298 IEEE TRANSACTIONS ON POWER APPARATUS AND SYSTEMS, VOL. PAS-90, NO.

1, JANUARY/FEBRUARY 1971

The Inductances Ld and Lq of Reluctance Machines

VERNON B. HONSINGER, MEMBER, IEEE

Abstract-Equations for the self-inductances Ld and La of a

reluctance machine having flux barriers and cutouts are developed. BARRIER
An electromagnetic model of the reluctance machine is postulated.
Flux densities are found in the air gap proper, the cutouts, and the
barriers. A comparison is made between measured and calculated
results.

INTRODUCTION
T HE self-inductances Ld and L. are defined as

Ld = Lmd+ la Lq Lmq + la
where la is the stator leakage inductance and Lmd,Lmq are the
direct-axis and quadrature-axis mutual inductances, respec-
tively. The rotor of this reluctance machine contains both
cutouts and flux barriers as shown by Fig. 1. The cutouts and
barriers serve the purpose of diminishing the quadrature-axis
flux flowing between poles and at the same time they permit Fig. 1. Rotor cross section of two-barrier per pole reluctance motor.
direct-axis flux to flow largely unimpeded directly through the
poles. This altogether has the effect of increasing pull-out torque,
which is proportional to 1/La - l/Ld. The parameters Ld coefficient. The rotor cutout and barrier configurations are
anid Lq also have powerful effects on motor synchronizing simulated by use of radial and circular lines as shown by Figs.
ability and on the matter of rotor oscillations that may appear 2 and 3. The actual rotor (Fig. 1) has a magnetic flux conducting
when the frequency of the applied voltage is small. bridge across the barrier at the air gap; the center portion of
The subject here discussed is divided into five categories. this bridge is saturated so that effectively the barrier width is
First, a model of the reluctance machine is postulated in which quite small at the air gap. Consequently, it is presumed that the
there is no relative motion between the physical parts; second, air-gap flux density is not interrupted by the barrier air-gap
the flux density fields are found in the air gap proper, the cut- opening, provided the barrier fringing flux is accounted for by
outs, and the barriers; third, the inductances Lmq and Lmd Carter's method. This being true, the barrier location at the air-
are found from the flux linkages; fourth, a method of measuring gap may be specified by an angle between the d axis and a radial
Lq and Ld is proposed; and fifth, a comparison is made between line through the barrier center equal to (ir/P)Pb Within the
measured and calculated results. barrier, it is assumed that lines of force are perpendicular to the
The analysis is confined to single- and two-barrier per pole boundaries of the barrier, i.e., radial in the circumferential
machines; these configurations are most commonly used. Ap- parts and circumferential in the radial parts.
parently the only other approach to this problem was made by The permeability of all iron is assumed to be infinite. How-
Lawrenson and his associates [1], [2], who investigated single- ever, to account for saturation of the iron, the air-gap length
barrier per pole machines having very wide barriers. is multiplied by a saturation factor. The net air gap g found from
the actual air gap g' is
BASIC RELUCTANCE MACHINE MODEL
g =g'kskcrkcfks
Fig. 1 represents the basic features of a rotor of a reluctance
machine having two flux barriers per pole and a cutout between where kC, and k,T are Carter coefficients for the stator and rotor
each pole. The stator laminations resemble those of an induction slotted surfaces, respectively, kC f is the pole edge fringing
motor. The stator has an m-phase winding in which there are coefficient, and ks is the saturation factor.
NkW/P effective turns per phase per pole. Currents ia ib,, The mattei of simulation is continued, except now from an
im flow through each of the phase windings. entirely different viewpoint. The transformation of voltage
For purposes of analysis, the slotted stator is simulated by a equations and inductances to a dq reference frame fixed in the
smooth -stator whose air-gap length is increased by Carter's rotor (Park's transformation) is well known. Quite aside from
pure electrical quantities, it will be presumed that the physical
parts of the machine (in some cases, the effects of the physical
Paper 70 TP 193-PWR, recommended and approved by the parts rather than the parts themselves) may be transformed
Rotating Machinery Committee of the IEEE Power Group for
presentation at the IEEE Winter Power iMeeting, New York, N. Y., into a physical model such that axis inductances Ld and La
January 25-30, 1970. Manuscript submitted August 18, 1969; can be directly deduced from the transformed model. The follow-
made available for printing December 3, 1969.
The author is with the Motor-Generator Divisiorn, Allis-Chal- ing is asserted to be characteristic of the transformed model
mers, Norwood, Ohio 45212. when the dq axes are fixed in the rotor.
HONSINGER: Ld AND L8 OF RELUCTANCE MACHINES 299

Fig. 2. Model of two-barrier per pole reluctance rotor used for Fig. 3. Model of single-barrier per pole reluctance rotor used for
mathematical analysis. mathematical analysis.

------------------

(a) (b)
Fig. 4. Flux maps. (a) q mode: lines of force resulting from sinusoidal MMF wave. (b) d mode: lines of force re-
sulting from cosinusoidal MMF wave.

1) The smooth stator and anisotropic rotor having relative This is a fairly accurate approximation, ignoring space har-
motion between them may be represented by configurations monics in the MMF wave caused by stator currents being
identical to the original stator and rotor but having no relative located at discrete points instead of being sinusoidally distrib-
motion. uted around the stator.
2) The m-phase stator winding may be represented by a d 6) According to the flux maps shown by Fig. 4, and except for
and a q winding, each being identical to the original m-phase fringing flux near the pole edges, the radial flux in a cutout in the
winding except that their axes lie respectively along the d and d mode caused by a cosinusoidal MMF wave is negligible, but
q axes, which are at rest with the other parts of the system. the radial flux in a cutout in the q mode caused by a sinusoidal
3) An external torque applied to the machine does not change MMF wave is not negligible. After decreasing the air-gap length
the position of the rotor relative to the stator d and q windings, for fringing flux at the pole edge (keeping the pole arc constant),
although this does change the magnitude of currents flowing in flux lines in a cutout are considered to be entirely radial in the q
these windings. mode and zero in the d mode.
4) The currents flowing in the d and q windings are id and i8, 7) All flux lines in the small air gap over the pole arc are
respectively, transformed from phase currents *, iim but considered to be radial.
harmonic axis currents are ignored. 8) The saturation factor k. ordinarily will have different values
5) The values of magnetomotive force (MMF) on the stator in-the d and q modes given by k1d and ksq, respectively.
surface at any space angle 0 from the d axis in d and q modes are 9) Each rotor segment A or B in the q mode (Fig. 2) is equi-
potential. Hence the value of magnetic potential or the product
4 m/2 (V\2I d)Nkw, P
Hw across a barrier is constant along the barrier. This applies to
Fd =-- cos-2 a barrier of any shape.
O.9mIdNk1c P
The model (Figs. 2 or 3), having now been reduced to the
=- p cos- case where the physical parts have zero relative motion and
where the direction of lines of force have been specified, permits
the magnitude of each line of force to be deduced in the air gap,
Fq m/2(V\21I)Nkw~~~sin -o
4 P
the cutouts, and the barriers, by taking line integrals around
various closed paths and determining the relation f H,ds =
O.9mIqNkT .P 0.4irI where I is the total current enclosed and H. is the force
p 2 intensity in the direction ds.
300 IEEE TRANSACTIONS ON POWER APPARATUS AND SYSTEMS, JANUARY/FEBRUARY 1971

In this endeavor it should be recognized that the amount of segment of it must leave the rotor or that segment with no gain
current enclosed between 0 = 0 and 0 = 0 (Fig. 2) of a line or loss. It is sufficient for the purpose to consider the segment in
integral extending into the stator iron is exactly Fd or Fq, Fig. 2, which contains a cutout, and then to evaluate the flux
previously described as the MMF. The Fd, Fq expressions are per unit length flowing into the outer boundary of the segment,
now written for convenience as equating this to the flux per unit length flowing out the inner
boundary or at the barrier interface. This is expressed by
Fa=0.9~nIdNkw
P
P
Fd = 0 lNWcos- 00
2
oP
Fdm cosP
2
(1) DeP//P (D ,/P
-r prb/P Bq2d
2 1
+ 2 -
- hc
rp/P
Bq3do B,h, + B2h2.
N sin P = FQm sin P
Fq -. (2) (5)
q
P 2 2
The right-hand side of (5) may be simplified by (3):
Since no current exists in the rotor, the value of any line
integral entirely confined within the rotor is zero. + B2h2 =B2W2 (2h,
BhB,h,B,h=B,, (-+ h2 (6)
When taking line integrals, it is understood that the integral 1 W2
through any iron part is zero since H = 0 because H = B/,u
where u is infinite. In the vacuous or air parts, B is taken to be When the values of B,2 and Bq3 given by the last two equations
numerically equal to H. in (4) are substituted in (5) and the integration carried out, the
value of B2w, so obtained is
FLUX DENSITY FIELDS IN AIR-GAP, CUTOUTS, AND BARRIERS
B2w2 = 0.4rFqm(f-.m) (7)
Since each d and q flux density field is symmetrical with respect
to its d and q axis, only the fields between the d and q axes need where
be found. Thus ordinarily the matter is confined to the wedge-
shaped region bounded by the angles 0 = 0 and 0 = ir/P in COS 7rPb/2 - cos 7rp/2 + c
Fig. 2. fb =

+ P -Pb + C2 (8)
2Pg f2h + h2
Quadrature-Axis Flux Densities (9)
7rD \rW W2J
It is convenient to restrict the line integrals to two types of
paths, with path vxyz in the rotor interior crossing only the Ci-
1-2hc g
cos-
rp
(10)
radial and circumferential parts of a barrier and path OstO
extending into the stator iron enclosing a current between
O = 0 and 0 = 0. (11)
Confining attention to the rotor interior of Fig. 2, it will be D g + hc
presumed that the radial part of the barrier may be represented The parts cl and c2 have to do with radial flux in the cutout,
by its average width w1 with an average flux density B, crossing which ordinarily is small; frequently cl and C2 may be ignored.
it, then any line integral around any path vxyz is B,w, - B2w2 = Consequently using (4) and (7), the quadrature-axis flux
0 (two barriers per pole) whereas for a single-barrier per pole rotor in the air gap and cutout are
shown by Fig. 3, the equation is Blw,/2 - B2w2 = 0 (one barrier
densities
per pole). Both equations may be expressed by one equation 0.47rFqm P
Bql = g sin- 0 0 < 0 < 7Pb
2'
B w - B2W2= 0 (3) - <0<
2 B2=0.47rFqm
.
g
siPo-
2
TPb -rp
f
W/- P -
7P< 0 -<-7
P
where r represents the number of barriers; in this case r = 1 or 2.
The line integrals into the stator iron along path OstO, including 0.47rFqm (. P 2
barriers and cutouts, enclose an amount of current Fqm sin Bq3 = h sin - fb )
P-< 0 < Pp. (12)
+
g b ( 2 7r )
(P/2)o. These integrals expressed piecewise in the angular
regions between 1) pole center and barrier, 2) barrier and pole And the flux densities B, and B2 along the barrier are found
edge, and 3) pole edge and cutout center, are given by from (6) and (7) to be
P
Bqlg =0.4TrFQm sin 2 o O <0<
0 < -~p B, = 0.4irFm
Wi
2
r
(') 7r) (13)

<0o <-"
7=Pb <<P 0.47rFqm (fb2)
Bq2g + B2W2 =0O.4TFqm, sinP2'
-
0) - +
P- - P
B
W2 T
(14)

P VrP 0 <-
7r
Bq3(g+ hc) + B2w2 =0.47rFqm sin 2 0, P < (4) Direct-Axis Flux Densities
It is evident that all flux in a barrier is caused by quadrature-
The equations in (4), when solved for Bqi, B,,2 and B.3 are axis MMF and that no flux in a barrier can be attributed to
flux densities in the air gap and cutout but with reference to an direct-axis MMF. In other words, the direct-axis flux flows
unknown quantity B2. The barrier density B2 is related to the through the iron without crossing a barrier. Since the radial
density B. by the condition that all flux entering the rotor or a component of Bd in a cutout has been assumed to&be negligible,
HONSINGER: Ld AND L8 OF RELUCTANCE MACHINES 301

(a) (b)
Fig. 5. (a) Flux whorls do not exist when barrier arc is large. (b) Flux whorls exist when barrier arc is small.

the direct-axis flux densities between 0 = 0 and 7r/P are INDUCTANCES Line AND Lmqe
The flux density equations previously derived have magnitudes
BdlO2 =
g
Fm COS
P

2
e <
-
a < 7rp which are maximum values in time. These time magnitudes
conform with the state of the art, which calculates flux from the
0, equation V = 4.44 fq$mNkw X 10-8 where q5m is the peak value
Bd3 =
P-
< 0 <-.
-P of a time-alternating flux in the equation k = 0m sin cot. In
other words, the currents used to express Fd and F, and sub-
The subscripts of Bd1,2 refer to the numbered angular regions sequently to express flux linkages are V2 Id and V\/2 Iq (see
previously used in the quadrature-field analysis. Since Bd1 has condition 5) in the Section Basic Reluctance Machine Model).
the same form as Bd2 between region 1, pole center and barrier Hence the inductance coefficients found from flux linkages are
and region 2, barrier and pole edge, the density in both regions obtained by dividing by \/2 Id and V'2 IQ according to the
is called Bd1,2 and the angular distances included by these equation L = X/V/2 I X 10-8.
regions are combined. This procedure, although unnecessary for inductance calcula-
Flux Whorls in Quadrature-Axis Field tion, yields a state of the art flux density magnitude.
With reference to the quadrature-axis field, it is instructive at Quadrature-Axis Mutual Inductance Line
this point to examine the flux maps shown by Fig. 5. eye-drawn The total linkage flux per pole is the flux integrated over the
from the possibilities expressed by the second equation in (12). stator inner surface between 0 = 0 and 27r/P:
Whorls of flux may exist in region 2, bounded by the barrier and
pole edge. For example, B.2 is the air-gap density in region 2: r27/P ID\ r/P
sq =LJ Br dO = 2Lkt ) JO B1 do. (18)
0.47rF0, siJsln.0-6-
P 2
(16)
Bq2 g =
\2 fb).
7r The flux linkages of all poles and the inductances per phase
As 0 traverses the angle between 7rpb/P and irp/P, the sign are given by
of Bq2 may change, thus indicating a flux whorl, depending Xm = kwNckq (19)
upon the relative magnitudes of sin P0/2 and 2 fb/lr.
The location of a flux whorl is specified by its kernel located in (20)
the air gap at angle 0, = 7rp6/P. The flux density at the kernel is Lmq =
,\/m21 X 10-8.

zero. Hence equating (16) to zero obtains the kernel position in

terms of Pc The evaluation of (18) depends upon whether flux whorls
exist or not. Consider first the case where flux whorls do not
2 2ft exist. Using the flux densities given by (12) to express (18)
P, (17) piecewise, the flux per pole is an ordinary substitution given by
\JPb/P oTP/P Jr/P
The existence criterion for a flux whorl is that PC found from
fb, which is a function of Pb, is greater than pb. Oa5=LD B1i1d0±J Bq2 do ±J B13ddo)
The boundary criterion between the existence and nonexistence (21)
of a flux whorl is given by
However, when a whorl exists, the lower integration limit of
2 . 2fb the second integral should be changed to -rpa/P since the flux,
Pb = - sin-'
7r 1r in the whorl, to the left of the kernel must not be included if it is
Ascertaining the existence and finding the location of the included to the right. The linkage flux now is expressed by
kernel in a flux whorl is very important. Equations that ignore WPb/P Sp/P +r/P
this will overestimate motor torque capability by as much as =LD l Bui dO +
kJ Or, %Pc/P
B,2 dO + J BQ.3 dS )
,p/P
75 percent. (22)
Incidentally, all single-barrier per pole machines have flux
whorls since: fb exists; hence from (17) PC exists and is a positive The flux densities B.1, Bq2, and B.3 are given by (12). Sub-
number, whereas Pb is zero. stituting these, performing the integration, and using (19) and
302 IEEE TRANSACTIONS ON POWER APPARATUS AND SYSTEMS, JANUARY/FEBRUARY 1971

(20), the quadrature-axis mutual inductance Lmq is found to be Case 3, Machine with Barriers but no Cutouts: Equate p = 1
in (24) and (25) noting that a disappears:
Lmq 1= 6mLD Nk, Cq X 10-8 (23) C = 1 - (1 -pb)fb, whorls nonexistent
In (23) C. is given by C8 = 1-P(1-P)fb -cos 2 + cos 2' whorls exist.
V

Cq = 1- 2
- (P - Pb)fb + (24) Case 4: A mathematical check on CQ may be made by trans-
forming the rotor shape to a cylindrical form having no barriers.
where flux whorl does not exist, and by This may be done by causing the pole arc/pole pitch ratio p
to approach both unity and zero; the air-gap length is g in the
Cq = 1 o
2
(P' - PA-ab c 2- + cos
2 2
(25) first case and g + h, in the second. It may be inferred from the
inductance equation for a cylindrical rotor with no barriers
where flux whorl exists, and a is previously mentioned that Cq should equal unity in the first
case and g/(g + h,) in the second. Using the equation given for
a = FOS 2
[o fb(1 P) (26) case 2 and setting p = 1 and p = 0, one does indeed get Cq = 1
gA-hc ~2j and Cq = g/(g + h,), respectively, for these two cases.

Direct-Axis Mutual Inductance Lmd Concerning Magnitude of C8 and Its Effect on Torque
The direct-axis flux linkage per pole analogous to (18) is The pull-out torque is proportional to (1/L,> - 1lLd). This
7'r/P quantity is maximized when Lq is made small and Ld is made
«d= DL BddO large. The same statement may be applied to Lmd and Lmq
and hence to Cd and Cq. Presuine because of a number of design
and the integral may be reduced by (15) to considerations that the pole arc/pole pitch ratio p is fixed,
noting that this determines Cd according to (28).
r_P/P It remains now to minimize C., which is a function primarily
kd = DL J Bdl,2dO of the barrier dimensions and barrier arc/pole pitch ratio Pb.
Ordinarily smaller values of C, are associated with values of Pb
where Bd1,2 is given by the first equation in (15). that cause whorls of flux to exist and barrier widths that are
Following the method just described for calculation of Lmq, made as large as possible without unduly increasing the flux
the direct-axis mutual inductance Lmd is found to be densities in the iron parts. Thus the matter may be confined to
the minimization of C, given by (25) for a constant p, focusing
attention particularly on the distance between barriers at the
Lmd = 1.6 dLD (Nk)2 C l-8 (27)
air-gap interface.
The minimum value of Cq with respect to Pb is given by the
where condition dCl/dpb = 0 applied to (25), in which the a term due
7.rp to cutout flux has been ignored:
Cd = sin -
dCq X .
--sin TPb
-
- - sin p
. aPc -- afb - Pc) C Pc = 0.
(p +- fb
dpb 2 2 2 aPb C3Pc aPb
Special Cases
Equations (23) and (27) for Lm8 and Lmd are expressed by a But since fb = 7r/2 sin (7r/2) PC, the equation reduces to
quantity (1.6 m LD/g)(Nk,,/P)2 multiplied by either C8 or Cd.
The just-mentioned quantity is the inductance of a cylindrical dC- -r
sin -pb - (P-Pc) = (29)
dpb 2 2 dpb
rotor having no barriers when enclosed by a cylindrical stator.
Hence the factors C. and Cd account solely for the presence of The preceding equation is best evaluated by numerical
barriers and cutouts; i.e., the anisotropic nature of a reluctance methods. However one point is noticeable. The derivative dC,/
machine rotor. dpb for a single-barrier per pole machine is given by the condition
Consider special cases caused by transforming the two-barrier Pb = 0:
per pole model of Fig. 2 into other shapes. The factor C. is of
particular interest since it alone contains the effect of barriers. d a
PC) (30)
d PbPb
- -

Case 1, Single-Barrier per Pole Machine with Cutouts: This dpb

case is found by setting Pb = 0 in (25): Evidently this derivative is minimum or zero only when fb = 0
or when the barrier widths are infinite. Since this situation
CM = cos -2 Cos - (p- Pc)fb + a. .may not be had, it is concluded that single-barrier per pole
2 2
machines do not produce the maximum obtainable torque. In
Case 2, Machine with Cutouts but no Barriers: Equate barrier fact, if C. is plotted versus Pb, (30) predicts that the slope is
widths to zero, hencefb = 0, and (24) gives negative at Pb = 0, becoming zero when (29) is satisfied and
becoming positive thereafter. This ordinarily is seen in actual
=
C8=-os----GOS ++ g hcs Cos-_. plots of C, versus Pb. Finally it should be noted that the optimum
2 gA- he 2 value of Pb becomes smaller as the barrier widths become larger.
HONSINGER: Ld AND Lq OF RELUCTANCE MACHINES 303

UNITS However, it is found that this equation simplifies when

All inductance equations listed use centimeters as the unit of impedance components Z, R, and X are used. The relations
length. Multiply all inductance equations by 2.54 to obtain between these types of components to admittance components
inductance in henrys per phase when length units are inches. Y, G, and B are
When flux density equations given by (12)-(15) are multiplied y 1 G fR B X=Z2
by 2.54, the result is flux density in lines per in2. z Z2'

MEASUREMENT OF Xd AND Xq Substituting these and also substituting for D and : their
equivalents obtains
In the Section No-Load Current of [3] it was shown that at no-
load xdXq + ra2 - 2Rrta - X(xd + Xq) + R2 + X2 = 0
Iq = 0 or
V (xd - X)x2 + (R - ra)2 -X(xd -X) = 0.
Id = 1ul
VdX52 + ra2 Therefore
and hence Xd may be found from no-load data according to (ft-Ta)2
XQ = X -
Xd - X (33)
Xd = -/ ( TV/I1)2 ra2 -
V/Inl (31)
Given a measured set of load curves consisting of, at least,
where the right-hand approximation applies to measurements amperes and power factors plotted versus horsepower or torque,
made around 60 Hz. Values of Ld = Xd/2irf found from 60-Hz then Xa may be determined at various load points by (33).
measurements can be used at other frequencies. Evidently [It is preferable to use (35).] The values of R and X are found
there is no way that xa can be measured directly at no load since from measured values of I and power factor angles p on the load
Iq is zero. curve according to
The measurement of x¢ may be made, by a method to be out-
lined, provided the value of Xd is known at all loadings. The V
Z = -, R = Zcosp, X = Zsino. (34)
magnitude of Xd decreases somewhat as the machine is loaded, I
but only by a small amount. Nevertheless the measurement of The value of xd is determined from (31). Equation (33) does
x2 will be based on a constant Xa whose value is given by (31). not, but should, include the effect of iron and rotational losses.
In [3], the machine admittance components G and B were The derivation therefore must be regarded as expository, a
shown to be means to explain the method most simply.
= 2ra + (Xd Xq) sin 28
-
Correction for Iron and Rotational Losses
2(xdXqa+ ra2)
It has been found that iron and rotational losses have negligible
B +
Xd ± q - (Xd -
Xq) cos 26 effect upon the measurement of Xd. On the other hand, these
2(xadx + ra2) losses appreciably affect the measurement of xa, which is a much
These equations have the form smaller reactance than xa.
When iron and rotational losses are included, the admittance
a ra + a sin 26 components are called Y1, G1, and B1. These are related to the just
D previously used components Y, G, and B by the relations [3]:

B a- cos 28 WO
D
g mV2' WO =
Wn1 -
mlnl2ra
where G1 = + go
B1 = B
XXd
-

2
X2 Xd +

2
Xq
D = xdxq + ra2.
D=aaT2
Y= GC - 1Bi.
Upon rearranging, It is only necessary to substitute G = G1 - go in (32) and then
GD - ra =+a sin 26 carry out the same procedure for finding Xq just described.
This obtains
BD-/ = -a cos 26. (32)
It is undesirable to measure torque angle; terms containing the X(Xd - X) - (R - ra)2 - go[(Z2go - 2R)ra2 + 2raZ2]
torque angle are eliminated by squaring both sides of each = [1 + go(goZ2 - 2Rf)]Xd- X
equation and adding. This obtains X(ad - X) - (R - ra)2 (35)
Y2D2- 2(Gra + BO)D + /2 - a2 + ra2 = 0. [1 + go(goZ2 - 2R)]Xd- X
But /2 - a2 + ra2= D, hence the equation reduces to where the Z, R, and X components have the same meaning as
given by (34).
Y2D - 2(GCa + B/) + 1 = 0. Ordinarily (35) is preferred over (33), even at pull-out torque.
304 IEEE TRANSACTIONS ON POWER APPARATUS AND SYSTEMS, JANUARY/FEBRUARY 1971

TABLE I
COMPARISON OF MEASURED AND CALCULATED RESULTS

Xd Xq
P D (inches) kd s Calculated Test Calculated Test
4 4.00 1.71 1.1 33.5 35.6 9.30 9.49
4 5.25 1.50 1.1 21.1 21.1 4.64 4.48
2 4.75 1.45 1.1 14.2 12.0 2.34 2.95
6 5.25 1.59 1.1 25.7 25.6 6.95 6.75

COMPARISON OF CALCULATED by Table I is in most cases close or at least tolerable, although

AND MEASURED RESULTS there is quite some discrepancy in the two-pole machine.
The measured results to be listed were obtained using (31) The greatest limitation in the calculations of Ld and Lq given
for xd; the value of xq was determined by (35) using pull-out by (23) and (27) is found in the estimation of saturation factors
data. ksd and k81, which appear as factors in the effective air-gap
The calculated data were found using (23) and (27) for Lmq length g.
and Lmd multiplied by 2.54 to convert from centimeter units to LIST OF PRINCIPAL SYMBOLS
inches. Equations (25) and (28) were used to find Cq and Cd.
The stator leakage reactance xa was calculated by methods well Ld direct-axis inductance per phase
known in the art. The total reactances are Xd = Xmd + Xa Lq quadrature-axis inductance per phase
and xq = Xmq + Xa. 9 effective air-gap length
In the process of this type of calculation, estimation of the N turns per phase in series
saturation factors is one of the greater detractions from accuracy. winding factor; ordinarily, product of pitch and
No matter how much care is taken elsewhere, saturation effects distribution factors
ordilnarily will dilute accuracy. In the magnetic model proposed number of poles
pP
herein, it is preferable to use a different saturation factor for the pole arc/pole pitch ratio
d mode than for the q mode. The saturation factors used to Pb barrier arc/pole pitch ratio
calculate Xd herein are factors measured from the no-load satura- Pc kernel arc/pole pitch ratio
tion curves of the various machines. These measured saturation D rotor outer diameter
coefficients kSd in the d mode are listed in Table I. The saturation r number of barriers
factors used to calculate xq are taken to be 1.1 for want of a fb see (8)
better method. m number of phases
A comparison of calculated and measured values of Xd and Xq L core length
at 60 Hz are given by Table I for several two-, four-, and six-pole WI, w2, hi, h2 barrier dimensions, see Fig. 2
motors whose rotor diameters are listed to give some concept of hc cutout depth
the size of these motors. The rotor lamination configuration is V volts per phase
alonig the lines of Fig. 1 and each motor has a 0.5 pole arc/pole Wnl no-load watts, all phases.
pitch ratio. REFERENCES
CONCLUSIONS [1] P. J. Lawrenson and L. A. Agu, "Theory and performance of
polyphase reluctance machines," Proc. Inst. Elec. Eng. (London),
The flux denisities anid inductances Ld anid Lq have been vol. 111, pp. 1435-1445, August 1964.
derived for a reluctan-ce machine having cutouts and flux [2] P. J. Lawrenson and S. K. Gupta, "Developments in the per-
formance and theory of segmental-rotor reluctance machines,"
barriers. This is accomplished by use of a reluctance machine Proc. Inst. Elec. Eng. (London), vol. 114, pp. 645-653, May
magnietic model having d and q windings in which altogether 1967.
there is no relative motion between the physical parts. [3] V. B. Honsinger, "Steady-state performance of reluctance
machines," this issue, pp. 305-317.
A method for measurement of Xd and xq is indicated and
comparisons made between measurements and calculations.
The agreement between calculated and measured results given For Combined Discussion see pp. 311-317.