Professional Documents
Culture Documents
Direct-Drive Applications
T. M. Jahns
Wisconsin Electric Machines and Power Electronics Consortium (WEMPEC)
University of Wisconsin - Madison
Madison, WI USA
E-mail: jahns@engr.wisc.edu
I. INTRODUCTION
Permanent magnet (PM) synchronous machines have been
adopted for a wide variety of high-performance applications
during the past 30 years since the development of modern
neodymium-iron-boron (NdFeB) magnets. These PM
machines are distinguished from competing machine types by
their ability to achieve superior combinations of high Fig. 1 Internal stator and external rotor assemblies of surface PM
torque/power density, high efficiency, and low rotor inertia machine in direct-drive washing machine [3].
characteristics. As a result, PM synchronous machines are
The engineering tradeoffs between the mass/volume/cost
used almost universally in demanding applications including
of the speed-changing component and the larger machine
electrically-powered servo machine tools and production
depend on each individual application. Fortunately, the key
hybrid-electric automobile drivetrains [1-2].
performance metrics of the PM machine are sufficiently
One of the broad application areas in which PM machines attractive that the direct-drive configuration is preferred in a
have been achieving growing success during recent years is growing number of applications. As a result, direct-drive
direct-drive (DD) machines. Direct-drive machines are distin- configurations using PM machines have become accepted, and
guished by the fact that intermediate gears, belts, or other in some cases, dominant solutions for a wide range of
types of mechanical transmission components are eliminated application areas including appliance white goods, HVAC
between the machine and the connected load. The reasons for (heating, ventilating and air conditioning), and electrified
the appeal of a direct-drive configuration depends on the transportation.
specific application, but typically involves system advantages
Figure 1 show an example of a direct-drive washing
derived from eliminating the weight, volume, cost, noise, or
machine using a surface PM machine with an external rotor
reliability risks associated with the speed-changing
configuration. In this case, the mechanical transmission that is
components.
used in conventional vertical-axis washing machines in the US
In exchange for these advantages, the PM machine is is replaced by this DD-PM configuration. It is worth noting
typically required to deliver higher torque at a lower speed, that the adoption of this direct-drive configuration makes it
resulting in a larger and heavier machine than what would possible to replace the fixed agitation and spin cycles that are
have been required if a gearbox (or equivalent) had been used.
designed into the hardware of the mechanical transmission 4) For convenience, the machine terminal voltage Vo [V]
with a highly programmable wash cycle control capability. and current Io [A] are assumed to be constant for rated
One of reasons for highlighting this application is that it is one power operation at Po as the radius r is varied
of the early commercial applications of DD-PM 5) The machine excitation frequency e [elec. rad/s] is held
configurations dating back to the 1980s when first introduced constant as the machines rated mechanical speed m
by Fisher & Paykel [4]. [mech. rad/s] is varied
But why are PM machines selected so often as the Although these assumptions are idealized approximations,
preferred machine type for these direct-drive applications they serve a very useful purpose for developing comparisons
when there are other types of ac machines that could be in this scaling exercise. One of the ramifications of holding
configured in the same way? The purpose of this paper is the machines voltage, current, and excitation frequency
two-fold: 1) to explore the underlying physical reasons why constant as the rotor speed is varied is that the same inverter
PM machines have an inherent advantage for direct-drive can be used for all of the evaluated machine design.
applications; and 2) to survey some of the key application B. Basic Machine Relationships
areas in which direct-drive PM (DD-PM) machine
Two of the basic machine relationships that result from
configurations are achieving success. It is both instructive and
these assumptions are:
fascinating to observe the wide range of factors, some rather
obvious on inspection and others far more subtle, that are Po = T eo m (1)
contributing to the progress of new DD-PM configurations. T eo = 2 r 2 l (2)
II. MACHINE DIRECT-DRIVE SCALING STUDY where Teo [Nm] is the rated torque of the machine and l [m] is
the active length of the stator core.
An inquiry into the fundamental reasons for the appeal of
PM machines for direct-drive applications compared to other Two more key relationships focused on the machine
types of machines leads to an investigation of the comparative frequency and surface speed are:
scaling characteristics of these machines to high-torque, low- = m r (3)
speed geometries. For this investigation, attention will be e
focused on developing a comparison of the differences m = (4)
p
between the scalability of PM machines and induction
machines. The induction machine was chosen for this where p is the number of pole pairs. Combining (3) and (4)
comparison for at least two reasons: 1) it is a popular choice leads to the following simple but significant relationship
for many of these application areas because of its appealing
performance vs. cost metrics; and 2) it is an excellent r= (5)
representative of the class of ac machines for which the source e p
of energy for establishing magnetic flux in the airgap is the which is important since, based on the adopted assumptions,
stator-side electrical excitation. both the rotor surface speed and the excitation frequency e
The second point turns out to be very important to the are constant, indicating that the machine radius is directly
results of this comparison. In this regard, induction machines proportional to the number of pole pairs p. Since p can only
can be contrasted with PM machines that depend on the rotor take on integral values (1, 2, ), this means that a comparison
magnets to establish the torque-producing component of is being formulated between a set of machine designs with
magnetic flux in the airgap instead of the stator current. varying discrete values of radii, torque ratings, and mechanical
speeds that are ultimately determined by (5) as the number of
A. Assumptions pole pairs is changed in integral steps.
In order to explore the relative scaling characteristics of the Next, (2) can be substitute for the rated torque Teo in (1) and
two types of machines from low-torque/high-speed to high- the mechanical rotor speed m can then be eliminated from the
torque/low-speed configurations, it is necessary to establish expression using (3), leading to
the ground rules for the investigation. They have been set as
follows: Po = 2 r l (6)
1) The mechanical power delivered to the load has a fixed which can be easily solved for the machine active length as
value Po [W] Po
l= r 1 (7)
2) The machines and speed-changing components are 2
assumed to be lossless
Similarly, m can be eliminated from (2) using (3), yielding
3) The airgap shear stress [N/m2] and rotor surface
Po
velocity [m/s] are assumed to be held constant as the T eo = r (8)
rotor radius r is varied
Since Po, , and are all assumed to be constant, (8) flux linkage . Although the full derivation will not be
indicates that the machine torque increases linearly with the provided here [5], the expression for the phase flux linkage
radius r while (7) indicates that the active machine length l can be conveniently expressed in terms of the machine
decreases inversely with increases in r as the machine torque parameters, leading to an expression for E [V]
rating is increased and the rated mechanical speed is decreased P N B
E = e = r l N B Bo e = o B o e (12)
to hold the power constant (according to (1)). 2
As a side note, it should be noted that the expression for the where the rightmost expression was developed using (7) to
machine pole pitch p [m] can be presented as eliminate the active length l from the expression which results
r in eliminating the radius r as well. This final expression is
p = = (9) independent of the radius and the number of pole pairs, and, in
p e
fact, consists entirely of constant parameters according to the
for which the rightmost expression makes use of (5) to study assumptions. This is very important because the
eliminate the radius r. Since all of the quantities in this final delivered power is proportional to the product of the back-emf
expression are assumed to be constant in this study, it is E and the stator current Io which is also constant.
interesting to note that the net effect is to hold the pole pitch
The conclusion of this analysis is that PM machines are
constant in the series of evaluated machines as p , r, Teo, and well-suited to high-torque, low-speed configurations in which
m are varied. the radius and the number of machine poles increase to high
C. PM Machine Scaling values. This analysis indicates that there is no degradation of
For simplicity, it is assumed that the PM machine uses a the airgap shear stress in the PM machine as the radius
surface magnet configuration and that both the magnet flux increases and that both E and Io are unaffected by the scaling.
density and the stator windings have ideally sinusoidal spatial The airgap stress issue deserves closer examination. It
distributions along the airgap. Furthermore, it is assumed that should be noted that the length of the airgap g [m] does not
peak amplitude of the magnet flux density in the airgap is Bo appear in any of the expressions and that the value of the
[T] and that this value is independent of the number of pole magnet flux density in the airgap is fixed at Bo. In actuality,
pairs. It is also assumed that the number of stator winding the length of the airgap must necessarily be increased as radius
turns in each phase per pole pair NB is constant and that all of is increased due to a combination of factors, including
the windings are connected in series, meaning that the total mechanical tolerances as one of the most important [6].
number of series turns in each phase winding No is Fortunately, the value of Bo is relatively insensitive to the
proportional to the number of pole pairs airgap length in PM machines [7] because the magnet flux
N o = p N B (10) density is dependent on the effective airgap length g that is
the sum of the physical airgap length g and the magnet length
Since it is assumed that each series turn is carrying the same
lm divided by the magnet relative recoil permeability rec
rated current Io, the average electric loading A, defined as the
current per unit distance along the airgap periphery [A/m], l
g = g + m (13)
will be proportional to the total number of Amp-turns divided rec
by the airgap circumference Since the magnet length is typically much larger than the
N I K N B I o e physical airgap length and rec 1, the effective airgap length
A=K o o = (11) g is relatively insensitive to g. As a result, the assumption of
2 r 2
constant Bo over a wide range of radius values, and, hence,
where K is a dimensionless constant, and the rightmost physical airgap lengths is generally well founded.
expression emerges by substituting (5) and (10) for r and No,
respectively. D. Induction Machine Scaling
The final expression in (11) indicates that the electric The same assumptions and approach are used to investigate
loading is constant and independent of the number of pole the scaling of induction machines with a fixed power rating
pairs. This is encouraging since the shear stress is the over a range of radii and pole pair numbers. All of the basic
product of the electric loading A and the magnetic loading B machine relationships derived in Section II.B still apply. In
which is directly proportional to Bo in this case. Therefore, addition, the assumption of series-connected sinusoidal stator
the analysis is leading to the conclusion that the airgap shear winding configurations with a constant number of stator turns
stress is constant, which is reassuring since it is consistent per pole pair NB as indicated in (10) still applies.
with the original study assumption. Unlike the PM machine, there is no rotor-based source of
Based on the sinusoidal flux density and winding magnetomotive force (MMF) to develop magnetic flux in the
distribution assumptions, it can be shown using Faradays airgap. A portion of the stator current that will be referred to
Law that the resulting phase back-emf amplitude E is equal to here as the magnetizing current Im is responsible for creating
the product of the excitation frequency and the total winding this magnetizing flux. Figure 2 provides a view of the
Fig. 2 Induction machine equivalent circuit identifying magnetizing
branch parameters