Modelling, Control and Experimental Investigation of a Five-Phase Series-Connected Two-Motor Drive with Single Inverter Supply

Emil Levi1A, Senior Member, Martin Jones1, Slobodan N. Vukosavic1, Member, Atif Iqbal1B

Hamid A. Toliyat2, Senior Member

1 Liverpool John Moores University, School of Engineering, Liverpool L3 3AF, UK 2 Texas A&M University, Dept. of Elec. Engineering, College Station, TX 77843-3128, USA

Abstract—The paper analyses a recently introduced two-motor five-phase drive system with series

connection of stator windings. It has been shown, using physical reasoning, that introduction of an

appropriate phase transposition in the series connection of two machines leads to a complete decoupling

of the flux/torque producing currents of one machine from the flux/torque producing currents of the

second machine. Consequently, independent vector control of the two machines becomes possible, while

using a single current-controlled five-phase voltage source inverter (VSI) as the supply. The drive system

modeling and control are elaborated at first in this paper, taking both machines as induction motors. It is

shown, using rigorous mathematical derivations, that the independent control of the two machines results

due to the placement of machines in two different sub-spaces of the five-dimensional space. This is

enabled by the phase transposition in the series connection. The models of the complete drive in the

stationary common reference frame and in the rotor flux oriented reference frames of the two machines

are developed. An associated vector control scheme for the two-motor drive is presented next. The second

part of the paper describes an experimental set-up, used further on to evaluate the dynamic behavior of

the two-motor drive. Performance is investigated by extensive experimentation for various transients

(acceleration, deceleration, reversing and disturbance rejection). Excellent decoupling of control of the

two machines is achieved. Both the concept of the drive and the approach to modeling and control are

thus fully verified experimentally. Finally, advantages and shortcomings of the series-connected five-

phase two-motor drive are discussed and potential application areas are highlighted.

Index Terms—Variable speed drives, Five-phase machines, Two-motor drives, Vector control,

Experimental investigation

I. INTRODUCTION Ever since the inception of the first five-phase variable speed drive in 1969 [1], five-phase machines have

been considered as a viable alternative to three-phase machines. This especially holds true for high-power and

safety-critical variable speed applications, where a five-phase drive can be realized using inverters with smaller

rating per leg while enabling fail-safe operation in redundancy mode [2,3]. Fault-tolerant properties are

especially important for the applications related to the concept of the ‘more-electric’ aircraft [4]. Five-phase

(and multi-phase in general) machines enable also an improvement in the noise characteristics of the drive [5,6], A Corresponding author: Liverpool John Moores University, School of Engineering, Byrom St, Liverpool L3 3AF, United Kingdom. Tel: +44-151 231 2257; Fax: +44-151 298 2624; e-mail: e.levi@ljmu.ac.uk. B A.Iqbal was with Liverpool John Moores University. He is now with Aligarh Muslim University, Aligarh, India.

1

a reduction in the stator winding losses and hence an improvement in the efficiency [7], and torque ripple

minimization [7,8]. More details regarding advantages and properties of multi-phase motor drives can be found

in a relatively recent survey of the state-of-the-art in the multi-phase induction motor drives [9].

A vector control scheme for a five-phase machine is in its basic form, regardless of the machine type,

identical to the corresponding vector control scheme for a three-phase machine [10,11]. However, since vector

control of an ac machine requires only two axis currents for decoupled flux and torque control, higher torque

density can be achieved in a five-phase machine by utilizing the remaining two degrees of freedom. The

injection of the third stator current harmonic enables utilization of the third spatial harmonic of the field for

torque production, in addition to the fundamental harmonic of the field [12-14].

A rather different use of these additional degrees of freedom was proposed in [15]. On the basis of

considerations related to the machine’s rotating field, it was suggested to connect two five-phase machines in

series and supply them from a single five-phase source. By introducing an appropriate phase transposition in this

series connection, it was reasoned that the two machines could be controlled completely independently, using

basic vector control schemes, although they are supplied from the common five-phase source. The major

advantage of such a two-motor drive system is the reduction of the number of required inverter legs, when

compared to an equivalent two-motor three-phase drive system (from six to five). This translates into increased

reliability at component level, due to a smaller number of components. However, it has to be pointed out that

one of the main advantages of multiphase machines, improved reliability at the system level due to fault tolerant

properties, is lost in this configuration. This is so since the available degrees of freedom, that can be used to

achieve fault-tolerant operation of a single multiphase machine, are used here to control other machine(s) of the

group.

The concept proposed in [15] for a two-motor series-connected five-phase drive has been further explored in

detail for all the possible system phase numbers greater than five. Detailed investigations, covering the principle

of required phase transposition, the number of connectable machines as a function of the system phase number,

connection diagrams, and verifications by means of simulation are available for various even and odd supply

phase numbers in [16] and [17], respectively. On the basis of theoretical studies, reported in [16,17], it was

concluded that, although the concept is applicable to any supply phase number, the best prospect for real-world

industrial applications hold the two-motor five-phase and six-phase series-connected drives. This is so since the

flow of flux/torque producing currents of one machine through the windings of the other machine(s) in the

group (where these currents are non-flux/torque producing) inevitably increases stator winding losses and

therefore jeopardizes the efficiency of the complete drive system. The two-motor six-phase drive, consisting of a

symmetrical six-phase (with 60° spatial displacement between any two consecutive phases) and a three-phase

machine connected in series and supplied from a six-phase inverter, has been examined in detail in [18,19]. Full

dynamic d-q model of the drive is given in [18] and detailed experimental studies, confirming existence of full

dynamic decoupling of the control of the two machines are reported in [18,19]. Additionally, using the same

principle of series connection, various drive configurations and control schemes are reported for an

asymmetrical six-phase drive system (two three-phase systems with 30° displacement) in [19]. One such

configuration, consisting of a series connection of two asymmetrical six-phase machines, has been tested

experimentally in the vector control mode of operation in [20].

A further, more detailed, analysis of the five-phase series-connected two-motor drive system of [15] has been

reported in [21], where once more physical reasoning rather than rigorous mathematical derivation was used to

2

underpin the concept and simulation rather than experimentation was used at the verification stage. In contrast to

[15], [21], where intuitive reasoning was predominantly used in the analysis of the five-phase two-motor drive,

strictly mathematical approach is adopted here. The concept is applicable to any type of ac machine with

sinusoidal spatial flux distribution. If the flux distribution contains higher spatial harmonics (as the case is in

concentrated winding machines), cross-coupling between the control of the two machines inevitably takes place,

since flux/torque producing currents of one machine interact with higher spatial harmonics of the magneto-

motive force of the other machine, leading to undesirable torque ripples [22]. Although it appears to be possible

to compensate for these harmonic torque ripples by using sophisticated modifications of the vector control

algorithm [23], the naturally decoupled and therefore simple control of the two machines, available with

sinusoidal flux distribution, is not realizable with concentrated winding machines. For the purposes of this paper

the two machines are taken as induction motors for both modeling and experimental studies and sinusoidal flux

distribution is assumed. Using the phase variable model of a five-phase machine as the starting point,

representation of the two-motor system in terms of d-q axis quantities is developed, at first in the stationary

reference frame and then in the rotor flux oriented reference frames of the two machines. It is shown that

specific method of stator winding series connection leads to the placement of the flux/torque producing

equivalent circuits of the two machines in two orthogonal and therefore mutually decoupled sub-spaces of the

five-phase system. On the basis of this d-q axis model of the series-connected two-motor drive system, an

indirect rotor flux oriented control scheme is designed. An experimental set-up is constructed and a detailed

experimental study is finally undertaken. Drive operation is investigated for a number of transients and the

presented results complement those of [24] with regard to acceleration, deceleration and speed reversal. In

addition to these transients, disturbance rejection properties during step loading/unloading of one of the

machines (not covered at all in [24]) are examined as well. It is shown that for all practical purposes the control

of the two machines is completely independent, although they are connected in series and a single five-phase

inverter is used as the supply. In many aspects this paper closely follows the approach of [18], but for the two-

motor five-phase series-connected drive. Due to the fact that here both machines are of the same phase number,

while in the six-phase two-motor drive one machine is six-phase while the other is three-phase (and hence the

three-phase machine is not adversely affected at all by the series connection), the perceived potential areas of

application are quite different.

The six-phase two-motor drive of [18,19] was suggested for applications requiring one high power machine

(six-phase) and one low-power machine (three-phase). In contrast to this, the five-phase two-motor series-

connected drive system is believed to hold a good prospect for industrial applications where two motors are

required to operate with constant output powers. Typical example would be a winder application. This issue is

discussed in detail in the last part of the paper, where advantages and shortcomings of this two-motor drive are

emphasized and the potential benefits of the application in constant power drives are underlined.

II. DESCRIPTION OF THE DRIVE The drive consists of two five-phase machines, which can be either induction or synchronous (permanent

magnet or synchronous reluctance) and which can be freely mixed within the system. As noted, it is assumed

here that the machines in question are both induction motors, without any loss of generality. Simplified cross-

section of the five-phase induction machine is shown in Fig. 1, together with an illustration of the connection of

the five-phase stator windings of the two machines in series, with an appropriate phase transposition. Phase

transposition in the series connection is a necessary prerequisite for independent vector control of the two

3

machines. Its purpose is to make flux/torque producing currents of one machine non flux/torque producing

currents in the second machine, and vice versa [15, 17, 21]. The two-motor drive is supplied from a single five-

phase VSI, which is current controlled. Current control is exercised upon phase currents in the stationary

reference frame, using either hysteresis or ramp-comparison current control. Inverter phase sequence is denoted

in Fig. 1 with capital letters A,B,C,D,E, while the phase sequence of the two machines, respecting the spatial

distribution of the windings (Fig. 1a), is identified with lower case letters a,b,c,d,e. Spatial displacement

between any two consecutive phases in the machines equals α = 2π/5.

According to the connection diagram of Fig. 1b, where phase voltages of the two machines are defined,

inverter phase-to-neutral voltages (A,B,C,D,E to neutral point n) and the correlation between inverter output

currents and machine phase currents are given with

as -as

bs

-es

cs

ds

-bs es

-cs

-ds

α = 72o

αs

αr θ

a.

A

B

C

D

E

vas1 vas2 iA

vbs1 vbs2 iB

vcs1 vcs2 iC n vds1 vds2 iD

ves1 ves2 iE

Stator of Stator of Machine 1 Machine 2

b.

Figure 1. Five-phase induction motor (a.) and five-phase two-motor drive with series connection of stator phase windings and an appropriate phase transposition (b.).

4

21

21

21

21

21

dsesE

bsdsD

escsC

csbsB

asasA

vvvvvvvvvvvvvvv

+=+=+=+=+=

(1) (2)

21

21

21

21

21

dsesE

bsdsD

escsC

csbsB

asasA

iiiiiiiiiiiiiii

==========

It is assumed for modeling purposes that all the standard assumptions of the general theory of electrical

machines apply [25], including the one related to sinusoidal distribution of the resulting field in the machine.

Rotor windings are initially taken as five-phase as well, for the sake of generality.

III. DRIVE MODELING

A. Phase-Variable Model Two machines of Fig. 1 are assumed to be of different parameters and ratings, for the sake of generality. The

electrical sub-system’s model of the drive in Fig. 1 is of the 15th order and it can be represented in matrix form

(underlined quantities) with

( )

dtiLd

iRv += (3)

where

⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡

=00

INVvv

⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡

=

2

1

r

r

INV

ii

ii , (4)

[ ][ ]TEDCBA

INV

TEDCBA

INV

iiiiii

vvvvvv

=

= (5)

[ ][ ]Terdrcrbrarr

Terdrcrbrarr

iiiiii

iiiiii

222222

111111

=

= (6)

⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡ +=

2

1

21

r

r

ss

RR

RRR ,

⎥⎥⎥

⎦

⎤

⎢⎢⎢

⎣

⎡ +=

22

11

2121

0'0

''

rrs

rrs

srsrss

LLLL

LLLLL (7)

Sub-matrices of the inductance matrix identified with the prime symbol are those whose form has been altered

through the phase transposition operation. Thus

⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡

++

++

+

=

111111

111111

111111

111111

111111

1

cos2cos2coscoscoscos2cos2cos

2coscoscos2cos2cos2coscoscos

cos2cos2coscos

MLMMMMMMLMMM

MMMLMMMMMMLMMMMMML

L

ls

ls

ls

ls

ls

s

αααααααααααααααααααα

(8)

⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡

++

++

+

=

222222

222222

222222

222222

222212

2

2coscoscos2cos2cos2coscoscos

cos2cos2coscoscoscos2cos2cos

2coscoscos2cos

'

MLMMMMMMLMMMMMMLMMMMMMLM

MMMMML

L

ls

ls

ls

ls

ls

s

αααααααααααααααααααα

and

5

( ) ( ) ( ) ( )( ) ( ) ( ) (( ) ( ) ( ) (( ) ( ) ( ) ( )( ) ( ) ( ) ( )

))

Tsrrs

sr

LL

ML

11

11111

11111

11111

11111

11111

11

coscos2cos2coscoscoscoscos2cos2cos

2coscoscoscos2cos2cos2coscoscoscos

cos2cos2coscoscos

=

⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡

−−+++−−+++−−−++−−−++

=

θαθαθαθαθαθθαθαθαθαθαθθαθαθαθαθαθθαθαθαθαθαθθ

(9a)

( ) ( ) ( ) ( )

( ) ( ) (( ) ( ) ( )( ) ( ) (( ) ( ) ( )

)

)

''

)cos(coscos2cos2cos2cos)2cos(coscoscos

coscos)2cos(2coscos2coscoscos)cos(2cos

cos2cos2coscoscos

'

22

22222

22222

22222

22222

22222

22

Tsrrs

sr

LL

ML

=

⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡

+−−+−++−

−−++++−−−−++

=

αθθαθαθαθαθαθαθθαθ

θαθαθαθαθαθαθθαθαθαθαθαθαθθ

(9b)

The rotor inductance sub-matrices of (7) are of the same form as 1sL of (8), while the resistance sub-matrices

are all five-by-five diagonal matrices.

Electromagnetic torques of the two machines can be expressed in terms of inverter currents as

( )( )( )( )( ) ⎪

⎪⎪

⎭

⎪⎪⎪

⎬

⎫

⎪⎪⎪

⎩

⎪⎪⎪

⎨

⎧

−+++++−++++++++++++++++

+++++

−=

)sin()2sin()2sin(

)sin(sin

111111

111111

111111

111111

111111

111

αθαθαθαθ

θ

erAdrEcrDbrCarB

erBdrAcrEbrDarC

erCdrBcrAbrEarD

erDdrCcrBbrAarE

erEdrDcrCbrBarA

e

iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

MPT (10a)

( )( )( )( )( ) ⎪

⎪⎪

⎭

⎪⎪⎪

⎬

⎫

⎪⎪⎪

⎩

⎪⎪⎪

⎨

⎧

−+++++−++++++++++++++++

+++++

−=

)sin()2sin()2sin(

)sin(sin

222222

222222

222222

222222

222222

222

αθαθαθαθ

θ

erAdrCcrEbrBarD

erDdrAcrCbrEarB

erBdrDcrAbrCarE

erEdrBcrDbrAarC

erCdrEcrBbrDarA

e

iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

MPT (10b)

B. Decoupling Transformation Decoupling (Clark’s) transformation matrix is applied first [25]. Let the correlation between original phase

variables and new (α-β-x-y-0) variables be given with abcde

fCf =αβ

, where C is the power-invariant

transformation matrix [12]

⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡

=

2/12/12/12/12/18sin6sin4sin2sin08cos6cos4cos2cos14sin3sin2sinsin04cos3cos2coscos1

0

52

αααααααααααααααα

βα

yxC (11)

Application of (11) in conjunction with inverter voltages yields axis components of the inverter voltages

⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡

=

⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡

E

D

C

B

A

INV

INVy

INVx

INV

INV

vvvvv

C

vvvvv

0

β

α

(12)

which can be further expressed, using correlation (1), as functions of the voltage axis components of the two

6

machines

⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡

++−+

=

⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡

+++++

=

⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡

021

21

21

21

21

21

21

21

21

0

sys

sxs

yss

xss

dses

bsds

escs

csbs

asas

INV

INVy

INVx

INV

INV

vvvvvvvv

vvvvvvvvvv

C

vvvvv

β

α

β

α

β

α

(13)

Due to the absence of the neutral conductor inverter zero-sequence voltage component must equal zero. The

correlation between inverter voltage axis components and individual machine’s voltage axis components implies

series connection between appropriate α-β and x-y circuits of the two machines. A corresponding correlation

between inverter output currents and α-β and x-y current components of the two machines is obtained by using

(11) in conjunction with (2),

21

21

21

21

sysINVy

sxsINVx

yssINV

xssINV

iii

iii

iii

iii

β

α

ββ

αα

==

==

−==

==

(14)

The zero-sequence component is omitted due to the star connection of the system without neutral conductor.

It follows from (14) that inverter α−β current components simultaneously represent α−β (flux/torque

producing) current components of machine 1, while they appear as x-y current components for machine 2. On

the other hand, inverter x-y current components are simultaneously equal to α−β (flux/torque producing) current

components of machine 2, while they flow through machine 1 as x-y current components. Since electromagnetic

torque and air-gap flux of a machine are produced solely by α−β current components, it follows that the

flux/torque of machine 1 can be controlled by means of inverter α−β current components, while flux/torque of

machine 2 can be controlled by means of inverter x-y current components. As α−β sub-space is orthogonal to x-

y sub-space, it follows that the specific method of series connection used in Fig. 1 will enable independent

vector control of the two machines. A similar explanation follows from (13), which shows that the α−β

(flux/torque producing) circuit of the first machine is placed into the inverter α−β sub-space. However, the α−β

circuit of the second machine is placed in the inverter x-y sub-space.

C. Model in the Stationary Common Reference Frame Upon application of the decoupling transformation matrix (11) onto inverter and rotor voltage equations of

(3), rotational transformation matrix, leading to the d-q system of equations, is applied in conjunction with rotor

equations:

11

1cossinsincos

⎥⎥⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢⎢⎢

⎣

⎡ −

=θθθθ

rD (15)

The angle θ in (15) is the instantaneous rotor position, which is different for the two machines and has values

21 and θθ , respectively. This means that different rotational transformation is applied to the two machines. This

7

is possible due to the decoupling of the equations of the two machines, achieved by application of (11). The

transformation (15) with the angle 1θ is applied to inverter α−β equations and rotor α−β equations of the first

machine, while the transformation with the angle 2θ is applied to the inverter x-y equations and rotor α−β

equations of the second machine.

The resulting system model is in the stationary common reference frame and is in general of the 15th order.

However, taking into account that rotor windings of the two machines are short-circuited, rotor x-y component

equations and rotor zero-sequence equation can be omitted from further consideration. Zero-sequence

component equation for the inverter can be omitted as well. The electromagnetic part of the drive system can

then be represented with eight first-order differential equations. The four inverter equations are

dtdi

Ldt

diLLiR

dtdi

LiRv

dtdi

Ldt

diLLiR

dtdi

LiRv

dtdi

LiRdt

diL

dtdi

LLiRv

dtdi

LiRdt

diL

dtdi

LLiRv

qrm

INVy

mlsINVys

INVy

lsINVys

INVy

drm

INVx

mlsINVxs

INVx

lsINVxs

INVx

INVq

lsINVqs

qrm

INVq

mlsINVqs

INVq

INVd

lsINVds

drm

INVd

mlsINVds

INVd

2222211

2222211

221

1111

221

1111

)(

)(

)(

)(

+++++=

+++++=

+++++=

+++++=

(16)

or, in terms of individual machine d-q axis stator voltage components (according to (13))

2121

2121

qsysINVydsxs

INVx

ysqsINVqxsds

INVd

vvvvvv

vvvvvv

+=+=

−=+= (17)

Rotor voltage equilibrium equations of the two machines are

( ) ( )( )

( ) ( )( )111111

11111

111111

11111

0

0

drmlrINVdm

qrmlr

INVq

mqrr

qrmlrINVqm

drmlr

INVd

mdrr

iLLiLdt

diLL

dtdi

LiR

iLLiLdt

diLLdt

diLiR

++−+++=

++++++=

ω

ω (18)

( ) ( )( )

( ) ( )( )222222

22222

222222

22222

0

0

drmlrINVxm

qrmlr

INVy

mqrr

qrmlrINVym

drmlr

INVx

mdrr

iLLiLdt

diLL

dtdi

LiR

iLLiLdt

diLLdt

diLiR

++−+++=

++++++=

ω

ω (19)

Finally, torque equations of the two series-connected machines are given in terms of inverter current axis

components with

[ ][ 22222

11111

qrINVx

INVydrme

qrINVd

INVqdrme

iiiiLPT

iiiiLPT

−=

−=

] (20)

Magnetising inductances in (16), (18)-(20) are defined as 11 5.2 MLm = , 22 5.2 MLm = .

D. Model in the Rotor Flux Oriented Reference Frames The simplest way to proceed further is by introduction of the space vectors, according to

222111

qrdrrqrdrr

INVy

INVx

INVxy

INVy

INVx

INVxy

INVq

INVd

INVdq

INVq

INVd

INVdq

jiiijiii

jiiijvvv

jiiijvvv

+=+=

+=+=

+=+=

(21)

8

Transformation into rotor flux oriented reference frames is done separately for the inverter d-q voltage equations

of (16) and rotor equations of the first machine (18) (variables will now have an additional superscript (1)), and

the inverter x-y voltage equations of (16) and rotor equations of the second machine (19) (superscript (2)). This

is possible since there exists full decoupling between the inverter d-q axis and x-y axis current components.

Rotor flux oriented reference frames are defined using and , respectively.

Correlation between inverter current and voltage space vectors in the stationary and in the rotor flux oriented

reference frames is

∫= dtrr 11 ωφ ∫= dtrr 22 ωφ

)exp(

)exp(

2)2(

1)1(

rINVxy

INVxy

rINVdq

INVdq

jff

jff

φ

φ

=

= (22)

where f stands for voltage or current. Transformation of (16), using (21) and (22) yields:

])[()()(

])[()()(

)2(22

)2(212

)2(2

2

)2(

21)2(

21)2(

)1(11

)1(211

)1(1

1

)1(

21)1(

21)1(

rmINVxyslsr

rm

INVxy

slsINVxyss

INVxy

rmINVdqlssr

rm

INVdq

lssINVdqss

INVdq

iLiLLjdtid

Ldt

idLLiRRv

iLiLLjdtid

Ldt

idLLiRRv

+++++++=

+++++++=

ω

ω (23)

while (18)-(19) take the same form as for a three-phase machine,

( ) ( )( )1(111

)1(111

)1(1

11

)1(

1)1(

11 )(0 rmlrINVdqmr

rmlr

INVdq

mrr iLLiLjdtid

LLdt

idLiR ++−++++= ωω ) (24)

( ) ( )( ))2(222

)2(222

)2(2

22

)2(

2)2(

22 )(0 rmlrINVxymr

rmlr

INVxy

mrr iLLiLjdtid

LLdt

idLiR ++−++++= ωω (25)

Torque equations (20) do not change the form,

[ ][ )2(

2)2()2()2(

2222

)1(1

)1()1()1(1111

qrINVx

INVydrme

qrINVd

INVqdrme

iiiiLPT

iiiiLPT

−=

−=

] (26)

Model (23)-(26) shows that independent rotor flux oriented control of the two series-connected machines is

possible and is a natural consequence of the way in which series connection of the two stator windings has been

done in Fig. 1. More specifically, rotor flux and torque of machine 1 can be controlled in the reference frame

attached to the rotor flux of machine 1, using inverter d−q current components (the first equation in (23) and

(26) plus (24)), while rotor flux and torque of machine 2 can be controlled in the reference frame aligned with

the rotor flux of machine 2, using inverter x−y current components (the second equation in (23) and (26) plus

(25)).

IV. VECTOR CONTROL OF THE TWO-MOTOR DRIVE Since, according to (23)-(26), the phase transposition in the series connection places stator d-q axis windings

of the second machine in series with x-y windings of the first machine (i.e. into the x-y sub-space of the inverter)

and vice versa, the independent vector control of the two machines can be realized using standard indirect

method of rotor flux oriented (RFO) control. The indirect RFO controller for each of the two machines is of the

same structure as for a three-phase machine or an asymmetrical six-phase machine [26] (the only difference is

that five, rather than three, phase current references are created at the output) and is shown in Fig. 2.

Two indirect RFO controllers operate in parallel and give at the output phase current references for the two

machines ( 5/2=k ; references for x−y stator current components are, according to Fig. 2, zero for both

machines):

9

( ) ( )

( ) ( )]4sin4cos[

)]3sin()3cos([

)]2sin()2cos([

)]sin()cos([

]sincos[

]4sin4cos[

)]3sin()3cos([

)]2sin()2cos([

)]sin()cos([

]sincos[

2*

22*

2*

2

2*

22*

2*

2

2*

22*

2*

2

2*

22*

2*

2

2*

22*

2*

2

1*

11*

1*

1

1*

11*

1*

1

1*

11*

1*

1

1*

11*

1*

1

1*

11*

1*

1

αφαφ

αφαφ

αφαφ

αφαφ

φφ

αφαφ

αφαφ

αφαφ

αφαφ

φφ

−−−=

−−−=

−−−=

−−−=

−=

−−−=

−−−=

−−−=

−−−=

−=

rqsrdses

rqsrdsds

rqsrdscs

rqsrdsbs

rqsrdsas

rqsrdses

rqsrdsds

rqsrdscs

rqsrdsbs

rqsrdsas

iiki

iiki

iiki

iiki

iiki

iiki

iiki

iiki

iiki

iiki

(27)

The currents of (27) are further summed, respecting the series connection with phase transposition of Fig. 1, in

order to create the overall inverter current references:

*2

*1

**2

*1

*

*2

*1

**2

*1

**2

*1

*

dsesEbsdsD

escsCcsbsBasasA

iiiiii

iiiiiiiii

+=+=

+=+=+= (28)

Closed loop phase current control in the stationary reference frame is finally applied to force the actual inverter

output currents of (2) to track the reference currents of (28). Assuming ideal current control, one has the equality

of the reference inverter currents (28) with actual inverter currents (2), so that actual machine currents are

related with reference machine currents of (27) through

*2

*121

*2

*121

*2

*121

*2

*121

*2

*121

dsesdses

bsdsbsds

escsescs

csbscsbs

asasasas

iiii

iiii

iiii

iiii

iiii

+==

+==

+==

+==

+==

(29)

PI

K1

1/s

jφr

e

2 5

ias*

ibs*

ics*

ids*

ies*

ids* = idsn

iqs* ω*

s

P

θ

φr

ω

ωsl*

Figure 2. Indirect RFO controller for a five-phase induction machine ( )(1 *1 dsr iTK = ).

Since the right-hand side of (29) contains in steady state operation two sets of sinusoidal currents of, in general,

different amplitudes and frequencies and these are summed according to the phase transposition in Fig. 1, it

10

follows from (29) that each of the five phases of any of the two machines carries simultaneously two sinusoidal

current components. One of these governs flux/torque producing (α−β) components while the other one is due

to the other machine in series and therefore it determines parasitic (x−y) current components.

V. EXPERIMENTAL SET-UP The experimental set-up is illustrated in Fig. 3. It utilizes two three-phase 14/42 A/A (continuous RMS /

transient peak) inverters with the common dc link, each of which is equipped with a Texas Instruments’

TMS320F240 DSP. The first three-phase inverter supplies phases A, C and E, while the second inverter supplies

phases B and D. All five currents are measured using LEM sensors and inverters’ DSPs perform closed loop

current control in the stationary reference frame, using digital form of the ramp-comparison method (in a custom

designed five-phase VSI measurement and control of four currents would suffice; however, since the set-up

utilizes internal DSPs of two commercial three-phase VSIs for current control, it was not possible to realize

current control using only four stator current component controllers). Current control rate and inverter switching

frequency are 10 kHz. PWM ripple is filtered out in the DSPs using FIR filters, which average 2n equidistant

samples taken during one switching period. Current signal, which is now PWM-ripple-free, is further used as the

input of the current controllers. The inverter current references of (28) are passed to the DSPs from a PC,

through a dedicated interface card. The control code is written in C. It performs closed loop speed control and

indirect rotor flux oriented control according to Fig. 2, in parallel for the two machines. Phase current references

are calculated for the two machines using (27) and inverter phase current references are then generated by

means of (28).

The results of the experiments, given in the next section, are obtained using two identical 4-pole, 60 Hz five-

phase induction machines with 40 slots on stator. These were obtained from 7.5 HP, 460 V, 60 Hz, 4-pole three-

phase induction machines by designing new stator laminations and a five-phase stator winding. The rotor is the

original three-phase machine rotor, unskewed, with 28 slots. The indirect vector controller for both machines is

the same and is the one shown in Fig. 2.

Various experimental tests are performed in order to verify the independence of the control of the two

machines. The results are reported in the following section. Operation in the base speed region only is

considered and the stator d-axis current references of both machines are constant at all times. Both machines are

running under no-load conditions in all tests, except for the disturbance rejection testing where one machine is

coupled to a dc generator (Fig. 3).

VI. EXPERIMENTAL RESULTS The approach, adopted in testing with the idea of proving the decoupling of control of the two machines, is

the same as the one used for six-phase two-motor drive testing in [18,19]. Both machines are initially brought to

a certain steady state speed. A speed transient is initiated next for one machine, while the reference speed of the

other machine remains unchanged. Full decoupling of control will exist if and only if the speed and, more

importantly, stator q-axis current command of the machine whose speed reference has not been altered, do not

change. The transients examined in experiments are acceleration, deceleration, and speed reversal under no-load

conditions. The machine running at constant speed is labeled as IM2, while the one undergoing a transient is

IM1. Since the machines are identical, it is irrelevant whether the transient is applied to the first or the second

machine and no distinction is made in this respect in the presentation of results.

11

Figure 3. Five-phase inverter (top), two series-connected five-phase induction machines (middle), and one of the two five-phase machines coupled to a dc generator for disturbance rejection testing (bottom).

Stator d-axis current reference for both machines is 2.5 A (RMS) and the stator q-axis current reference limit

is set to 5 A (RMS). The transient behavior is illustrated by means of speed responses, stator q-axis current

references (peak values are shown), phase current reference for one phase of each machine, and by comparison

of inverter reference and measured current for one phase.

In the first test IM2 runs at 700 rpm. The speed reference of IM1 is initially 0 rpm and is then stepped to 500

rpm. Fig. 4 illustrates the responses. It can be seen from Fig. 4 that initiation of the acceleration transient for

IM1 does not affect the operational speed of IM2, which stays at 700 rpm. Stator q-axis current reference of IM2

in Fig. 4 has some inevitable noise, but what matters the most is that there is practically no change whatsoever

in this trace during the acceleration of IM1 (note that the scale for stator q-axis current reference of IM2 is very

small and this current does not exceed a few hundreds of mA). Thus it follows that interaction between the two

machines is practically negligible. Further evidence of undisturbed operation of IM2 during this transient is the

12

phase current reference in Fig. 4, which does not exhibit any change during acceleration of IM1. The achieved

quality of current control is excellent, as evidenced by the comparison of the measured and reference currents

for inverter phase D in Fig. 4. It should be noted that the illustration in Fig. 4 includes drive operation at zero

speed (initial steady state of IM1), which presents no difficulties whatsoever in this drive configuration since

both machines are equipped with the position sensor.

The second transient, illustrated in Fig. 5, is deceleration of IM1 from 600 to 200 rpm, with IM2 still running

at 700 rpm. Very much the same conclusions apply as for the acceleration transient. The ripple in the stator q-

axis current reference of IM2 in Fig. 5 is again negligibly small, with no evident change during the transient of

IM1. Hence it follows that for all practical purposes control of the two machines is perfectly decoupled,

although they are connected in series and supplied from a single five-phase VSI.

The third transient is the reversing transient of IM1 from 800 to −800 rpm, with IM2 running at 500 rpm. The

results are shown in Fig. 6. Once more, excellent decoupling of control of the two machines is achieved.

Reversal of IM1 does not have any impact on the operating speed of IM2. Moreover, stator q-axis current

reference of IM2 does not exhibit any noticeable variation during the transient either. Consequently, phase

current reference of IM2 is with constant amplitude and frequency during the whole observation interval.

Inverter measured and reference currents are once more in excellent agreement.

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

500

0

400

600

Time (s)

Spe

ed IM

1 (rp

m)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

700

800

600

IM1

IM2

Spe

ed IM

2 (rp

m)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-4

-2

0

2

4

6

8

Time (s)

Sta

tor q

-axi

s cu

rrent

refe

renc

e IM

1 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-1

0

1

IM1

IM2

Sta

tor q

-axi

s cu

rrent

refe

renc

e IM

2 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

Time (s)

Inve

rter p

hase

"d"

cur

rent

(A)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

measured

reference

Inve

rter p

hase

"d"

cur

rent

refe

renc

e (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

Time (s)

Sta

tor p

hase

"d"

cur

rent

refe

renc

e IM

1 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

IM1

IM2

Sta

tor p

hase

"d"

cur

rent

refe

renc

e IM

2 (A

)

Figure 4. Speed responses, stator q–axis current references, comparison of measured and reference current for one inverter phase, and current references for one phase of the two machines: IM2 runs at 700 rpm, while IM1

accelerates from 0 to 500 rpm.

13

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

0

200

500600700

Time (s)

Spe

ed IM

1 (rp

m)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

700

800

600

IM1

IM2

Spe

ed IM

2 (rp

m)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-4

-2

0

2

4

-6

-8

Time (s)

Sta

tor q

-axi

s cu

rrent

refe

renc

e IM

1 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-1

0

1

IM1

IM2

Sta

tor q

-axi

s cu

rrent

refe

renc

e IM

2 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

Time (s)

Inve

rter p

hase

"d"

cur

rent

(A)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

measured

reference

Inve

rter p

hase

"d"

cur

rent

refe

renc

e (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

Time (s)

Sta

tor p

hase

"d"

cur

rent

refe

renc

e IM

1 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

IM1

IM2

Sta

tor p

hase

"d"

cur

rent

refe

renc

e IM

2 (A

)

Figure 5. Speed responses, stator q–axis current references, comparison of measured and reference current for one inverter phase, and current references for one phase of the two machines: IM2 runs at 700 rpm, while IM1

decelerates from 600 to 200 rpm.

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

1200

800

-800

0

Time (s)

Spe

ed IM

1 (rp

m)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

400

500

600

IM1

IM2 Spe

ed IM

2 (rp

m)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-8

-6

-4

-2

0

2

4

Time (s)

Sta

tor q

-axi

s cu

rrent

refe

renc

e IM

1 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-1

0

1

IM1

IM2

Sta

tor q

-axi

s cu

rrent

refe

renc

e IM

2 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

Time (s)

Inve

rter p

hase

"d"

cur

rent

(A)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

measured

reference

Inve

rter p

hase

"d"

cur

rent

refe

renc

e (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

Time (s)

Sta

tor p

hase

"d"

cur

rent

refe

renc

e IM

1 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

IM1

IM2

Sta

tor p

hase

"d"

cur

rent

refe

renc

e IM

2 (A

)

Figure 6. Speed responses, stator q–axis current references, comparison of measured and reference current for one inverter phase, and current references for one phase of the two machines: IM2 runs at 500 rpm, while IM1

reverses from 800 to −800 rpm.

14

Finally, to test the disturbance properties of the drive, one machine is connected to a dc generator, as

illustrated in Fig. 3. Step load is applied at certain steady state no-load speed of operation. The results for this

transient are shown in Fig. 7, where the machine that is being loaded runs with 500 rpm speed reference, while

the other machine has 300 rpm speed reference. As can be seen from Fig. 7, application of the load to IM1 has

no consequence whatsoever on behavior of IM2, since neither speed nor stator q-axis current reference of IM2

experience any change during the loading of IM1. Thus the control of two machines remains practically

completely decoupled in loaded operation as well (since the speed controllers were tuned under no-load

conditions and the inertia is substantially increased by connecting the dc generator, the responses of IM1 are

now somewhat more oscillatory). Very much the same conclusions follow from Fig. 8 as well, where step

unloading is illustrated with initially higher loading torque and with IM1 running at 300 rpm (IM2 now at 500

rpm).

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

400500600

Time (s)

Spe

ed IM

1 (rp

m)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9-4

-2

0

2

4

6

Speed

q-axis current reference

Sta

tor q

-axi

s cu

rrent

refe

renc

e IM

1 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

200300

400

Time (s)

Spe

ed IM

2 (rp

m)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-1

0

1

Speed

q-axis current reference

Sta

tor q

-axi

s cu

rrent

refe

renc

e IM

2 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-8-6-4-202468

Time (s)

Inve

rter p

hase

"d"

cur

rent

(A)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-8-6-4-202468

measured

reference

Inve

rter p

hase

"d"

cur

rent

refe

renc

e (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

Time (s)

Sta

tor p

hase

"d"

cur

rent

refe

renc

e IM

1 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

IM1

IM2

Sta

tor p

hase

"b"

cur

rent

refe

renc

e IM

2 (A

)

Figure 7. Speed responses, stator q–axis current references, comparison of measured and reference current for

one inverter phase, and current references for one phase of the two machines: IM2 runs at 300 rpm, while IM1 is suddenly loaded during running at 500 rpm.

VII. APPLICATION RELATED ISSUES

When compared to an equivalent two-motor three-phase drive, series-connected five-phase two-motor drive

provides three main benefits: a saving of one inverter leg, easiness of the complete vector control algorithm

implementation within a single DSP, and possibility for direct utilization of the braking energy that does not

have to circulate through the inverter. The only but serious disadvantage is an increase in the stator winding

losses in each of the two machines, since flux/torque producing currents of both machines flow through the

windings of both machines. The problem of the increase in the stator winding loss has been discussed in detail

in [21], where it was shown that, for two identical five-phase machines operating under the same conditions, the

stator winding losses will double. This shortcoming is likely to outweigh the advantages for general-purpose

applications.

15

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

200300

400

Time (s)

Spe

ed IM

1 (rp

m)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9-4

-2

0

2

4

6

Speed

q-axis current reference

Sta

tor q

-axi

s cu

rrent

refe

renc

e IM

1 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

400500600

Time (s)

Spe

ed IM

2 (rp

m)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-1

0

1

Speed

q-axis current reference

Sta

tor q

-axi

s cu

rrent

refe

renc

e IM

2 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-8-6-4-202468

Time (s)

Inve

rter p

hase

"d"

cur

rent

(A)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-8-6-4-202468

measured

reference

Inve

rter p

hase

"d"

cur

rent

refe

renc

e (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

Time (s)

Sta

tor p

hase

"d"

cur

rent

refe

renc

e IM

1 (A

)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-6-4-20246

IM1

IM2

Sta

tor p

hase

"b"

cur

rent

refe

renc

e IM

2 (A

)

Figure 8. Speed responses, stator q–axis current references, comparison of measured and reference current for one inverter phase, and current references for one phase of the two machines: IM2 runs at 500 rpm, while IM1 is

suddenly unloaded during running at 300 rpm.

However, it is believed that the scheme of Fig. 1 may offer a considerable saving in the installed inverter

power (compared to the standard solution with two three-phase motors and two three-phase VSIs) if the two

motors are required to operate in the constant power mode, with opposing requirements on rotational speeds and

torques. Such a situation arises in winders. For example, a typical paper machine involves nowadays a number

of separate drives for various sections, which are independently controlled with an adequate synchronization.

The material is transported from one drum, driven by one machine, to another drum, driven by the other

machine and a certain technological process takes place in between. Typically, force and linear speed at which

rewinding takes place have to be kept constant, so that a constant power regime of operation results. Since all

the material is initially at one drum, while at the end of the process it is at the other drum, the two consecutive

sections of a winder system are typically with very different operating conditions. While one machine (driving

the drum at which all the material initially is) runs at low speed (low voltage) with high torque (current), the

other machine (to which the material is transported) runs at high speed (high voltage) and low torque (low

current). The situation is reversed at the end of the process. It is precisely these very different operating

conditions, which may make the drive structure of Fig. 1 a viable solution, especially if surface-mounted

permanent magnet synchronous machines are used rather than induction motors (since the absence of the

requirement for stator d−axis current component makes total stator current directly proportional to the required

torque; hence at very low load torque stator current is practically zero, while in the induction machine it is

almost equal to the stator d−axis current component). Due to the very different torque and speed requirements

on the two motors, it could be possible to attain operation at all speeds/torques with the total stator winding loss

16

in each machine that does not exceed the rated one. In other words, there could be no requirement for de-rating

of the motors due to the increased stator winding loss, although the total loss would be still higher (and

efficiency therefore lower) than with two single independently-controlled three-phase drives. The loss in

efficiency could however be offset by the reduction in the capital outlay for the inverter supply. Once more, due

to the operation of the two motors under opposing speed/torque requirements, the rating of the inverter for the

series-connected two-motor drive can be practically equal to the rating of just one motor. This applies to both

voltage and current, and hence total power, ratings. This contrasts favorably with the current situation, where

two fully-rated three-phase inverters are required for independent supply and control of the two-motor drives in

winders.

VIII. CONCLUSION

The paper deals with a series-connected five-phase two-motor drive. A brief review of the operating

principles is provided first. This is followed by the detailed mathematical modeling of the complete drive

system. Models in the stationary common reference frame and the rotor flux oriented reference frames of the

two machines are developed. It is thus shown mathematically that the specific method of the stator winding

series connection enables independent vector control of the two motors although they are supplied from a single

five-phase inverter.

An experimental set-up is further described and the emphasis is placed on presentation of experimental results

for various transients (acceleration, deceleration and speed reversal). It is thus verified experimentally that the

control of the two series-connected machines is practically completely decoupled.

The investigated drive structure is applicable to all types of five-phase ac machine with sinusoidal flux

distribution. It is believed that the best prospect for real-world industrial applications exists in the winder area,

which belongs to the category of constant power drives. Here the series-connected two-motor drive could

provide a substantial saving on the capital outlay, especially if permanent magnet synchronous machines are

used, when compared to the standard two-motor three-phase solution.

ACKNOWLEDGEMENT The authors gratefully acknowledge support provided by the EPSRC, under the standard research grant

numbers GR/R64452/01 and EP/C007395, and by Semikron – UK, MOOG – Italy and Verteco − Finland.

REFERENCES [1] E.E. Ward, H. Härer, “Preliminary investigation of an invertor-fed 5-phase induction motor,” Proc. IEE,

vol. 116, no. 6, 1969, pp. 980-984. [2] T.M. Jahns, “Improved reliability in solid-state AC drives by means of multiple independent phase-drive

units,” IEEE Trans. on Industry Applications, vol. IA-16, no. 3, 1980, pp. 321-331. [3] J.R. Fu, T.A. Lipo, “Disturbance-free operation of a multiphase current-regulated motor drive with an

opened phase,” IEEE Trans. on Industry Applications, vol. 30, no. 5, 1994, pp. 1267-1274. [4] J.B. Wang, K. Atallah, D.Howe, “Optimal torque control of fault-tolerant permanent magnet brushless

machines,” IEEE Trans. on Magnetics, vol. 39, no. 5, 2003, pp. 2962-2964. [5] A.N. Golubev, S.V. Ignatenko, “Influence of number of stator–winding phases on the noise characteristics

of an asynchronous motor,” Russian Electrical Engineering, vol. 71, no. 6, 2000, pp. 41-46. [6] B. Zhang, H. Bai, S.D. Pekarek, W. Eversman, R. Krefta, G. Holbrook, D. Buening, “Comparison of 3-, 5-,

and 6-phase machines for automotive charging applications,” in Proc. IEEE Int. Electric Machines and Drives Conf. IEMDC, Madison, WI, 2003, CD-ROM paper 10642.

[7] S. Williamson, S. Smith, “Pulsating torques and losses in multiphase induction machines,” IEEE Trans. on Industry Applications, vol. 39, no. 4, 2003, pp. 986-993.

17

[8] K. Atallah, J.B. Wang, D. Howe, “Torque-ripple minimization in modular permanent-magnet brushless machines,” IEEE Trans. on Industry Applications, vol. 39, no. 6, 2003, pp. 1689-1695.

[9] G.K. Singh, “Multi-phase induction machine drive research – A survey,” Electric Power Systems Research, vol. 61, 2002, pp. 139-147.

[10] H.A. Toliyat, R. Shi, H. Xu, “A DSP-based vector control of five-phase synchronous reluctance motor,” in Conf. Rec. IEEE Ind. Appl. Soc. Annual Meeting IAS, Rome, Italy, 2000, CD-ROM Paper No. 40_05.

[11] H. Xu, H.A. Toliyat, L.J. Petersen, “Five-phase induction motor drives with DSP-based control system,” in Proc. IEEE Int. Elec. Mach. & Drives Conf. IEMDC2001, Cambridge, MA, 2001, pp. 304-309.

[12] H.A. Toliyat, M.M. Rahimian, T.A. Lipo, “A five phase reluctance motor with high specific torque,” IEEE Trans. Industry Applications, vol. 28, no. 3, 1992, pp. 659-667.

[13] H. Xu, H.A. Toliyat, L.J. Petersen, “Rotor field oriented control of a five-phase induction motor with the combined fundamental and third harmonic injection,” in Conf. Rec. IEEE Applied Power Elec. Conf. APEC, Anaheim, CA, 2001, pp. 608-614.

[14] L.Y. Xu, W.N.N. Fu, “Evaluation of third harmonic component effects in five-phase synchronous reluctance motor drive using time-stepping finite-element method,” IEEE Trans. on Industry Applications, vol. 38, no. 3, 2002, pp. 638-644.

[15] S. Gataric, “A polyphase Cartesian vector approach to control of polyphase AC machines,” in Proc. IEEE Ind. Appl. Soc. Annual Meeting IAS, Rome, Italy, 2000, Paper no. 38-02.

[16] E. Levi, M. Jones and S.N. Vukosavic, “Even-phase multi-motor vector controlled drive with single inverter supply and series connection of stator windings,” IEE Proc. - Electric Power Applications, vol. 150, pp. 580-590, 2003.

[17] E. Levi, M. Jones, S.N. Vukosavic and H.A. Toliyat, “A novel concept of a multiphase, mulit-motor vector controlled drive system supplied from a single voltage source inverter,” IEEE Trans. Power Electronics, vol. 19, pp. 320-335, 2004.

[18] M. Jones, S.N. Vukosavic, E. Levi, A. Iqbal, “A six-phase series-connected two-motor drive with decoupled dynamic control,” IEEE Trans. on Industry Applications, vol. 41, no. 4, 2005, pp. 1056-1066.

[19] E. Levi, S.N. Vukosavic, M. Jones, “Vector control schemes for series-connected six-phase two-motor drive systems,” IEE Proc. – Electric Power Applications, vol. 152, no. 2, 2005, pp. 226-238.

[20] K.K. Mohapatra, R.S. Kanchan, M.R. Baiju, P.N. Tekwani, and K. Gopakumar, “Independent field-oriented control of two split-phase induction motors from a single six-phase inverter,” IEEE Trans. on Industrial Electronics, vol. 52, no. 5, 2005, pp. 1372-1382.

[21] E. Levi, M. Jones, S.N. Vukosavic and H.A. Toliyat, “A five-phase two-machine vector controlled induction motor drive supplied from a single inverter,” European Power Electronics (EPE) and Drives Journal, vol. 14, no. 3, 2004, pp. 38-48.

[22] E. Levi, M. Jones, S.N. Vukosavic and H.A. Toliyat, “Stator winding design for multi-phase two-motor drives with single VSI supply”, in Proc. Int. Conf. on Electrical Machines ICEM, Chania, Greece, 2006, CD-ROM.

[23] E. Semail, E. Levi, A. Bouscayrol and X. Kestelyn, “Multi-machine modeling of two series connected 5-phase synchronous machines: Effects of harmonics on control”, in Proc. European Power Electronics and Applications Conf. EPE, Dresden, Germany, 2005, CD-ROM paper 398.

[24] A. Iqbal, S. Vukosavic, E. Levi, M. Jones and H.A.Toliyat, “Dynamics of a series-connected two-motor five-phase drive system with a single-inverter supply,” IEEE Industry Applications Society Annual Meeting IAS, Hong Kong, 2005, pp. 1081-1088.

[25] D.C. White, H.H. Woodson, “Electromechanical Energy Conversion,” New York, NY: John Wiley and Sons, 1959.

[26] G.K. Singh, K. Nam, and S.K. Lim, “A simple indirect field-oriented control scheme for multiphase induction machine,” IEEE Trans. on Industrial Electronics, vol. 52, no. 4, 2005, pp. 1177-1184.

Emil Levi (S’89, M’92, SM’99) was born in 1958 in Zrenjanin, Yugoslavia. He received the Dipl. Ing. degree from the University of Novi Sad, Yugoslavia, and the MSc and the PhD degree from the University of Belgrade, Yugoslavia in 1982, 1986 and 1990, respectively. From 1982 till 1992 he was with the Dept. of Elec. Engineering, University of Novi Sad. He joined Liverpool John Moores University, UK in May 1992 and is since September 2000 Professor of Electric Machines and Drives. He has published extensively in major journals and conference proceedings and serves as an Associate Editor of the IEEE Transactions on Industrial Electronics and a member of the Editorial Board of the IEE Proc. – Electr. Power Applications. Martin Jones was born in 1970 in Liverpool, UK. He received his BEng degree (First Class Honors) from the Liverpool John Moores University, UK in 2001. He has been a research student at the Liverpool John Moores University from September 2001 till Spring 2005, when he received his PhD degree. Dr Jones was a recipient of the IEE Robinson Research Scholarship for his PhD studies and is currently with Liverpool John Moores University as a post-doctoral research associate. His research interests include inverter and electric drive control.

18

Slobodan N. Vukosavic (M’ 93) was born in Sarajevo, Bosnia and Herzegovina, Yugoslavia, in 1962. He received his B.S., M.S., and PhD degrees from the University of Belgrade, in 1985, 1987, and 1989, respectively. He was employed in the Nikola Tesla Institute, Belgrade until 1988, when he joined the ESCD Laboratory of Emerson Electric, St. Louis, MO. Since 1991, he has held a number of industrial and academic positions, including the project leader with the Vickers Co., Milano, Italy and Professor at the University of Belgrade, Serbia. He has published extensively and has completed over 40 large R/D and industrial projects. Atif Iqbal was born in 1971 in India. He received his B.Sc. and M.Sc. degrees in 1991 and 1996, respectively, from the Aligarh Muslim University, Aligarh, India. He was employed as Lecturer in Department of Electrical Engineering, Aligarh Muslim University, from 1991 till 2002, when he joined Liverpool John Moores University, Liverpool, UK as a PhD student. Dr Iqbal received his PhD in December 2005 and is now employed again by Aligarh Muslim University, Aligarh, India as a Reader. His principal area of research interest is induction motor drives. Hamid A. Toliyat (S’87, M’91, SM’96) received the Ph.D. degree from University of Wisconsin-Madison, Madison, WI in 1991, in electrical engineering. He is currently a Professor in the Department of Electrical Engineering, Texas A&M University. He has received the Texas A&M Select Young Investigator Award in 1999, Eugene Webb Faculty Fellow Award in 2000, NASA Space Act Award in 1999, and the Schlumberger Foundation Technical Award in 2000 and 2001. He is the recipient of the 1996 IEEE Power Eng. Society Prize Paper Award. His main research interests include multi-phase variable speed drives, fault diagnosis of electric machinery, and analysis and design of electrical machines.

19