You are on page 1of 20

The Calculation of the Armature Reactance of Synchronous Machines

BY P. L. ALGER*
Member, A. I. E. E.

Synopsis.-This paper presents new and simplifiedformulasfor the armature leakage reactance of synchronous machines, and compares the results obtained with tests on 100 machines of varied types. The new form?nulas are characterized by novel expressions for the end leakage and for the "air-gap" or "differential leakage" reaclances. Forimulas for zero phase-sequence reactances are also

given. The definitions of armature leakage and armature reaction reactances used are those proposed by Doherty and Nickle in 1926, and as a result the leakage reactance has a much smaller value than has heretofore been assigned to it, thus requiring a rather fundamental revision of the ordinary conceptions of the flux densities existing in the various parts of a loaded synchronous machine.

I. INTRODUCTION T HE growing interest of operating engineers in machine reactances, (as system stability and shortcircuit phenomena become continually more important), together with the recent advances in the theory of synchronous machines, make renewed consideration of the subject of reactance calculation opportune at the present time. Papers on this subject have appeared in the JOURNAL'S pages at intervals over a period of more than 20 years, each paper marking a further refinement of methods and an advance in accuracy. At first, it was customary merely to estimate the numbers of leakage lines per ampere inch of embedded and free conductor, and multiply them by the respective lengths to obtain the reactance. Later, the reactance was segregated into several distinct elements that were separately calculated. These calculations due to their greater importance and in this case, the possibility of more accurate tests have reached greater precision in connection with induction machines than with synchronous machines. Adams"11'3'14 first developed formulas for induction motor reactances, and later Fechheimer9 and Doherty and Shirley7 developed formulas for synchronous machines. In 1926 Doherty and Nickle3 presented some fundamental extensions of the theory of synchronous machines, and proposed new, more precise, definitions of armature leakage reactance and of armature reaction reactance. The present paper adopts these new definitions and develops simple formulas for the accurate calculation of the two quantities, for salient pole machines. The reactance of armature reaction is here defined as the reactance due to the fundamental sine wave of air-gap flux produced by the armature current acting alone. It is calculated by means of the curves derived by Wieseman2 from flux plots. There are two distinct for it values occurring when the it, the direct axis value oceurring values for axis of armature m. m. f. coincides with the pole axis; and the quadrature axis value, occurring when the axis

thecaldzgganbetekg,rsptilyinoero

of armature m. m. f., is midway of the interpolar space. The armature leakage reactance is defined as the difference between the total, or synchronous, armature reactance and the above defined reactance of armature reaction. The distinction between the leakage and the armature reaction components of the total reactance is quite an artitrary one, since no winding can have a definite value of leakage reactance except with respect to another winding. Some distinction of this character is very desirable, however, for convenience in calculating the transient reactance, and for other purposes. The armature leakage reactance as above defined, is as nearly equal as may be, to that part of the total armature reactance which remains the same under transient as under steady conditions of operation. Following Adams' treatment of the induction motor, the armature leakage reactance of a synchronous maehine will be divided into four parts; the slot, end, zigzag, and belt leakages. Formulas for each of these will be derived in turn. As the slot leakage has been quite satisfactorily treated by Adams"1 already, the discussion of it here will be abbreviated as much as possible, consistent with the derivation of a little more exact formula, and its extension to apply to zero phasesequence currents. None of the other three elements of the leakage has been adequately treated by previous writers, so they will be given more detailed consideration. The four divisions of the reactance found useful for induction machines are followed for the two reasons that they fit in very well with the new conceptions of leakage reactance, and that their adoption will tend to wind their unify the particular the total "air-gap leakage, due to harmonics of. the air-gap flux, is segregated into the two parts called zigzag and belt leakage, respectively, in order to permit the effects of the independent variables, number of slots, and numbDer of phase belts to be separately taken into account.

tyption

to

*Assistant Engineer, A-C. Engineering Department, General All previously published formulas for zigzag leakage Electric CSompany, Schenectady, N. Y. hv novdtecnieaino vrapn rmr h osdraino vrapn rmr For numbered references see bibliography,.aeivle Presented at the Winter Convention of the A. I. E. E., New York, and secondary teeth, and so have inadequately taken

N. Y., Feb1ruary 13-17, 1928.

493

into account the effects of flux fringing in the air-gap,

28-30

494

ALGER: SYNCHRONOUS MACHINES

Transactions A. I. E. E.

and have been inapplicable to synchronous machines. Also, older formulas for end leakage have been frankly empirical in nearly all cases. All new formula for these elements will be derived by considering the leakage flux to consist of revolving magnetic fields, divided into a fundamental and a series of harmonics. In each case, the coefficient of the resulting formula will be derived from theoretical considerations alone, and no empirical multipliers will be used to bring the results into agreement with practise, although simplifying assumptions will be freely made as the occasion demands. Before proceeding to the derivation of the formulas, it is desirable to get clearly in mind the distinctions between the four elements of the leakage that have been specified. The end leakage reactance is due to the flux linking the end windings alone. The slot leakage comprises all of the flux crossing the slots due to the armature current, but does not include flux passing from tooth to tooth in the air-gap space. Evidently there is a slight error here, because the flux lines near the mouth of the slot do not pass straight across, but bulge outwards into the gap. The zigzag leakage comprises all of the space harmonics of the air-gap flux, due to the armature current in a one slot per pole per phase winding which induces fundamental frequency voltages in the armature. This includes the flux which crosses from tooth to tooth in the air-gap and interpolar spaces without actually reaching the field surface, and so takes up the leakage flux at the point at which the slot leakage lef t it. The belt leakage then comprises all the remaining fundamental frequency voltage producing space harmonics of the air-gap flux due to armature current. This last reactance is the additional reactance that an actual winding has above that which it would have if there were as many phases as slots per pole. Thus, the zigzag lealt age reactance is due to the deviation of the armature m. m. f. wave from a sinusoid, caused by the limited number of slots, and the belt leakage reactance is due to thQ further deviation caused by the limited number of phases. The belt and zigzag reactances together have been appropriately called5 the "differential leakage," since they are due to the deviations of the armature m. m. f. from the ideal sine wave. They may also be called "air-gap leakage," since they represent the difference between the total and the armature reaction components of the fundamental frequency voltage producing air-gap flux. As the differential leakage reactance varies with the relative positions of the axes of the armature m. m. f. and the poles, the armature leakage reactance varies likewise, and has a slightly higher value in the quadrature than in the direct axis.3 Formerly, the air-gap flux due to the armature current was divided into two parts, that due to the armature reaction, and that due to "tooth tip leakage."7 The armature reaction part was intended to represent the air-gap flux produced by the armature that links the

field, but the determination of this flux was erroneously based on the linkages of the armature by the air-gap flux due to the field.7 The remaining part of the total air-gap flux produced by the armature was then called tooth tip leakage. As the new definition of armature reaction includes most of this latter flux, we have abandoned the name "tooth tip" and called the small remaining part, the synchronous frequency voltageproducing harmonic fluxes, "differential leakage," as just described. Bearing in mind these conceptions of the leakage reactances which we are going to calculate, we can now proceed to the derivation of the formulas. The four components of the reactance will be taken up in the order of their difficulty, and subsequently the results of their application to actual machines will be compared with tests. II. SLOT LEAKAGE REACTANCE Only two-layer barrel-type armature windings in

op slots

il

cnd inu this paper, isice

X I

t I

FIG. 1-SLOT LEAKAGE FLUX PATHS of single layer windings have reduced them to a place of only historical importance in polyphase a-c. machines of American manufacture. Consider the slot shown in Fig. 1. All the flux crossing the slot returns below its closed end through a path of negligible reluctance, and none of it links any of the other slots. The leakage reactance in ohms per phase, due to this flux is then equal to the product of the following factors: 1. The permeance of the path across unit length of slot. 2. The embedded length of slot, or L 3. The number of slots in series per phase, or S/q 4. The square of the number of series connected / qZ \2 conductors in each slot, or ( S / 5. The rate of change of the flux, or 2 r ft 6. A constant to reduce linkages per second to ohms, or 10-8

Feb. 1928

ALGER: SYNCHRONOUS MACHINES

495

The permeance of the path for flux A linking all the and that of II is proportional to: d1- d2 d1-d21 1 r conductors in the slot of Fig. 1 is:

40

wd

lines per ampere per cm. length of slot

4 ld3+

2
I

1 dll - [d3 + And if the winding is full pitch, so that the currents 3 in coil sides I and II are identical in magnitude and to: proportional is inductance phase, the linkages per ampere produced by the flux in while their mutual [ 1 paths B and D are equal to the expression: (di - d2) 4 Ld+ 4 7r 4 d-d2 di d Hence, the total linkages due to the two currents ' 10 ( d w since with uniform current distribution over the height 00 out of phase acting together are proportional to: d2 2 di 1 - d2 di, the flux density distribution is linear and the linkage 1 di 6 + + d3 + distribution is parabolic, while the average height of a - d3 + 3 6 3 41 parabola is one-third its maximum height. As all the 4 (4 d3 + d- d2) cos 6 flux through path C links one-half the total current, the + 8 linkages due to it are one-quarter as much as they would be if this flux linked all the current. The linkages due Or, the slot constant for this case is: to C are, therefore:

l10 ( 4 w)
and so the total slot reactance of a balanced, full pitch, polyphase winding is equal to:

2 (w 1+ cos 0) + 2w (5 + 3 cos 0)

d2 24w (3cos 0-1).

When the winding pitch is 100 per cent, all the slots / q Z \ 12 d3 + 4 d -d ] ] have 0 = 0 deg.; when the pitch is q- 1 , all the 12w S / L q 180 deg. 0.79 f L q Z2 [ 12 d3+ 4 di - d2 1 , and when the pitch is j 12w slots have 0 = L 10'S q (1) ohms per phase all have 0 g For intermediate For dimensions measured in inches, the constant 0.79 becomes 2.006. The expression outside the bracket is a quantity containing factors common to pitches, the slot constant varies linearly between many of the elements of reactance, and so it is con- these values. Thus, by plotting the function KS venient to give it a special symbol, M, in future equa- against per cent pitch, where KS is a broken line tions. The bracketed expression is the "slot constant." function havihg values equal to For a fractional pitch winding, the currents in the q-m m 7r q at (1+cos two coil sides in some slots are not in phase, and so, a pitch, (2) KS= 2 q more detailed analysis must be made. If the current in coil side II lags that in I by 00, then in the slot carrying m being any integer, (as shown in Fig. 2), and substitutthe return conductor of I, in the bottom of another ing it in the expression for the slot constant, we have slot, there will be (under balanced conditions) a top as the final equation for slot reactance: conductor carrying current 0Q ahead of the current inl. d3 __ di + 8 + = M KS As the mutual inductance of I, with respect to II, is (1-Ks) 3w J 12w ( W equal to that of II with respect to I, the out-of-phase component of voltage induced in each coil side by thle -___ {Ks- ~-()J other will be canceled by the corresponding out-of 3JJ 4w phase voltage in the return coil side, leaving only the Th thr ter in thi exrsio.snglgbe in-phase components of voltage to be considered. The otalsel-indctane o I i proortonalto: except for zero phase-sequence currents, unless there iS a large space between the top and bottom coil sides, d1-d2 1r J as in the case of a multi-speed machine with two 4 d+ . ~~~~primary windings. The second term is very small,

(Sq =10, = 109(fL) (S/q) t Xslot

8 wr2

12'

Xsi0t

496

ALGER: SYNCHRONOUS MACHINES

Transactions A. I. E. E.

TABLE I unless the pitch is below two-thirds, so that only the VALUES OF Ks FOR USE IN EQUATIO-N (3) first term is ordinarily of importance. Winding pitch The slot reactance given by (3), with the coefficients Connection Type of winding 0 1/3 2/3 1 Ks of Fig. 2, applies when balanced polyphase currents Ki values flow in the windings. It is sometimes desirable to know the slot reactance for other conditions, however, Balanced threephase......... three-phase, 60-deg. belts 0 0.250 0.750 1.00 as for single-phase, or for zero phase-sequence currents. to neutral, three-phase, 60-deg. belts 0 0.500 0.500 1.00 These reactances can all be obtained from equation (3), Line one-phase... if proper values of Ks are used. The appropriate val- Line to line, ones

phase......... three-phase, 60-deg. belts Zero phase ....... three-phase, 60-deg. belts Balanced, threephase.......three-phase, 120-deg. belts Line to neutral,
Line to line, onephase ......three-phase, 120-deg. belts Zero phase.....three-phase, 120-deg. belts
twoBalanced, two-phase, 9C-deg. belts phase ..;.. Line to line, onetwo-phase, 90-deg. belts phase....

0 0 0

0.250 0.750 1 .00 1.00 0 1.00

.375 0.750 0.750

one-phase. one-three-phase, 120-deg. belts

0
0 0

0.250 0.500 0.500


0 .375 0.750 0.750 0 0

0
0 0

0.333 0.667 1 .00


0.333 0.667 1.00 0.333 0.667 1.00

Zero phase.....two-phase, 90-deg. belts

For intermeldiate lated values.

va1i s of pitch. Kh

varies linearly between the tabu-

v I

~~~~reactance.

reactance, plus one-third of the zero phase sequence


The leakage reactance of the end windings will here
iII.
END LEAKAGE REACTANCE

it it j iz1 + < 91|

kind is a revolving magnetic field of the same nature as those that are produced in the core length. This field flows in planes perpendicular to the shaft, and is

be considered as due to two kinds of field. The first

FIG.

2X/)(~

ues can readily be derived by the methods already employed, and therefore to save space the details of the work will be omitted here. The following table of special values of Ks covers most of the cases that are met with in practise. It will be noted that the single-phase line to neutral slot reactance of a three-phase winding is not half the reactance between lines, unless the winding pitch is exactly 0, 50 per cent, or 100 per cent. This is so because the mutual slot reactance between any two phases, A and B, is only zero when there are as many slots containing an A coil side in the bottom and a B in the top as there are slots containing a B coil side in the bottom and an A in the top; a condition that always exists with two-phase, but that only occurs with pitches of 0, 50, and 100 per cent with threephase windings. As shown by Park,1 the line to neutral reactance of a three-phase winding is always equal to two-thirds of the positive phase sequence

opc.7 t.,8r

:_:-

-c

C0,9X/,V4

eZ,9rlv/-

zwv

4/,

FIG. 3 AXIAL LEAKAGE FLUX OF END VVINDINGS

due to the axially directed and winding currents only. Hereafter it will be designated as the "peripheral leakage field." The second kind is due to the peripherally directed end winding currents, shown in Fig. 4, which produce a leakage field flowing in radial planes, as indicated in Fig. 3. This will hereafter be called the "axial leakage field." The peripheral leakage predominates for synchronous machines with barrel typ\e windings. With single layer chain type windings, the axial leakage flux is greatly increased, and so becomes the greater of the two. For induction machines, the peripheral leakage is greatly reduced by the proximity

Feb. 1928

ALGER: SYNCHRONOUS MACHINES

497

The first of these factors is given by the familiar of the primary and secondary windings, and so the expression: axial leakage usually predominates. V The procedure will be, then, first to find the per7 (8) f Kp Kd Z 10-8 meance of a cylindrical air core to a sinusoidally dis2 tributed encircling m. m. f., second to find the effective The second factor is: m. m. f. of this character produced by the end windings, 87 DL third to determine thence the reactance due to the = (9) peripheral end leakage flux; and, finally, to determine A gP g the additional reactance due to the axial leakage. This completes the story for a salient-pole synchronous and the third* is: 2 q Z Kp Kd A machine, where there is no mutual inductance between (10) the armature and field end windings. For induction 7r P / 2 machines, however, the further step of subtracting the The product of these three factors gives the magnetmutual reactance of the two windings from the sum of their total reactance is necessary. As shown by earlier writers3'6 l0, the total flux per - -pole produced in a circular air core of one cm. axial length by a sinusoidal m. m. f. distributed around the periphery is equal to just twice the maximum m. m. f. , A Or: in gilberts. 4w A 4 7r

=2 (10)

(4)

| l This equation assumes the flux to lie entirely in air %-and of the entirely core, periphery Cr in iron of high permeability on the other side. It makes _ the outor inside the is on air no difference whether the side of the core, and neither does it make any difference what the radius of the periphery of the core is. If there _ is no iron present, so the flux paths lie entirely in air, both inside and outside of the m. m. f., the reluctance is doubled, and so:
on one side of the
7 -<

--

.9
4'O

(.1)2=

10

(5)

PLO

The flux per pole produced by the same m. m. f. acting on a uniform air-gap of length g is equal to:
3=
.8 7r A D

0W //A. U19Lh94, YS'P?9c _o 8649'S 9A,9q,Vqg ,,v 73*7o 1-WIRAS

,9

ws ACE WZ/M o ?flhVy1 0 f,aurA-oe: ,3'r A ,9,V;' ?'E1 Po,4r /i v0 7Wo Co,'vPa rsWr

S/NU4

/DA91.

4,qrA

,7.

';:/ 'Acz
LS<44y

gP

(6)

FIG. 4 END WINDING CUTRRENT RELATIONSHIP

r4'P"

a given sinusoidally distributed m. m. f. acting on the periphery of an air cylinder of diameter D, with air outside, as well as inside the cylinder, to the flux produced by the same m. m. f. acting on a uniform airgap of length g, is equal to:

And, therefore, the ratio of the flux produced by izing reactance of a polyphase armature winding in ohms per phase: 8 7r q f D LZ2 KP Kd2 X0= P2 g 109
0.319 Kp2 Kd2 D SM on 'The ratio of the axial length of the end windings is:
(11)

-1)2

Pg(7) 2 D q53

both ends of the machine to the core length where P is the number of poles of the m. m. f. iv p D tan ae The magnetizing reactance due to the air-gap flux P L'(12) of an induction motor with uniform air-gap g is equal to the product of the following factors: 1. The volts per phase due to unit flux per pole; where p is the fractional pitch, and at is the angle 2. The flux per pole due to unit armature reaction; indicated in Fig. 4. *R. R. Lawrence, "Frinciples of Alternating-Current 3. The maximum of the sine wave of armature page 108. Machines," phase. in one reaction per ampere

498

ALGER: SYNCHRONOUS MACHINES

Transactions A. I. E. E.

Multiplying together (7), (11), and (12), the leakage reactance of the end windings due to the fundamental of the peripheral field only is: Pg D [0.319Kd2DS1 9 XEP M L p2g Jt 2 D )
7r p D tan

al

P L L

[pD S Kd2 tan al jF (P) J F (p) = M L 2 p2 L

where F (p) is that function of the winding pitch that corresponds for the coil end projection length to the function Kp2 over the core length. The distribution factor Kd is the same for the end windings as for the The enter of gravity of this current area core portion, due to the unvarying widths of the phase cos P 2 belts. To evaluate F (p), consider an elementary section of is accordingly located at a distance from the end of the the end windings at a distance x from the outer end, as core equal to indicated in Fig. 4. If x is expressed as a fraction of 1 the total axial winding projection on one end, the pitch 7 x dx of the winding at any point is equal top x. Therefore, x Cos p 6 the square of the ratio of the m. m. f. at any point to its r pD tan a 10 2P value in the core portion is, Cos P 7 x dx sin2 P
2 J 7r-p D tan ae /2tan p 7r/4 tapW/4) 2P
p

(13)

harmonics may be approximately taken care of by simply omitting the factor Kd2 in the first member of equation (13). The component of end leakage reactance due to axial flux may be evaluated by considering the peripheral components of the end winding currents at any fractional distance x from the outer end of the winding, as indicated in Fig. 4. The peripheral currents in the upper and lower layers of the winding are out of phase in time by an angle equal to p 7r x, which differs by 7r from the phase difference between the axially directed components of the same currents. Thus, the resultant peripheral current at any point is proportional to

and accordingly the value of F (p) is:


F (p)

sin2
0

dx=

2pw

(14)

(14)

sinpDpp2

pD tan a

1 +

P5

To evaluate the additional end reactance due to the irregular peripheral space distribution of the end winding m. m. f., we may proceed as follows: The ratio of the nth harmonic flux per pole in an air core to the fundamental flux per pole is n times greater than the same ratio for a uniform air-gap, (from equations (5) and (6) ). For a full pitch winding with a large number of slots per pole, the reactance due to the nth harmonic is therefore approximately 1/n2 times that due to the fundamental for an air core, and 1/n4 times the fundamental for a uniform air-gap, the latter ratio being derived in the section of this paper on belt leakage. The sum of the values of 1/n2 for all the harmonics present in a full-pitch polyphase winding with an infinite number of slots is equal to 1/Kd2.t The actual effect of the harmonics is less than this for windings with pitches a little less than 100 per cent, and greater for very short pitch windings, but these errors are not

(15) approx. We will now make the assumption that the axial end leakage reactance is the same as if the peripheral current were concentrated in a conductor of circular crosssection at each end of the machine at the distance from the core ends given by (15). The proper diameter of cross section to assume for these conductors is evidently greater than the depth of primary slot, and less than the axial length of the end projection on one end. For reasons of simplicity, we will assume this diameter to be one-half the axial projection at each end, or, from

(12):

7r

p D tan a 4p

(16)

important. The tooth harmonics due to the concentration of the m. m. f. in a small number of conductors are smaller in the end windings than in the core, since the finite slot depth makes the actual length of flux path much greater for the higher harmonics than in the ideal case, for which the formulas are derived, where the m. m. f. is concentrated in a cylinder of zero thickness. Thus, the increase in end leakage reactance due to this effective diameter enters the final equation, a ISilberstein, "Synopsis of Applicable Mathematics," p. 87.

As the peripheral current extends over a radial depth equal to the depth of primary slot, the equivalent diameter cannot be less than this. Hence (16) is really based on the assumption that the primary slot depth is about equal to one-third the axial end projection, and if wide variations from this relationship occur, then (16) will be correspondingly in error. The effect of this error is to make the calculated reactance too high for very deep primary slots and too low for very shallow slots. However, since only the logarithm of

Feb. 1928

ALGER: SYNCHRONOUS MACHINES

499

considerable error in the diameter makes only a small error in the result. The current in this conductor varies cyclically around the periphery, its r. m. s. value for a full-pitch winding

very closely. The substitution of 0.3 (3 p - 1) for the complicated function of pitch given in (20) is justified by the comparison of the two values shown in Fig. 5. Over the ordinary range of pitches from 0.5 to 1.2, the

being

, 2times the total axially directed current in adopt (21) as our final expression for the end leakage 2 v/ 2 reactance of a synchronous machine with a barrel-type primary winding. one pole pitch, or: Equation (21) gives the end leakage reactance, under \ {2 a/2) (2 1) ( Kp Kd q I Z) ( P Jt 7 2 -\/ 2 J

1 being

agreement is very satisfactory. We will, therefore,

qIZK,Kd
largely canceled by opposing eddy currents in the punchings, that are entered broadside on. Then, the peripheral currents at the two ends may be considered as flowing in a pair of parallel wires separated by a distance twice that given by (15).
with this distance, the problem reduces to that of calculating the reactive kv-a. of a single-phase transmission line consisting of round wires of a diameter given by (16), separated by a distance twice that givenl by (15), of a length w D, and carrying a current per conductor given by (17). The inductance of such a line is equal to: We will neglect the leakage flux that enters the ends of the armature core in an axial direction, as it is

_ _

Assuming the air-gap diameter, D, large compared

LeS | |7/|

Y
I

2+ 2 1 n 4 (1 + p 2/5)] henrys 20D [1


which is practically equal to: (8 4wrD (8.2 +p2) (5) 109 The total reactive volt-amperes due to the axial end leakage flux are, therefore, equal to: (19) 13.1 f q D Kp2 Kd2 Z2 (1 + 0.12 p2) XEA = P2 Adding (19) to (13), and dropping the Kd2 factor of (13) as previously explained, we obtain for the total end leakage reactance: p D S tan a F (p) 2 p2 L XEnd = M p

FIG. 5-END LEAKAGE REACTANCE COEFFICIENTS

polyphase currents flowing in the armature winding. As the derivation of this equation is based on the assumption that the end leakage fields are due to revolving m. m. fs., with nearly negligible space harmonies, the reactances for proportional other winding connections may be taken as to the corresponding values of armature reaction. On this basis, the coefficient 0.3 of equation (21) be changed in accordance with the following table may (1 + S p2) 0.12 Kp2 Kd2 D in the (20) + special connection cases: p2 J 6 L

10

the

condition

of

balaneed

TABLE__________II__ We will take a = 60 deg., a value higher than the ||Value of coefficient usual angle of the ends, but one that checks the mean for equation (21) Winding Connection length of conductor on most windings, and so one that . 0.3 orthree-phase allows something for the straight portion beyond the Balanced polyphase................... ........|two 0.23 Single-phase, line tO lineutral] two or three-phase core. Equation (20) then reduces to:
0.3 1M D 5 (3 p- 1) Zero phase .................three-phase, 60 or 120 Zero phase

End = p2 L

.... |two-phase

....|0.3

500

ALGER: SYNCHRONOUS MACHINES

Transactions A. I. E. E.

IV. ZIGZAG LEAKAGE REACTANCE Inasmuch as the stator windings of polyphase machines consist of coils arranged in a finite number of slots and connected in a finite number of phases, usually two or three, the shape of the wave of magnetomotive force along the air-gap surface is not sinusoidal, even though the winding carries balanced sinusoidal polyphase currents. At any particular instant the shape of the m. m. f. wave is stepped, the agreement between the stepped figure and a true sinusoid becoming closer and closer as the numbers of slots and phases are made larger and larger, until perfect agreement results when both the number of slots and the number of phases are infinite. As shown in reference 3, the air-gap flux produced by this stepped m. m. f. wave in a salient-pole synchronous machine consists of three elements: 1. A fundamental sine wave stationary on the rotor, inducing fundamental frequency voltages; 2. A series of harmonics moving with respect to the rotor and also inducing fundamental frequency voltages; and 3. Other harmonics inducing other than fundamental frequency voltages in the armature winding. Of these, 1 is the effect of the armature reaction, 2 constitutes an element of the leakage reactance, and 3 may be disregarded. Thus, if we now derive a formula for 2, we will complete the work of determining the armature leakage reactance. In the present discussion, to begin with, an ideal type of machine will be considered. The air-gap length will be assumed uniform, and the flux will be assumed to cross the gap radially. The slots will be assumed to have no real physical existence which might entail a variation in air-gap permeance, but will be assumed to be merely the points along the periphery of the machine at which the current is concentrated and at which, therefore, the m. m. f. changes. Under these conditions, the air-gap flux will vary along the periphery of the machine at any instant in a stepped wave shape exactly proportional to the m. m. f. at each point. If the flux wave produced by the stator currents alone be considered, it has been shown that the stepped wave of flux may be analyzed into a fundamental sinusoidal distribution of flux having the same number of poles as that for which the stator winding is connected, plus an infinite number of harmonic flux waves having different numbers of poles.3 There being no rotor currents by assumption, the magnitudes of the fundamental, and of all of the harmonic fields, are proportional to the current in the windings, the back e. m. f. produced by these fluxes in the stator winding is proportional to the current, and the fluxes are truly self-inductive. The reactance due to these harmonic fields constitutes the "air-gap leakage," or "differential leakage" reactance of the machine, which is usually divided into two parts, the "belt leakage" and the "zigzag leakage."

The distinction between belt- leakage and zigzag leakage is that the former is due to the concentration of the m. m. f. in a definite number of phase belts, while the latter is due to its concentration in a definite number of slots. No accurate dividing line can be drawn between the two kinds of leakage, as they are mutually dependent on each other, and as in the case of one slot per pole per phase, for example, they coalesce into the same thing. The reason for making a distinction is that, for ordinary design proportions, a change in the number of slots affects only the zigzag leakage, while a change in the winding pitch affects only the belt leakage, so that the independent effects of number of slots and of pitch can be more easily dealt with when the two parts are considered separately. Independent formulas will be developed for the belt and zigzag reactances, therefore, but it must be recognized that, whenever the number of slots per pole is small, the two are not separable, and the formulas become inaccurate. The only accurate way to obtain the total differential reactance is to compute it for each case separately, and tabulate the results. The computation of the total for any regular winding can be simply carried out by the means suggested by Chapman.' This has been done in Appendix A, wherein tables for some of the more usual regular windings are given. The definitions adopted will be as follows: The per cent differential leakage reactance is the total per cent reactance due to harmonic fluxes crossing the air-gap that are not of the fundamental number of poles. Or, it is the total excess per cent reactance, due to air-gap fluxes, above that of a similar machine, with an infinite number of slots and of phases. The per cent belt leakage reactance is the total excess per cent differential leakage reactance of a polyphase armature winding above that of a squirrel-cage winding with the same number of slots. The per cent zigzag leakage reactance is the total per cent differential leakage reactance of a squirrelcage winding, or of a winding with as many phases as it has slots per pole. In this section we will consider the zigzag leakage, or the harmonic fluxes produced by a squirrel-cage winding. It is shown in Appendix A by Chapman's method that the total inductance of a full pitch winding with one slot per pole per phase, as a ratio to the inductance of one full-pitch coil is: 1 Q =- CSC2 (22) s 2s where s is the number of slots per pole. And, it is also shown in Appendix A that the total useful inductance of such a winding, due to the fundamental flux it produces, as a ratio to the inductance of one full pitch coil is: 4s (23) Qu=

Feb. 1928

ALGER: SYNCHRONOUS MACHINES

501

Hence, the zigzag leakage, Xt, is equal to:


Xe-

KP =iKdn :=2 (7 CcS1r2s-) (f7 -d p 2N) 92d Po' + 1/2( Pi d Xt=Xa( ~- CSC2 ir/2 s -4 s2(24)pN+n where the plus signs are taken if the nth harmonic where X,, is given by (11). rotates backwards with respect to the fundamental, As 7r/2 s is a small angle, (24) may be developed into and vice versa. For the quadrature axis, the signs of a series of which only the first two terms need be the P2 terms in both the fundamental and the harmonic considered, so that: magnetizing reactance expressions change sign. 4 2 Xa 7.2 It may easily be shown that the product KP KPn2 V 7 (25) is always either negative or very small, and that xt = 102 l1+ 20 2 + 4S KdnKdn,2is always negative for the low order (belt or, approximately: leakage) harmonies, but is positive for the tooth 5 Xa 5 ' P X (zigzag leakage) harmonics. Hence, the numerical xt = 6S2 =6 S2 J (26) values of the two terms in the foregoing expression are additive for the belt leakage, but are subtractive for Equation (26), then, gives the desired result, the the zigzag leakage, and the contrary is true for the total differential leakage reactance of a regular winding quadrature axis. with one slot per pole per phase, or, by definition, the It is quite feasible to carry through the indicated zigzag reactance of any winding with the same number numerical work for any particular case, using these of slots per pole. Clearly the error in the formula is general methods, but to do so for the general case is small for large values of s, but it makes the calculated quite impractical. We will, therefore, make some value a little low when s becomes less than 3. The bold assumptions at this point, with the object of value of Xt thus found applies to a single winding, and obtaining a simple formula accurate enough for general to the case of a uniform air-gap length. use. First, we will assume average values for the For the case of a salient pole synchronous machine, permeance coefficients, corresponding to a two-thirds the zigzag leakage can be found from the same formulas ratio of pole arc to pole pitch, a minimum air-gap two if an equivalent uniform value of air-gap length can per cent of the pole pitch, and a maximum air gap 1.5 be ascertained. It has been shown by Doherty and timesthe minimum, as follows: 2/3 pmn Iu 12 nd N =13 N Nickle3 that the reactance due to the nth harmonic of pil On3this bs, h nt2hroni anetiin reactaNc primary m. m. f. depenids on the average permeance, On this basis, the nth harmonic magnetizing reactance and on the second harmonic of permeance variation only, and that these two permeances have different varies about 15 per cent, above and below its average, values for each harmonic. The effect of the variation of between the direct and quadrature axes. We shall permeance due to the salient poles is to make the leave this relatively small variation out of account, fundamental armature flux and the belt leakage flux therefore, and calculate only the average reactance, greater in the main axis than in the quadrature axis, assuming it to be the same in both axes. Then, the but to make the zigzag harmonic leakage fluxes greater ratio of the magnetizing reactance for the nth harmonic in the quadrature axis than in the main axis. This is to the direct axis magnetizing reactance of the fundatrue because the predominating portions of the high mental is approximately equal to: order harmonic fluxes are produced at the points of .9 Kpn2 Kdn2 Xan maximum armature current, and these come over the 2 K2 2 n K K Xad pole faces (low reluctance) when the poles are in the quadrature axis, while they come over the interpolar Finally, we must take into account the fact that our spaces (high reluctance) when the poles are in the main ideal assumption of true rectangular m. m. f. waves for each coil is not exact. Actually, the m. m. f. of each axis. The average permeance increases with the order of slot is distributed across the slot opening, instead of the harmonic, since the shorter the harmonic pole pitch, concentrated at a point in the center of the slot, and so the greater the relative permeance for the peripheral the steps in the m. m. f. waves are not vertical, but flux that reenters the primary iron without ever reach- slightly sloping. This reduces all the harmonies aping the pole face. For the same reason, the second preciably, especially the higher order ones. We shall harmonic of the permeance variation is less the greater assume that the average effect of this error is to reduce thevalue of n. the magnetizing reactance to 5/6 of its ideal value. The same authors have shown that the fundamental Multiplying the last expression by this factor, therefore, magnetizing reactance in the direct axis is proportional we find the proper average value of Xa to substitute in to: equation (26) is: 3 Kp,2 Kd,2( Po' 1/2 p2'), and that on the same basis the direct axis magnetizing Xa = Xad (27)
I

KPnKdn (KPnKdn

reactance for the nth harmonic is proportional to:

502

ALGER: SYNCHRONOUS MACHINES

Transactions A. I. E. E.

and rotor belt leakage must be added to obtain the V. BELT LEAKAGE REACTANCE The belt leakage reactance is due to the increase total. As a first step, it will be necessary to calculate the in differential leakage caused by reducing the number of phase belts per pole, q, to a small number. When the Xai number of slots per pole is very large, this effect can ratio Xa where X is the magnetizing reactance be readily calculated, but as the slots are made fewer, tmesitsof the nth harmonic. A measure of the effectiveness of or thee three times to two wo o the belt ncresesto eltleakge increases theleakage limiting value. When the slots finally become equal the stator winding in produeing fundamental flux is to one per pole per phase, the belt leakage becomes zero, the number of turns in series per phase times the pitch by definition. As the effect of the number of slots is and distribution factors for the fundamental number of further complicated by variations in the winding pitch, poles or N1 =K1 Kd1 N it is not considered practicable to present any adequate formula for the belt leakage in the general case. How- in the same way the effectiveness of the stator winding ever, as the belt leakage is small by comparison with the in producing nth harmonic flux is measured by Nn = KPr, Kd N, zigzag leakage, unless s is large, and as in this case the are where are errors these accurate, is fairly formula belt leakage KP,,. Kdn the pitch and distribution factors, of the stator winding for the nth harmonic respectively, to necessary is it not important. For accurate results, compute the total differential leakage reactance, and flux. tabulate the results for all useful windings, as outlined and iS to the iS applied a winding primaryby voltage to ~~~~~~~~~If zen 1X in Appendix A. fundamental entirely be opposed assumed first This is especially necessary in the case of fractional flux and then entirely by nth harmonic flux, the ratio slots per pole, as here the belt leakage increases greatly, of the magnetizing currents in the two cases will be the and as it varies considerably, depending on the partic- ratio of Xa to Xa. n total flux per pole, then ular sequence of large and small numbers of slots per If is the phase belt employed. = 4 N1 The harmonic fields corresponding to the belt leakage n are of lower orders than those of the zigzag leakage, as the former are primarily due to the phase belts which and if B is the air-gap flux density, then extend over several of the slots to which the latter are N1 due. Further, since all the harmonic fields must = B B1N in the producing voltages fundamental frequency create winding, and since the nth harmonic has n times the fundamental number of poles, it follows that each since the area of each pole for the nth harmonic is 1/n harmonic must revolve with respect to the producing timesthatforthefundamental. Since for the nth harmonic there are n times as many winding at one nth of fundamental speed. Therefore, the harmonic fields due to the primary winding induce poles to magnetize as there are for the fundamental, high frequency voltages in the secondary, and, if the the magnetizing current in the second case is: latter is short-circuited, produce damping currents in it. A regular phase winding, or a salient-pole field winding, n2 Imn N2n has so little admittance for these voltages that this damping effect can be neglected. But, in the case of a And, as the magnetizing reactance is equal to the squirrel-cage, the induced currents are sufficient to greatly reduce the belt leakage harmonic fluxes. The ratio of the applied voltage to the magnetizing current, zigzag leakage fluxes are only slightly reduced in usual t (2 1 / KPn2Kd,z2 cases, as the squirrel-cage admittance for these high aK 222 2 %X order harmonies is small. Therefore, the primary belt leakage can be practically The total belt leakage reactance is equal to the sum neglected in all machines with integral slots per pole per phase, having squirrel-cage secondary windings, and of the magnetizing reactances of all the harmonics existing with a definite number of phases, a definite so we will assume for this case: and an infinite number of slots. It is (28) winding pitch, XB =O()well known that a winding with q phase belts per pole For regular windings with an infinite number of slots produces only those harmonics whose orders are equal per pole, the belt leakage reactance is the sum of the to 2 k q 1. And, for all these harmonics, the dismagnetizing reactances of all the harmonic m. m. fs. tribution factor with an infinite number of slots is For synchronous machines the stator only need be simply equal to one nth of the fundamental distribution considered, while for induction machines the stator factor. Hence, we may write:

itsthsaorwnignpodcgfuamtlfuxs

Feb. 1928
=

ALGER: SYNCHRONOUS MACHINES

503

actual value of XB for any pitch for each of XB Xa > (30) type winding can not be expressed by any simple P1 formula, but must be found by actually carrying where n takes only the values 2 k q 1; k = 1, 2, 3, ... out the summation indicated in (30). This has been For a full pitch two-phase winding with 90 deg. phase done, and the results are shown in Fig. 6. belts, this is: The value of XB for a salient pole synchronous machine is also approximately given by the foregoing )equation, if 3/4 the direct axis magnetizing reactance is XB90o = Xa ( 3 + 54 + 74 + * used for Xa, as shown in the previous section. Summing up, therefore, the belt leakage reactance is equal 7F4 \to: (31) XB = 0 96 - 1 Xa = 0.0147 Xa (28) for any machine with a squirrel-cage winding and inteFor a full-pitch three-phase winding with 60 deg. gral slots per pole per phase; phase belts, it is: 3 XB - Xad (KB1) 1 1 (33) 1 (33 + XB60= Xa + 1 13 /74 14for salient-pole synchronous machine without squirrel1B (1 1 ) cage; and with integral slots per pole per phase; and 7r14)X
Go

ne

'

2The

14

=(I + XBg0) t1- 34

Xal

486

= 0.00214 Xa
eza _a

(32)

t tJW

X_

I 1B 114It

[tl

satisfactory until such time as an adequate table of differential leakage reactance values for irregular __ Q _ C _ _ _ r - - - -1-Nwindings is available. Since some of the harmonics of belt leakage flux for fractional slot windings are of fractional, and many are of rather low, orders, a squirrel-cage winding will have To<'- 1"1 t _ l / _ _ _ _ L t _ C I __ l _ _ _demagnetizing currents induced in it, which may reduce I4L4- 1t1t 1 111 \1ll F 4 ZFXXF the belt leakage reactance considerably. The magni~~~~~~~tude of this effect is extremely variable, however, as it _00161, XI I depends on many factors. It is the writer's opinion that the demagnetizing action of a squirrel-cage on the belt leakage reactance due to fractional slot windings is A V small for usual proportions, and so it will be IrbrEi+tl ll [ 3l/ nreglected in the design present paper. In these equations, we have derived expressions for /Po all the elements of the reactance, and we may now turn O /40 lo * Jo SX /AoF to their application to actual machines.
lav,
3 0

Xd 5 (34) S 8 for a salient-pole synchronous machine with fractional ~~~~slots per pole per phase. This last equation is arbitrary, asi XB for fractional slot windings varies very widely, but in a general way XB in such cases is larger the larger XT S, and it is on t e average of the same order of magnitude. From some unpublished work of Professor A. A. Bennett, the differential leakage reactance for six different fractional slot windings averaged 1.82 times XT. Hence, our assumption of 2 for this factor in the general case is not unreasonable, and should be
=

XB

XT=

p2

.5-

40

FIG. 6 CHART OF

VALUIES

OF KB THE BELT LEAKAGE

VI. APPLICATION TO SALIENT POLE SYNCHRONOUS

CONSTANT

For a full-pitch three-phase winding with 120 deg. phase belts, it is also 0.00214 Xa, but in this case it increases rapidly as the pitch changes from unity, due to the resulting dissymmetry of the winding, while in the other cases XB decreases as the pitch changes from

unitJr.~

The leakage reactance in ohms of a salient pole synchronous machine with integral slots per pole, and with a squirrel-cage winding, has been determined by (3), (21), (26), (27), and (28), to be: r S-_ ___d \ 0.3 (3 p-i) D) S 271= M LK5 I + wJ+ 3 p ~w wjP2

MACHINES

504

ALGER: SYNCHRONOUS MACHINES

Transactions A. I. E. E.

P 5 p (35) )2 Xad + 8 8( / The belt leakage given by equation (33) or (34) should be added in case the machine has fractional slots or has no squirrel-cage winding. The second order terms of the slot reactance (equation (3)) have been omitted as being non-essential for normal synchronous machines. Expressed as a fractional voltage drop due to fullload current, (34) becomes by (8), (10), and (11): 7.90 A r PLKS / d3 di K 2K2 + 31 j XI

Finally, therefore, our formula for the per cent leakage reactance, expressed as a decimal, of a salient-pole synchronous machine with a barrel-type armature winding, and fractional slots per pole, in inch units, is: P L 3p + 1\ d3 20 A di LS 4 A w + 3w) KP2Kd2 XI
P
d

0.3 (3 p- 1) D ] + + p

1.1 A
F

((P/S)2 + 0.6 KB)

(37)

If integral slots per pole are used, the (P/S)2 term should be multiplied by 0.5 If in addition the machine has a squirrel-cage winding, the KB term is to be omitted. The reactance is assumed to be the same in 2 both axes, but actually it is greater in the quadrature A K\ p 0.3 (3 p- 1) D 5 axis than in the direct axis by about 1/4 of the A/F F--) ( S 8p ]+ ( a (36) i + / F Sf 8 term. As the value of X1 found from test is totally different, for lengths measured in cm., the constant 7.90 becoming 20.0 if inch units are used. The value of F to be used depending on how the armature reaction is conin the last term of (36) is the number of normal voltage verted into equivalent field ampere-turns, it is necessary to consider the matter before proceeding to "ef=/o I ZLUX A2NDF,TetqL ,q () the comparison of test reactances with those given by

A/E ecF&i ,F0 227 r0e.,s^6 0J7 0oe9,* sea~ ,qae-:


AI

__

70

1.04

_ 74.2-I~ ~ ~<t-

/_ 7- _Z7 11}>sE Z& ^fJ 0 /I

.W3Wieseman2 has derived accurate coefficients by flux


X

_ _

VAO

,-

_> _ 4$ ~ wave _ of flux produced in a salient pole machine by the field turns, and by a sinusoidally disconcentrated t=t

plotting for the calculation of the fundamental sine


coefficients, the necessary factor, Ka,, to

tributed armature reaction. By taking the ratio of

convert the armature reaction ampere-turns into field ampere-turns, can be derived. This has t lr, B,06 e-equivalent ! > | 0ov been done by Mr. Wieseman, and the result is shown in T s_ - 0& SelFig. 7. By substituting the quadrature axis coefficients, /.021 _ _ 0. 060 0_ a 0.l75 also given in reference No. 2, for those of the direct axis, the armature reaction and the total, or synchronous, reactance in either axis can be accurately calculated. The usual way to determine the armature leakage reactance by test is first to take open and short-circuit ao3 o.07 voa 07e characteristic curves, which give armature voltage on open-circuit and armature current on sustained short FIG. 7 circuit, respectively, as functions of field current. field ampere-turns, and Ka is the constant required to Then, the field current corresponding to full load convert armature reaction ampere-turns into equiva- armature current on short circuit, divided by the field lent field ampere-turns. Ka depends on the particular current corresponding to normal voltage on the air-gap shape of field pole and length of air-gap used, but it has line, is equal to the synchronous reactance, expressed as an average value a little less than 0.9, and is given by a fraction. By subtracting the calculated armature reaction, also expressed as a fraction of the no load airFig. 7. A further simplification of (36) can be made by gap field ampere-turns, from the test value of synchronous reactance, the test value of armature leakage 3p + 1 _, which is exactly correct for reactance is found. The calculated armature reaction putting Ks = ampere-turns are given by (10), multiplied by the value three-phase windings with pitches between 2/3 and 1, of Ka from Fig. 7. As the errors of the test value of synchronous reacand which is slightly higher than correct for all other windings, (Fig. 2). As two-phase windings have higher tance probably average one per cent, and as Fig. 7 also belt leakages, and as the neglected second order terms involves slight errors, it is evident that this method of in the slot reactance become appreciable when the pitch obtaining the leakage reactance is inherently very is below 2/3, the small errors so introduced are on the inaccurate. On high-speed machines, the leakage reactance is very small by comparison with the total right side.
_

these two

004

005

Feb. 1928

ALGER: SYNCHRONOUS MACHINES

505

TABLE IV reactance, so that the results are particularly inaccurate MEDIUM HIGH SPEED MIACHINES case.TetX in this Tes TestXd ______Calculated It is possible to measure the total leakage reactance Xd Calc. Xd Xd Xid Xad Freq. Kv-a. Poles an for as of a synchronous machine at standstill, just 0.815 60 0.882 8 560 0.884 1.002 0.067 induction machine, but this requires the inclusion of the 800 1.416 1.001 1.415 1.338 0.077 60 8 reactances of the field winding and squirrel-cage, if 0.992 1.150 1.159 1.073 1200 60 0.086 8

state performance. It is also possible to measure the armature reactance with the rotor removed and to determine the leakage reactance as the difference between the total and the calculated reactance due to the fundamental of the flux produced in the air core. This last method involves large errors in the amounts of zigzag and end leakage, however, and is not as accurate as the usual short circuit test method of measurement. Finally, it is possible to insert exploring coils on the armature surface and by their means measure the net flux existing during a short-circuit test, which gives a measure of the leakage reactance. The dissymmetry and wave form errors introduced, unless the exploring coils exactly follow the armature coil grouping, together with the fact that the slot leakage part of this flux links only a portion of the winding, make this method of no practical use. For these reasons, the usual method of finding X1 from open and short-circuit tests has been followed in ehecking the results obtained with the formulas derived in this paper. HIGH_SPEED_____ACHI_ES
Calculated

the calculation of do not pre-sentwhich d te n o enter present,

steady

8 10 10
10

2500

60

0.143

2.100

2.243

2.230

0.994

5000
7500

1880

1100

'60

60

0.161

60
60

0.061

0.097

0.983
0.853

1.750

1.911

1.928
0.922

0.080

1.139

1.080
0.933

1.200

1.214

1.098

0.988
0.994

1.009 1.019 1.012

12 12 12

12 12

1000 1900 2400

25 25

12

2500 5000

60

0.081 0.089

60
60

0.159 0.101

0.810 0.758

12 12

12 12

12500
15000

5300 7000

5000

60 50

60

0.093 0.132

1.103

0.891 0.847

12

15000

50
60

0.135 0.115
0.137

0.750
1.280 0.900 0.667
1.065

1.452 1.560

1.262

0.842

0.905

0.851

1.692

1.545

1.245

1.564
1.667

1.016 0.986
0.985

1.012

1.415
1.015

0.852

60

0.086

0.110

1.048

0.753
1.202

1.158

1.414 0.987

0.766
1.208

1.172

1.001 0.999 0.972 1.012 1.017


1.005

16

3500

50

1.083 1.071 0.989 0.969 0.114Average ...........1.000 -.010 Average error from mean ......3

Averageerrorfrom 1..010

MIEDIUM LOW-SPEED MACHINES


Poles Kv_a.
18 18

TABLE V

Freq.
40 60 40 50
60

__i _Calculated
Xld
Xad
0.108 0.123 0.148 0.123
0.123

Xd
0. 9 52 1.084
1.123

Test -

Test Nd

Nd

Cale.

Xd

1.650
1,650

0.844

TABLE III HIGH SPEED M\ACHIN'ES

0.961
0.975

0.993 1.100 1.136 1.023 1.188 1.198

1.043 1.015

Poles

Xd Xad 22 Xld Ky-a. Freq. -_____.........22 _______ ____ .2____________ 1 0.984 1.043 1.060 0.955 0.105 60 4 90 0.981 1.250 24 1.274 1.220 0.054 50 105 4 24 0.971 1.032 1.063 1.002 0.061 60 1S0 4 24 0.987 2.195 2.223 2.163 0.060 60 600 4 24 0.998 1.895 1.898 1.770 0.128 940 I 25 4 24 1.018 1.793 1.762 1.562 25 0.198 2000 C 4 0.974 2.355 2.417 2.300 0.117 25 3000 4 28 0.980 1.937 1.966 1.857 0.109 25 5700 4 0.954 1.092 1.142 1.040 0.102 6 25 150 1.007 30 1.068 1.061 0.998 0.063 69 187 6 1.015 0.872 0.859 0.804 0.055 6 60 2C0 0.967 1.275 32 1.319 1.250 0.069 60 6 4Cl00 32 0.973 1.172 1.204 1.127 0.077 60 435;) 6 0.998 32 2.212 2.216 2.065 0.151 60 500 6 0.978 1.678 1.715 1.610 0.105 60 500 6 36 0.991 2.040 2.C59 1.932 0.127 60 1000 6 1.013 36 2.145 2.117 2.025 0.C92 60 6 1500 1.014 1.133 1.019 1.117 0.098 25 :315 6 1.003 1.161 1.157 1.060 40 25 0.097 5000 6 40 1.039 1.745 1.680 1.592 0.088 25 5000 6 Average ..................... 0.999 0 ).017 Average error from mean ..... Average error from I ......... 09.012

Test Xd

Test Xd __ _ Calc. Xd

20 20

14,444

2,250

0.896
1.078 1 017

1.019

1.011 1.004
0.989

2,000 7,900
1,100 4,000

60

0.162

1.201 1.179

1.016

30,00

tO,700

6.880

50 50 50

60

0.132 0.136
0.131 0.120 0.128

0.992
0.6 18 0.617
0.919 0.973 0.820 0.918 1.064

0.688

1.027

60

1.163 0.819 0.768


0.775

1.124

1.152

1.178

1.025

0.826 0.761 0.803

0.991

1.013 1.009
1.036

10,125 3.125 2,250 9,375

60
60

0.147 0.106 0.138 0.180

I.Q66
1.079

1.089
1.080

1.022
1.001

750

60 40 60
60

0.958
1.098

0.951
1.113

0.142
0.117

1.206
1.008

1.202

0.993 1.014 0.997

1,50()
750
1,375

500

60
60 60

0.155

0.853 0.881

0.998

1.003 0.988

0.995 0.990
0.987 1.013

0.918 0.906 0.756 0.162 0.716 0.707 0.619 0.088 Average ............. mean . Average error from .0 Average error from 1

1.008 0. 013 .014

In Tables III to VII inclusive, the calculated and test values of synchronous reactance are compared for five groups of machines of different types. The machines were selected at random, and the tests were all made in the regular commercial routine. The average absolute error for the entire list ofl100machines is 0.4 per cent, and the average numerical error is 1.7 per cent, so that the results are as accurate as could reasonably be expected when the errors of test and those

inherent in Fig. 7 are considered. If the value of the armature reaction reactance, Xad, and the test value of total reactance, Xd, are assumed to be exact, the test value of leakage reactance can be found by subtracting the one from the other. This has been done for each group of machines, and the average errors between the test and calculated values of X1 for each case are given in Table VIII, together with the corresponding errors in Xd.

506

ALGER: SYNCHRONOUS MACHINES

Transactions A. I. E. E.

The average absolute error of the leakage reactance for the 100 machines on this basis is 0.6 per cent, and the average numerical error is 14.6 per cent. The worst errors occur for the high speed machines, and inspection of Table VIII indicates that the calculated value of Xd is appreciably too high for high speed machines and a little too low for low speed machines. By empirical
TABLE VI LOW SPEED MACHINES
Poles
56 56 56 60 60 60 60 64 64 64 68
72 72

each one uses. The formulas as given were derived straightforwardly from theoretical considerations, and no attempt has been made to make empirical corrections, although simplifying assumptions have been freely used. The causes for the varying magnitudes of error shown in Table VIII are brought out more clearly by a study
TABLE VIII
Percent Numerical
error

Kv-a.
3,500 4,500 5,720 750 1,500 1,500 2,500

Freq.
60 60 60 60 60 60 60 60 60 60 60

l__l_________________d__
Xld Xad
0.144 0.147 0.159 0.177 0.155 0.231 0.130

Calculated

Tes

0.798 0.860 0.648 0.699 0.871 0. 587 0.724

Xd

Test
Xd

Test Xd

Test
-

Calc. Xd
1.019 1.006 0.960 1.030 1.015 1.078 0.994 0.955 1.015 1.013 1.003 1.008 1.026
-

0.942 1.007 0.807 0. 876 1.026 0.818 0. 854 0.950 0.998 1.030 0.958 1.057 1.041

0.960 1.013 0.775 0.902 1.041 0. 882 0. 849


0.907 1.013 1.043 0.961

Type of machine

Calc.

_ Per cent -_ Absolute error From 1


31 11 11 12 8
1.9 1.0 1.4 2.2 1.7 1.9

_ From mean 30 11 10 10 7
1.7 1.0 1.8

625 6,250 8,000 1,500

0.168 0.128 0.254 0.192


0.239 0.223

0.782 0.870 0.776 0. 766


0.818 0.818

of the relative proportions of the various elements of the total reactance for the different types of machines. Table IX gives the average ratios of slot, end, and differential leakage to the total leakage reactance; and also the average ratio of the total leakage reactance to 0.510 0.652 1.024 0.127 0.637 60 100 1,250 the armature reaction reactance for each case. Since in Average ..................... 1.011 Average error from mean ........ 0.018 the high-speed machines, Xi averages only 6 per cent of Average error from 1 .......... -0 .022 Xd, while in the small low-speed machines it averages it is clear that a given error in the meaadjustments of the coefficients of the end and differential 25 per cent, leakage components of the reactance, it is evident that surement of Xi makes more than four times as great an these errors could be reduced. Such adjustments of the apparent error in Xi in the former as in the latter case. constants should be made by each manufacturer for _TABLE_IX himself, however, as they will depend upon the parAverage ratios ticular end winding constructions, and so forth, that
72

72 72 72 80 82

1,000 12,250 32,500 3,500 2,250


2,250

750 875

60 60

60 60 60 60 60 60

0.228 0.147 0.213 0.166 0.187 0.162

0.740 0.813 0.782 0.926 0.810 0.745

0.968 0.960 0.995 1.092 0.997 0.907

0.962 0.983 1.015 1.130 1.020 0.898

1.065 1.068

1.06 5.7 ......... Low-speed (V) ......... 1.04 3.6 ... Small low-speed (VI) ........... Errors in values of Xd - 0.8 0.992 High-speed (II) ......... ........ 0.0 Medium high-speed (III) ......... 1.000 0.8 Medium low-speed (IV) .......... 1.008 1.1 1.011 ......... Low-speed (V) ......... 1 .019 1.0 ..... Small low-speed (VI) ........

........ High-speed (II) ......... Medium high-speed (III) ......... Medium low-speed lV) ..........

Errors in values of XI -13.6 0.86 1.01 1.0 6.3 1.06

1.3

0.994 1.024 1.020 1.035 1.023 0.990

X slot TABLE VII SMALL LOW SPEED -MACHINES

XEnd

XDit-

Xl
_

XI new
Xi old
0.56 0.58 0.66

Type of machine

Xl

XI

Xl
0.03 0.08 0.08 0.17

Xd
0.059 0.087 0.132 0.189 022

Calculated

poles
Oe 36 40 44

Kv-a.
200 240 80

Freq.
60 60 60 60

Xld
0.249 0.244 0.281 0.284

Xad
0.970 0.833 0.660 0.576

Xd
1.219 1.077 0.941 0.860

Xd

Test Xd r Xd PoesK Cale.

Xd

36

42 44

40

80

100

62

360

120

60

160

60

60

0.302

0.245

0.704

44

50

5406 52 60 60 62 74

44

1040

62.5

62

60 60

60

60

280
375

140 60 250

0.224 0.781 1.015 1.005 1.010 . 0.017 1.010 . ....... Average mean..........~. Average error from 1 from ____ ___________ 0.019 Average error ....... =

660 60 60 . 60 60 50

0.293 0.430 0.251 0.143

0.208

0.793 1.013

1.006
1.038 1.221

1 07 1.007 1.2281 228 1.076 1.001 0.962 1.022 0.888 1.033

~Low-speed

0.60 High-speed ............. 0.37 0.35 . ........0.57 XMedium high-speed. 0.32 Medium low-speed...... 0.60
. Small Saio-pe

low-speed

....

-.....

--

........ .

0.67 6 .58 .8

0.16 0.16 01

0.261 .6

0.2521

0.72 07

0.70

1. OCS

0.1008 0.3012 0.409 0.o404 0.988 0 .2909 0 .9012 1.192 1.183 0.992 0.195 0.6579 0.8521 0.826 0.9 69

0. 277l 0.207

0.961 0.972 0.619 0.417

0.4698 0.69475 0.69658


0.611 0.818 0.838

1.254 1.402 0.870 0.560

1.307 1.460 0.906 0.552

1.017 1.256

0.980 1.029

1.002

1.042 1.041 1.041 0.986

dma in erosiXifrhg-p errorsinX1forhigh-speedmachines.

This partially accounts for the relatively large numerical


The last column of Table IX gives the average ratio of the new value of leakage reactance, to the value given by the older formulas derived by Doherty and Shirley.7 This ratio varies from a little over one-half for the high speed, to nearly three-quarters for the lowspeed machines. The percentage error between test and calculated values of Xl iS thus magnlified on the new b.asis, so that a direct comparison of the accuracy given by the two formulas cannot be made. The older of about 16 per formula had an average apaeterror aprn cent, so that taking into account the test errors, which

1.010

1.024

Feb. 1928

ALGER: SYNCRHONOUS MACHINES

507

cause a considerable part of the dispersion, the new machines, and the values compared with tests, with the formula is indicated to have about half the true error following results: of the old. Zero-phase Assumed Sequence Reactance VII. ZERO PHASE-SEQUENCE REACTANCE value CalcuPer cent CALCULATIONS CALCULATIONS error of Ko lated Test pitch Freq. ~~~~Poles Kv-a. PoWinding The zero phase reactance is important in the calcula- --6 -24 435 60 0.026 0.021 2/3 4 60 0.80 0.25 0.070 0.0675 tion of single phase short circuits, circulating currents in 36 20,000 delta windings, and in other cases. The slot portion VIII. CONCLUSIONS of this reactance is given by the complete equation (3), It is believed that equation (37) is at once the most using the appropriate values of KS from Fig. 2. For usual three-phase windings, KS for the zero phase accurate and the simplest comprehensive formula for sequence reactance is equal to 3 p - 2. The end the armature leakage reactance of a synchronous maleakage portion is very small, and will be neglected. chine that has been published. It requires no curves, The differential leakage portion is given by the ratio of no logarithms, and no tables for its use, but only a few equation (12a) or (13a) to (14a) in Appendix A. Com- slide rule operations, and it gives a value of leakage bining these expressions, we derive the following formula reactance which, added to the armature reaction for the zero phase reactance of a three-phase, 60 deg. reactance derived from flux plots, quite accurately belt winding, with a pitch between 2/3 and unity, checks the test values of synchronous reactance for the entire range of salient pole synchronous machines in expressed as a decimal: use. commercial PL 20 A The leakage reactance so determined averages about K Xzero phase K2 ad W oS,/two-thirds of that given by the widely used formulas np (p derived in reference No. 7, since the latter included as leakage reactance a part of the fundamental sine wave (3 p -2) (d3) + (9 p 5) 12 -(9 p -8) 12 1 of air-gap flux due to the armature, which links the field winding in the direct axis, and thus constitutes a 4 A Ko (p-2/3) part of the true armature reaction. The new value of F Kp2 Kd2 leakage reactance, added to the squirrel-cage reactance, checks observed standstill reactances of synchronous 2 (P motors; and added to the field reactance checks the + + 7/18 ( -2/3)-( -2/3)2 (38) values s of transient reactance found from oscillographic 27 tests on synchronous generators. With the old reThe corresponding formula for a winng pitch actance formulas, rather arbitrary reductions of the between 1/3 and 2/3 is: calculated field reactance and tooth tip reactance were necessary before checks with transient and stand/ L 20 A still reactances could be secured. Finally, with the old Xzero phase = ) formulas, the calculated increase in internal voltage of a synchronous generator under load gave much higher d di [(2 - 3 p) d3 + (7-9 P) 12 -(4 - 9 p) 12 flux densities in low power factor machines than actually do occur, so that arbitrarily reduced values of L field leakage under load were used in calculating satura4 A Ko (2/3 p) tion curves, thus establishing a series of compensating / ( + A F K 2 K,2 errors. There is no space to demonstrate the validity of 2 \ statements here, and they are simply made to these (39) ( ) 2 S show that the acceptance of the new values of leakage reactance involves a fundamental revision of the The coefficient Ko is introduced in (38) and (39) to generally adopted design constants of synchronous allow forthereductionoftheharmonicfluxes,especially machines. In writing the paper, an effort has been the third, by induced currents in the rotor circuits. made to make the new formulas and the new definitions As all the terms except the first in the last brackets of such that the further refinements which the future will these equations represent third harmonic fluxes, thesebrneabededwtotginleigouen terms are generally reduced to a fraction of their appar- brpiong can bhe addeds wtout magainud althern ouaracon-sti ent values, and K0 should ordinarily be taken as less ceptonsts or teoders o antdeo h haatrsi than 0.5. The coefficient 4 of the last bracket in (38) cntnso ein NOMENCLATURE includes the factor of 3/4 for the average permeance of A = Maximum ampere-turns per pole of a salient pole machine, as previously derived. armature reaction The zero phase reactance has been calculated for two

-(2/3-p)2l

508

ALGER: SYNCHRONOUS MACHINES

Transactions A. I. E. E.

di, d2, d3, w = Slot dimensions shown in Fig. 1


f
D
F = Gap diameter
= Frequency = No-load air-gap field ampere-turns per

KB
Ks
L

= Air-gap length (radial)


=

pole

KP, Kd

= Pitch and distribution factors of armature winding (less than unity) = A function of winding pitch given by

Belt Leakage constant, given by Fig. 6

M
p P q s S V Xa

Fig. 2 = Gross core length 2f L q Z2 - 1O7 S for inch units

XB Xd

= Winding pitch expressed as a decimal = Number of poles Number of phases = Slots per pole = SIP = Total number of slots Volts per phase = Armature reaction reactance

ductance of a regular, distributed, polyphase winding with s slots and q phase belts per pole (s/q a whole number), and a uniform air-gap length, following Chapman's method. In doing this, the self-inductive voltage produced in the set of coils forming the inner halves of two adjacent phase belts in the same phase by their air-gap flux is considered as the sum of the s voltages produced by the fluxes in the s teeth under one pole. The voltage due to any tooth is proportional to the square of the number of turns linking it, as the reluctance of each tooth pitch is the same, while both the flux per ampere and the volts per unit of flux are proportional to the number of turns. It is most convenient to make all the calculations in relative terms, and, therefore, to take as the unit of inductance the self-inductance which the two adjacent half-phase belts of one phase would have if all their turns were concentrated in one pair of slots exactly one pole pitch apart. Assuming the winding pitch to be b slots short of
-, -llI
P_j 2345
17

=Belt leakage reactanceofpiA

XI XI
Z
a

or synchronous, reactance in direct axis Total armature leakage reactanceFlxwt PasB = Zigzag leakage reactance - Series connected conductors per phase =Angle of end windings as shown in Fig. 4 = Angular phase difference between currents in upper and lower coil sides in one slot =Flux per pole in c. g. s. lines

Total,

76''4 3

7t

321

A
:~_PaeB

Appendix A
EXACT CALCULATION OF DIFFERENTIAL LEAKAGE REACTANCE has shown that the total inductance of a Chapman* winding due to the air-gap flux can be very conveniently calculated by adding the inductances due to the rectangles of flux produced by successive pairs of symmetrically located slots. With this method the smallest unit of flux is that produced by a uniform m. m. f. acting over one slot pitch, so that the permeance variations due to the slot openings are quite satisfactorily taken into account by merely using the average permeance, as in calculating the magnetizing reactance. Thus, the point of view leads to the immediate conclusion that all the harmonic fluxes; (hence the zigzag and belt leakage reactances) are exactly proportional to the fundamental magnetizing reactance. And so the elaborate analyses of the permeance for the zigzag leakage flux as distinct from that for the fundamental flux, that have frequently been employed heretofore, are shown to be unnecessary. We shall now proceed to calculate the total in*Reference No. 13.
FIG. 8

IDFALMM\F.

DIAGRAMS FOR

POLYPHASE WINDINGS

full

pitch,

(b being not greater than

s/q), and referring

+ 1b central teeth q that are linkedj2by all the turns in the two half belts of phase A, and hence the total inductance due to these teeth will be: 1 (q 1) 1 (q-b q s\ ,
s

to Fig. 8 there will be (q

s wl aeaprene2so h oa emac,s ta hywl otiuet h nutne


- ) 2 (s_ S Similarly, consecutive pairs of teeth will each have

onswl.elne by s-q ofteoalursan

Feb. 1928

ALGER: SYNCHRONOUS MACHINES

509

qls less turns than before, until the bth pair is reached.

The next pair of teeth after that (1 and 1') will be s- 2q- bq linked by of the total turns, so that they will contribute an inductance equal to: 2 S3(s - 2 q-b q)2

is:

The following pairs of teeth will have consecutively 2 q/s less turns than the last, until the outermost pair is reached. Hence, the total self-inductance of Phase A
L =

S2 (1- b) +2

(s-q)2 + 2 (s-2q)21

1 [ S3 (q1)

Of the remaining teeth there are b - 1 that carry full flux, but are linked by successively fewer coils of the B phase. These contribute an inductance equal to: 1 co COS [(S -q) + (S -2 q) [ + cs + . . . + (s -(b -1) q)] There are also b - 1 pairs of teeth, (3' and 3") one of each pair of which carries positive flux and the other an equal negative flux, the former of which is linked by all, and the latter by successively less of the coils of phase B. Each pair is linked by successively less of the coils of phase A. Hence, these teeth contribute +an inductance to phase B of: S3 s3 cos
q

* . . + (b -1) q (s -q)] . ~~~~~~+ The remaining teeth consists of pairs (6, 5-7, 4"As the sum of the squares of the first n integers is 1', 1"- 2',2") carrying equal and opposite fluxes and n (n + 1) (2 n + 1) linking the same numbers of coils of phase B, so that to: be , may this reduced 6 they contribute nothing to the inductance of phase B. Hence, the total increase of inductance of phase B r 6 2 z1 S bi due to phase A is: L 2 S s- q + 1- b [63
+22

+...-+ 2(s-bq)2+2 (s-2q-bq)2+2

(s-4q-bq)2+...I

[q (s- (b- 1) q) + 2 q (s- (b- 2) q)

(S 1)

(1)

(281)

+2q2 (S _ b-b) (s

b)b)s -b-)]

b+1 Cos q c r/q[S2 (S+ s(s-q+s-2 q + . . . + s- (b-1) q) +w q(s-o(b-)q)+2q(s-(b-2)q)

su
-

an thi reue to 2
andthisredueesto:

+ 1) q (s-q) + 3 + (b)- [ q-2 bb q s owq LAq


(2s - 3b2S- b q +
<

b2b21
(b2 -1)] ob

LAA

21

qS

Vq),

~~~~~~~~LAB cos r/q [


=

-6

q(a

, q
q1

zero if it is negative, since the inductance is always


positive.
s

all for a winding pitch not less than

The linkages produced by phase A with phase B are similarly found, except that, as the currents in the two phases are out of time phase by 7r/q degrees, and as conditions of symmetry make the out-of-phase components of the linkages cancel with the corresponding components due to the other phases, the effective and 3b q current in phase A must be reduced by a factor cos ir/q. r 1 The number of teeth that are linked by all of phase B LAB = cos 7r1/q L 2 q + 2 82 + 28 as well asby allofphaseAis (s
q -b +

The corresponding expressions for values of b between q and 2 slq, corresponding to winding pitches between one and two phase belts short of full pitch, are, if q is greater than 2: 1 1 2s (3a) LAA 1-3 q - S2 (b s-q/3), s/q < b 7r q

11, and

bsq J b2q

3S 1 bq

(4a)

these will create linkages with B equal to:

(+)ceos

(s-

-b-+Fl)

s/'q <b <

510

ALGER: SYNCHRONOUS MACHINES

Transactions A. I. E. E.

And, the value of the self-inductance of one phase for winding pitches greater than zero by not more than one-phase belt width is:

3 7r / 2 7r 2 = 1-Cos 7r/s + 2 cos s + 3 cos s


+

LAA= q (s

) [q+(s-b)(3s-qs+b
s

q)]
(5a)

.+(s-1)eo (s) w)
= cs c2 7r/2 s
(a)

<bs

It is interesting to compare the total reactances of the There is no need of considering values of q greater than 3, except for full pitch windings, as three-phase several types of windings by means of these formulas. belts per pole are all that are used in practise. A For a regular three-phase winding, equations (9a) and squirrel-cage winding always has one slot per pole per (lOa) show that: 2 10 phase, so that its differential leakage ratio is the same S" 9 for full pitch = + = 1, and For case of this pitch. values for all L60 s,/q (2a) reduces to: 7 149 s- 2 S + (6a) LAB = 144 4 s2 for 5, 6 pitch COSwr, S

and the corresponding expression for the nth phase is: (n-1)w s-2 (n-1) (7a) COS LAN = 8 s The total inductances per phase are obtained by adding the self and effective mutual inductances, as given by the foregoing expressions. For q = 2, (2a) is zero, so that the total inductance of a two-phase winding is from (la): 1 (8a) 3 (4s-6b2s-4 b + 4 b3) Lgoo =- 2/3 For q = 2, (la) and (5a) are identical, so that the single equation (8a) is valid over the entire range of pitches from 0 to 1, for two-phase windings. For a three-phase winding, q = 3, and the total inductance of one phase is given by the sum of (la) and twice (2a) for pitches between 2/3 and 1, and by the sum of (3a) and twice (4a) for pitches between 1/3 and 2,/3:

3 5 6 + 282 for 2,3 pitch


5 9

for 1 2 pitch

1 5 18 + 2 2 for 1/3 pitch For


a

regular two-phase winding (8a) shows that:


Lgoo =
4 2 3 + 3 2 for full pitch
s

1 9 16 + 2 for 3//A pitch 1


+ 3

2
s

for 1,2 pitch

48 + 32 for4pitch For a single-phase winding made by using two legs 0 < b < s/3 (9a) of a regular three-phase winding, the total inductance is found by substituting 1 for cos 7r,/2 in (2a) and adding and (la) to it, which gives identically the same equation 1 L60= 19/18 + 2 2 (5 s + b S2-6 b -9 b2 s + 6 b3), (9a) as for the original three-phase winding. Thus, sl the per cent line-to-line reactance of a three-phase winding is always identical with the per cent threephase reactance. For a single-phasewinding consisting 2s (lOa) of one leg of a three-phase winding only, the total s/3 < b < 3 inductance is given by (la), (3a), and (5a) ,whence: Line to neutral (9a) and (lOa) become identical if b = s/3* 2 7 For a squirrel-cage winding, the total inductance of = + 9 s2 for full pitch L600 one phase is, from (la) and (7a), placing b = 0 and q = s in the former: 3 17 s-4 2wr s-2 - 24 + 2 s2 for 5/6 pitch + LK =1 +. <;COSw/S + S COSs

L60

3 10/9 + 2s 2sl (4s-6b2s-3 b+ b3),

Feb. 1928

ALGER: SYNCHRONOUS MACHINES

511

1 2 for 2/3 pirch

1 for + s2 18 7

1/2 pitch

2 1 9 + s2 for 1/3 pitch


1 5 72 + 282 for1/6pitch Thus, the line-to-neutral air-gap flux reactance of a three-phase winding operated single-phase is 7/10 of its value in three-phase operation for full and half-pitch windings, but is only 2/3 of its three-phase value for 2/3 pitch windings. For a 1/3 pitch winding, it is 4/5 of its three-phase value. These variations are due to the triple harmonic fields that are produced by each phase winding separately, but that are canceled out by a 2/3 pitch, or by a three-phase connection. Since the zero phase sequence reactance is equal to three times the line-to-neutral, less twice the balanced three-phase reactance, of a three-phase winding, the foregoing expressions lead to the following expressions for the zero phase sequence inductance: Zero phase

reactance will give the desired values of total differential leakage reactance. The ratio of the useful to the total flux generated by a single full-pitch coil is 8/7r2, as this is the ratio of the area of a sine wave to the area of a rectangle whose fundamental it is. This factor, multiplied by the squares of the pitch and distribution constants of the winding, and by the ratio, q/2, of the total revolving field produced by all the phases acting conjointly to the field made by a single phase, gives an expression for the reactance due to the useful flux. Or:

Lm

4 q3 sin 27r/2 q sin2 7r/2 (1 - b/s) S2 sin2 ,/2 2

a)

reactance:

Putting q = 2 in (14a), dividing it into (8a), and subtracting unity from the result, we have for the total differential leakage reactance of a two-phase winding, expressed as a ratio to the fundamental magnetizing

XD90 1r2 (s-b) (S2+ b s-2 b2 + 2) sin2 w/2 s -1 24 s sin2 7r/2 (1 - b/s) Xm'
If s is large, and b is zero, corresponding to the case of a full-pitch two-phase winding, (15a) gives:

(lSa)

L60,.

XD,, w2 (82 + sin2ir/2s1 s- 3 b2s- 6 b + 6 b3), + n r=9 9 s3 (2 2s-1 96 24 0 b s/q (12a) 74 . 687 1 5 - +61 -5 (16a) (.412 +4882 (2s 3 +4bs2-6b-9b2S+6b3) .S2 s348s 9
-

~~~~~~~Xm

-= q
Hence:

<

8
<

(13

Zero phase L60

+-~2 for full pitch Zero phase L60~ =+ s2 for full pitch 9
- 18 +

If s is very large, (16a) reduces exactly to (31), thus confirming the latter result. For small values of s, equation (15a) is the only convenient means of obtaining exact values for XD60. Numerical values' of it

for some usual windings are given in Table X.


TABLE X
DIFFERENTIAL LEAKAGE REACTANCE OF

18
1

s2

for 5/6 pitch

WINDING EQUATION (l5a)


2 3 4

TWO-PHASE

0
1

for 2/3 pitch

4 s2 for -18 +
1

1/2pic / pitch

Pitch deficiency 15
0 1 2 3

s/2 = slots per pole per phase


1
0 2337

6 0 0229 0.0193 0.0138 0.0096

0.2337

0.0840

0.0583 0.0840 0.0840

-2 for 1/3 pitch s-2

4
6

0.1019 0.2337 ----5


-

0.0334 0.0285 0.0468

0.0468

0.03:30

0.0251 0.0177 0.0189


0 .1222

0.0330 0.0649

0.0264 0.0213 0.0147 0.0116

0.2337 0.0839 0.0398 7 l All of this zero phase inductance is leakage flux, since 0.1382 0.0650 the fundamental m. m. fs. of the three-phases cancel 8___ - ___ with this connection. Putting q = 3 in (15a), dividing it into (lOa) and The useful, or fundamental, flux produced by the same windings can be readily calculated from the usual (lla), and subtractx,ng unity from the result, the total pitch and distribution factors, and the difference re- differential leakage reactance for a three-phase, 60 deg. maining after the useful is subtracted from the total phase belt winding is:

0.0148 0.0265 0 .0486

0.0089. 0.0130
0.0229

512

ALGER: SYNCHRONOUS MACHINES

Transactions A. I. E. E.

XD6, XD60
Xm

xm

Is -S- 1, 7r2(20 s3+36 s-54 b's-27 b+27 b3 sin'7r/2 +3 -4


486 s sin'2 r/2 (1 -

bl/s)
=

nous Machine Design, TRANS., A. 1. E. E., Vol. XLVI, p. 141.

1. 1928, Park, R. H., and Robertson, B. L. The Reactances of Synchronous Machines, Winter Convention of the A. I. E. E., 1928. 2. 1927, Wieseman, R. W., Graphical Determination of Magnetic Fields; Practical Applications to Salient-Pole Synchro3. 1926, Doherty, R. E. and Nickle, C. A., Synchronous Machines-Part I. An Extension of Blondel's Two-Reaction Theory, TRANS. A. I. E. E., 1926, Vol. 45. p. 912.

Bibliography

and

<b

O = b = s 17

( 7a)

XD60
xm
t3\ 2(19 S3 2S i7r2(19 S3+45t +9 b S2-54 b-81 bs+54 s)1n1 486 S Sjn2 7r/2 (1 - b/s)

2s 1

4. 1926, Gray, A., "Electrical Machine Design, 2d. ed. Rev." P. by M. Lincoln, p. 523, 1926, McGraw, N. Y. 5. 1922, Siegel, E., "Leakage Calculations for Three-Phase Induction Motors with Phase-Wound Rotors," (in German), ~Elekt. u. Masch., Vol. 40, pp. 217-22, May 7, 1922. Sci. Abstr.,

See. B, Vol. 25, pp. 502-03, Oct 1922, Abstract translation.

S/3 < b

2S 3

(18a)

:For a full-pitch three-phase winding with a large 8. 1916, Chapman, F. T., "Air-gap Field of the Polyphase Induction Motor," Electrician, Lond., Vol. 77, pp. 663-67, 705-08, number of slots per pole, (17a) may be reduced to:
XD6O_ 2wr2 (5 + 9)in2i-/2s~ 1 Xm 243
1 S + - 1 (19a) = 486 j4 s2 0543If s is very large, (19a) reduces exactly to (32). Numerical values of (17a) and (18a) for some usual three-phase windings are given in Table XI. 50 For a Forsquirrel-cage a quirrl-cag winding, =s s ad and b b = 0, so =0, windng, q q=

1918, Vol. 37, pp. 1209-97; discussion, pp. 1298-1340.

Synchronous Machines and Its Applications, A. I. E. E. TRANS.,

Core Losses, A. I. E. E. JOUR., 1920, Vol. 39, October, pp. 906-20. 7. 1918, Doherty, R. E. and Shirley, O. E., Reactance of

6. 1920, Alger, P. L. and Eksergian, R., Induction Motor

5 w4

w44

0.672

730-34, 904-06, 936-38; Vol. 78, pp. 9-12, 49-50. Aug 18Sept. 1, Sept. 22-29, Oct. 6-13, 1916. 9. 1912, Fechheimer, C. J., Self-Starting Synchronous Motors, A. I. E. E. TRANS., 1912, Vol. 31, pp. 529-85; Discussion, pp. 586-604.

TABLE XI
DIFFERENTIAL LEAKAGE REACTANCE OF THREE-PHASE WINDING S''

10. 1909, Schenkel, M., "Determination of Leakage Flux in Vol. 27, pp. 201-08, Feb. 28, 1909. Sci. Abstr., Sec. B, Vol. 12, pp. 160-61, Apr. 1909, Abstract-translation. 11. 1907, Adams, C. A. and others, Fractional Pitch Windings for Induction Motors, A. I. E. E. TRANS., 1907, Vol. 26, pp. 1525-26.
Alternating-Current Machines," (in German), Elekt. u. Masch.,

14805-1503; Discussion, pp.

deficiency
b

Pitch

EQUATIONS (17a) AND (18a) s/3 = slots per pole per phase
2

1
0.0966

4
0.0074 0.0063

5
0.0055 0.0044

12. 1907, Hellmund, R. E., Zigzag Leakage of Induction Motors, A. I. E. E. TRANS., 1907, Vol. 26, pp. 1505-24. ' 13. 1905, Adams, C. A., The Design of Induction MotorsWith Special Reference to Magnetic Leakage, A. I. E. E. TRANS., 1905, Vol. 24, pp. 649-84; Discussion, pp. 685-87. 14. 1904, Adams, C. A., "The Leakage Reactance of Induc15.

0
2
3 6 7 8 9 1

0.0966

0.0284

0.0966

0.0284

0.0235

0.0140
0.0111

0.0115

0.0089

0.0064

0.0052

0.0045 0.0035

706-24; Discussion, pp. 725-28, 1904.

tion Motors," Int. Elee. Congress, St. Louis, Trans., Vol. pp.

0.0284

0.0140
0.0137
0.0140

0.0089
0.0092
0.0089 0.0083
0.0089

0.0041
0.0064
0.0068 0.0067
0.0062

0.0052
0.0052

0.0040

0.0030
0.0055

1904, Hobart, H. M. and Punga, F., A Contribution to the Theory of the Regulation of Alternators, A. I. E. E. TRANS., 1904, Vol.. 23, pp. 291-322; Discussion, pp. 323-336, 340-43.
Discussion

0.0056

______0_____0057 _________0____0052_______ that by substituting these values in (15a) and dividing


is found to be:
X
2
Xm

____

0.0057

xm

= 4 4 S2

CS C2

7r/2 s

(20a)

B. A. Behrend: In connection with Mr. Alger's very complete investigation of the factors which determine the armature reactance of a synchronous machine the question may be raised whether it is more satisfactory for the designer to derive his characteristic data from comparison with known similar designs or by a fresh calculation for each particular case. As a rule
the designs vary comparatively little, thus making it possible

which only the first term need be considered:


XDK
T2 / T2 K | 1=22 Xm 12S\ 20 S
504 S

If~ ~

bedevelped nto Slret1 a~aserie of noasre f ~ ~this ~ ay ~ ~ edvlpd ~ ~ i ~~~~~~~~i

to obtain design data by comparison. The analysis of the onstituent parts of the tota.l leakage field as carried out by M.Agr1n keeping with the desire to leave no part of the

theory unexplored and to link together the leakage of syn-

(21a)

cli\ eronous and induction machines. In this very painstaking a.nd intricate procedure I doubt a method which is is evolved by one designer and which suits that his method of working likely
to appeal equally to another designer. I am inclined to view

For large values of s, (21a) agrees exactly with (25), thus confirming the previously derived formula for zigzag leakage reactance.

such methods as to a great extent individual though notwithsta,nding this they are of the greatest importance in clarifying obscure problems and showing us the empirical basis of our

You might also like