Electrimacs 2002, August 18-21, Power systems III 1

Abstract -- The paper analyses generator operations of

asynchronous machines when they are connected to a passive
electrical network either directly or by an intermediate static
power converter. The analysis is performed by taking into
account the saturation phenomena of the main magnetic
circuits of the machine. Mathematical models of both physical
configurations are given and their non-linearities are
evidenced. The paper inve stigates the existence of periodical
sol utions related to stable operations of the systems. Numerical
and experimental results support the theoretical analysis.

Index Terms Operations, Generator, Asynchronous
machines, Saturation, Mathematical models.
I. NOMENCLATURE
i
r
space vector of rotor currents, referred to stator
windings;
i
s
space vector of stator currents;
l leakage inductance with reference to stator phase;
p pole-pair number;
r
r
rotor phase resistance;
r
s
stator phase resistance;
I
r
steady state vector of rotor current referred to stator
windings;
I
s
steady state vector of stator current;
t
m
L
trigonometric air-gap inductance for saturated
machines;
g
m
L
differential air-gap inductance for saturated machines;
T
el
electromagnetic torque;
T
R
rated value of T
el
;
R
u
load resistance;
v space vector of stator voltages;
angular rotor position;
?
r
angular speed of rotor;
?
n
rated value of ?
r
;

E. Pagano, D. Iannuzzi, L. Piegari are with the Department of
Electrical Engineering, University, of Naples, Via Claudio, 21 - I80125
Naples ITALY (e-mail: enpagano@unina.it ).
O. Veneri is with the Istituto Motori-CNR , Via G. Marconi, 8 -
I80125 Naples ITALY (e-mail: o.veneri@im.na.cnr.it ).

m
amplitude of air-gap flux space vector
O armature voltage pulsation.
II. INTRODUCTION
n recent years generator operations of asynchronous
machines have been the subject of some of our applied
researches. The idea has been suggested by various
considerations. The evolution of alternative sources of
electrical energy (for example solar and wind energy) that imply
weak distributed and disconnected electrical networks
requiring self adjusting operating conditions, suggests the use
of low maintenance, high reliability and low cost machines. The
growing use of asynchronous machines in electrical drives for
industrial and traction applications implies not only traditional
motor operations but also generator and plug braking
operations.
An analysis of technical literature has shown that motor
operations of asynchronous induction machines have been
widely studied both when they operate with or without
intermediate electronic power converters. By contrast,
generator operations have been dealt with less often. This lack
of knowledge may, perhaps, also be the cause of the poor
practical use of asynchronous generators. For this reason we
have prepared a theoretical and experimental research
programme, devoted to a thorough investigation of the
different aspects of generator behaviour, both on passive and
active electrical networks, i.e. self-excitation, self-sustaining
oscillations and steady-state operating frequencies on passive
networks; slip regions, air-gap magnetic flux densities and
variable frequency operating conditions on active networks.
The research programme is now in progress. We are convinced
that the results so far obtained will be useful for explaining the
most interesting aspects of generator behaviour of
asynchronous induction machines and, therefore, for
increasing their use in those practical applications where these
machines appear to better satisfy users requirements related
to performances, buying prices and maintenance costs.
Basically, we have found that in all cases generator operations
imply air-gap magnetic flux densities that are higher than those
corresponding to motor operations (see for example [1]). This
situation implies that when traditionally designed machines are
used, saturation may occur in their main magnetic circuits.
Therefore, theoretically, analysis of generator operations
should be performed by means of suitable mathematical models
Generator Operations of Asynchronous Induction
Machines Connected to AC or DC Active/Passive
Electrical Networks
D. Iannuzzi, E. Pagano, L. Piegari, O. Veneri
I
c
Electrimacs 2002, August 18-21, Power systems III 2
that take into account saturation phenomena. We have found
that it is possible to carry out a model which is very simple but
at the same time fully satisfactory for practical purposes. It has
been presented in a previous paper [2]. The model gives,
moreover, rise to simple feeding algorithms for digital control
applications. By means of this mathematical model we have
evaluated the necessary conditions for obtaining stable
generator operations on passive electrical networks. These
operations can, however, be really attained only if the system
is capable of self-excitation. Self-excitation occurs only if rest
conditions of the system are not all identically nil. The physical
system, represented by an asynchronous generator and its
passive electrical network, has a mathematical model set-up by
non-linear differential equations. On the basis of finite initial
conditions, their solutions can lead to limit cycles or focal
points [3], according to whether the necessary conditions for
the existence of stable operations which are quasi-stationary
are satisfied or not. Pseudo-sinusoidal steady-state operations
are obviously obtained by system parameters and boundary
conditions that imply solutions of the former kind. Theoretical
results related to this topic have been experimentally validated.
Some sample phase-plane pictures are shown in the paper to
give readers reasons for the solutions obtained. The behaviour
of machines operating in electrical drives, provided by
electronic power converters, can be analysed by referring to
the same mathematical model of the machine. It should,
however, be considered together with the mathematical model
of the converter. The mathematical model of the whole physical
system is once again set up by non-linear differential
equations, representing the non-linear behaviour of both
machine and converter. Solutions of this set of differential
equations are, obviously, depending on the operating
frequencies of the converter.
We have investigated the relationship between passive
network parameters and driving frequencies of the converters,
both related to the asynchronous machine shaft speed. As a
result we have experimentally verified the validity of the
relationships, theoretically found, that relate the above
mentioned quantities to each other. The analysis has been
devoted both to constant speed operations and to dynamic
behaviour of the system. In the paper we shortly summarise
some basic results previously published in other papers [4]
and then we give the results of the performance analysis of
asynchronous generators operating on passive electrical
networks both with or without intermediate converters. It is
clear that only in the former case can real generator operations
be obtained. In the latter case generator operations are mainly
used for braking mechanical loads.
III. BASIC PRELIMINARY CONSIDERATIONS
For sake of simplicity we initially refer to an asynchronous
machine which is supplied by a traditional sinusoidal voltage
with constant maximum amplitude and frequency. In this case it
is easily seen that the air-gap magnetic flux density has quite
different maximum amplitudes when the machine either
operates as a motor or as a generator (see for example fig. 1).
The differences are easily explained if we consider that, due to
the different sign of the real parts of the machine impedance,
e.m.f. in generator operations are greater than those in motor
operations. Saturation phenomena of the main magnetic
circuits, which can be neglected during motor operations of
traditionally designed machines, take on, by contrast,
significance during generator operations.


0 0.05 0.1 0.15 0.2
0.85
0.95
1.05
1.15
motor
generator

B/B
n
[p.u.]

Fig.1 Sample air-gap magnetic flux density in motor and generator
operations

We have experimentally verified [2] that saturation does
not give rise to noteworthy stator current distortion as,
perhaps, traditional theoretical considerations would lead to
expect. For this reason we have found that an improved
Kovacss mathematical model of the machine could be
successfully used for practical applications. Therefore, we
have introduced two different air-gap inductances, i.e:

( )

I
I
m
t
m
L and
( )

dI
I
m
d
g
m
L

and we have carried out and experimentally validated a
mathematical model, which has been presented in a previous
paper [2]. It is as follows:

( ) ; i
dt
d
t
m
L
g
m
L
i dt
d
t
m
L
s
dt
d
s
l
s
r

+ + +

,
_

i i
i v
( )
( ) ;
dt
i d
t
m
L
g
m
L
i dt
d
t
m
L
s m
L
s
r r
L
r
p j
s
r
dt
d
'
r
l
'
r
r

+ +
+ + +

,
_

i i
i i i 0

(1)
;
el
T
m
T
dt
r
d
0 +

( ) { };
r
jp exp
'
r s
m
m
pL
el
T i i
2
3

Electrimacs 2002, August 18-21, Power systems III 3
( ) [ ] ( ) [ ] .
s
s
r
m
s
s
r
e
s
s
r m
i
2 2
i i i i i i i + + + +

Form a practical point of view the mathematical model can
be easily used. In effect it requires only a preliminary
experimental evaluation of the no-load magnetic characteristic
of the machine. The evaluation can be made by measuring the
corresponding pairs of armature voltage and current values.
The non-linearities of the suggested mathematical model
disappear when the machine operates at steady-state.
Therefore, also when saturation phenomena occur, a
sinusoidal steady-state can be obtained for generator
operations on passive, linear electrical networks.
IV. GENERATOR OPERATIONS ON PASSIVE AC ELECTRICAL
NETWORKS
It is well-known from technical literature that generator
operations of asynchronous machines on passive electrical
networks can occur only on external networks provided by
capacitors. This condition is easily and simply understood by
considering that at steady-state no external voltages are
impressed on the physical electrical system and, therefore, real
and imaginary parts of its internal impedance must be nil. Due
to the inductance of asynchronous machines, only capacitors
can obtain the compensation of the imaginary part. Real parts
are compensated by negative equivalent resistance of the
rotor.
C R
i i
C R
s
i
3
G
T
m

Fig. 2 Schematical electrical configuration of a passive electrical
network fed by an asynchronous generator

From a general point of view the mathematical model of
the whole system, schematically represented by fig. 2, is given
by eq.s (1) together with

C R C dt
d
u
s
v i v
. (2)

Eq. s (1) and (2) represent a system of non-linear differential
equations, where the driving mechanical torque is given and all
currents are unknown. The torque is also the one and only
forcing cause of the physical system. It is easily verified that if
all initial conditions of electrical quantities are nil, also the
currents are always identically nil. By contrast, the presence of
a residual flux density or an initial capacitor charge gives rise
to not nil solutions for unknown electrical currents. From a
physical point of view the result is easily understood in both
cases. It is also easy to explain the analytical result connected
with initial capacitor charges. Residual flux density can be
mathematically considered as a Diracs pulse at the initial time
when defining ( ) ( ) 0 0

i d / d
g
m
L
m
by means of the
theory of generalised functions.
It is well-known that solutions of non-linear differential
equations can be of different kinds. Our problem can be set in
the following terms. Do the set of eq. s (1) and (2) admit a
periodic solution and, if yes, when? A theoretical answer to
this question can be given by considering whether the
Fouriers series expansions of all unknown currents represent a
set of solutions to the equations. This t heoretical procedure is,
indeed, very difficult to apply in practice also because the
function ( )

i
m
cant be successfully approximated with
simple analytical functions in the whole domain of variation of
m
and

i . Therefore, a numerical procedure has been

followed. It is based on the conditions that if the system of
eq.s (1) and (2) admits a periodic solution all its phase-
trajectories are closed. The results of the numerical procedure
are, indeed, limited to dominia of investigation assumed for the
range of the external parameters i.e. of R and C and for each
given machine. We have applied the numerical procedure to
different asynchronous machines. As an example in the
following, results referred to the sample machine of Tab. I are
given. By using our test bench we have driven the
asynchronous machine by applying different driving shaft
torque. Fig.s 3 refer to torque of 10 and 20 Nm. The value of
the load resistance is 67.8 / phase and capacitors have 210
F / phase.

0 0.2 0.4 0.6 0.8
0
10
20
30
Experimental Is
Simulated Is
Magnetizing current
t [s]
[A]
s

Electrimacs 2002, August 18-21, Power systems III 4
0 0.2 0.4 0.6 0.8
0
50
100
150
Experimental speed
Simulated speed
[rad/s]

r
t [s]

0 0.2 0.4 0.6 0.8
0
10
20
Experimental Is
Simulated Is
Magnetizing current
[A]
t [s]
s
0 0.2 0.4 0.6 0.8
0
50
100
Experimental speed
Simulated speed

r [rad/s]
t [s]


Fig. 3 - Theoretical and experimental amplitudes of machine angular
speeds and of armature current space vectors for different driving torque
(M = 10 Nm; M = 20 Nm)

As can easily be seen, numerical and experimental results
correspond very well. Both give evidence of the high value of
magnetising current which reaches values of up to 96% of the
load currents. Results confirm the necessity to refer to
mathematical models of the machine which takes into account
saturation phenomena of main magnetic circuits.
Theoretical and experimental results also allow us to
better understand physical phenomena, which occur during
self-excitation. For this reason phase trajectories of state
variables can be drawn and examined. In another paper [4] we
have dealt more extensively with this problem. For sake of
completeness fig. 4 shows phase-trajectories of armature
currents and machine shaft speed. They are drawn for two
different load circuits. The former one implies a steady-state,
the latter one gives rise to unstable operations. The results can
be explained simply if we analyse the behaviour of the system
at sinusoidal steady-state by means of eq.s (1) and (2). The
existence of such a state implies that a particular integral of the
system of differential eq.s (1) and (2) is represented by a set of
sinusoidal functions, when the shaft speed is constant, i.e. by:

;
t j
e
s s
~
,
t j
e
r
'
r
~
,
t j
e
s s
~
V v I i I i (3)

If eq.s (3) are put into eq.s (1) and (2), for each given
asynchronous machine, a system of seven algebraic equations
is obtained, where the values of O, ?
r
, C, R
u
, I
s
, I

r
, V, T
m
are
connected to each another:

C
s

u
CR
1
j V
I
+

,
_

;
( ) ;
r t
L
r
jp -
"
r m
L
2
3

t
VL

s
2
m
L
4
9
r
jp
s
r
t
L
s
j
I i
I I

+
+ +
,
_

(4)
( )
( ) ;
s t
L
r
jp
s
r
m
L
2
3
V
m
L
2
3
-

r t
L
t
L
r
jp -
'
r
r
r
j
I
I I

+ +
+

{ }
r
I
s
I m A
l e
T
(
.

The situation implies that, given three values of the above
mentioned quantities, the other seven can be univocally
evaluated.
Electrimacs 2002, August 18-21, Power systems III 5
Is /In
[p.u.]
d Is /dt [A /s] A

r /

n
[p.u.]
d

r /dt [rad/s]
B

Is /In
[p.u.]
d Is /dt [A /s] C

r /n
dr /dt [rad /s]
[p.u.]
D

Fig. 4 Phase-plane trajectories of current amplitude and speed for
steady-state (4A, 4B) and unstable (4C, 4D) operations.

When the system of fig. 2 is used in practice, the values
of O, I
s
and R
u
are generally given. The other corresponding
ones are obtained by solving the system of eq.s (4). It might be
interesting to examine how C and ?
r
vary in a given range. As
an example fig. 5 shows the function C (R
u
) for =314 rad
sec
-1
and V=380V.

10 20 30 40 50
-1
-0.5
0
0.5
1
[]
R u
u C /C b [p.u]
m b P /P
2 . 17 k
s
s
I
V
k
5 . 15 k 7 . 14 k
4 . 12 k
5 . 11 k
3 . 10 k

Fig. 5 - Function C (R
u
) for =314 rad sec
-
1 and V=380V obtaining
self-sustained steady-state generator operations of the asynchronous
machine of Tab.1.

As an exa mple fig. 3 compares theoretical and experimental
results achieved by a mass production asynchronous motor,
whose rating and main parameters an given in Tab. I.
For testing we have used a bench where the shaft of the
asynchronous machine is connected to a dc motor, supplied
by variable voltage sources, both for armature and excitation
windings. The armature windings of the asynchronous
machine are connected to a three-phase RC network, whose
resistances can be changed simply, ranging from 10 to 250.
Capacitors are connected to the machine by means of a two-
way switch, so that they can either operate in parallel to the
resistances or be charged separately.

TABLE I
Main parameters and rating of the tested asynchronous induction motor
Electrimacs 2002, August 18-21, Power systems III 6

P
R
[kW]
f
R
[Hz]
V
R
[V]
p r
s

[]
r
r

[]
l
s

[mH]
l
r

[mH]
L
m

[mH]

[kgm
2
]
11 50 380 2 0.4 0.4 3.0 3.0 69.7 0.23

Previous tests have been made without pre-charging
capacitors. We have verified that the decaying time of the
residual flux density is very short, so that self-excitation of the
system, i.e. generator operation of the machine, occurs only if
its armature windings are suddenly switched from the ac
feeding active network to the RC passive one. In this case
steady-state operations will take place for the pair of values of
R and C that have been theoretically predicted (see fig.5).
When armature windings are kept even for short times
disconnected from external networks, self-excitation does not
occur. However, if a small charge is given to capacitance (we
have used a dc source of 12 V) self-excitation occurs and
higher values of voltages across capacitors are reached at
steady-state. This experience shows that the system is self-
adjusting and, due to its non-linear behaviour, final stable
operating conditions can be reached.
V. GENERATOR OPERATION ON DC PASSIVE ELECTRICAL
NETWORK
In some electrical drives that require braking operations,
asynchronous machines operate as generators on a dc passive
electrical network by means of an intermediate inverter,
according to the schematically electrical configuration shown
in fig. 7. The mathematical model of the physical system is now
represented by eq.s (1), and by:

. e e i
;
C
i
R C
V
dt
dV
T
t j
dc
dc
u
ab ab

'

+
1
]
1

6
3

s
i
(3)

The system of equations (1) and (3) is still nonlinear, but now
nonlinearities not only occur for the saturation and for the
presence of product between unknown, but also for the non-
linear characteristic of the inverter-components. A preliminary
analysis

3
M
i
A B
R
C
u
a

Fig. 6 - Schematically electrical configuration of asynchronous machine
operating as generator on a dc passive electrical network by means of
an intermediate inverter

immediately shows that the circuit configuration does not
imply sinusoidal steady-state operations of the machine,
because the system of eq.s (1) and (3) does not have solutions
with sinusoidal armature currents and for constant i
dc
current.
Periodic solutions can, however, still be found. They imply
that:

1
O
1
O
k
t jk
sk s
k
t jk
sk s
s s
e V ; e I v i ;
(4)
. e dt e ; e I '
k k
t jk
k
t jk
rk r
k
t jk
rk r
r

0 0
O O
1
O

i

For sake of simplicity we put:

.
s
;
r
jp
e
r
' q i Q i q +



By taking into account eq.s (4), the second of the eq.s (1) can
be written as:

. e L jp e l jp
e
dt
d
L
e
dt
d
l e r
k
t jk
k m r
k
t jk
k r r
k
t jk
k m
k
t jk
k r
k
t jk
k r
s s
s
s s

+ +
+ +
1 1
1
1 1
2
3
2
3
0

Q q
Q
q q
(5)

For non periodical variation of the load torque, the frequency
of the speed variation is only due to the frequency of the
electromagnetic torque. In particular, because the torque is

'

q i
s
m k
el
T

Electrimacs 2002, August 18-21, Power systems III 7
the speed frequency is always an exact multiple of the stator
current frequency; i.e. it is

s
m

.

If we consider that speed can be substituted with its Fouriers
series and that according to Cauchys Rule for the product of
sums, eq (3) can be written as:

[ ]
[ ]

+
+

+
+

,
_

,
_

0
0
0
k
k
k
j
e
t
s
m ) k ( j
e
k
j
e
t
s
m ) k ( j
e
k , r
a
k
N
t
s
jk
e
k
M

(6)

From the analysis of eq. (6) it is clear that the variation of
speed produces voltage on armature windings at the same
frequency of the harmonic generated by the feeder; for this
reason the reaction of the armature to the speed variation can
be compensated by the feeder and a steady state solution for
the system can exist as normal practice confirms.
VI. CONCLUSIONS
The paper has analysed the behaviour of asynchronous
machines where they operate as generators on passive
electrical network, directly or by intermediate static power
converters. The analysis has been performed by means of
suitable mathematical models of machine and networks.
Emphasis has been given to self-excitation phenomena and to
the existence of periodical solutions. Theoretical results have
been validated by experimental tests.
VII. REFERENCES
Periodicals:
[1] Pagano E.; Piegari L.;Veneri O.: Generator and plug brake
operations of asynchronous machines Accepted for publication
in the book Jubil Power Electrical Engineering edit by G. A.
Capolino and R. Goyet.
[2] Pagano E.; Veneri O.: Generator operations of saturated
induction machines IEE Proc.-Electr. Power Appl. 2002.

Books:
[3] Davis, H. T. : Introduction to nonlinear differential and integral
equations. Dover Publ., New York 1962.

Papers from Conference Proceedings (Published):
[4] D. Iannuzzi - E. Pagano - L. Piegari - O. Veneri: Braking
operations of electrical drives for road vehicles. EDPE2001 3-5
Oct. Slovak Republic 2001.


VIII. BIOGRAPHIES

Enrico Pagano was born in Milan,
Italy on December 27
th
, 1939. He
graduated from the University of Naples
in December 1963. Since January 1964
he has been a lecturer in
Electromechanical Design at the
University of Naples and full professor
of Electric Machines at the University
of Naples since February 1975.
Founder and President of ANAE (Italian Association for Electrical
Drives), he is also Director of CRIAT (Italian Center on Traction
Drives). He is a member of VDE, AEI, IEEE Japan, IEEE USA and of
the New York Academy of Science. He is Director of the revue EERR
(Electrical Engineering Research Report).

Diego Iannuzzi was born in Naples Italy on
October 3
rd
1972. He graduated from the
University of Naples in February 1998. He is, at
present, a PhD at the University of Naples in the
field of Electrical drives and machines. His
main fields of interest are the asynchronous
machine either as motor or generator, traction
systems and alternative electrical supplies.


Luigi Piegari was born in Naples, Italy on April
2
nd
1975. He graduated from the University of
Naples in February 1999. He is, at present, a PhD
at the University of Naples in the field of
"Electrical drives and machines". His main fields
of interest are the asynchronous machine either
as motor or as generator, and alternative
electrical supplies (batteries, fuel cells) and
Cableway plants.


Ottorino Veneri was born in Naples, Italy on
February 14
th
, 1965. He graduated from the
University of Naples in January 1996 and
received a Ph.D in Electrical Engineering in
March 2000. He is, at present, a researcher at the
Istituto Motori of the National Research Council
CNR in the field of "Electrical drives and
machines". His main fields of interest are the
mathematical model of electrical machines,
control and feeding algorithms for electrical traction systems.