the value of C]l was decreased, wrfiile L]_ remained at the

lower setting. These runs were made at unity power factor

and, for comparison, are plotted with the unity power factor

result of Run I. The data appears in the Appendix, and is

plotted in Figures b, 6, and 7,

For all the above laboratory vrork, three-phase balance

was maintained at generator and load within b percent. The

generated waveform was observed on an oscilloscope.

The transient behavior of the induction generator

with inductive conipensation was investigated. A 3-bIade

''guillotine" switch was inserted in the circuit of Figure

A-i at S3 with its secondary connected to provide single-

and three-phase short circuits. Voltage and current

transients were recorded using a V^estinghouse portable

oscillograph, with six recording elements* The traces

were roughly adjusted to indicate the balanced conditions

existing in the circuit, A time trace was not included

since the steady state appearing before the interruption

is a 60 cycle trace vsdiich can be used for timing. To

accomplish sudden application of load, a 1 HP, three-phase

induction motor was started under load by connecting it

through the "guillotine^' switch. For unloading character-

istics, the value of P...^ was increased as a step function

by using a relay-controlled "guillotine" which removed

parallel resistors from the circuit.

Data compiled for the transient study are included

in the /^Dpendix and oscillograms are shown in Figures

8 - 12,

. Xo-isq

B> DATA

nm^ RLAff PAiA or w^^im v^^ 4s a wg r^fi^AXQi^

M. I. T. Iriduction Machine No. 704 (Squirrel Cage)Westinghouse Type CS Induction Motox^

Frame 485C Serial No. 4884645 Style 890120

7.b HP 220 volts 60 CVZ 3 phase

Poles 4 6 8 12

Amps per Terminal 19,7 19.3 25,3 33.

B

Full load RPM 1710 1130 860 D70

Temperature rise bO^ in one hour at 100 percent load.

^T

<iI

^sn N» o» <i "o

Q qj <o V, <(| «w

-v \i «o <i

^-1 J» ^ <»> « * ^ ^•) «« <^ Ol ^ ^

. 'O O I^ O ^ n

I(U "VJ fg -. ^ ^

V, '1 0| «< N "

^i « « ^ «

t>^ e\t ri nl ^<n ^ ;. \» •* '-

^ S ^ Q (5

<^

"^ f':: * "0 5 - N ^

W * N •. *, 01 ay

6 ci e> 6t '4 4 vl >«

i ^ Q 6

1 . <l| .n "« '

i ^ «w ><. <v <

( <^ n *>• *M

. !? ^ "o ^ ^ -^ -- ^

5 «*)»1NNN\N ^ iv^in'VN'V"' N<\J'V'>J^Q 1 50 N •«•

"

«) (.) '>1 <\» N «\i -xJ -) ^ ft! <*! •

« tS « O ^o ^ o ^ I

«V. ^^ • O O o nD "^ I

,o o

St 5 o '->

1 P U «^ N K »« o

- O O o O

ci o "^ ^ "^ "^

t^ rn t^ n t*i '< fi

o o « pa. N N «0 «0

2_cvJ <) <* n Jl

'

Iti !i ^ J^ ^ 4

J> Ci -J

"2 ^

V* K> <M vO •.& »» ^

C ^ 'i ij:v (^ «,

>^ '

N »J -V ^i

<M <« M I

ni »t •! <

IK * «^ - - O

4?

s^'

* ^<o,,^^v^„^^f- :^ i i S ' J

, .^ ^ S. «6 O ^

, O to N- * "*

^i «J rr^ <- 'n Vi

,0 rt «^ «<i *i N-f- "

4 «i

<^ ^--

^ Q V, ,» ^ «D -M

<i < < oi ry< rv -^

<"^ ** '^

;» i; ; Jl 5 ^ :j ^ 5:

">•>'- ^ - Qvj »c «c -:. "^ ->* +i

*

S ;>;3 ^ 4 J V,0) .^ N <i «

J:5 r< (>« .4 oJ ri -< c< >. <^ ol ri ,i A r* d -

, W, ^ -^ - N ^ ^ ^ «0

r"^ ^ i - = = ^**^<\/ «\ o

~ ^ £N 5 Ps^^-^: >?

« * ~ »^ »^

•V «• N "f 13 «J •<

fi <^ '^ H. nt ni ni

I ( ^ K v^ m c* - -IK * -^ ~ - d

?^

*9i ^ -n 5

•Xi -O <0 4. It)

Q o «0 r^ *• «Vi ^ ro 4- -^ *<»

; Is ;^ Jl 5 ^ :^"5

5:

kt^ „

^ Q ^ >» ^ «0 -M

tj < < «i ryi r^ r.,i

-n <>; ^ ^ - Qvi « «C ^ ^ M +:

^ -'^ N N <>< •>*

5^

, <^ ^ i" - N ^ * ^ «0

o <5 O ^kU

»iU

« « «^ >^

N "* N <y 2 "• ••

H. '4 '^ >< ^ f^l ni

•^ ^ ^ k

IX ro^^-^t"^.^*:*e^ rr, Kr, 9> 1 .

^ ^

oOOOOOQO

^''"^5; S 2 5 N :3 ; t "^

<^^

i" .V -^ "^ ' -^

^ nj nt <\t '

Jt 0, « X •S<^

Si>

^fVlA

7jf,^ri.§i.^n;U

Three Phase Short Circuitc = 196 mf - L ~ 9,o mil

Before Transient; 'g^ 2bb voits

^9= 29 amps

\ = 212 voits

^L==13 amps

4.0 KW

f - 60 c.p.s.

Single Phase ohort Circuitc = 196 mf - L = 9.8 ?nh

Before Transient: V = 255 voltsQ

I. = 29 amps

V, « 212 volts

I^ = 13 amps

Pj = 4,8 KW

f = 60 c.p.s.

starting Induction Motor - Vs/ith Loaded Generatorc - 196 rnf - L - 9.8 mh

ieiore Transierit'.g

260 volts

T -; 28 amps

\- 218 vroits

\- 11 amps

\ ^^ 4.4 iCW

f = 60 c.p.s.

^"15^

JLk'Z'i.iu J 3 oar f.3jErr-'

rim 8,$ » J - %m d'^i

sJiov cr: 3X8n«iT ©ao:

3 Cii:\D V. ".

• 3.q..3 Cc

itiuo^iO ;h«OfB e8*fi<l dXgnia .S

8 OBIS Ql

.a.q»v t/w

After Transient: 4.1 KW

Induction Motor Data:

i h.p,3 amp.

220 V, 3 phase60 c.p.s 4 pole

Starting Induction Motor - With Unloaded Generatorc = 196 mf - L == 9,8 mh

Before Transient*.

After Transient:

^9 = 300 volts

^9 = 34 amps

JL.

300 volts

I^ * J

f « 60 c.p.s.

P. = 600 watts

Induction Motor Data as directly above.

Unloadingc = 196 mf

Full Load:

- L 27 mh

Half Load:

^ a 188 vol t s

II - 6.4 ar/rps

Pt = 2700 vvatts

f = 60 c.p.s.

^^g=: 294 volts

^9- 34 amp s

\ s 2S0 volts

\ = 3.0 amps

^L= 1440 watts

f ^ 60 c.p.s.

S3'

1» J» B, Friauf , "Calculation of Capacitor-Excited InductionGenerator Performance ", Unpublished* 23 June 1948,(Not9S at Code 650, Bureau of Ships, Navy Department,Washington, D* C.

}

2* C« S, Swift, 'V\n Investigation of Capacitor-ExcitedInduction Generator Performance and a Verification ofa Method of Performance Calculation*', Mnapolis, Md,

,

1950. (Thesis at U, S. Naval Postgraduata School,Monterey, California,)

3, 0. T. Estes and Hi. J, Hussong, Jr., "Study of Capacitor-Excited Induction Generators; Parallel (^©ration andTransient Loading**, Cambridge, Mass,, 1951, (Tnesis atMassachusetts Institute of Technology.

4, A, S» MacAllister, "Excitation of Asynchronous Generatorsby means of Static Condensers '% ELECTRICAL liORLD .AMD

ENGINEERS, Vol. 41, 17 January 1902.

5, km S, Mac/Ulister, "/\synchronous Generators", Af/IERICAI^

ELECTRICI/\N, Vol. 15, November 1903,

6, Stanley and Faccioli, "A-C Machinery - Induction Alter-nators**, Trans* AIEE, Vol, 24, 1905,

7, Hobart and ICnowlton, "I'he Squirrel-Cage InductionGenerator", Trans, AIEE, Vol, 31, Part II, 1912,

8, D, F, Alexander, "The Polyphase Induction Generator**,THE ELECTRIC JOURNAL, Vol. 25, 1928.

9, F, 0. Stebbins, "Effects of Capacitors and Varied VoltagesUpon Induction Motor Characteristics", G. E. REVIEVi,April 1934.

10. Basset and Porter, "Capacitiv© Excitation for InductionGenerators" , Trans. AIEE, Vol. 54, May 1935,

11. Kurtz and Corcoran "Introduction to Electric Transients",John Wiley and Sons, New York, 1935.

12. H. P. Hoya, "Self Excitation of Induction Generators*,Cambridge, Mass., 1938. (Thesis at MassachusettsInstitute of Technology.)

13. C. F. Wagner, «Self Excitation of Induction Motors",Trans. AIEE, Vol. 58, 1939.

.^-?.

Y>!^/^^oDijflia

nolizi

» loortacJ •JiLJt

brtfi noi:f£'TG<:<^ l3llr^«q is^iotBiftnec

llAOinm

-^S3lA aolfoijtr^J. - Y^^n^rfos' ^^aoe^ bnfi v«Xfi»*3 •d

,,:i;3bnx>xeIA

3&0;.::7i. :

jowi nox.-^

noxc^aubfil 'lo

"ajnsisr.

."attoi^oM

ioitB^i»x3 "ties'* ,&XoH .'•? .H .s:.

•8£^I ,.««ftM , •gfoiadinjsD

14, C» F, Wagner, "Self Excitation of Induction Motorswith Series Capacitors*** Trans. AIEE, SupplementVol, 50, 1941.

15, "Self Exciting Alternator", EMGIIMEERXNG, Vol, 168,October 21, 1949.

16, 0, J, M. Sraith, "Generator Rating of Induction f.-otors",

Trans, AIEE, Vol* 69, 1950,

17, M, Walker, "Asynchronous Generators; Sir/pie Theoryvathout Mathematical Coiaplication", ELECTRICAL REVIEW,Vol. 147, t4ov8mber 1950,

18, H, H. Roth, "Induction Generators for Small Loads",Paima, Vol, 94, December 1950,

19, Tsao and Tsang, ^Squirrel Cage Induction Generatorfor Power Generation*', AIEE Misc, Paper 51-158,ELECTRICAL ENGINEERING, Vol, 70, September 19131,

ao,d. ""8The experiinental deter-

mination of the perform-ance of a capacitor-excitedinduction gener?.tor withan inductive reactance inseries with the load.

17118

Tniesis Coode056 The experiment !*1 deterrtln?*.*^ on

of the performance of a capacitor-excited induction genetator withrn inductive reactance in seriesvith the load.

U. S. Naval Postgraduate Schoo]Monterey, California