Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
\
|
|
.
|
\
|
|
.
|
\
|
=
1
2
3
2
. (2.6)
The real power at the machine terminals is calculated as
( )
ds ds qs qs e
i v i v P +
|
.
|
\
|
=
2
3
. (2.7)
The above equations are realized in the qd model of the induction machine as shown in
Figure 2.1
23
Figure 2.1 Control schematics for an induction model based on qd reference frame.
24
2.3 Variable Frequency Drive Modeling
To study the operation of multi-motor systems with VFDs, modeling is indispensable [17].
The basic VFD with V/F control is used in the industry in most of the cases because to its
low cost, and this drive is assumed in this thesis. The control operates under two principles:
1. The rotor speed in the motoring region of an IM is proportional to the input electrical
frequency
2. To avoid machine saturation, the voltage-per-frequency ratio input to the motor
should be constant (uncompensated approach).
These principles result in the following relationships for the voltage and frequency:
*
e
b
b
s
V
V e
e
(
= , (2.8)
here
* *
2
rm e
P
e e = ,
(2.9)
where
b
V and
b
e are the rated base voltage and base electrical frequency of the IM;
*
e
e and
*
rm
e are the commanded electrical frequency and approximate mechanical speed. The model
considered in this thesis is implemented according to the methodology defined in [[17],
Chap. 14] and the diagram shown in Figure 2.2.
25
Figure 2.2 Control schematics for V/F controller.
26
3 Load Sharing under Rotor Resistance Variation
3.1 Introduction
The traditional/basic V/F IM drives operate only using speed command signals, and the
developed torque is consequently determined according to the torque-speed characteristics of
the machine. As the developed torque is a function of rotor resistance, in load-sharing
applications, particularly, deviations among the rotor resistance values is probable and will
result in disproportionate sharing of the mechanical load and hence overloading of one or
several machines. In this thesis, a solution for the above issue is presented in the form of an
improved V/F scheme, which compensates for the parameter change by adjusting the
command speed reference signals accordingly. The proposed method is shown to be effective
and easy to implement, and may be readily extended to an arbitrary number of motors or to
the case where dissimilar motors are coupled and the mechanical torque is intentionally
distributed unevenly among the machines according to their ratings.
3.2 Induction Machine Steady State Torque
The steady state electromagnetic torque developed by an IM is given by [23]
( ) ( )
2 2
2
2
3
lr th r th
r
e
th
e
X X s r R
s r V P
T
' + + ' +
'
=
e
, (3.1)
where
th
V ,
th
R , and
th
X are the Thevenin equivalent circuit parameters obtained from the
steady state equivalent circuit of Figure 3.1 and s is the rotor slip.
27
r
s
L
ls
V
s
Thevenin
Equivalent
L
lr
L
m
'
r
r
s
I
s
I'
r
I
m
Figure 3.1 Steady state equivalent circuit of an induction machine.
The stator side of the induction machine can be combined with the magnetizing inductance
using Thevenin equivalence. The resulting voltage, resistance and reactance can be
calculated as follows:
( )
( )
M ls s
M s
th
X X j r
X j V
V
+ +
= ,
(3.2)
( )
M ls
M s
th
X X
X r
R
+
= ,
(3.3)
( )
M ls
M ls
th
X X
X X
X
+
= ,
(3.4)
where
s
V is the voltage input (per phase) to the IM stator; and
e
e is the stator electrical
frequency.
In the motoring region, where the slip is typically low, (3.1) may be approximated as
( )
r e
th
r
r
e
th
e
r
s V P
s r
s r V P
T
'
=
'
'
~
e e
2
2
2
2
3
2
3 . (3.5)
Equation (3.5) shows that the torque has strong dependence on the rotor resistance and slip
and the Thevenin voltage. Changes in these variables will affect the developed torque. Slip is
28
mechanical load dependent. The next step would be to find the sensitivity of the torque
toward a change in the remaining electrical parameters
th
V and
r
r , the two variables in (3.5).
The Thevenin voltage
th
V depends on the motor parameters
s
r ,
ls
X ,
M
X and the source
voltage
s
V as seen in (3.2). The source voltage
s
V is normally regulated by the drive and
hence can be considered as constant. Thus to measure the sensitivity dependence of
e
T on
th
V and
r
r' , the effects of variation of the motor parameters
s
r ,
r
r' ,
ls
X , and
M
X on
e
T have
to be studied. The parameters of 3HP and 500HP IMs (given in the Appendix A) from [17]
and [24] were considered here, and the steady-state torque speed curves were plotted for
parameter variations of 130% to 70%. The parameters were changed sequentially and the
results are plotted in Figure 3.2 and Figure 3.3.
29
Figure 3.2 Torque-speed characteristics for parameter variation in a 3HP IM: (a) rotor resistance;
(b) stator resistance; (c) stator leakage inductance; and (d) magnetizing inductance.
30
Figure 3.3 Torque-speed characteristics for parameter variation in a 50HP IM: (a) rotor
resistance; (b) stator resistance; (c) stator leakage inductance; and (d) magnetizing
inductance.
31
In Figure 3.2 and Figure 3.3, it is seen that in the motoring region, rotor resistance variation
had the most noticeable effect with the operating torque increasing with the decrease in the
rotor resistance. This is seen in Figure 3.2 (a) and Figure 3.3 (a). Similar studies were also
done on 50HP and 2250HP IMs, which also showed comparable results. Thus, it can be
deduced that rotor resistance variation has the most observable impact on the motoring
torque and hence this variation was considered for the thesis. The variations in the rotor
resistance may naturally occur due to temperature variations, replacement of motors with
previously repaired motors, use of non-identical motors, etc.
The sensitivity of the torque also depends on the slope of the torque-speed curve in the
motoring region. The IMs with smaller rotor resistance and steeper slopes have higher
sensitivity and hence are more vulnerable to large torque fluctuations under small changes in
the rotor resistances. Therefore, motors with higher rotor resistance are preferred for load
sharing applications as they have higher slip and hence much gentler slope in the motoring
region [25]. However, high-slip motors have higher copper loss and are less efficient.
Instead, machines with lower rotor resistance are generally preferred to reduce the losses, and
such IMs have steeper slopes in the motoring region.
3.3 Rotor Resistance Variation and Load Sharing Disparity
The rotor resistance of the two rigidly-coupled IMs may not be similar. It is highly unlikely
that even the motors coming from the same manufacturer will have exactly the same rotor
resistances. Moreover, the rotor resistance changes with loading and temperature. The
equivalent rotor resistance also changes with the frequency of the rotor currents and slip due
32
to the deep-rotor-bar effect, which depends on the rotor design [23], [26].
The diagram of Figure 2.2 is simplified here and depicted in Figure 3.4. The ramp, PI
controller and the speed limit blocks of Figure 2.2 is masked under the Speed-Control-
Regulator block. The VSI block is the actual hardware of the drive consisting of the
switching devices which was shown as qd to abc transformation block in Figure 2.2.
V/F Control Speed Command IM
Speed
Control
Regulator
VSI
Speed feedback
Figure 3.4 Block diagram of Volt / Hertz control scheme.
A conventional scheme of multi-motor speed referencing is shown in Figure 3.5. Two motors
are rigidly coupled to share the common mechanical load. The input to the V/F control block
is the PI controlled difference between the speed feedback and the speed command. In rigidly
coupled multi-motor applications, all motors are coupled to each other and hence have a
common speed feedback. The PI corrected speed commands for both the VFDs are the same
and similar voltages and frequencies are injected into the coupled IMs. Thus, the similarity in
the torque developed by the two motors will depend on the rotor resistance of each IM.
33
LOAD
V/F Control Speed Command
V/F Control
IM1
IM2
Speed feedback
Speed
Control
Regulator
Speed
Control
Regulator
VSI
VSI
Figure 3.5 Block diagram for load sharing between two V/F controlled induction motors under
conventional speed referencing.
In the following discussion, two IMs rated 1HP each (as per Appendix A) are simulated to
share a mechanical load of m N. 1 . 8 at
sec
188
rad
. The V/F drive and IM are modeled using
(2.1) to (2.9). The inertia of the load is considered to be
2
. 02 . 0 m kg . The machines (without
rotor resistance variation) are both rated to generate a full load torque of m N. 05 . 4 . Both IMs
are identical in all respects, except the rotor resistances which are set to O 06 . 5 and O 41 . 7
for IM1 and IM2, respectively. The resultant torque-speed characteristic of each IM is shown
in Figure 3.6. It is observed that, as predicted by (3.5), the motor with the smaller rotor
resistance, IM1, carries a higher percentage of the load than the other motor. The values of
electromagnetic torque developed by the machines are provided in Table 3-1. It can be seen
that IM1 is overloaded to 118% whereas IM2 is operating below the rated torque by 18%.
The problem may become more severe from the process point of view, because the
overloading may force one or more IMs to operate closer to the breakdown region. In this
case, the entire process may become susceptible to breakdown as there is a possibility that
the overloaded machine would be isolated either by circuit tripping or by breakdown. The
34
rest of the system might not be able to carry the extra load and leading to stoppage of the
production line.
Figure 3.6 Torque-speed characteristics of two coupled induction motors, with rotor resistance
variation, sharing a common load without compensation.
Table 3-1 Electromagnetic torque developed by IM1 and IM2 under rotor resistance variation
using the conventional V/F scheme without compensation.
IM1 ( O = 06 . 5
1 r
r ) IM2 ( O = 41 . 7
2 r
r )
( ) Nm T
e1
1
1
rated
e
T
T
( ) Nm T
e2
2
2
rated
e
T
T
4.77 118% 3.33 82%
35
3.4 Load Sharing Using Speed Reference Compensation Method Based on Rotor
Resistance
In the previous Section, it was shown that the electromagnetic torque is sensitive to the rotor
resistance, the input voltage and the input frequency. Thus, the participating torques in the
load sharing are seldom even or constant. In order to keep the torques of the coupled motors
equal under rotor resistance variation, we should change the input voltage and frequency
equivalently to compensate the system. This Section describes a methodology of changing
the input frequency (the speed reference) so as to vary the torque-speed curve and thus
achieve balanced load sharing among the coupled IMs. It is assumed that the rotor resistance
variation is known or can be estimated using either temperature-dependent look-up tables or
the online estimation methods [23]. Various methods have been proposed in the literature for
estimating the rotor resistance, but otherwise such methods are outside of the scope of this
thesis.
For two coupled IMs with different rotor resistances, (3.5) can be written as
1 1
1 1
2
1
1
2
3
r e
th
e
r
s
V
P
T
'
=
e
,
(3.6)
2 2
2 2
2
2
2
2
3
r e
th
e
r
s
V
P
T
'
=
e
.
(3.7)
Herein, the suffix 1 and 2 correspond to IM1 and IM2, respectively. From (3.6) and (3.7) we
can see that when
2 1 r r
r r = , the operating torques of these IMs are not the same. For these
IMs to have the same operating torque, the right hand side of (3.6) and (3.7) should be equal.
36
2
2
2
2
2
1
1
1
2
1
2
3
2
3
r e
th
r e
th
r
s V P
r
s V P
'
=
' e e
. (3.8)
Solving (2.8), (3.2) and (3.8) for the electrical frequency of the second drive,
2 e
e , we get:
( )
r
S
S
r
r
M
M
r e e
X
X
r
r
X
X
e e e e +
(
(
|
|
.
|
\
|
|
|
.
|
\
|
'
'
|
|
.
|
\
|
=
1
2
1
2
2
1
1 2
,
(3.9)
where
r
e
is the rotor electrical frequency, and
ls M S
X X X + = .
If (3.9) is satisfied then both the IMs will generate the same electromechanical torque and
equally share the load. Then, (3.9) is used to generate the corrected reference speed for the
second drive in the proposed load sharing scheme. The new scheme is shown in Figure 3.7
where the Speed reference compensation block is formed using (3.9). The second drive is
now operating without the speed control regulator block and the actual speed feedback is
taken to the Speed reference compensation block. The speed reference to the second drive
is varied as per the change in the rotor resistance. Employing the proposed approach, the
improved load sharing is shown in Figure 3.8. Notice that the maximum torque is now
different indicating that different voltages and frequencies are injected into the machines. It is
also observed that the machines are running with different synchronous speeds unlike the
case in Figure 3.6. Most importantly, the torque-speed characteristics of the machines now
intersect near the commanded speed which results in almost equal values for torque.
37
Speed Command
IM1
IM2
Speed feedback
Speed feedback
Speed
Control
Regulator
Speed Reference
compensation block
V/F Control
V/F Control
VSI
VSI
LOAD
Figure 3.7 Block diagram of the proposed scheme for load sharing between two V/F controlled
induction motors.
Figure 3.8 Torque-speed characteristics of two coupled induction motors, with rotor resistance
variation, sharing a common load with compensation.
38
Comparing the torque values in Table 3-2 with those of Table 3-1, it is seen that, using the
proposed scheme, the load sharing between the machines has become more symmetrical and
the overloading on IM1 is removed. Each of the motors now shares approximately 50% of
the load. The proposed scheme can be readily extended to multiple motors (more than two).
It can be realized in practice, by implementing the Speed reference compensation block in
a PLC or similar logical device (assuming that the equivalent rotor resistance may be
appropriately estimated online). The first drive is speed referenced as per the required speed
and the second drive is speed referenced by the Speed reference compensation block.
Table 3-2 Electromagnetic torque developed by IM1 and IM2 under rotor resistance variation
using the proposed load sharing compensation scheme.
IM1 ( O = ' 06 . 5
1 r
r ) IM2 ( O = ' 41 . 7
2 r
r )
( ) Nm T
e1
1
1
rated
e
T
T
( ) Nm T
e2
2
2
rated
e
T
T
4.1 101% 4 99%
It is well understood that (3.5) holds true only when the slip is small and the machines
operate near the synchronous speed. As load and slip increases, the torque-speed curve
gradually loses its linearity. It is then expected that the performance of the proposed load
sharing scheme would be very good under lighter loads and will start to deteriorate under
heavily loads. In order to investigate the above mentioned characteristic, the IMs considered
in this paper are subjected to 25%, 50% and 100% loading. The torque developed by the
machines using the conventional and proposed V/F schemes has been superimposed in
Figure 3.9.
39
As seen in this figure, using the proposed load sharing scheme, the loading of the machines is
essentially equal at 25% load and has a difference of 2% at full load. However, using the
conventional/uncompensated method, the load is never shared equally and the difference
ranges from 10% at 25% load to 36% at full load. The results are summarized Table 3-3 and
Table 3-4. As can be seen in these tables, the proposed compensation scheme significantly
improves the load balancing between the considered IMs under light and heavy loads.
Figure 3.9 Load sharing of two coupled IMs with rotor resistance variation under different
loading conditions.
40
Table 3-3 Individual motor torques under different loading conditions for conventional scheme
without load sharing compensation scheme.
Loading IM1 ( O = ' 06 . 5
1 r
r ) IM2 ( O = ' 41 . 7
2 r
r )
( ) Nm T
e1
1
1
rated
e
T
T
( ) Nm T
e2
2
2
rated
e
T
T
25% 1.21 30% 0.82 20%
50% 2.40 59% 1.65 41%
100% 4.77 118% 3.33 82%
Table 3-4 Individual motor torques under different loading conditions for proposed load sharing
compensation scheme.
Loading IM1 ( O = ' 06 . 5
1 r
r ) IM2 ( O = ' 41 . 7
2 r
r )
( ) Nm T
e1
1
1
rated
e
T
T
( ) Nm T
e2
2
2
rated
e
T
T
25% 1.01 25% 1.01 25%
50% 2.03 50% 2.02 50%
100% 4.10 101% 4.00 99%
3.5 Load Sharing Using Speed Reference Compensation Method Based on Current
Feedback
The scheme described above provides very good accuracy in torque sharing but requires
rotor resistance estimation using various methods [23], which would add to the complexity
and cost of the VFDs. A more practical and simpler approach may be required to share the
torque under the rotor resistance variations. A simple but practical approach may be derived
based equating the real component of the phase current (which sometimes is referred to as
the torque current). The torque currents of the individual motors can be taken as a feedback
and the signal can be used to alter the speed referencing of the drives. As the generated
torque is approximately proportional to the torque current (real component of the phase
41
current), adjusting the torque current can control the generated torque of the IMs. A block
diagram depicting this approach is shown in Figure 3.10.
IM2 current
IM1 current
LOAD
V/F Control Speed Command
V/F Control
IM1
IM2
Speed feedback
Speed
Control
Regulator
Speed
Control
Regulator
VSI
VSI
Speed feedback
PID
Torque
current
estimator
Torque
current
estimator
Figure 3.10 Proposed load sharing compensation scheme under rotor resistance variation.
Under normal operation, both motors IM1 and IM2 have equal rotor resistances and take
equal torque currents. When the rotor resistance of IM1 drops from the rated O = ' 41 . 7
r
r to
O = ' 06 . 5
r
r , it starts taking higher percentage of the load. The difference in the currents
(error current) are fed back to alter the speed reference of the VFDs such that the error
current decreases. The simulation results are plotted in Figure 3.11. Also, in Figure 3.11 we
have plotted the results from the speed compensation using resistance block scheme. In the
considered study, the motors are assumed to be working with the same rotor resistance till
sec 5 = t . At sec 5 = t , the rotor resistance of IM1 decreases from O = ' 41 . 7
r
r to
O = ' 06 . 5
r
r . In practice, this may be a much slower change, but the study is focused on
steady state before and after the change. It is seen that when this happens under the
traditional method the motor torques start differing away from each other leading to
42
unbalanced loading. However, with the proposed compensation methods, the IMs remain in
good balancing throughout the study.
Figure 3.11 Mechanical load of 8.1 m N. shared between the coupled IMs: (a) without
compensation; and using proposed speed compensation based on (b) rotor resistances;
and (c) current feedback
43
3.6 Experimental Verification of the Proposed Load Sharing Scheme
3.6.1 Physical test bench set-up
To check the validity of the proposed load sharing compensation scheme, we used the
following experimental set up available in the Alpha Technology Lab, Kaiser 3075. The
overview schematic of the test bench is depicted in Figure 3.12. In the considered motor
bench with three machines shown in detail in Figure 3.13, two IMs rated 1HP and rated 5HP
were coupled on a common shaft and assumed to be driving a common mechanical load
emulated by the remaining third motor. The proposed scheme is implemented by using one
of the motors, Motor 3 (1HP), as a load and the other motors (Motor 1 (1HP) and Motor 2
(5HP)) for driving this load. As two identical motors with dissimilar rotor resistances were
not available, we used two dissimilar motors, 1HP and rated 5HP, for the experiment. These
motors are powered using VFDs which are run in V/F control. The specifications of the
VFDs used in the set-up are summarized in the Appendix B. The PLC is used to emulate the
control as explained in the previous sections. The details of the PLCs hardware and software
are listed in Appendix C. The experimental set up of the VFDs and PLC boxes are shown in
Figure 3.14, and Figure 3.15, respectively.
In the considered configuration, Motor 1 (1HP) and Motor 2 (5HP) are speed referenced with
1507.3 rpm anti-clockwise direction; while the Motor 3 (1HP) is torque referenced to run at
100% loading clockwise to emulate a common mechanical load of 4.05 m N. . The excess
energy from Motor 3 (1HP) is being dumped through the VFD 3 into an auxiliary breaking
resistor box. However, as the PLC was not programmed during the course of the experiment,
the above control (changing speed reference) was done manually using a computer
44
(Programming Station) over Ethernet/IP. A customized Rockwell Automation Software
DriveExecutive and DriveExplorer were used for online monitoring of the drive parameters.
Figure 3.12 Physical set-up for verification of the proposed scheme.
45
Figure 3.13 Load sharing motor test bench used for the practical observations.
Motor #1
Motor #2
Motor #3
46
Figure 3.14 VFDs used for powering the load sharing motors.
47
Figure 3.15 PLC used to control the drives running the load sharing motors.
48
3.6.2 Experimental procedure of load sharing and results from VFD interface
The current taken in by an IM comprises of a reactive component and a real component. The
reactive component of the current is due to the leakage flux and magnetizing current,
whereas the real component of the current is spend on the losses in the resistances and the
rotor current which produces the torque (torque current). The real current is in-phase with the
supplied voltage as depicted in Figure 3.16. In the following simplified analysis, it is
assumed that the IM torque is generated by the real component of the current. Hence, to share
the load, the Motor 1 (1HP) and Motor 2 (5HP) should carry the same torque current.
Figure 3.16 Torque and flux component of the motor current.
To emulate the needed conditions, the experimental set-up is run twice: once with identical
speed reference for both the drives without the proposed load sharing scheme; and second
time with the proposed improved load sharing scheme. To implement the proposed scheme,
the speed reference of Motor 2 (5HP) is decreased to a point where the torque currents of
both motors are similar. The torque currents are monitored in the DriveExplorer software
(see Figure 3.17 and Figure 3.18). Based on results shown in Figure 3.17, we can see that
initially both the VFDs were referenced with the same speed, and the individual load torque
current for the Motor 1 (1HP) and Motor 2 (5HP) are unequal with second motor taking more
load. The results of this study are summarized in Table 3-5, which shows that torque current
49
of Motor 2 (5HP) is 0.9A, whereas the torque current of Motor 1 (1HP) is around 0.2 A. This
happens because in the motoring region of the torque speed curve at a shaft speed, the Motor
2 (5HP) has higher electromagnetic torque compared to the Motor 1 (1HP).
In the second study, with the speed reference compensation, these motors run with different
operating frequency. The frequency change brings about a change in the torque-speed curve
of the IMs. It is seen from Figure 3.18 that the load is now better shared between Motor 1
(1HP) and Motor 2 (5HP) and the torque currents are equalized. The motors now take 0.6 A
and 0.7 A of the real currents, respectively. Please note that the total phase current of each
motor will be different even with the proposed scheme due to the dissimilar motors that
require different flux current. Another interesting observation is that initially the power-
factor of Motor 1 and Motor 2 were 0.16 and 0.29, respectively. With the proposed load
sharing, the power-factor of Motor 1 (1HP) has improved to 0.49 due to the increase in the
torque current of the motor, and Motor 2 (5HP) power-factor has dropped to 0.21 doe to
reduction in load.
Table 3-5 Torque current read from the DriveExplorer software interface: (a) without
compensation; and (b) with proposed compensation for load sharing.
Motor 1 (1HP) Motor 2 (5HP)
Speed
Reference
( rpm)
Torque
Current
(A)
Speed
Reference
( rpm)
Torque
Current
(A)
Shaft Speed
( rpm)
Without
compensation
1507.3 0.2 1507.3 0.9 1490
With
compensation
1507.3 0.6 1478.0 0.7 1460
50
Figure 3.17 Screenshots of the DriveExplorer software interface for the VFD 1 and VFD 2 without
load sharing compensation.
51
Figure 3.18 Screenshots of the DriveExplorer software interface for the VFD 1 and VFD 2 with
load sharing compensation.
52
3.6.3 Steady state measurements and calculations
The individual motor currents were measured using Yokogawa WT1600 (specification is
summarized in the Appendix D). The measured phase currents were captured, saved, and are
plotted in Figure 3.19 for the steady state conditions without and with the proposed
compensation method. As we can see in Figure 3.19, the currents in the Motor 1 (1HP)
increases and the currents in the Motor 2 (5HP) decreases when the proposed speed reference
compensation is enabled to improve the load sharing. However, in the given experimental
set-up, the currents should not be equal to each other as the machines and their magnetizing
currents are different.
Figure 3.19 Measured line currents for Motor 1 (1HP) and Motor 2 (5HP): (a) without
compensation; and (b) with proposed load sharing compensation.
53
To further validate the results, we have calculated the real component of the measured
currents. This is done by calculating the phase difference between the respective phase
voltages and phase currents and then calculating that component of the current which is in
phase with the phase voltage. The results are summarized in Table 3-6. These results show
that the balance between the torque currents and subsequently the load sharing has improved.
Table 3-6 Torque current calculated from the measured data: (a) without compensation; and (b)
with proposed compensation.
Motor 1 (1HP) Motor 2 (5HP)
Speed
reference
( rpm)
Torque
current (A)
Speed
reference
( rpm)
Torque
current (A)
Shaft Speed
( rpm)
Without
compensation
1507.3 0.3 1507.3 1.4 1490
With
compensation
1507.3 0.7 1478.0 1.1 1460
The corresponding voltage and current phasors are also plotted for Motor 1 (1HP) and Motor
2 (5HP) with and without the proposed compensation. The plots are shown in Figure 3.20
and Figure 3.21, respectively. As the torque sharing improves, the load on the previously
lightly loaded Motor 1 (1HP) increases. This increases the real current of the machine,
improving the power factor of the machine. This can be seen in Figure 3.20, where the angle
between the voltage and current phasor corresponding to a particular phase has decreased
from
1
u to
2
u . Thus, the proposed scheme, if used wisely, can improve the sharing and also
improve the power factor.
54
Figure 3.20 Phasor diagram of voltage and current for Motor 1 (1HP) with and without the
proposed load sharing compensation.
Figure 3.21 Phasor diagram of voltage and current for Motor 2 (5HP) with and without the
proposed load sharing compensation.
55
3.6.4 Verification of model using simulation of the physical set-up
In this Section the previously developed model of the multi-motor drive system with verified
against the physical set-up for two dissimilar motors. A load of 4.05 Nm equivalent to the
load used in the physical set-up is used to load Motor 1 (1HP) and Motor 2 (5HP). The motor
parameters are exactly equal to the ones used in the actual set-up. The measured and
simulated voltage and current waveforms are shown in Figure 3.22 through Figure 3.25. It is
seen that the simulated results match the measured data very well.
Figure 3.22 Measured and simulated line voltages and currents for Motor 1 (1HP) without load
sharing compensation.
56
Figure 3.23 Measured and simulated line voltages and currents for Motor 2 (5HP) without load
sharing compensation.
57
Figure 3.24 Measured and simulated line voltages and currents for Motor 1 (5HP) with load sharing
compensation.
58
Figure 3.25 Measured and simulated line voltages and currents for Motor 2 (5HP) with load sharing
compensation.
The simulation results are shown in Figure 3.26 and the corresponding values of the torques
and currents are summarized in Table 3-7. Initially, the motors are assumed to run from the
VFDs without the proposed compensation. Both the motors VFDs are speed referenced with
1507.3 rpm, similar to the physical set-up. It is seen that Motor 2 (5HP) takes most of the
load and gets 3.5 m N. ; whereas the Motor 1 (1HP) gets loaded to 0.5 m N. . The torque
59
current of the motors are calculated using the Fourier transform of the current signal and then
calculating the component of the current that is in phase with the voltage. These values are
close to the ones observed from the test bench tabulated in Table 3-6. At sec 2 = t the
proposed load sharing scheme is switched on. It is seen that the frequency input for Motor 2
(5HP) decreases until the load torques are matched. The load sharing compensator changes
the speed reference of the Motor 2 (5HP) such that the torque speed curve at the motoring
region is changed and the motors start sharing the load between them equally. It is seen in
Figure 3.26 that with enabled proposed control, the revised speed reference for the second
motor has changed to 1484.3rpm and the torque balancing is improved significantly. The
load on the Motor 2 (5HP) reduces whereas the 1HP motor load increases to achieve the
desired balance. The revised speed reference is not exactly equal to the ones in the set-up
because in the physical model the speed is manually changed using the DriveExecutive
software where the speed change is restricted to step size of around 15 rpm.
The motor real power mostly comprise of the resistive losses and the electromagnetic torque.
Thus the real power reflects the torque loading of the machines. It is seen that the real power
of the machines are different without compensation as seen in Figure 3.26. But when the
compensator is switched on, the set-up starts to share the load. The real power of the Motor 2
(5HP) decreases as Motor 1 (1HP) becomes equally loaded.
60
Table 3-7 Torque current calculated from the simulated results with and without proposed load
sharing compensation.
Motor 1 (1HP) Motor 2 (5HP)
Speed
Reference
( rpm)
Torque
Current (A)
Speed
Reference
( rpm)
Torque
Current (A)
Shaft
Speed
( rpm)
Without
compensation
1507.3 0.3 1507.3 1.2 1500
With
compensation
1507.3 0.7 1484.3 0.7 1480
Figure 3.26 Simulation results for the load sharing without compensation and with the proposed
compensation.
61
Based on the simulation studies and the experimental results, the proposed method is seen to
not only improve the load sharing between similar motors but also dissimilar motors. The
scheme also improves the power-factor of the system which is a considerable contribution.
The proposed scheme should be wisely used when implementing for dissimilar motors
keeping in mind the loading capacity of the individual motors. One way to generalize the
proposed scheme is to share the load in relative proportion to the capacity of each
participating IMs.
62
4 Load Sharing under Wheel Slippage in Vehicular Application
4.1 Motivation
It was seen in Chapter 3 that load sharing is a function of individual motor parameters, the
load distribution among the motors, and the commanded speed of each motor. In this Chapter
we extend the concept of load sharing control under the change in the load distribution that
may occur in multi-motor systems such as vehicular platforms, rail cranes, etc. due to
external factors such as wheel slippage.
IMs driven by basic Volts/Hertz controller are commonly used for driving the wheels of
gantry cranes such as the one shown in Figure 1.1(d) and many other industrial applications.
The torque developed by the propulsion motors generates the tractive force that moves the
vehicular or a crane platform. If the contact between the surface or rail and the driving wheel
is not the same among all participating wheels, then the adhesion and the tractive force
generated by each wheel will be different leading to unequal loading of the driving IMs. In
addition to the degradation of the vehicular/platform/crane performance, the result is
undesirable leading to some motors being under loaded and others overloaded. In this
Chapter, a method is proposed for sharing the torques equally between the wheels of a gantry
crane, driven by IMs with basic VFDs operating in V/F mode. The proposed control strategy
is explained for two coupled motors but may be readily extended to a number of coupled
motors and other similar applications where the coupling between the motors is not rigid and
the torque passed to each IM may vary by the virtue of adhesion.
63
4.2 Concept of Torque Transfer and Wheel Slippage
The torque generated by the IM needs a medium to get transferred from the machine to the
load. The medium can be in the form of [27]:
1. Rigid mechanical coupling between the load and the motor,
2. Resilient connection or viscous damping coupling, where the interconnection is either
by long shafts, chains or belts, or by the material being processed where twisting and
elongation becomes significant, and
3. Friction-based tractive coupling in which changing mechanical surfaces influence the
values of the tractive parameters and forces.
The problem discussed in Chapter 3 considered a rigid coupling between the load and the IM.
In such coupling, the loss of mechanical torque is minimal as there exists a direct coupling.
However, some application makes use of resilient or friction coupling in which the medium
of transfer is a contact surface between the machine and the material. Rolling Mill
(Steel/Paper/etc.), tube mill, gantry crane wheel motion, roller table application etc. are some
of the applications that make use of such coupling. The torque generated by the IM is
transferred to the material or the ground/rail by the virtue of contact between the two
surfaces. Hence, to have efficient torque transfer, the surfaces should not be slippery and
remain always be in contact with each other.
Without loss of generality, in this thesis a gantry crane operation shown in Figure 4.1 is
considered. The gantry crane runs on steel rails and typically has multiple wheels driven by
IMs. For effective motion of the gantry crane, the torque generated by the wheels should be
64
transferred to the rail to generate the propulsion force. For this to happen there should be
active bonding between the wheels and the rail. Such a bonding between the two surfaces is
known as adhesion. If the bonding breaks, then the torque does not get transferred and the
crane may stop moving in the desired direction with the wheels spinning rapidly (wheel
slippage). The bonding can break when the crane is accelerating from standstill or is trying to
stop rapidly, for example. The bonding may also break because of wear-out of the wheels
and the rails or some external contamination like water or oil spillage, etc.
Induction
Motor
Steel Wheel
Steel Rail
Trolley hook for
lifting loads
Figure 4.1 Gantry crane running on a set of rails with IM driven wheels.
4.3 Adhesion, Tractive Force, and Vehicular Motion
The total adhesion coefficient can be considered as a bonding between the wheel and the
rail. The bonding is dependent on two factors, namely on the surface adhesion factor
a
,
and the speed correction factor
c
. Both factors changes the total adhesion coefficient as
follows
65
c a
- = , (4.1)
where
a
is the surface adhesion factor, and
c
is the speed correction factor.
During the normal operation, the gantry (bridge) crane speed is very close but never equal to
the linear speed of the wheel. The difference in the speeds is known as the slip-speed , and
is given by
t w
V V = , (4.2)
where
w
V is the linear circumferential wheel velocity and given by
w w w
r V e = ; and
t
V
is the
vehicular velocity; and
w
r is the radius of the wheel.
The slip-speed , along with other environmental factors [28], determines the surface
adhesion factor
a
, between the wheel and the surface. The surface adhesion coefficient
determines the amount of tractive force that can be transferred from the wheel onto the rail
without sliding [29]. Higher coefficient of surface adhesion factor means better transfer with
less energy loss in sliding. Surface adhesion factor points out to the phenomena of bonding
between the surfaces in contact. This is a surface phenomenon and is highly dependent on the
nature of the surface; such as its roughness, contamination, hardness, chemical and physical
composition, etc. The relationship between the surface adhesion factor and the slip-speed is
not theoretical and hence there is no standard equation encapsulating this relationship.
Significant research has been done to generalize this relationship and today many empirical
formulas exist to show this relationship [30], [31]. This relationship is defined for different
surfaces in contact and hence there exists more than one equation formulizing this
relationship. A sample graph for tire-road adhesion on dry and wet road condition is shown
66
in Figure 4.2. The pattern can be divided into two regions:
1. Stable region (where the adhesion increases with increase in slip also known as Creep
Region)
2. Unstable region (where the adhesion decreases with increase in slip know as
Slip/Spin Region)
Dry Condition
Wet Condition - 1
Wet Condition - 2
Stable Unstable
Slip-speed (m/s)
S
u
r
f
a
c
e
a
d
h
e
s
i
o
n
f
a
c
t
o
r
a
Figure 4.2 Graph depicting general relationship between surface adhesion factor
a
and slip-
speed for different road conditions.
It was also noticed in series of conducted experiments that the total adhesion coefficient
decreased with increase in the material speed / vehicle velocity [31]. This is because as the
vehicular speed is increased the dwell time is decreased, decreasing the molecular level
interlocks and hence decreasing the adhesion energy required to break the interlock. This
factor is known as the speed correction factor
c
. Similar to the relationship between surface
adhesion factor and slip speed, the relationship between the adhesion coefficient and the
vehicle speed could only be stated empirically. A sample pattern is as shown in Figure 4.3.
This relationship was seen prominent under high velocities.
67
Vehicle speed - V
t
(m/s)
S
p
e
e
d
c
o
r
r
e
c
t
i
o
n
f
a
c
t
o
r
c
Figure 4.3 Graph depicting general relationship between speed correction factor
c
and vehicle
speed.
The total adhesion coefficient defined in (4.1) is responsible for the generation of traction
force
N F
T
= , (4.3)
where N is the normal force due to mass on the driving wheel.
For the purpose of further discussion, a wheel with mass M is considered to be rolling on a
flat surface as depicted in Figure 4.4. The wheel is rolling with a linear speed of
t
v and a
linear circumferential wheel velocity given by
w w w
r V e = , under a tractive effort of
T
F given
by (4.3). This tractive effort produces vehicular motion against the rolling resistance
Rolling
R .
This motion can be described by the following differential equation
Rolling
t
T
R
dt
dv
M F + = . (4.4)
68
If the rail surface were on a slope with respect to the ground, (4.4) would need to include the
gravitational component, ( ) u sin Mg , which is neglected for our studies as we consider the
normal operation where the gantry crane is running on a perfect horizontal rail.
The rolling resistance
Rolling
R consists of different resistive forces. These forces can be in
form of micro-slip (adhesion component), surface deformations (hysteresis component) due
to the compressive force of the body [32], [33], [34] and also due to the aerodynamics [35],
[36]. The rolling resistance,
Rolling
R in (4.4), is thus a combination of different resistive forces.
air n deformatio adhesion Rolling
R R R R + + = . (4.5)
Figure 4.4 Diagram showing different speeds and forces acting on the wheel.
Adhesion is important in transfer of generated torque to the rail to overcome the surface
friction. Adhesion and friction are related and it is helpful to have an overview of the relation
between the Adhesion and Friction. Many papers, e.g., [37] and [38], have explained the
relationship between adhesion and friction in greater depths. To put it in simpler term,
friction can be defined as an undesirable opposition to motion of the body while adhesion can
69
be defined as a force of attraction between the two surfaces [29],[39], [40].
Although (4.4) suggests that increase in tractive effort will result in higher linear velocity of
the vehicle, there is an upper limit beyond which the wheel will start spinning and the vehicle
will lose its linear velocity. Spin is a phenomenon where the vehicle wheel accelerates
more than the vehicle; whereas Slip is a phenomenon where the vehicle wheel decelerates
more than the vehicle [41]. When a wheel moves on a surface, it undergoes micro-slip.
Micro-slip, as the name suggests, is a phenomena where the contact surface of the wheel
undergoes random slip and stick. The distribution of this slip and stick instances depend on
the slip-speed . A sample slip-stick phenomenon is plotted against the slip-speed in Figure
4.5. When the tractive effort is rapidly increased, the vehicle speed cannot keep up with the
wheel speed due to inertia and the slip-speed increases. If the difference is too high, then the
slip region in the slip-stick region increases and the vehicle starts slipping [39], [42].
S
T
I
C
K SLIP STICK SLIP SLIP
Low Slip Speed
Moderate Slip
Speed
Sliding Motion
Rolling Direction Rolling Direction Rolling Direction
Figure 4.5 Slip-stick phenomenon at the contact surface of wheel and rail.
70
4.4 Induction Motor Load in a Gantry Crane Application
The angular motion of the wheel is generated by the wheel that is driven by IM. In this
Section, we discuss the role of an IM in the vehicular motion. The top view of the general
case of gantry crane of Figure 4.1 is reproduced in Figure 4.6 for better clarity. For the
purpose of this thesis, the gantry crane is assumed to have four wheels, out of which two
wheels are driving wheels (Wheel 1 and Wheel 2) coupled to the two separate IMs (IM 1 and
IM 2), and the remaining two are passive driven wheels (Wheel 3 and Wheel 4). The total
vehicle mass M is considered to be distributed equally amongst the four wheels, with each
wheel carrying an equivalent mass of
4
M
. Let
1
M , and
2
M be the mass acting on the
Wheel 1 and Wheel 2. The vehicle is considered to be moving linearly on a perfect plane
surface with a linear speed of
t
V . The wheels have linear speed of
1 w
V (Wheel 1) and
2 w
V
(Wheel 2) with a total adhesion of
1
and
2
for Wheel 1 and Wheel 2, respectively. Each
wheel contributes a tractive force
1 T
F (Wheel 1) and
2 T
F (Wheel 2), where each force is
g M F
T 1 1 1
= , (4.6)
g M F
T 2 2 2
= .
(4.7)
It should be noted that only the driving wheels generates the tractive force. This means that
for the gantry crane in the Figure 4.6, the tractive effort of the total vehicle will be
=
= + = + =
2
1
2 2 1 1 2 1
i
i i T T total T
g M g M g M F F F . (4.8)
71
Driving
Wheel 1
Driving
Wheel 2
Driven
Wheel 3
Driven
Wheel 4
V
w1
V
w2
V
t
F
T1
F
T2
F
T total
I
M
1
I
M
2
Figure 4.6 Top view of the gantry crane showing the different speeds and tractive forces.
The tractive force
T
F acts as load torque given by
w T load
r F T = . Thus the individual traction
loads will be given by the following
w T load
r F T
1 1
= , (4.9)
w T load
r F T
2 2
= .
(4.10)
The traction loads along with the individual frictional torque
1 fric
T and
2 fric
T , and the inertia
1
J and
2
J act as a load on the connected IMs. The torque of IM 1 and IM 2 can be given by
dt
d
J T T T
rm
fric load e
1
1 1 1
e
+ + = ,
(4.11)
dt
d
J T T T
rm
fric load e
2
2 2 2
e
+ + = ,
(4.12)
where
w
r
is the radius of the driving wheel; and
1 rm
e , and
2 rm
e are the rotational speeds of
the IM 1 and IM 2. The motor torques
1 e
T and
2 e
T produces the angular motion of the wheels
which helps in generating the micro-slip and further generates the tractive effort as explained
72
in Section 4.3.
Assuming equal mass distribution and an ideal condition where the adhesion under each
wheel is the same and neglecting the frictional load, the load on each IM will be equal ((4.11)
and (4.12)). However, in reality it is highly unlikely that the adhesion will be the same for
both the wheels because of its dependency on various variables. A difference in surface
adhesion factor under the wheels (oil/water spillage, snow, surface deformity and abrasion)
during the process is also inevitable and will result in a change in tractive effort and thus a
change in the loading of the IMs ((4.9) and (4.10)). The loading of the IMs is hence highly
susceptible and depends a great deal on the external environmental conditions. The
conventional scheme of powering the IMs for the gantry crane application is shown in Figure
4.7. It is seen that similar to Figure 3.5, the speed referencing of both VFDs is the same and
hence the input voltage and frequency to both the IMs is the same. Under the conventional
scheme, it is impossible to achieve equal load sharing between the wheel motors. In most
cases, slip occurs temporarily because of occasional oil slippage or occurrence of
water/snow. Though temporarily, such random occurrence of slip can overload the respective
IM high enough for circuit tripping and stoppage of the process.
Figure 4.7 Block diagram for conventional load sharing between two V/F controlled induction
motors under resilient or stochastic coupling.
WHEEL-1 V/F Control Speed Command
V/F Control
IM1
IM2
Speed feedback
Speed
Control
Regulator
Speed
Control
Regulator
VSI
VSI WHEEL-2
RAIL-1
RAIL-2
73
4.5 System Model and Performance under Wheel Slippage
A slow moving gantry crane is considered for the study in this thesis. The gantry crane is
considered to be moving with a speed of 0.5
s
m
or 30
min
m
. The simulation parameters
are listed in Table 4-1. The model used for the simulation is shown in Figure 4.9 and is
simulated in Matlab/Simulink [12], [13]. In the model, the Coriolis-effect of the gantry
crane hook and the suspended body, the thermal dependence of adhesion energy and
frictional parameters are neglected. For steel wheel on steel rail, the rolling friction is due to
the plastic deformation of the contact surfaces and is scarcely affected by lubricants [33].
Hence, the rolling friction is considered constant. Assuming that the gantry crane is moving
on planar surface, we can also neglect the effect of gravity. The rolling resistance thus is
considered constant at 0.0025/ m N. on each wheel.
Table 4-1 Sample parameters of the gantry crane used in the simulation.
Simulation Parameters
Mass (kg) M
16,000 kg
Moment of inertia J
0.1
2
m kg
Wheel radius / gear ratio
G
r
w
0.05 m
Linear speed of the crane
t
V
0.5
s
m
Total number of wheels 4
Number of driving motors (Specifications
are given in Appendix A)
2 x 5HP
74
The surface adhesion factor for the wheel-rail system considered here is modeled as [43]
b a
a
de ce
= , (4.13)
where a , b , c , and d are the parameters defining the surface adhesion factor depending
on the surface condition. For the same considered wheel-rail system, the speed correction
factor is modeled as [31]
v
c
8 100
8
24 . 0
+
+ = , (4.14)
where v is the vehicle linear speed in
h
km
.
The typical parameters needed for (4.13) to represent the dry and slippery conditions [43] are
listed in Table 4-2, and the resultant adhesion is plotted against the slip speed in Figure 4.8.
Table 4-2 Surface adhesion factor parameters for steel wheel on steel rail in dry and slippery
conditions.
Rail Condition a b c d
Dry 0.54 1.2 0.29 0.29
Slippery 0.54 1.2 1.0 1.0
75
Figure 4.8 Surface adhesion factor vs. slip speed for parameters listed in Table 4-2.
76
Figure 4.9 Block diagram of the simulation model for the wheel and vehicle dynamics of the gantry
crane.
77
4.5.1 System performance without load sharing
To study the system performance in this Subsection, it is assumed that the driving conditions
of one of the wheels, Wheel 1, is changed from dry to slippery for a period of time, and then
is changed to dry again. This situation may emulate accidental appearance of ice, oil, etc. on
one of the rails, which is passed by the crane platform. The model is run with a speed
reference of 10
sec
rad
to have a linear speed of 0.5
s
m
. At time sec 8 = t , Wheel 1
encounters slippage. This is modeled by changing the surface adhesion factor parameters as
shown in Table 4-2 from dry to slippery. This condition prevails for the next 7secs, and at
sec 15 = t , the rail condition changes back to dry. The results of such a drop in surface
adhesion factor predicted by the model are shown in Figure 4.10. Figure 4.10(a) shows that
the vehicle speed decreases when the adhesion on Wheel 1 drops. This is because the tractive
effort
total T
F momentarily falls as shown in Figure 4.10(b) b the driving force is reduced and
the vehicle speed drops as per (4.4). The corresponding surface adhesion factor change is
shown in Figure 4.10(c). The slip speed of both wheels increases identically as both the IMs
are referenced with the same speed command signal. As seen in Figure 4.10(c), the increased
slip speed also increases the surface adhesion factor on the Wheel 2. The difference in the
adhesion between the wheels gives rise to a variation in the load torque on the motors as seen
in Figure 4.10(d). Table 4-3 summarizes the difference in the load torques during this
scenario. From Table 4-3, we can see that during the course of disturbance the motor on
Wheel 2 becomes overloaded to around 150% while the load on the Wheel 1 motor drops to
around 50%, respectively.
78
Figure 4.10 Results of wheel slip under conventional V/F control without the proposed load sharing
scheme: (a) wheel and vehicle speed; (b) total tractive effort; (c) surface adhesion factor
of Wheel 1 and Wheel 2; and (d) load torque on Wheel 1-IM1 and Wheel 2-IM2.
79
Table 4-3 Load on Wheel 1-IM1 and Wheel 2-IM2 under wheel slippage without using the
proposed load sharing compensation scheme.
Adhesion
a
Load torque
e
T ( Nm)
Time Wheel 1 Wheel 2 Wheel 1-IM1 Wheel 2-IM2
sec 8 < 0.01 0.01 19.6 19.6
sec 15 sec 8 0.005 0.015 8.82 30.38
sec 15 > 0.01 0.01 19.6 19.6
Moreover, the difference in the tractive force (~431N) between the wheels will generate a
deformation torque
def
T around the center of the axle connecting the two wheels given by
( )l F F T
T T def 2 1
= ,
(4.15)
where l is the distance of the wheels from the center of the axle. The formation of
deformation torque
def
T is shown in Figure 4.11. This torque will generate a sideward stress
on the rail and the wheel, potentially causing faster wear or even blocking of the crane cart in
between the rails, which may be catastrophic for the system.
T
d
e
f
=
0
N
m
F
T1
F
T2
Axle
F
T1
=F
T2
F
T1
F
T2
F
T1
F
T2
Axle
T
d
e
f
>
0
N
m
Figure 4.11 Moment of force vector on the axle under wheel slip with conventional V/F control
without the proposed load sharing scheme.
80
4.6 Proposed Methodology for Improved Load Sharing under Wheel Slippage
In a simple V/F control technique, the VFD only has control over the speed of the IM. Thus
the reference speed is the only controllable variable that can be taken into consideration. In
this Section, a methodology for changing the speed reference of the VFDs to achieve load
sharing during wheel slippage is discussed and implemented. The considered adhesion is
plotted against the slip speed in Figure 4.12. The adhesions at the different steady state slip
speed are indicated with points a 1 and b 1 , a 2 and b 2 , and a 3 and b 3 , respectively. In
particular, for the considered points we have:
1. Point a 1 and b 1 : Steady state operating points of Wheel 1 and Wheel 2 before the
slip occurs,
2. Point a 2 and b 2 : Steady state operating points of Wheel 1 and Wheel 2 after the slip
occurred,
3. Point a 3 and b 3 : Desired steady state operating points for the Wheel 1 and Wheel 2.
It is desirable for the wheels to have the same adhesion hence the same torque after
the slip occurs.
81
Figure 4.12 Trace of adhesion at different rail conditions vs. slip speed.
When both VFDs have the same speed reference, the slip-speeds of both the wheels are
identical. From Figure 4.12, it can be seen that when this happens point a 2 and b 2 gives
different adhesion because of dissimilar characteristics of the rail under each wheel. To have
similar adhesion from different road characteristics, we need to alter the wheel speeds
differently to give slips (point a 3 and b 3 ) corresponding to the same adhesion.
From Figure 4.12, it is clear that adhesion from the two rail surfaces can be made equal only
by increasing the slip of the Wheel 1 while at the same time decreasing the slip of the Wheel
2. When this is achieved, point a 2 moves to a 3 while b 2 moves to b 3 in Figure 4.12. The
wheel slip is increased or decreased by increasing or decreasing the wheel speed. The desired
speed reference would try to reduce the difference in torque which demonstrates the idea of
proposed method for adjusting the speed reference. The resulting proposed control scheme is
82
depicted in Figure 4.13. In this approach, the speed loop alters the reference speed of each
wheel to achieve equal torques. In practice, torque may be estimated from measured currents
and voltages. The control block diagram of Figure 4.13 is added to Figure 4.9 across the
points A and B to give final proposed control scheme shown in Figure 4.14.
Figure 4.13 Proposed change in the speed referencing of the drive powering the wheels of a gantry
crane under wheel slippage.
WHEEL-1 V/F Control Speed Command
V/F Control
IM1
IM2
Speed feedback
Speed
Control
Regulator
Speed
Control
Regulator
VSI
VSI WHEEL-2
RAIL-1
RAIL-2
Speed feedback
Motor
Torque-1
Motor
Torque-2
controller
Figure 4.14 Block diagram of the proposed scheme for improved load sharing between two V/F
controlled induction motors under wheel slippage.
83
4.6.1 Computer studies demonstrating the proposed methodology
The proposed methodology is implemented using the block diagram depicted in Figure 4.14
and the conditions similar to the previous case, wherein the Wheel 1 slipping at time
sec 8 = t . The results achieved by the proposed method predicted by the simulation are
shown in Figure 4.15. From Figure 4.15(a), it is seen that the slip velocity of the wheels are
different unlike in Figure 4.10(a). The slip velocity is automatically adjusted such that the
surface adhesion factors [see Figure 4.15(c)] for both the wheels remains the same. Although
the wheels are running with different velocities due to the difference in the slip-speed and the
road conditions, the adhesion under Wheel 1 and Wheel 2 are the same. Thus, the tractive
force for both wheels are equal, which assures that the load torques on both wheels are
properly shared as shown in Figure 4.15(d). The results are also summarized in Table 4-4.
84
Figure 4.15 Results of wheel slip under the proposed load sharing scheme: (a) wheel and vehicle
speed; (b) total tractive effort; (c) surface adhesion factor of Wheel 1 and Wheel 2; and
(d) load torque on Wheel 1-IM1 and Wheel 2-IM2.
85
Table 4-4 Load on Wheel 1-IM1 and Wheel 2-IM2 under wheel slippage using the proposed load
sharing compensation scheme.
Adhesion
a
Load torque ( Nm)
Time Wheel 1 Wheel 2 Wheel 1-IM1 Wheel 2-IM2
sec 8 < 0.01 0.01 19.6 19.6
sec 15 sec 8
0.01 0.01 19.6 19.6
sec 15 > 0.01 0.01 19.6 19.6
To further modify the proposed methodology, the torque signal in Figure 4.13 can be
replaced by the real components of the phase current (the torque current). This would be
similar to the methodology described in Chapter 3. Also, since the IMs are typically of the
same base rating with the same magnetizing current, it may be possible to simply use the rms
currents of each motor and use these signals to implement a similar torque balancing
approach. It is seen that with the proposed load sharing scheme, the load balancing between
the motors has improved significantly even under the wheel slippage. The proposed scheme
is simple to implement on existing VFDs with minimal additional hardware.
86
5 Conclusion
In this thesis, two practical cases of load sharing in systems with multiple induction motor
drives where studied and their solutions were discussed in detail. The provided solutions
have the element of novelty, are simple and cost-effective to implement considering the same
existing basic equipment and V/F VFDs, which may be appealing to many industrial
applications.
5.1 Summary
Chapter 3 gave analyzes the load sharing among rigidly-coupled induction motors with
variations of internal parameters such as rotor resistance, and proposes a solution for load
improving sharing under rotor resistance variations. The rigidly-coupled induction motors are
common in industrial applications such as roller table, conveyor belts, etc. The proposed
solution was designed assuming a basic V/F control VFD which is widely used in the
industry and was intended to give a cost effective economical solution to a generic problem.
Chapter 4 extended the problem of load sharing to a practical case of moving vehicular or
crane platform with multiple wheels driven by several induction motors and corresponding
VFDs. In such common applications, oil / water spillage, occurrence of snow, etc., may cause
variations in adhesion and tractive forces. Slipping and/or spinning of wheels are a common
phenomenon in such applications leading to uneven distribution of load passed to the
individual induction motors. Under traditional V/F divers it was not possible to maintain load
balancing under such conditions. A simple solution was provided to improve the load/torque
sharing among the participating induction motors, which may be practically realized based
87
on current measurements. Although, the proposed method was developed and presented for a
gantry crane application, the proposed methodology with appropriate modifications can be
extended to other applications with similar properties, e.g., trains, trolley, vehicles, etc.
5.2 Future Research
Chapter 3 considered knowledge of the rotor resistance variation. In practice, this method
may be more difficult to realize effectively. The solution can be suitably modified to include
the online parameter identification to identify the rotor resistances and other parameter
variations and then change the speed reference accordingly. The online parameter
identification is a vast topic of research [44], [45], [46], [47]. A further research would be
required to find the most suitable and at the same time simple-to-implement approaches that
can be considered together with the existing basic VFDs. The alternative approach proposed
in Chapter 3 uses the current measurement which may be more practical for implementation
and applicable to larger number of applications with similar problem. This approach can be
further generalized to include proportional load sharing among dissimilar motors.
Chapter 4 dealt with load sharing between the motors under wheel slippage. The general
idea can be also studied from a point of view of operating the motor at the peak of the
adhesion-slip curve (Figure 4.8). This ensures the maximum use of available adhesion
resulting in maximizing the tractive effort. This property may be particular desirable for in
automotive applications for maximizing the vehicular acceleration for the given surface and
improving the driving traction and stability of the vehicles under different road conditions.
This involves online identification of the surface adhesion [48], [49], [50].
88
References
[1] Danieli. Danieli Centro Combustion. [Online]. http://www.danieli.com
[2] Eastport Trading Co., Inc. Rolling Mill Rolls & Spares. [Online].
http://www.eastporttrading.com/
[3] Danieli. Danieli Wean United. [Online]. http://www.danieli.com/
[4] ThyssenKrupp Steel USA. ThyssenKrupp Construction. [Online].
http://www.thyssenkruppsteelusa.com/
[5] Rockwell Automation, "Rockwell Automation-Load Sharing Applications for AC
Drive," DRIVES-WP001A-EN-P, June 2006.
[6] Rockwell Automation, "Allen Bradley: PowerFlex 700 AC Drives Technical Data,"
20B-TD001F-EN-P, July 2009.
[7] ABB, Direct torque control - the worlds most advanced the worlds most advanced,
2001, 3AFE58056685 REV C 6.6.2011.
[8] N. Mitrovic, V. Kostic, M. Petronijevic, and B. Jeftenic, "Multi-Motor Drives for
Crane Application," Advances in Electrical and Computer Engineering, vol. 9, no. 3,
pp. 57 - 62, October 2009.
[9] The MathWorks, Inc. (2006) SimPower Systems: Model and simulate electrical power
systems User's Guide. [Online]. www.mathworks.com
[10] Plexim GmbH. Piecewise Linear Electrical Circuit Simulation (PLECS) User
manual. [Online]. www.plexim.com
[11] Krause & Associates, Inc. (2003) Automated State Model generator (ASMG)
Reference Manual. [Online]. www.pcka.com
89
[12] The MathWorks, Inc. (2009) MATLAB - The Language of Technical Computing.
[Online]. www.mathworks.com
[13] The MathWorks, Inc. (2009) SIMULINK - Simulation and Model-Based Design.
[Online]. www.mathworks.com
[14] Cadence Design Systems. Personal Simulation Program with Integrated Circuit
Emphasis (PSpice). [Online]. www.cadence.com
[15] J. S. Carson, "Introduction to modeling and simulation," Engineering Management
Review, IEEE, vol. 37, no. 2, p. 13, 2009.
[16] A. Maria, "Introduction To Modeling And Simulation," in Proceedings of the 1997
Winter Simulation Conference, 1997, pp. 7-13.
[17] P. C. Krause, O. Wasynczuk, and S. D. Sudhoff, Analysis of electric machinery and
drive systems, 2nd ed.: John Wiley & Sons Inc., 2002.
[18] R. H. Park, "Two-Reaction Theory of Synchronous Machines-Generalized method of
analysis," AIEE Transactions, vol. 48, pp. 716-727, July 1929.
[19] H. C. Stanley, "An Analysis of the Induction Motor," AIEE Transactions, vol. 57, pp.
751-755, 1938.
[20] G. Kron, Equivalent Circuits of Electric Machinery. New York: John Wiley and
Sons, 1951.
[21] D. S. Brereton, D. G. Lewis, and C. G. Young, "Representation of Induction Motor
Loads During System Stability Studies," AIEE Transactions, vol. 76, pp. 451-461,
August 1957.
[22] P. C. Krause, and C. H. Thomas, "Simulation of Symmetrical Induction Machinery,"
90
IEEE Transactions on Power Apparatus and Systems, vol. 84, pp. 1038-1053,
November 1965.
[23] P. C. Sen, Principles of electric machines and power electronics, 2nd ed.: John Wiley
& Sons, 1997.
[24] J. J. Cathey, R. K. Cavin, and A. K. Ayoub, "Transient Load Model of an Induction
Motor," IEEE Transactions on Power Apparatus and Systems, vol. 92, pp. 1399-
1406, July/August 1973.
[25] G. E. Paulson, "Motor Selection for Belt-Conveyor Drives," in 10th CIM
Maintenance/Engineering Conference, Saskatoon, Saskatchewan, September 13-16,
1998.
[26] S.C. Foroosh, L. Wang, and J. Jatskevich, "A simple induction machine model for
predicting low frequency dynamics," in Canadian Conference on Electrical and
Computer Engineering, 4-7 May 2008, pp. 001655-001660.
[27] B. Jeftenic, M. Bebic, and S. Statkic, "Controlled multi-motor drives," in
International Symposium on Power Electronics, Electrical Drives, Automation and
Motion (SPEEDAM), Taormina (Sicily)- ITALY, 23-26 May 2006, pp. 1392-1398.
[28] S. Senini, F. Flinder, and W. Oghanna, "Dynamic simulation of wheel-rail interaction
for locomotive traction studies," in Proceedings of the 1993 IEEE/ASME Joint
Railroad Conference, Pittsburgh, PA , USA, 6-8 Apr 1993, pp. 27-34.
[29] K. P. Ramalingam, "Improving Adhesion Characteristics of electric locomotives with
DC traction Motors," Rail International, vol. 9, no. 8/9, pp. 41-48, October 1988.
[30] F. Gustafsson, "Monitoring tire-road friction using the wheel slip," Control Systems,
91
IEEE, vol. 18, no. 4, pp. 42-49, August 1998.
[31] W. Zhang, J. Chen, X. Wu, and X. Jin, "Wheel/rail adhesion and analysis by using
full scale roller rig," Wear - An International Journal on the Science and Technology
of Friction, Lubrication and Wear, vol. 253, no. 1-2, pp. 82-88, July 2002.
[32] K. Kendall, "Rolling friction and adhesion between smooth solids," Wear - An
International Journal on the Science and Technology of Friction, Lubrication and
Wear, vol. 33, no. 2, pp. 351-358, July 1975.
[33] K. R. Eldredge, and D. Tabor, "The mechanism of rolling friction. I. The plastic
range," Proceedings of the Royal Society of London-Series A, Mathematical and
Physical Sciences, vol. 229, no. 1177, pp. 181-198, April 1955.
[34] R. Szoszkiewicz, B. Bhushan, B. D. Huey, A. J. Kulik, and G. Gremaud,
"Correlations between adhesion hysteresis and friction at molecular scales," Journal
of Chemical Physics, vol. 122, no. 14, pp. 144708 (1-6), April 2005.
[35] C. D. Yang, and Y. P. Sun, "Mixed H2/H cruise controller design for high speed
train," International Journal of Control, vol. 74, no. 9, pp. 905-920, June 2001.
[36] S. C. Wang , and X. Xia, "Mathematical modeling heavy ore load train equipped with
electronically controlled pneumatic brake," in 1st African Control Conference, Cape
Town, South Africa, 3-5 December 2003, pp. 74-79.
[37] H. Yoshizawa, Y. L. Chen, and J. Israelachvilli, "Fundamental mechanisms of
interfacial friction. 1. Relation between adhesion and friction," The Journal of
Physical Chemistry, vol. 97, no. 16, pp. 41284140, April 1993.
[38] Q. Zhang, Y. Qi, L. G. Hector, T. Cagin, and W. Goddard, "Origin of static friction
92
and its relationship to adhesion at the atomic scale," Phys. Rev. B, vol. 75, no. 14, pp.
144114 (1-7), April 2007.
[39] U. Olofsson, and R. Lewis, "Tribology of the Wheel Rail Contact," in Handbook of
Railway Vehicle Dynamics, S Iwnicki, Ed.: Taylor & Francis Group, 2006, ch. 5, pp.
121-140.
[40] G. W. Stachowiak, and A. W. Batchelor, Engineering Tribology, 3rd ed.: Elsevier
Inc., 1993.
[41] C. W. Sheppard, "Combined Anti-slip and Anti-Spin Control for vehicle wheels,"
3,482,887, December 9, 1969.
[42] T. Ohyama, "Tribological studies on adhesion phenomena between wheel and rail at
high speeds," Wear-An International Journal on the Science and Technology of
Friction, Lubrication and Wear, vol. 144, no. 1-2, pp. 263-275, April 1991.
[43] Y. Ishikawa, and A. Kawamura , "Maximum adhesive force control in super high
speed train," Proceedings of the Power Conversion Conference - Nagaoka 1997.,
vol.2, no., pp.951-954 vol.2, 3-6 Aug 1997.
[44] H. A. Toliyat, E. Levi, and M. Raina, "A review of RFO induction motor parameter
estimation techniques," IEEE Transactions on Energy Conversion, vol. 18, no. 2, pp.
271-283, June 2003.
[45] M. J. Duran, J. L. Duran, F. Perez, and J. Fernandez, "Induction-motor sensorless
vector control with online parameter estimation and overcurrent protection," IEEE
Transactions on Industrial Electronics, vol. 53, no. 1, pp. 154-161, Feb 2006.
[46] B. Karanayil, M. F. Rahman, and C. Grantham, "Online Stator and Rotor Resistance
93
Estimation Scheme Using Artificial Neural Networks for Vector Controlled Speed
Sensorless Induction Motor Drive," IEEE Transactions on Industrial Electronics,
vol. 54, no. 1, pp. 167-176, Feb 2007.
[47] C. Moons, and B. De Moor, "Parameter identification of induction motor drives,"
Automatica, vol. 31, no. 8, pp. 1137-1147, August 1995.
[48] A. Kawamura, et al., "Maximum adhesion control for Shinkansen using the tractive
force tester," in IEEE 28th Annual Conference of the Industrial Electronics Society
(IECON 02), 2002, pp. 567-572.
[49] A. Kawamura, et al., "Measurement of the tractive force and the new adhesion
control by the newly developed tractive force measurement equipment ," in
Proceedings of the Power Conversion Conference, 2002. PCC Osaka 2002, Japan,
2002, pp. 876-884.
[50] Y. Matsumoto, N. Eguchi, and A. Kawamura, "Novel Re-adhesion Control for Train
Traction System of the Shinkansen with the Estimation of Wheel-to-Rail Adhesive
Force," in The 27th Annual Conference of the IEEE Industrial Electronics Society
(IECON'01), 2001, pp. 1207-1212.
94
Appendices
Appendix A
Motor Specification
A. 5HP Hyundai, 460 V, 60 Hz, 1750 rpm, 4 Pole,
Model: HNV413-CC-DBLS, Type: HJS184SR235-HNV413-CC-DBLS,
O = 503 . 1
s
r , O = 147 . 1
r
r , O = 665 . 3
ls
X , O = 786 . 4
lr
X , O = 38 . 101
m
X ,
m N T
rated
. 25 . 20 = , m N T . 75 . 60
max
= ,
2
. 105 . 0 m kg J = .
B. 1HP Baldor Reliance, 460 V, 60 Hz, 1725 rpm, 4 Pole,
Catalogue No.: CM3546, Spec No.: 34G795X269,
O = 98 . 6
s
r , O = 41 . 7
r
r , O = 84 . 11
ls
X , O = 03 . 11
lr
X , O = 23 . 207
m
X ,
m N T
rated
. 05 . 4 = , m N T . 15 . 17
max
= ,
2
. 00261 . 0 m kg J = .
C. 3HP, 220 V, 60 Hz, 1710 rpm, 4 Pole,
O = 435 . 0
s
r , O = 816 . 0
r
r , O = 754 . 0
ls
X , O = 754 . 0
lr
X , O = 13 . 26
m
X ,
2
. 089 . 0 m kg J = .
D. 3HP, 460 V, 60 Hz, 1705 rpm, 4 Pole,
O = 087 . 0
s
r , O = 228 . 0
r
r , O = 302 . 0
ls
X , O = 302 . 0
lr
X , O = 08 . 13
m
X ,
2
. 662 . 1 m kg J = .
E. 500HP, 2300 V, 60 Hz, 1773 rpm, 4 Pole,
95
O = 262 . 0
s
r , O = 187 . 0
r
r , O = 206 . 1
ls
X , O = 206 . 1
lr
X , O = 02 . 54
m
X ,
2
. 06 . 11 m kg J = .
F. 2250HP, 2300 V, 60 Hz, 1786 rpm, 4 Pole,
O = 029 . 0
s
r , O = 022 . 0
r
r , O = 226 . 0
ls
X , O = 226 . 0
lr
X , O = 04 . 13
m
X ,
2
. 87 . 63 m kg J = .
96
Appendix B
Variable Frequency Drive Specification
Motor
Rating
Drive Rating Specifications
1HP
5 HP Normal
Duty, 3 HP
Heavy Duty
PowerFlex700 AC Drive, 480 VAC, 3 PH, 8 Amps, 5 HP
Normal Duty, 3 HP Heavy Duty, IP20 / Type 1, No HIM
(Blank Plate), Brake IGBT Installed, Without Drive
Mounted Brake Resistor, Second Environment Filter per
CE EMC directive (89/336/EEC), No Communication
Module, Vector Control with 120V I/O, No Feedback
Input Voltage 480 VAC, 3 PH
Current Rating 8 Amps
Enclosure IP20 / Type 1
Frame Size Frame Size 0
Output Current Information
Output Amps: 8A Cont,
8.8A 1 Min, 12A 3 Sec
I/O Options
Vector Control with 120V
I/O
Brake IGBT Brake IGBT Installed
Filter Options
Second Environment
Filter per CE EMC
directive (89/336/EEC)
5HP
10 HP Normal
Duty, 7.5 HP
Heavy Duty
PowerFlex700 AC Drive, 480 VAC, 3 PH, 14 Amps, 10
HP Normal Duty, 7.5 HP Heavy Duty, IP20 / Type 1, No
HIM (Blank Plate), Brake IGBT Installed, Without Drive
Mounted Brake Resistor, Second Environment Filter per
CE EMC directive (89/336/EEC), No Communication
Module, Vector Control with 24V I/O, No Feedback
Input Voltage 480 VAC, 3 PH
Current Rating 14 Amps
Enclosure IP20 / Type 1
Frame Size Frame Size 1
Output Current Information
Output Amps: 14A Cont,
16.5A 1 Min, 22A 3 Sec
I/O Options
Vector Control with 24V
I/O
Brake IGBT Brake IGBT Installed
Filter Options
Second Environment
Filter per CE EMC
directive (89/336/EEC)
97
Appendix C
Software details
Name: DriveExecutive
Description: This software is an online/offline drive and adapter configuration tool that
leverages Windows Explorer-style navigation, built-in html product help, and handy
diagnostic and setup wizards. A state-of-the-art comparison tool lets you look at differences
and make two devices/files the same.
Source: Rockwell Automation
Link: http://www.ab.com/en/epub/catalogs/36265/1323285/9616672/9616694/index.html
Name: DriveExplorer
Description: DriveExplorer Software is an easy-to-use, cost effective application for
monitoring and online configuration of your PowerFlex drives and communication
adapters. It makes drive set-up easy and faster than using a Human Interface Module (HIM).
Source: Rockwell Automation
Link: http://ab.rockwellautomation.com/Drives/Software/9306-DriveExplorer
98
Appendix D
Yokogawa WT1600 Digital Power Meter
Description: The WT1600 is power meter designed for measurement of extremely small
currents in energy-saving equipment, as well as measurement of large currents for evaluating
large-sized loads. The WT1600 works with voltages ranging from 1.5 V up to 1000 V,
supporting a wide range of applications. Because it can accept signal inputs for up to six
phases, a signal WT1600 unit can measure I/O signals on inverters.
Source: Yokogawa
Link: http://tmi.yokogawa.com/discontinued-products/digital-power-analyzers/digital-power-
analyzers/wt1600-digital-power-meter/