Escolar Documentos
Profissional Documentos
Cultura Documentos
q
), the voltage can be defined as
q e d d S d
dt
d
i R v e + =
(1)
d e q q S q
dt
d
i R v e + + =
(2)
where the flux linkage components are functions of
d- and q-axis current as well as the rotor angle ().
(3)
(4)
From the above mentioned d- and q-axis quantities
it is possible to obtain the voltage space phasor (v
s
),
current space phasor (i
s
), as well as the flux linkage
space phasor (
s
). These vectors can also be
represented in a phasor diagram (Figure 3) which
gives an intuitive understanding as to how they
relate to each other.
Fig. 3 Space phasor diagram
In this diagram, is angle between the voltage and
current space phasor and is the angle between the
flux linkage and current space phasor. The
remaining two angles, and 0, are the angular
distances from the d-axis for theflux linkage and
current space phasors, respectively. Here the flux
linkage excludes any leakages in order to maintain
simplicity but these will be covered in a later
section. With the knowledge of these vectors, the
power factor can be determined using eq. (5).
(5)
And the developed torque, using eq. (6)
( )
s s
i T =
2
3
(6)
The RSM simulated in the FE model, however, uses
the Maxwell Stress Tensor (MST) method to
calculate the instantaneous torque, T. The MST
equation is defined as
}
I
I =
s n t
d B rB T
0
1
(7)
In eq. (7) the normal and tangential flux density
components, B
n
and B
t
are integrated along a closed
contour, I through the centre of the air gap having a
radius, r. Furthermore,
s
is the motors effective
stack length and
0
the permeability of free space.
C. Finite Element model
Since, in a voltage-fed machine, all the individual
voltage drops must be simulated, the end-windings,
which are not included in any two-dimensional
simulation, must be considered separately. This is
achieved by including an external equivalent circuit
(Figure 4) into the FE calculation [13].
Fig. 4 Fundamental equivalent circuit of per phase RSM
The equivalent circuit in Figure 4 consists of an
applied voltage (v), stator resistance (R
S
), end-
winding leakage reactance (L
e
), core loss resistance
(R
C
) and magnetising inductive reactance (L
m
),
producing an emf, e. The value of L
m
is calculated
by the FE model while the values of v, R
C
, L
e
and R
S
are selected by the user. Since a rough estimate of
the core loss is attainable through the use of a
datasheet and the resistance through measurement,
)) ( ), ( ), ( ( t t i t i
q d d d
u =
)) ( ), ( ), ( ( t t i t i
d q q q
u =
s
d q
v
v v 0 0
cos sin
cos
=
0
R
s
L
e
R
c
L
m
dt
d
e
m
=
i
v
the challenge lies in the calculation of the end-
winding leakage reactance.
C.1. End-winding leakage reactance
One of the methods of calculating the end-winding
leakage reactance is based on a formula given by
Honsinger [14] which was later simplified by
Kamper (eq. 8) [13].
ph H k
p
k Wk
md V L
p e
u p d
i u e
/ 10
8
) (
2
) (
) (
= (8)
The subscript u in V
(u)
represents a constant that
refers to the shape of the end-winding with
V
1
=600 (V-shaped coils),
V
2
=1180 (elliptical shaped coils), and
V
3
=1400 (rectangular coils).
Since the RSM investigated in this study has an
elliptical shaped end-winding, the co-efficient V
2
=
1180 is selected for calculation. Furthermore, m
represents the number of phases, d
i
the stators inner
diameter, W the number of turns in series per phase,
p the number of pole pairs, k
d
the distribution factor,
given by
|
|
.
|
\
|
|
.
|
\
|
=
q
q
k
d
6
sin
6
sin
t
t
(9)
and k
p(u)
the end-winding pitch factor. Similar to
V
(u)
, the factors k
p(1)
, k
p(2)
, and k
p(3)
refer to a V-
shaped, elliptical-shaped, and rectangular coil
respectively. The factors can be calculated as
follows:
|
|
.
|
\
|
|
|
.
|
\
|
=
q
c
q
c
k
p
3
4
6
sin 3
) 1 (
t
(10)
|
|
.
|
\
|
=
q
c
k
p
6
sin
) 3 (
t
(11)
2
) 3 ( ) 1 (
) 2 (
p p
p
k k
k
+
~ (12)
where c is the coil-pitch and q is the number of coils
per pole phase group, in which case k
p(2)
= 1.
Finally, k
e(p)
represents the end-winding factor for a
machine with p pole-pairs. The factors are specified
as
k
e(1)
=0.51,
k
e(2)
=0.595,
k
e(3)
=0.64, and
k
e(4)
=0.785.
For a 3-phase, 36 slot, 4-pole machine with an inner
stator diameter of 126.98mm, having 210 turns in
series/phase, with 3 coils/ppg, a coil-pitch of 9 and
a distribution factor of 0.9597, an end-winding
leakage inductance of L
e
= 9.778 mH/phase is
determined.
C.2. Combining field and circuit equations
The simulation is then made up of circuit as well
as field equations. The terminal voltage is supplied
by the user, which, in effect, calculates the vector
potentials and magnetising currents from the circuit
parameters using the combined field and circuit
equations [15]. The calculation for the voltage in an
electrical machine can be described in its simplest
form as
dt
d
iR
v + =
(13)
Since the FE program in this study performs its
calculations in a 2D mesh, the end-windings are
included as a separate, series inductance in the
equivalent circuit. This study therefore assumes the
end-winding flux to be a function of the current
only (
e
=
e
(i)) and the main flux to be a function of
the current and rotor angle (
m
=
m
(i,)). Eq. (14) is
consequently expressed as
( ) ( ) ( ) ( ) i i
dt
d
iR
dt
d
iR v
e m e m
u + + = + + = , (14)
dt
di
i dt
d
dt
di
i
iR v
e m m
c
c
+
c
c
+
c
c
+ =
u
u
(15)
The end-winding inductance is a constant value and
can be written as a voltage-drop across an inductor,
while the remaining magnetising flux-component
responsible for torque production is defined as
m
.
e
m
L i
dt
d
iR e
v + + =
(16)
The instantaneous current and flux linkage can be
written in its integral form to represent the current
density J and the magnetic flux density B.
e r
a a
s
L i a d B
dt
d
a d J R v
s r
e + + =
}} }}
(17)
It is important to realise that the current density
component is only integrated over the surface of the
slot
s
while the flux density is integrated over the
surface area
r
having the radius from the centre of
the rotor to the centre of the air gap. The current
density can now be written in terms of the electric
scalar potential V, the vector potential and the
conductivity .
e r
a a
s
L i a d B
dt
d
a d gradV
t
A
R v
s r
e o + +
|
|
.
|
\
|
c
c
=
}} }}
(18)
Since B=curl
e r
a a
s
L i a d A curl
dt
d
a d gradV
t
A
R v
s r
e o + +
|
|
.
|
\
|
c
c
=
}} }}
(19)
Using Stokes law,
} }}
=
C F
r d F F d F curl
(20)
the second term of (18) is written as
}} }
+ +
|
|
.
|
\
|
c
c
=
s
a C
e r s
L i r d A
dt
d
a d gradV
t
A
R v e o
(21)
The motor windings in this study are simulated to
have a solid conductor in each slot. The voltage V
1
across a conductor with the length l is therefore
written as
( ) Vl dl gradV V
l
= =
}
0
1
(22)
Using eq. (22), the currents in the windings are
therefore defined as
}} }} }}
+
c
c
= =
s s s
a
s
a
s
z
a
s
a d
l
V
a d
t
A
a d J I
1
1
o o (23)
s a
s
z
R
V
a d
t
A
I
s
1
1
+
c
c
=
}}
o (24)
Since the size of the slots cross-section is known,
the individual currents can be calculated through the
use of the vector potentials and stator voltage [15].
IV. MEASURED AND SIMULATED RESULTS
A. Measured results
The test bench consists of a RSM having the same
specifications as used in the FE simulation. By
definition, the voltage-fed RSM is not controlled in
any way and runs from the frequency and voltage
supplied by the network. It is therefore started and
driven to its synchronous speed using an external
DC motor. Once the motor has reached its
synchronous speed of 1500rpm, it will be switched
onto a sinusoidal voltage which will, in turn, drive
the motor from that point onwards. The method of
starting the RSM, however, is of no particular
importance to this thesis, as only the RSMs
operation under constant voltage supply is
considered. The motors various load conditions are
obtained through a 22kW DC motor which is
controlled by a four-quadrant DC-drive. The
measured results are taken for loads varying from
no-load to 150% of full-load. The full-load setting
(100%) is defined as the full rated output power of
the motor. In this case, a full-load motor has an
output power of 3kW. The measurements in this
study are therefore taken up to 4.5kW (28.65Nm).
These loads are varied by adjusting the DC-drives
current limiting capability, allowing the DC motor
to deliver the exact amount of reaction torque
needed to obtain the desired results. Referring to
Figure 5, it can be seen that the voltage-fed RSM
has a fairly low pull-out torque when compared to
the current-fed RSM. This is a slight disadvantage
for the voltage-fed RSM in terms of its practicality
in industry. The remainder of the measured
performance parameters for the voltage-fed RSM
are consequently taken up to a load torque of 22Nm.
Fig. 5 Average torque comparison vs. input power
Figure 6 displays a slight advantage for the voltage-
fed RSM in which it produces a slightly lower
torque ripple for its lower range of loads.
Fig. 6 Percentage torque ripple comparison vs. output torque
Figure 7 reveals a slight difference in the applied
voltage, along with a difference in the voltage-fed
RSMs stator current behaviour. The increase in the
voltage-fed RSMs current/torque rate of change
after 15Nm is indicative to its low pull-out torque.
The current-fed RSM, however, maintains a steady
supply of current increase to the stator as part of its
torque per ampere functionality. The rate of change
for the voltage-fed RSMs power factor vs. torque
graph in Figure 8, however, shows an opposite
behaviour when compared to that of Figure 7. The
combination of these two observations causes the
voltage-fed RSM to have a 4% overall lower
efficiency (Figure 9).
Fig. 7 Current and voltage comparisons vs. output torque
Fig. 8 Power factor comparison vs. output torque
Fig. 9 Efficiency comparison vs. output torque
Finally, it is interesting to note how the current-fed
RSMs current angle remains fairly constant in the
initial increase in applied load and then gradually
lags a few degrees behind the voltage-fed RSMs
current angle. This deviation is a result of the
torque-current estimator technique which controls
the amount of flux in the air gap for the current-fed
RSM.
0
5
10
15
20
25
30
0 1000 2000 3000 4000 5000 6000
A
v
e
r
a
g
e
T
o
r
q
u
e
(
N
m
)
Input power (W)
Measured torque (V-fed) Measured torque (C-fed)
0
0.5
1
1.5
2
0 5 10 15 20 25 30
%
T
o
r
q
u
e
r
i
p
p
l
e
Torque (Nm)
Torque ripple (C-fed) Torque ripple (V-fed)
0
5
10
15
20
25
0
100
200
300
400
500
0 5 10 15 20 25 30
R
m
s
l
i
n
e
c
u
r
r
e
n
t
(
A
)
R
m
s
l
i
n
e
v
o
l
t
a
g
e
(
V
)
Torque (Nm)
Voltage (C-fed) Voltage (V-fed)
Current (C-fed) Current (V-fed)
0
0.1
0.2
0.3
0.4
0.5
0.6
0 5 10 15 20 25 30
P
o
w
e
r
F
a
c
t
o
r
Torque (Nm)
PF (C-fed) PF (V-fed)
0%
20%
40%
60%
80%
100%
0 5 10 15 20 25 30
E
f
f
i
c
i
e
n
c
y
(
%
)
Torque (Nm)
Efficiency (C-fed) Efficiency (V-fed)
Fig. 10 Efficiency comparison vs. output torque
B. Finite Element results
B.1. FE model accuracy
One of the objectives of this study is to find an
acceptable coherency between the measured and
simulated results for both the voltage- and current-
fed RSM. Since the RSMs efficiency (eq. 25) is
comprised of torque, angular velocity, line voltage,
line current and power factor, it is a sufficient
method of establishing the FE models accuracy.
% 100
cos 3
% 100 % = =
e
q
L L
IN
OUT
I V
T
P
P
(25)
This coherency is shown by calculating the
percentage error between the measured and
simulated results (eq. 26).
% 100
1
%
2
1
|
|
.
|
\
|
=
=
n
i m
s m
i
i i
n
c (26)
This error calculation uses the square of the
difference between the sum of the measured values
(i
m
) and the sum of the simulated values (is) to find
the percentage error over the whole range of values
[16]. Referring to Figures 11 and 12, it is clear that
the measured and simulated results coincide well.
Using eq. (26) a percentage error of 0.24% and
0.79% can be found for the voltage- and current-fed
models, respectively. Since the two methods for
both these models coincide to a reasonable degree, a
deeper investigation can be made into the RSM.
Fig. 11 Simulated and measured efficiency vs. torque
for voltage-fed RSM
Fig. 12 Simulated and measured efficiency vs. torque
for current-fed RSM
B.2. D-axis flux linkage
Figures 13 and 14 represent the voltage- and
current-fed RSMs change in flux linkage,
respectively. This difference is due to the manner in
which the two RSMs were simulated. For the
current-fed RSM, the input variables consisted of an
rms current as well as the applied load, while for the
voltage-fed RSM, only the applied load was varied
while the voltage was kept constant (slight voltage
deviations can be neglected in this case). The FE
software was therefore able to calculate the voltage-
fed RSM flux linkages from a constant voltage
source, without the intervention of any external
parameters, explaining the steady decrease in d-axis
flux linkage (Figure 13). The current-fed RSM, on
the other hand, seemed to produce a more constant
flux linkage (Figure 14) which averaged at around
0.945Vs. This is because the stator current input
parameter was selected according to the measured
current produced by the RSMs sensorless vector
drive, consequently having an influence on the
overall calculated flux linkage.
0
10
20
30
40
50
60
0 10 20 30
C
u
r
r
e
n
t
a
n
g
l
e
(
)
Torque (Nm)
Current angle (C-fed) Current angle (V-fed)
0%
20%
40%
60%
80%
100%
0 5 10 15 20 25
%
E
f
f
i
c
i
e
n
c
y
Torque (Nm)
Simulated efficiency (V-fed)
Measured efficiency (V-fed)
0%
20%
40%
60%
80%
100%
0 5 10 15 20 25 30
%
E
f
f
i
c
i
e
n
c
y
Torque (Nm)
Simulated efficiency (C-fed)
Measured efficiency (C-fed)
Fig. 13 Flux linkage in the d-axis for voltage-fed RSM
Fig. 14 Flux linkage in the d-axis for current-fed RSM
B.3. Harmonics
The harmonic distortion produced by the RSM
geometry is of interest. Since sinusoidal voltages
and currents are used to drive the RSM, it is
necessary to obtain the voltage harmonics produced
by the current-fed RSM, and the current harmonics
produced by the voltage-fed RSM. From the
harmonic results found in the FE calculations, the
odd harmonics are taken up to the 7
th
order as there
are no even harmonics present and magnitude of
harmonic orders higher than the 7
th
becomes
negligible. Each harmonic orders percentage is
related to the fundamental harmonics rms value.
% 100
1
%
=
h
h
h
n
(27)
Where h
%
, h
h
and h
1
represent are percentage, n
th
harmonic, and fundamental harmonics respectively.
Figure 15 represents the change in the voltage
harmonic distortion, for the three harmonic orders,
as the load is increased. Here the current-fed RSM
produces a 3
rd
order harmonic distortion of 9.47% at
full-load, while the voltage-fed RSM (Figure 16)
produces 5.9% current distortion for the 3
rd
harmonic.
Fig. 15 Voltage harmonic spectrum for current-fed RSM vs.
output torque
Fig. 16 Current harmonic spectrum for voltage-fed RSM vs.
output torque
B.4. Core loss
As a final addition to the voltage-fed RSMs
performance study, the core loss, P
c
is investigated
as a function of the load. To analyse this loss, it is
broken up into its three main components, namely;
eddy current (P
e
), hysteresis (P
h
) and excess loss
(P
ex
). Using the FE software, it is possible to
determine an accurate method of obtaining the core
loss by deriving it from the motors flux density.
According to the separation model proposed by
[17], these three components can be written in terms
of flux density and frequency,
( ) ( )
5 . 1 2
B f e B f b B f a P
x
c
+ + = (28)
where a, b and e represent the hysteresis, eddy
current, and excess coefficients, respectively, and x
0.9
0.92
0.94
0.96
0.98
1
0 5 10 15 20 25
D
-
a
x
i
s
f
l
u
x
-
l
i
n
k
a
g
e
(
V
s
)
Torque (Nm)
0.89
0.9
0.91
0.92
0.93
0.94
0.95
0.96
0 5 10 15 20 25 30
D
-
a
x
i
s
f
l
u
x
-
l
i
n
k
a
g
e
(
V
s
)
Torque (Nm)
3rd
5th
7th
0%
5%
10%
15%
2
6
10
13
19
27
Harmonic order
V
o
l
t
a
g
e
h
a
r
m
o
n
i
c
s
(
%
)
Torque (Nm)
0%-5% 5%-10% 10%-15%
3rd
5th
7th
0%
2%
4%
6%
1
3
6
10
14
18
20
Harmonic order
C
u
r
r
e
n
t
h
a
r
m
o
n
i
c
s
(
%
)
Torque (Nm)
0%-2% 2%-4% 4%-6%
Average:
0.945Vs
h
3
=9.47% @19.1Nm
h
3
=5.9% @ 19.1Nm
is the Steinmetz coefficient. These parameters are
mathematically calculated using a nonlinear
regression analysis in MATLAB [18]. The results
from the FEM are shown in Figure 17.
Fig. 17 Core loss and its three main components vs. output
torque
It is clear that the current-fed RSM produces an
overall higher core loss than the voltage-fed RSM.
Although it does not make a big contribution to the
total core loss, an exception is found in the
hysteresis loss. Here, the voltage-fed RSM remains
slightly higher for the majority of the load torque. It
is also interesting to note how the hysteresis loss in
the voltage-fed RSM decreases with an increase in
load in comparison with the rising hysteresis loss
for the current-fed RSM. Furthermore, it can be
seen that, for both motors, the losses due to eddy
currents comprise the majority of the motors core
loss.
V. CONCLUSION
The FE model for both the voltage- and current-
fed RSM proved to be accurate to within 0.52%,
making it possible to obtain a deeper investigation
into the RSM. To conclude from these measured
and simulated results, the following disadvantages
were discovered for the voltage-fed RSM with
regard to its performance characteristics.
Low starting torque (requires cage or
external/soft start),
Low pull-out torque,
4% Lower overall efficiency,
No speed control and
No torque control
However, there are some advantages. These
include:
Cheaper operation (does not require a drive),
Overall lower torque ripple,
Lower voltage harmonic distortion and
Lower core losses
From these results, it can be concluded that the
disadvantages of the voltage-fed RSM outweigh the
advantages, making the current-fed RSM the
favourable alternative. However, despite the
aforementioned conclusion, the results found in
Figures 5, 8 and 9 clearly show that the torque,
power factor and efficiency are quite compatible
within the voltage-fed RSMs operating range,
making the voltage-fed RSM a viable alternative to
the current-fed RSM for fixed speed applications.
It is of the authors opinion that, although the
voltage-fed RSM appears to be outdated
technology, there are still numerous possibilities for
its application in industry. The technology growth
in RSMs is increasing by the day and is being
utilised in various complex applications throughout
industry. Some of these RSMs are used in an
environment that requires less complicated tasks.
The argument effectively leads towards the same
ultimate goal to reduce the complexities and
expenses related towards a current-fed operation.
Researchers and engineers are continuously
attempting to solve problems by complicating
solutions that could have been simplified. Complex
electronic filter circuits in electrical machines, for
instance, can be simplified by short-pitching the
stator windings thus eliminating harmonic content.
It is pointed out that electronics is not always
needed and, to the authors opinion, a physical
method is rather more elegant.
13.00
13.50
14.00
14.50
0 10 20 30
C
o
r
e
l
o
s
s
(
W
)
Torque (Nm)
Core loss (Cfed) Core loss (Vfed)
12.75
13.00
13.25
13.50
13.75
14.00
14.25
0 10 20 30
0.21
0.21
0.22
0.22
0.23
0.23
0.24
0.24
0 10 20 30
8.50E-03
8.60E-03
8.70E-03
8.80E-03
8.90E-03
9.00E-03
0 10 20 30
Eddy current loss (W) Excess loss (W) Hysteresis loss (W)
VI. REFERENCES
[1] C.A.M.D. Feraz and C.R. de Souza,
"Measuring the parameters of a cage-rotor
reluctance synchronous motor," in Canadian
Conference on Electrical and Computer
Engineering, 2001, pp. 775 780.
[2] C.A.M.D. Ferraz and C.R. de Souza,
"Reluctance synchronous motor asynchronous
operation," in Canadian Conference on
Electrical and Computer Engineering, 2002,
pp. 195 200.
[3] P.J . Lawrenson and R.M. Mathur,
"Asynchronous performance of reluctance
machines allowing for irregular distributions of
rotor conductors," in Proceedings of the
Institution of Electrical Engineers, 1972, pp.
318 - 324.
[4] H. Nam et al., "Design to improve starting
performance of line-start synchronous
reluctance motor for household appliances," in
Industry Applications Conference 39th IAS
Annual Meeting, 2004, pp. 79-85.
[5] Alwyn Nicolaas Hanekom, A torque ripple
analysis on reluctance synchronous machines,
1st ed., Voss. E, Ed. Cape Town, Western
Cape, South Africa: Cape Peninsula University
of Technology, 2006.
[6] S. Williamson, T.J . Flack, and A.F. Volschenk,
"Representation of skew in time-stepped two-
dimensional finite-element models of electrical
machines," in IEEE Transactions on Industry
Applications, 1995, pp. 1009 - 1015.
[7] P. L. Alger, The Nature of Polyphase Induction
Machines. New York: J ohn Wiley and Sons
Inc., 1951.
[8] P. J . Lawrenson, "Magnetic Field of the End-
Windings of Turbo Generators," in Proc IEE
108A, 1961, pp. 538-554.
[9] M. Liwschitz, Alternating Current Machines.:
D. van Nostrand Co Inc., 1961.
[10] A. B. J . Reece and A, Pramanick, "Calculation
of the End-Region of A.C. Machines," in Proc
IEE 112 No. 7, 1965, pp. 1255-1368.
[11] W. Schuisky, Beregnung Elektrischer
Maschinen.: Wien Springer-Verlag, 1960.
[12] S. Williamson and M. A. Mueller, "Induction
Motor End-Winding Leakage Reactance
Calculation using the Biot Savart Method,
Taking Rotor Currents Into Account," in Proc
Int. Conf. Elec Machines, Cambridge, 1990,
pp. 480-484.
[13] M.J . Kamper, "Design Optimization of
Cageless Flux Barrier Rotor Reluctance
Synchronous Machine," University of
Stellenbosch, Stellenbosch, PhD Thesis 1996.
[14] V.B. Honsinger, "Theory of end-winding
leakage reactance," IEEE Transactions, vol.
90, no. 1, pp. 417-426, 1959.
[15] Joo Pedro A. Bastos and Nelson Sadowski,
Electromagnetic Modeling by Finite Element
Methods, 1st ed.: CRC Press , 2003.
[16] Yalin Nie, Qiansheng Hu, and Yunkai Huang,
"The Measurement and Prediction of Iron Loss
Under Nonsinusoidal Voltage Waveform with
Arbitrary Frequency," in International
Conference on Electrical Machines and
Systems, Nanjing, 2008, pp. 232 - 236.
[17] G. Bertotti, "The Prediction of power losses in
soft magnetic materials," Journal de Physique,
vol. 49, no. 8, pp. 1915-1919, 1988.
[18] J emimah C. Akiror, "Model for Core Loss
Prediction at High Frequency and High Flux
Density," Concordia University, Montreal,
Canada, Degree of Master of Applied Science
2012.
Robert Smith was born in Cape Town, South Africa
and completed his National Diploma and BTech
(Electrical Engineering) in 2009 and 2011, respectively,
at the Cape Peninsula University of Technology
(CPUT). He is currently completing his MTech degree
at CPUT Cape Town, South Africa and wishes to
specialise in motor and generator design engineering.
Email: robertalbertsmith@gmail.com
Egon Voss was born in Hamburg (Germany) and
received his National Diploma inElectrical engineering
from the Fachhochschule Hamburg in 1978 and the
MSc. fromthe University of Cape Town in 1988. He
has worked as a consulting engineer in South Africa
and overseas. He is a senior lecturer at the Cape
Peninsula University of Technology in Cape Town,
South Africa and head of the electrical machines section
in the electrical engineering department. His main interest is the analysis,
design and finiteelement modeling of electrical machines.