A Novel Analytical Model of

Switched Reluctance Machine

Topic number: 3 (or 4)

I. INTRODUCTION
Switched reluctance machine (SRM) has highly
saturated doubly salient pole structure. The requirement
for predicting the steady state and dynamic performance
of SRM is to generate its magnetization characteristic
(I, ) and torque lookup table T(I, ), where is the
flux linkage, I is the excitation current, is the rotor
angular position, and T is the torque.
The magnetization characteristic can be obtained by
magnetostatic FEA [1][3] or by measurements done on
the existing machine. However, both methods are not
particularly suitable to implement during initial stages of
the machine's design process.
An alternative approach is to develop analytical model
of SRM to compute its magnetization characteristic. This
model has to be faster than FEA but still accurate enough.
The SRM modeling using analytical methods have
been reported in [4][18]. The most common models are
based on the use of: magnetic equivalent circuit (MEC)
[4][6]; implementation of basic laws of physics on
simplified motor geometry [7][9]; analytical equations
that approximate magnetization characteristic [10][15];
several predefined flux-tubes [16][18].
This paper presents a novel analytical model of SRM,
which is intended to be a part of a sizing-design
estimation process of the machine. The analytical model
will be discussed in Section II. Proposed model combines
two already known techniques, flux-tube and gage-curve.
Flux-tube method is used to compute (I) characteristics
at the aligned and the unaligned rotor position. Gagecurve method is used to compute (I) characteristics at
other rotor positions.
The comparison of the analytical results to FEA is
given in Section III for a 12/8 SRM.
II. ANALYTICAL MODEL
A. Computation of (I) characteristics using flux-tube
The flux-tube based models that enable computation of
flux linkage characteristics in arbitrary rotor positions are
reported in literature, but the achieved level of accuracy
in intermediate rotor positions can be a quite low [19].
In this paper flux-tube method is used only for
computation of (I) characteristics at the aligned and the
unaligned rotor position, where substantial accuracy can
be obtained.
As in [16], two flux-tubes are assumed to be sufficient
to represent the actual flux paths at the aligned position

and seven flux-tubes are used at the unaligned position.

The considered flux-tubes are shown in Fig. 1.
9

8
7
56
4
3

a. aligned position

b. unaligned position

Fig. 1. Assumed flux tubes aligned and unaligned position.

Flowchart for the computation of (I) characteristics

at the aligned and the unaligned position is shown in Fig.
2.
Model input data are: BH curve of the lamination
material, numbers of stator and rotor poles Ns and Nr,
stator and rotor poles arc angles s and r, stator and rotor
outer radii Rs and Rr, air gap length , stator and rotor
yokes radii Rsy and Rry, shaft radius Rsh, stack length lstk,
number of turns per phase winding Nph, and number of
phases m.
The results of FEA or measurements in several
characteristic points are not needed, which is the
advantage in comparison to classical gage-curve models
described in [10] and [12]. Also, implementation of fluxtube method is not based on the use of iterative
computation like in [16][18]. Instead of using numerical
iteration with a prescribed value of error to compute
magnetomotive force, the flux linkage contributed by
each flux-tube in computed for the assumed values of
stator pole magnetic flux density.
Usually, only a small number of points k define
original BH curve of the lamination material. In order to
achieve an acceptable level of computational accuracy, it
was necessary to use cubic spline interpolation to
generate interpolated BH curve with more defined
points n (n > k). Interpolated magnetization curve is
recalculated to generate int = f(Bint) curve. Flux linkage
characteristics are computed for n values of magnetic flux
density. Since the computational time is rising with n,
this number should be chosen carefully.
Equations derived for flux-tube No. 1 are given in
order to explain the computation procedure. The list of
equations used for all defined flux-tubes is given in [20].

Enter k points of BH magnetization curve of

the lamination material and machine geometry data
Generate n points (n > k) of interpolated curve int = f(Bint)
using cubic spline interpolation of original BH curve
Compute area of the stator pole cross section for
considered flux tube j
Assume that values of magnetic flux density in the stator pole
are equal to values Bint of interpolated curve int = f(Bint)
Compute ndimensional array of stator pole flux values
[1 2 ... n] corresponding to array of magnetic flux density
values Bint for considered flux tube j
Compute indimensional matrix of flux values
in the path of tube j for i various machine segments
Compute idimensional array of cross section area values
of i various machine segments encountered by flux tube j
j=j+1

Compute indimensional matrix of magnetic flux

density values in i various machine segments
Using curve int = f(Bint) find the indimensional matrix
of magnetic permeability values corresponding to
indimensional matrix of magnetic flux density values
Compute idimensional array of path lengths
of tube j in each of i machine segments
Compute indimensional matrix of reluctances
contributed by considered flux tube j
Derive MEC and using Ampere's law compute
ndimensional array of currents [I1 I2 ... In] corresponding
to array of flux values [1 2 ... n] and matrix of reluctances
Compute ndimensional array of flux linkage
values [1 2 ... n] for considered flux tube j

No

Computation
finished for
all defined
flux tubes?

Yes
Interpolate (I) characteristics obtained for tubes j = 1, ..., 9
using linear and cubic spline interpolation. These
characteristics are interpolated at ncurrent points. This step
ensure that final characteristics for all tubes are computed
for the same currents, so they can be added in the next step.
Add flux linkage values obtained for tubes j = 1, 2 to
generate (I) characteristic at the aligned rotor position.
Add flux linkage values obtained for tubes j = 3, 4, ..., 9 to
generate (I) characteristic at the unaligned rotor position.

Fig. 2. Flowchart for the computation of flux linkage at the

aligned and unaligned rotor position.

Flux-tube No. 1
The areas of cross-section penetrated by the flux-tube
No. 1 are given in Table I. The lengths of the individual
machine segments encountered by flux-tube No. 1 are
given in Table II.

TABLE I
AREAS OF CROSS-SECTION FOR FLUX-TUBE NO. 1
Machine segment
Equation
Stator pole

Asp1 = ( Rr + ) s lstk

Stator yoke

Asy1 = Rs Rsy lstk

Rotor pole

Arp1 = Rr r lstk

Rotor yoke

Ary1 = Rry Rsh lstk

Air gap

A1 =

Asp1 + Arp1
2

TABLE II
LENGTHS OF MACHINE SEGMENTS FOR FLUX-TUBE NO. 1
Machine segment
Equation

Rs + Rsy

( Rr + )
2
Rs + Rsy 2 m
=

2
Ns

One stator pole

lsp1 =

One quarter of stator

yoke

lsy1

One rotor pole

lrp1 = Rr

One quarter of rotor

yoke

lry1

One air gap

l1 =

Rry + Rsh

2
Rry + Rsh 2 m
=

2
Ns

The array of magnetic flux density values in the stator

pole [Bsp1]1n is assumed to be equal to the magnetic flux
density array [Bint]1n of interpolated curve int = f(Bint)

Bsp1 = Bsp11

Bsp12

... Bsp1n = Bint

(1)

The array of magnetic flux values in the stator pole

[1]1n is calculated as

[ 1 ] = Bsp1 Asp1

(2)

The arrays of magnetic flux density values in other

parts of SRM are determined as
[ 1 ] [1 ]
Bsy1 =

2 A ; Brp1 = A ;
sy1
rp1
(3)
1 ]
1 ]
[
[
Bry1 =

2 A ; B1 = A .
ry1

Then on the basis of int = f(Bint) curve, the arrays of

magnetic permeability [sp1]1n, [sy1]1n, [rp1]1n, and
[ry1]1n, corresponding to the arrays of magnetic flux
density values computed by (1) and (3), are obtained.
Now it is possible to compute the arrays of reluctances
related to various laminated parts of the magnetic circuit
[Rsp1]1n, [Rsy1]1n, [Rrp1]1n, and [Rry1]1n. These arrays are
computed using arraywise right division in Matlab as
Lsp1
Lsy1
Rsp1 =

A ; Rsy1 = A ;
sp1 sp1
sy1 sy1
(4)
Lrp1
Lry1
Rrp1 =

A ; Rry1 = A .
rp1 rp1
ry1 ry1
The reluctance in the air gap is computed as follows

R1 =

L1
0 A1

, inductance

(5)

[ ]
[ I1 ] = 2 N 1

pole

Rsy1 Rry1
+
2 Rsp1 + Rrp1 + R1 +
(6)
2
2

[ 1 ] = [1 ] Nph

(7)

B. Computation of (I) characteristics using gage-curve

In this paper, gage-curve method is used for analytical
computation of flux linkage characteristics in
intermediate rotor positions.
Gage-curve method is introduced in [10]. The (I)
characteristics at the unaligned and the aligned position
are approximated using three precalculated points, firstorder, and second-order functions. These characteristics
are used to determine flux linkage characteristics at other
rotor positions.
In this paper, (I) characteristics at the unaligned and
the aligned position are calculated by flux-tube method
instead of using their approximation. Then, these
characteristics are exploited to compute (I,) curves at
intermediate rotor positions using equations given in [10].
These equations are based on the analysis of () curve
corresponding to a fixed value of current I (Fig. 3.).
The rotor positions are divided into three regions. The
equations for modeling (I,) characteristics for any
chosen value of excitation current in these regions are:
- region I (ua < < 1)
( 1 )
( I i , ) = kai (1 0i ) + Lu I i + i
(8)
i ( 1 )
- region II (1 2)

( I i , ) = uai + ki ali uai

- region III (2 < al)

( I i , ) = kai ( 2 0i ) + Lu I i +

(9)

i ( 2 )
i ( 2 )

(10)

where Ii (i = 1, 2, ..., ncurrent) represent actual value of

excitation current selected for computation, while uai
and ali are the flux linkages in the unaligned and the
aligned position, respectively, corresponding to Ii.
The boundaries between regions are defined as
r s
r + s

ua = 0 ; 1 =
; 3 =
;
Nr
2
Nr
2

1 + 3
2

; al =

.
Nr

Region III

Idealised
variation of
the phase
inductance

ua

where Npole is number of turns per pole (Npole = Nphm/Ns).

The array of flux linkage values of flux-tube No. 1
corresponding to the array of currents [I1] is

2 =

Region II

Region I

al

The array of currents [I1]1n corresponding to the array

of magnetic flux density values [Bsp1]1n is computed
using binary addition and arraywise multiplication in
Matlab by means of expression

ua

3 al

Stator pole
Rotor pole

Fig. 3. () curve corresponding to a fixed value of current I.

The coefficients in (8)(10) are determined as

i =

kai (1 0i ) + Lu I i uai (1 ua )
kai (1 ua ) kai (1 0i ) + Lu I i uai

i = kai i
ki =

0i
3 1

i =

ali kai ( 2 0i ) Lu I i ( al 2 )
kai ( al 2 ) ali kai ( 2 0i ) Lu I i

i = kai i
The unaligned inductance Lu can be found from the

(I) characteristic at the unaligned position.

The offset angle 0i is given as
( I i-1 , ) al
0i = 1
m

where m is maximum flux linkage at the aligned

position corresponding to maximum value of excitation
current, is an empirical coefficient with the value in the
range from 8 to 12, and (Ii1,) is the flux linkage in the
position obtained for excitation current Ii1 selected in
the previous step of computation.
Parameter kai is given as
uai
kai = ali
3 1

C. Computation of torque lookup table T(I,)

The torque lookup table can be calculated on the basis
of magnetic coenergy Wm , using first central difference
approximation method of numerical differentiation, as:

T=

Wm ( I , )

(11)
I=const.

where the magnetic coenergy can be obtained using

numerical integration method based on Simpson's rule as
I

Wm = ( I , ) dI
0

(12)

III. COMPARISON OF ANALYTICAL MODEL AND FEA

To investigate whether the proposed analytical model
is suitable to predict SRM magnetization characteristic
and torque lookup table, an experimental 12/8 SRM is
used for analytical and numerical computation. The basic
parameters of 12/8 SRM are given in [21].
Fig. 4 shows the flux linkage and static torque
characteristics computed by both analytical model and 2D FEA. Ansoft Maxwell software is used for numerical
computation.
The magnetization characteristics obtained using
analytical model is fairly matching with those obtained
by FEA. However, there is a discrepancy between
analytical and FEA results for higher torques. Actual
static torque curves tend not to be flat in the region II.
The flat part of torque curves corresponding to the linear
part of (I,) characteristics modeled by (9). It is
necessary to improve modeling of (I,) curve in the
region II in order to boost the accuracy of proposed
model. The analytical model is proved as computationally
very fast in comparison to FEA.

[2]

[3]

[4]

[5]

[6]
[7]
[8]

[9]

12/8 SRM
1.5

22.5

Solid lines - Analytical model

Marked lines - FEA

16.5

[10]

Flux Linkage (Weber)

1.2

12.5

[11]

0.9

8.5
0.6

[12]
4.5

0.3

10

15

20

25

[13]
30

Current (Ampere)

12/8 SRM
140

30 A

Solid lines - Analytical model

Marked lines - FEA

[14]

Torque (Newton Meter)

120

25 A

100

[15]

80

20 A

60

15 A
40

[16]

10 A
20

5A
0

2.5

7.5

10

[17]
12.5

15

17.5

20

22.5

Rotor Position (Mechanical Degree)

Fig. 4. Flux linkage and static torque characteristics computed

by analytical model and FEA for 12/8 SRM.

Experimental verification of proposed analytical

model is currently being implemented. The final version
of the paper will compare the analytical results to
experiment for 12/8 SRM.

[18]

[19]

[20]

REFERENCES
[1]

R. Arumugam, D. A. Lowther, R. Krishnan, J. F. Lindsay,

''Magnetic Field Analysis of a Switched Reluctance Motor Using a
Two Dimensional Finite Element Model,'' IEEE Transactions on
Magnetics, Vol. Mag-21, No. 5, September 1985, pp. 18831885.

[21]

M. Moallem, C. M. Ong, ''Predicting the Steady-State

Performance of a Switched Reluctance Machine,'' IEEE
Transactions on Industry Applications, Vol. 27, No. 6,
November/December 1991, pp. 10871097
K. N. Srinivas, R. Arumugam, ''Analysis and Characterization of
Switched Reluctance Motors: Part I Dynamic, Static, and
Frequency Spectrum Analyses,'' IEEE Transactions on Magnetics,
Vol. 41, No. 4, April 2005, pp. 13061320.
A. Deihimi, S. Farhangi, G. Henneberger, ''A general nonlinear
model of switched reluctance motor with mutual coupling and
multiphase excitation,'' Electrical Engineering 84 (2002), Springer
Verlag, pp. 143158.
S. H. Mao, D. Dorrell, M. C. Tsai, ''Fast Analytical Determination
of Aligned and Unaligned Flux Linkage in Switched Reluctance
Motors Based on a Magnetic Circuit Model,'' IEEE Transactions
on Magnetics, Vol. 45, No. 7, July 2009, pp. 29352942.
D. Lin, P. Zhou, S. Stanton, Z. J. Cendes, ''An Analytical Circuit
Model of Switched Reluctance Motors,'' IEEE Transactions on
Magnetics, Vol. 45, No. 12, December 2009, pp. 53685375.
A. Radun, ''Analytical Calculation of the Switched Reluctance
Motors Unaligned Inductance,'' IEEE Transactions on Magnetics,
Vol. 35, No. 6, November 1999, pp. 44734481.
A. Radun, ''Analytically Computing the Flux Linked by a
Switched Reluctance Motor Phase When the Stator and Rotor
Poles Overlap,'' IEEE Transactions on Magnetics, Vol. 36, No. 4,
July 2000, pp. 19962003.
H. C. Lovatt, ''Analytical model of a classical switched-reluctance
motor,'' IEE Proc.-Electr. Power Appl., IEE Proceedings online
no. 20040545, 2004, pp. 17.
T. J. E. Miller, M. McGilp, ''Nonlinear theory of the switched
reluctance motor for rapid computer-aided design,'' IEE
Proceedings, Vol. 137, Pt. B, No. 6, November 1990, pp.
337347.
D. A. Torrey, X.-M. Niu, E. J. Unkauf, ''Analytical modelling of
variable-reluctance machine magnetization characteristics,'' IEE
Proc.-Elect. Power Appl., Vol. 42, No.1, January 1995, pp. 1422.
C. Roux, M. M. Morcos, ''A Simple Model for Switched
Reluctance Motors,'' IEEE Power Engineering Review, October
2000, pp. 4952.
H. Le-Huy, P. Brunelle, ''A Versatile Nonlinear SwitchedReluctance Motor Model in Simulink using Realistic and
Analytical Magnetization Characteristics,'' Proceedings of 31st
Annual Conference of IEEE Industrial Electronics Society IECON
2005, paper number RF-001757, November 2005.
B. Schinnerl, D. Gerling, ''Novel Analytical Calculation Method
for the Non-Linear -i Characteristics of Switched-ReluctanceMachines in the Aligned Rotor Position,'' Proceedings of the IEEE
International Electric Machines and Drives Conference IEMDC
2006, 2006, pp. 793796.
B. Schinnerl, D. Gerling, ''Novel Analytical Calculation Method
for the Non-Linear -i Characteristic of Switched-ReluctanceMachines in Arbitrary Rotor Positions,'' Proceedings of the
European Conference on Power Electronics and Applications
EPE 2007, ISBN 9789075815108, 2007.
R. Krishnan, ''Switched Reluctance Motor Drives: Modeling,
Simulation, Analysis, Design, and Applications,'' CRC Press LLC,
2001, Chapter 2.3.
P. Rafajdus, I. Zrak, V. Hrabovcov, ''Analysis of the Switched
Reluctance Motor (SRM) Parameters,'' Journal of Electrical
Engineering, Vol. 55, No. 78, July 2004, pp. 195200.
N. K Sheth, K. R. Rajagopal, ''Calculation of the Flux-Linkage
Characteristics of a Switched Reluctance Motor by Flux Tube
Method,'' IEEE Transactions on Magnetics, Vol. 41, No. 10,
October 2005, pp. 40694071.
E. Pdurariu, L. Somean, I.-A. Viorel, ''Switched Reluctance
Motor Analytical Models, Comparative Analysis,'' Proceedings of
the 12th International Conference on Optimization of Electrical
and Electronic Equipment OPTIM 2010, 2010, pp. 285290.
S. Smaka, '''Optimizacija dizajna elektrinog motora za pogon
hibridnog vozila,'' Ph.D. Disertation, Univerzitet u Sarajevu,
Elektrotehniki Fakultet, Sarajevo, 2012.
H. J. Wehner, ''Betriebseigenschaften, Ausnutzung und
Schwingungsverhalten bei geschalteten Reluktanzmotoren,'' Ph.D.
Dissertation, Der Technischen Fakultt der Universitt ErlangenNrnberg, Erlangen, Germany, 1997.