5715ch11 PDF

11
Transverse Flux and

Flux Reversal
Permanent Magnet
Generator Systems
11.1 Introduction ..................................................................... 11-1

11.2 The Three-Phase Transverse Flux Machine (TFM):
Magnetic Circuit Design.................................................. 11-6
The Phase Inductance Ls • Phase Resistance and Slot Area
11.3 TFM — the d–q Model and Steady State .................... 11-15
11.4 The Three-Phase Flux Reversal Permanent Magnet
Generator: Magnetic and Electric Circuit Design ....... 11-18
Preliminary Geometry for 200 Nm at 128 rpm via Conceptual
Design • FEM Analysis of Pole-PM FRM at No Load • FEM
Analysis at Steady State on Load • FEM Computation of
Inductances • Inductances and the Circuit Model of
FRM • The d–q Model of FRM • Notes on Flux Reversal
Generator (FRG) Control
11.5 Summary......................................................................... 11-42
References .................................................................................. 11-43
11.1 Introduction
There are certain applications, such as direct-driven wind generators, that have very low speeds (15 to
50 rpm) and microhydrogenerators with speeds in the range of up to 500 rpm and power up to a few
megawatts (MW) for which permanent magnet (PM) generators are strong candidates, provided the size
and the costs are reasonable.
Even for higher speed applications, but for lower power, today’s power electronics allow for acceptable
current waveforms up to 1 to 2.5 kHz fundamental frequency f1n.
Increasing the number of PM poles 2p1 in the PM generator to fulfill the standard condition
f1n = p1 ⋅ nn ; nn − speed (rps) (11.1)
thus becomes necessary.
Then, the question arises as to whether the PM synchronous generators (SGs) with distributed wind-
ings are the only solution for such applications, when the lowest pole pitch for which such windings can
be built is about τmin ≈ 30 mm, for three slots/pole. And, even at τ = 30 to 60 mm, is the slot aspect ratio
large enough to allow for a high enough electrical loading to provide for high torque density? The apparent
answer to this latter question is negative.
11-1
© 2006 by Taylor & Francis Group, LLC

11-2 Variable Speed Generators
(a) (b)
FIGURE 11.1 The single-sided transverse flux permanent magnet (PM) machine: (a) with surface PM pole rotor
and (b) with rotor PM flux concentration (interior PM poles).
The nonoverlapping coil winding concept (detailed in Chapter 10) is the first candidate that comes to
mind for pole pitches τ > 20 mm for large torque machines (hundreds of Newton meters [Nm]), but when
the pole pitch τ ≈ 10 mm, they are again limited in electrical loading per pole, though there is about one
slot per pole (Ns ≈ 2p1). In an effort to increase the torque density, the concept of a multipole span coil
winding can be used, which becomes practical, especially when the number of PM poles 2p1 > 10 to 12.
Two main breeds of PM machines were proposed for high numbers of pole applications: transverse
flux machines (TFMs) and flux reversal machines (FRMs).
The TFMs are basically single-phase configurations with single circumferential coil per phase in the
stator, embraced by U-shaped cores that create a variable reluctance structure with 2p1 poles. A 2p1 pole
surface or an interior pole PM rotor is added (Figure 11.1a and Figure 11.1b). Two or three such
configurations placed along the shaft direction would make a two-phase or three-phase machine [1].
There are many other ways to embody the TFM, but the principle is the same. For example, the concept
can be extended by placing the PM pole structure in the stator, around the circumferential stator coil, while
the rotor is a passive variable reluctance structure with axial or radial airgap (Figure 11.2a and Figure 11.2b) [2].
The PM flux concentration is performed in the stator, in Figure 11.2 configurations, but the PM flux
paths in the rotor run both axially and radially; thus, the rotor has to be made of a composite magnetic
material (magnetic powder).
Stator pole
Magnet Magnet
Stator pole
S Magnet
N
Winding
Winding
Rotor
S
N
Rotor pole
(a) (b)
FIGURE 11.2 Transverse flux machine (TFM) with stator permanent magnets (PMs): (a) with axial airgap and
(b) with radial airgap.

Transverse Flux and Flux Reversal Permanent Magnet Generator Systems 11-3
The TFMs with rotor or stator PM poles are characterized by the fact that the PM fluxes of all North
Poles add up at one time in the circumferential coil, and then, after the rotor travels one PM pole angle,
all South Poles add up their flux in the coil. Thus, the PM flux linkage in the coil reverses polarity 2p
times per rotor revolution and produces an electromagnetic field (emf) Es:
∂φ PM
t
dθ ∂φ t
Es = −W1 ⋅ p ⋅ ⋅ r ; PM ≈ Bg p1lstack ⋅ τ ⋅ sin pθr (11.2)
∂θr dt ∂θr
where
lstack is the axial length of the U-shape core leg
θr is the mechanical angle
Bg is the PM airgap flux density
τ is the pole pitch
ΦPMt is the total flux per one turn coil
From Equation 11.2,
Es ≈ −W1 ⋅ p12 ⋅ Bg τ ⋅ lstack ⋅ 2 ⋅ π ⋅ n = −W1 ⋅ Bg ⋅ τ ⋅ p1 ⋅ lstack ⋅ ω1 ⋅ sin pθr (11.2)
Equation (11.2) serves to prove that for the same coil and machine diameter (2p1τ = constant), the
number of turns, and stator core stack length, if the number p1 of U cores is increased, the emf is increased,
for a given speed. This effect may be called torque magnification [2], as torque Te per phase (coil) is as
follows:
Es (θr ) ⋅ I1(θr )
Te = ; ω1 = 2 ⋅ π ⋅ n ⋅ p1 (11.3)
phase 2 ⋅π ⋅n
The structure of the magnetic circuit of the TFM is complex, as the PM flux paths are three-dimensional
either in the stator or in the rotor or in both.
Soft composite materials may be used for the scope, as their core losses are smaller than those in silicon
laminations for frequencies above 600 Hz, but their relative magnetic permeability is below 500 µ0 at 1.0 T.
This reduces the magnetic anisotropy effect, which is so important in TFMs. This is why, so far, the
external rotor TFM with the U-shape and I-shape stator cores located in an aluminum hub is considered
the most manufacturable (Figure 11.3a and Figure 11.3b) [3]. Still, the additional eddy current losses in
the aluminum hub carrier are notable.
Though double-sided TFMs with rotor PM flux concentration were proposed (Figure 11.4a and
Figure 11.4b) to increase the torque per PM volume ratio, they prove to be difficult to manufacture.
While increasing the torque/volume and decreasing the losses per torque are key design factors, the
power factor of the machine is essential, as it defines the kilovoltampere of the converter associated with
the TFM for motoring and generating.
(a) (b)
FIGURE 11.3 The three-phase external permanent magnet (PM)-rotor transverse flux machine (TFM): (a) internal
stator and (b) external rotor.

(a) (b)
FIGURE 11.4 Double-sided transverse flux machine (TFM) (a) without and (b) with rotor permanent magnet (PM)
flux concentration.
The ideal power factor angle ϕ1 of the TFM (or of any PM synchronous machine [SM]) may be defined,
in general, as follows:
ω1 ⋅ Ls ⋅ I sn
ϕ1n = tan −1 (11.4)
Es (ω1)
where
Ls is the synchronous inductance of the machine
Is is the phase current
Equation 11.4 is valid for a nonsalient pole machine behavior (surface PM pole machines).
For the interior PM pole TFM with flux concentration and a diode rectifier, the machine is forced to
operate at about unity power factor. In such conditions, what is the importance of ϕ1n as defined in
Equation 11.4? It is to show the power factor angle of the machine for peak torque (pure Iq current
control for the nonsalient pole machine). For generality, the presence of a front-end pulse-width mod-
ulated (PWM) converter is necessary to extract the maximum power and experience the ϕ1n power factor
angle conditions. The power factor at rated current cosϕ1n is a crucial performance index.
In any case, the larger the machine inductance voltage drop per emf, the lower the power factor
goodness of the machine. From this point of view, the PM flux concentration, though it leads to larger
torque/volume, also means a higher inductance, inevitably. The claw-pole stator cores were proposed to
improve the torque/volume, but the result is still modest [3], due to low power factor. Consequently,
only at the same torque density as in the surface PM pole machine, can the TFM with PM flux concen-
tration eventually produce the same power factor at better efficiency and with a better PM usage. While
the above rationale in power factor is valid for all PM generators, the problem of manufacturability
remains heavy with TFMs.
In order to produce a more manufacturable machine, the three-phase flux reversal PM machine (FRM)
was introduced [4]. FRM stems from the single-phase flux-switch generator [5] and is basically a doubly
salient stator PM machine. The three-phase FRM uses a standard silicon laminated core with 6k large
semiclosed slots that hold 6k nonoverlapping coils for the three phases.
Within each stator coil large pole, there are 2np PM poles of alternate polarity. Each such PM pole
spans τPM. The large slot opening in the stator spans 2/3 τPM. The rotor has a passive salient-pole laminated
core with Nr poles. To produce a symmetric, basically synchronous, machine,
 2
2 ⋅ N r ⋅ τ PM =  2 ⋅ np ⋅ 6 ⋅ k + 6 ⋅ k ⋅  ⋅ τ PM (11.5)
 3

(a) (b)
FIGURE 11.5 Three-phase flux reversal machine (FRM) with (a) stator surface permanent magnets (PMs)
(Ns = 12; np = 2; Nr = 28) and (b) inset PMs (Ns = 12; np = 3; Nr = 40).
where
k = 1, 2, …
np = 1, 2, …
A typical three-phase FRM is shown in Figure 11.5a and Figure 11.5b.

As for the TFM, the PM flux linkage in the phase coils changes polarity (reverses sign) when the rotor
moves along a PM pole span angle. The structure is fully manufacturable, as it “borrows” the magnetic
circuit of a switched reluctance machine. However, the problem is, as for the TFM, that the PM flux
fringing reduces the ideal PM flux linkage in the coils to around 30 to 60%, in general. The smaller the
pole pitch τPM, the larger the fringing and, thus, the smaller the output. There seems to be an optimum
thickness of the PM hPM to pole pitch τPM ratio, for minimum fringing.
On the other hand, the machine inductance Ls tends to be reasonably small, as end connections are
reasonable (nonoverlapping coils) and the surface PM poles secure a notably large magnetic airgap.
Still, in Reference 4, for a 700 Nm peak torque at 7 N/cm2 force density, the power factor would be
around 0.3. In order to increase the torque density for reasonable power factor, PM flux concentration
may be performed in the stator (Figure 11.6) or in the rotor (Figure 11.7).
Air
Winding
PM
Rotor
FIGURE 11.6 Three-phase flux reversal machine (FRM) with stator permanent magnet (PM) flux concentration.

Winding
Stator
SS NN SS NN SS
NN NN
SS SS
N N NN
SS
SS Rotor
NN
N
SS
N
SS
NN
NN
N SS NN SS
SS NN SS NN S
Secondary
stator
SS N
NN
Shaft
S
NN
SS
SS
N
N
N
SS
N
SS
NN NN
SS N SS
N SS NN SS NN
FIGURE 11.7 Three-phase flux reversal machine (FRM) with rotor permanent magnet (PM) flux concentration
and dual stator.
The FRM with stator PM flux concentration is highly manufacturable, but as the pole pitch τPM gets
smaller, because the coil slot width ws is less than τPM, the power factor tends to be smaller. For τPM = 10
mm and ws = τPM and a 4ws height, with jpeak = 10 A/mm2 and slot filling factor kfill = 0.4, the slot
magnetomotive force (mmf) Wc Ipeak is as follows:
Wc ⋅ I peak = w s [mm]⋅ 4 ⋅ w s [mm]⋅ k fill ⋅ j peak = 10 ⋅ 4 ⋅100 ⋅ 0.4 ⋅10 = 1600 Aturns/slot
Larger slot mmfs could be provided for the TFM and FRM with stator surface PM poles. However,
the configuration in Figure 11.7 allows for the highest PM flux concentration, which may compensate
for the lower Wc Ipeak and allow for lower-cost PMs, because the radial PM height is generally larger than
3 to 4 τPM. As with any PM flux concentration scheme, the machine inductance remains large. But, for
not so large a number of poles, the machine’s easy manufacturability may pay off. On the other hand,
the FRM with rotor PM flux concentration configuration (Figure 11.7) provides for large torque density,
because, additionally, the allowable peak coil mmf may be notably larger than that for the configurations
with stator PM flux concentration (Figure 11.6).
The rotor mechanical rigidity appears, however, to be lower, and the dual stator makes manufacturability
a bit more difficult. Still, conventional stamped laminations can be used for both stator and rotor cores.
To provide more generality to the analysis that follows, we will consider only the three-phase TFMs
and FRMs, which may work both as generators and motors with standard PWM converters and position-
triggered control.
11.2 The Three-Phase Transverse Flux Machine (TFM): Magnetic

Circuit Design
The configuration for one phase may be reduced to the one in Figure 11.1a, but inverted, to have an
external rotor. The iron behind the PMs on the rotor may be, in principle, solid iron, which is a great
advantage when building the rotor.

I-shaped
core
Armature
winding
U-shaped
core
Aluminum
carrier
FIGURE 11.8 Transverse flux machine (TFM) — aluminum carrier with interior stator cores and coil.
Also, the stator U-shape and I-cores (Figure 11.4a) may be made of silicon laminations. The aluminum
carriers that hold tight the stator I- and U-cores are the main new frame elements that have to be
fabricated by precision casting.
Apparently, the circumferential coil has to be wound turn by turn on a machine tool after the U-cores
have been implanted in the aluminum carrier. Then, the I-cores are placed one by one in their locations
on top of the coil (Figure 11.8).
It was shown that, in order to reduce the cogging torque, the stator U- and I-cores of the three phases
have to be shifted by 120 electrical degrees with respect to each other [6]. In such a case, the three phases
are magnetically independent, though the PMs are axially aligned on the rotor for all three phases.
As seen in Figure 11.3a, the PM flux paths are basically three-dimensional in the rotor but only bidi-
mensional in the stator. However, as the flux paths are basically radial in the PMs and circumferential-radial
in the rotor back iron, the rotor back iron may be made of mild solid steel. The bidimensional flux paths
in the stator allow for the use of transformer laminations in the U- and I-cores.
There is, however, substantial PM flux fringing, which crosses the airgap and closes the path between
the stator U- and I-cores in the circumferential direction. To reduce it, both U- and I-cores should expand
circumferentially less than a PM pole pitch τPM: bu, bi < τPM. Also, to reduce axial PM flux fringing, the
axial distance between the U-core legs lslot should be equal to or larger than the magnetic airgap (lslot > g +
hPM). But, lslot is, in fact, equal to the width of the open slot where the stator coil is placed.
The U-core yoke hyu and the I-core hyi heights should be about equal to each other and around the
value of 2/3 lstack in order to secure uniform and mild magnetic saturation in the stator iron cores. Also,
the rotor yoke radial thickness hyr should be around half the PM pole pitch, as half of the PM flux goes
through the rotor magnetic yoke (Figure 11.9).
We will now approach performance through an analytical method, accounting approximately for
magnetic saturation in the stator and iron cores.
Each PM equivalent mmf FPM = HchPM (Hc is the coercive field of the PM material) “is responsible”
for one magnetic airgap hgM = g + hPM. This way, the equivalent magnetic circuit on no load (zero current)
is as in Figure 11.9:
hPM + g
RPM + g = (11.6)
µ0 ⋅ bPM ⋅ lstack ⋅ (bPM + bu )/2

FPM FPM
RPM+g RPM+g
F PMax
Rfringe
2 Ryr Ryi + Ryu
Rfringe
RPM+g RPM+g
F PMax
FPM FPM
FIGURE 11.9 The magnetic equivalent circuit for permanent magnets (PMs) in the position for maximum flux in
the stator coil.
where
RPM+g is the PM and airgap magnetic reluctance
bPM is the PM width (bPM/τPM = 0.66 to 1.0)
bu is the U-core width
Rfringe is the magnetic reluctance of the PM fringing flux between the stator U- and I-cores through
the airgap, corresponding to one leg of the U- and I-cores.
To a first approximation,
 h τ PM − b yu  1
R fringe ≈  PMb + ⋅ ; b yu = b yi (11.7)
 µ0 ⋅ PM
2 µ0 ⋅ 2yi  lstack
h
Straight-line magnetic flux paths are considered between U- and I-cores up to the height of the I-core
(Figure 11.8). The axial fringing may be added as a reluctance in parallel to Rfringe. The stator U- and I-core
reluctances Ryu and Ryi are as follows:
2 ⋅(hslot + hyi ) lstack + lslot

Ryu = + (11.8)
µcu ⋅ bu ⋅ lstack µ yu ⋅ hyu ⋅ bu
lslot 4 ⋅ lstack
Ryi ≈ + ; b ≈b (11.9)
µ yi ⋅ hyi ⋅ bi µ yi ⋅ hyi ⋅ bi i u
Also,
τ PM
Ryr = (11.10)
µ yr ⋅ b yr ⋅ lstack
where µcu, µyu, µyi, and µyr are the magnetic permeabilities dependent on magnetic saturation. As the PM
equivalent mmf Fc is given, once all the PM geometry and material properties are known, an iterative
procedure is required to solve the magnetic circuit in Figure 11.9. To start with, initial values are given
to the four permeabilities in the iron parts: µcu(0), µyu(0), µyi(0), and µyr(0). With these values, the flux
in the stator and rotor core parts and ΦPMax are computed.

But, the average flux densities in various core parts are straightforward, once ΦPMax is known:
φ PMax = b yu ⋅ lstack ⋅ Bcu = b yu ⋅ hyu ⋅ B yu = hyi ⋅ bi ⋅ B yi

(11.11)
φ PMax
= B yr ⋅ hyr ⋅ lstack
2
Once the average values of flux densities Bcu, Byu, Byi, and Byr are computed, and from the magnetization
curve of the core materials, new values of permeabilities µcu(1), µyu(1), µyi(1), and µyr(1) are calculated.
The computation process is reinitiated with renewed permeabilities:
µ(1) = µi (0) + c(µi (1) − µi (0)); c = 0.2 − 0.3 (11.12)
such as to speed up convergence.

The computation is ended when the largest relative permeability error between two successive com-
putation cycles is smaller than a given value (say, 0.01). This way, the maximum value of the PM flux
per pole (U-core), ΦPMax , in the coil, is obtained.
The PM flux per pole varies from + ΦPMax to − ΦPMax when the rotor moves along a PM pole angle
(2π /2p1 mechanical radians). What is difficult to find out analytically, even with a complicated magnetic
circuit model with rotor position permeances, is the variation of ΦPM with rotor position from +ΦPMax
to −ΦPMax and back.
It was shown by three-dimensional (3D) finite element method (FEM) that the PM flux per pole varies
trapezoidally with rotor position (Figure 11.10).
We may consider, approximately, that the electrical angle θeconst, along which the PM flux stays constant
(Figure 11.10) is rather small and is as follows [3]:
 1 (b + b 
θ econst ≈  − PM u)  ⋅ π [rad] (11.13)
2 4 ⋅τ PM 
Magnetic saturation alters not only ΦPMax but also the value of θeconst , tending to flatten the trapezoidal
waveform and bringing it closer to a rectangular waveform of lower height. But, local magnetic saturation
plays an even more important role when the machine is under load.
FPM (qer) 1 (bPM + bu ) p =q

− const
2 4tPM
FPMax
p qer = p1qr
2p
-FP Max
FIGURE 11.10 Permanent magnet (PM) flux per stator pole vs. rotor position.

ft N
cm2
6
Ipeak
2
Irated
qer
0
p p
2 I=0
−2
FIGURE 11.11 Typical force density (N/cm2) vs. rotor position and various constant current values.
In general, the computation of the force density (in N/cm2) on the rotor surface is done via the Maxwell
stress tensor through FEM. The magnetic saturation presence is evident when the force is calculated for
various rotor positions for constant phase current (Figure 11.11).
The PM flux per pole and the emf, obtained through 3D FEM, look as shown in Figure 11.12.
The instantaneous emf per phase E is as follows (Equation 11.2 and Equation 11.3):
dφ PM (θ er )
E = W1 ⋅ p ⋅ ⋅ 2 ⋅π ⋅n ⋅ p (11.14)
dθ er
θ er = ω r ⋅ t (11.15)
So, the emf per phase has a waveform in time, which emulates the waveform of dΦPM(θer)/dθer. The above
derivative maximum decreases when the PM pole pitch decreases due to the increase in fringing flux,
dFPM/dqer
10−4
FPM (Wb)
0
90° 180°
−10−4
FIGURE 11.12 Typical finite element method (FEM)-extracted permanent magnet (PM) flux and its derivative vs.
rotor position.

when the number of poles increases. Consequently, there should be an optimum number of PM poles
for a given rotor diameter that produces maximum emf for given W1, stack length, and mechanical gap g.
If the phase emf is considered through its fundamental, the average torque per phase, when the emf
and current are phase shifted by angle γ1, is as follows:
Es ⋅ I1 ⋅ cos γ 1 W1 ⋅ p 2 ⋅ φ PMax ⋅ I1 ⋅ cos γ 1

(Te )aph = = (11.16)
2 ⋅π ⋅n 2
where I1 is the root mean squared (RMS) phase current, and
φ PM (θr ) = φ PMax ⋅ cos( p ⋅θr ) (11.17)
11.2.1 The Phase Inductance Ls

The phase inductance Ls is composed from the leakage inductance Lsl and the main path inductance Lm.
As the magnetic circuits of the three phases are separated, there is no coupling inductance between phases.
Ls refers to one phase only. The airgap inductance Lm is, approximately,
b +b
W12 W12 ⋅ p ⋅ u 2 PM ⋅ lstack
Lm = µ0 ⋅ ≈ µ0 ⋅ (11.18)
Rmag 4 ⋅( g + hPM ) ⋅(1 + ks ) ⋅(1 + k f )
where ks is an equivalent magnetic saturation coefficient that considers the contribution of the iron parts
to the total mmf along a flux path. kf is a fringing coefficient that takes care of the fringing flux in the
airgap: kf < 0.2 in general.
In a similar manner, we may treat the maximum flux per pole:
bPM + bu 1
φ PMax = Bgi ⋅ ⋅ lstack ⋅ (11.19)
2 (1 + ks )(1 + k fringe )
kfringe is the PM fringing flux coefficient (very different from kf ) that can go as high as two (kfringe = 0.7
to 2), while ks is the magnetic saturation coefficient. In general, ks has to be calculated when current is
present in stator phases. The leakage inductance may be calculated approximately from slot leakage flux,
extended between U-cores, because I-cores “create” a moderate “slot leakage effect”:
hs
Lsl ≈ µ0 ⋅ W12 ⋅ ⋅ p ⋅ bu ⋅ (1 + ki ) (11.20)
3 ⋅ lslot
ki accounts for the leakage inductance between the U-cores. In general, ki < 0.2 to 0.3.
As the total iron area, as seen along the airgap, by the stator coil, is independent of the number of
poles, both Lm and Lsl are independent of the number of poles 2p1. At the same time, the emf E and the
torque (Te)aphase per phase are proportional to the number of poles if the increasing fringing (kfringe) with
the number of poles is not considered.
11.2.2 Phase Resistance and Slot Area

The phase resistance Rs is straightforward:
π ⋅(Dis + 2 ⋅ hyi + hs ) ⋅W12

Rs ≈ W1⋅I1
(11.21)
σ copper ⋅ jcon
The ampereturns per phase (RMS value), W1I1, may be calculated from Equation 11.16 once the torque
per phase requirement is given.

The rated (continuous) current density jcon depends on the machine duty cycle, type of cooling, and
design optimization criterion (maximum efficiency or minimum machine cost, etc.). In general, jcon = 4
to 12 A/mm2. The window area in the slot Aslot is as follows:
W1 ⋅ I1
Aslot = hs ⋅ lslot = (11.22)
jcon ⋅ k fill
The total slot filling factor kfill is, in general, kfill = 0.4 to 0.6. The larger values correspond to the preformed
coils eventually made of rectangular cross-sectional conductors.
Example 11.1
Consider sizing a three-phase TFM generator with surface PM interior rotor and single-sided stator
(with U- and I-cores) that has the following specifications: Ten = 200 kNm and nn = 30 rpm.
Solution
While part of the design formulas are included in the previous paragraphs, some new ones are
introduced here. They are mainly related to the stator bore diameter Dis with given lstack/Dis ratio (λ
= lstack /Dis = 0.05 to 0.1).
We make use of the force density ft = 2 to 8 N/cm2 to determine the interior stator diameter Dis:
Te D
Dis = 3 ; Te = is ⋅ f t ⋅ π ⋅ Dis ⋅ 3 ⋅ 2 ⋅ lstack (11.23)
3 ⋅π ⋅ λ ⋅ ft 2
In our case, with λ = 0.05, ft = 2.66 N/cm2:
200000
Dis = 3 = 2.525 m (11.24)
3 ⋅ π ⋅ 0.05 ⋅ 2.66 ⋅104
The stack length is as follows:

lstack = λDis = 0.05·2.515 = 0.1257 m (11.25)
We now have to choose the airgap g, first.
For such a large diameter, an airgap of 1.5 to 5 mm is required for mechanical rigidity reasons. Let
us consider g = 4·10−3 m. The ideal airgap flux density Bgi produced by the PM magnets in the airgap
is as follows:
hPM
Bgi = Br ⋅ (11.26)
hPM + g
with Br = 1.3 T, µrem = 1.04 µrem at 100°C (very good NeFeB magnets: Br0 = 1.37 T at 20°C).
We need to choose a large PM height hPM, because the actual airgap flux density will be notably
reduced by the fringing (kfringe ≈ 2). So, hPM = g = 3 · 4 · 10−3 = 12 · 10−3 = 1.2 · 10−2 m. Consequently,
from Equation 11.26,
1.3 ⋅12
Bgi = = 0.975 T (11.27)
12 + 4
Now the maximum flux per pole may be calculated if we adopt the number of poles 2p1.

In general, the PM pole pitch τPM should be higher than the total (magnetic) airgap: g + hPM =
1.6 · 10−2 m. The number of poles 2p1 is, thus,
π ⋅ Dis π ⋅ 2.515
2⋅ p = = = 394.8 (11.28)
τ PM 0.02
Let us consider 2p = 400 poles and τPM = 1.974 · 10−2 m. The fundamental frequency at 30 rpm is
fn = p1 · nn = 200 · 30 / 60 = 100 Hz, which is a reasonable value in terms of core losses. Now,
the maximum PM flux per pole from Equation 11.19, with kfringe = 2.33; ks = 0.1; bPM/τPM = 0.85;
bu/τPM = 0.8 is as follows:
bPM + bu 1
φ PMax = Bgi ⋅ ⋅ lstack ⋅
2 (1 + k fringe ) ⋅ (1 + ks )
(11.29)
0.2515
−2 1
= 0.975 ⋅1.628 ⋅10 ⋅ ⋅ = 0.5986 ⋅10−3 Wb
2 (1 + 2.33) ⋅ (1 + 0.1)
We now turn directly to the torque expression to find the ampere turns per phase W1I1 (RMS value)
from Equation 11.16, with γ1 = 0 (emf and current in phase):
Ten ⋅ 2 200 ⋅103 ⋅ 2

W1 ⋅ I1 = = = 3885 Aturns (11.30)
3 ⋅ p 2 ⋅ φ PMax 3 ⋅ 2002 ⋅ 0.5986 ⋅10−3
This is a reasonable value.

The slot area required for the coil for j = 6 A/mm2 and kfill = 0.6, from Equation 11.22, is as follows:
W1 ⋅ I1 3885
Aslot = = = 1079.16 mm 2 (11.31)
jcon ⋅ k fill 6 ⋅ 0.6
Taking
lslot = 2(g + hPM) = 32 mm (11.32)
the slot height hs is
Aslot 1079
hs = = = 33.72 mm (11.33)
lslot 32
The fact that hs /lslot ≈ 1 will lead to a smaller leakage inductance, which is favorable for a good power
factor.
Consider hyi = hyu = 2/3 · lstack = 0.083 m. Then, from Equation 11.21, the stator phase resistance Rs is
2.3 ⋅10−8 ⋅ π ⋅(2.515 + 2 ⋅ 0.083 + 0.03372)

Rs = 3885
⋅W12 = 0.3026 ⋅10−3 ⋅W12 (11.34)
6⋅10 6
The total winding losses for the machine, Pcon, are as follows:
Pcon = 3 ⋅ Rs ⋅ I12 = 3 ⋅ 0.3206 ⋅10−3 ⋅ 38852 = 13.70 kW (11.35)

The electromagnetic power, Pelm, is
30
Pelm = Te ⋅ 2 ⋅ π ⋅ nn = 200 ⋅103 ⋅ 2 ⋅ π ⋅ = 628 kW (11.36)
60
The winding losses are about 2% of input power.
Now we need to calculate the synchronous inductance Ls = Lm + Lsl. From Equation 11.19,
b +b
µ0 ⋅ W12 ⋅ p ⋅ u 2 PM ⋅ lstack
Lm =
4 ⋅ ( g + hPM ) ⋅ (1 + ks ) ⋅ (1 + k f )
(11.37)
1.256 ⋅10−6 ⋅ 200 ⋅1.62 ⋅10−2 ⋅
0.2515
1.321 ⋅10−5
= ⋅ 2
W 2
= ⋅ W12[H ]
4 ⋅1.6 ⋅10−2 ⋅ (1 + 0.1) ⋅ (1 + 0.1) 1
2
Also, from Equation 11.20, Lse, the leakage inductance is as follows:
hs
Lsl = µ0 ⋅ W12 ⋅ ⋅ (1 + ki ) ⋅ p1 ⋅ bu
3 ⋅ lslot
(11.38)
0.03372
= 1.256 ⋅10−6 ⋅ ⋅ (1 + 0.3) ⋅ 200 ⋅ 0.016 ⋅ W12 = 1.834 ⋅10−6 ⋅ W12
3 ⋅ 0.032
So,
 1.321 ⋅10−5 
Lsl = Lm + Lse =  + 1.834 ⋅10−6  ⋅ W12 = 0.844 ⋅10−5 ⋅ W12 (11.39)
 2 
The emf E (rms value) is as follows:
Ten ⋅ 2 ⋅ π ⋅ n 628 ⋅103 209.3 ⋅103

E= = = (11.40)
3 ⋅ I1 ⋅ cos γ 3 ⋅ I1 I1
The ideal power factor angle ϕ 1 (for γ = 0) is
ω1 ⋅ Ls ⋅ I1  2 ⋅ π ⋅100 ⋅ 0.844 ⋅10−5 

ϕ1n = tan −1 = tan −1  ⋅ W12 ⋅ I12 
E  208.33 ⋅10 3
 (11.41)
−1 −8 −1
= tan (2.544 ⋅10 ⋅ 3885 ) = tan 0.384 = 21°; cos ϕ1n = 0.933
2
The very good value of the power factor angle for E and I in phase is a clear indication of machine
volume reduction reserve (the specific tangential force is only 2.66 N/cm2).
The data of the preliminary design are given in Table 11.1.
11.3 TFM — the d–q Model and Steady State

Let us consider here that TFM is provided with a surface PM rotor, and thus, the synchronous inductance
Ls is independent of rotor position. We adopt this configuration, as the power factor may be higher due
to smaller inductance Ls, even though at slightly smaller torque density than for the PM flux concentration
configurations. Though the emf waveform is not quite sinusoidal, we consider it here as sinusoidal.

TABLE 11.1 Example Design Summary

Electromagnetic torque Te = 200 kNm
Speed nn = 30 rpm
Force density ft = 2.66 N/cm2
Current density jcon = 9 A/mm2
Stator interior diameter Dis = 2.515 m
Stator stack U-core leg lstack = 0.1257 m
Mechanical gap g = 4·10−3 m
Permanent magnet radial thickness hPM = 12·10−3 m
Number of permanent magnet poles 2·p = 400
Fundamental frequency fn = 100 Hz
Permanent magnet pole pitch τPM = 1.974·10−2 m
Axial room between U-core legs lslot = 32·10−3 m
U-core width bu/τPM = 0.8
Permanent magnet width per pole bPM/τPM = 0.85
U-core yoke height hyu = 2/3·lstack
I-core yoke height hyi = hyu
Slot useful height hs = 33.72·10−3 m
Total U-core interior height hs + hyi
External stator diameter Dos = 2.621 m
Total axial length per phase lslot + 2·lstack = 0.032 + 0.2515 m = 0.2835 m
Total axial length for three phases laxial = 3·(lslot + 2·lstack) = 0.85 m
Reduction of permanent magnet flux due to fringing 1/(1 + kfringe) = 1/3
Stator resistance per phase Rs = 0.454·10−3·W12
Stator inductance per phase Ls = 0.844·10−5·W12
Emf at f1 = 100 Hz E = 53.61·W1 (rms)
Stator copper losses Pcon = 20.55 kW
Electromagnetic power Te·2·π·nn = 628 kW
Power factor angle ϕ1n ≈ 21°; cos ϕn = 0.933
(with E and I in phase)
The d–q model is thus straightforward, if the core losses are neglected for the time being:
dψ s
I s ⋅ Rs + V s = − − j ⋅ωr ⋅ψ s
dt
ψ s = ψ d + j ⋅ ψ q ; ψ d = ψ PM + Ls ⋅ I d ; ψ q = Ls ⋅ I q (11.42)
I s = I d + j ⋅ I q ; V s = Vd + j ⋅ Vq ; ψ PM = φ PMax ⋅ W1 ⋅ p1
where
W1 is the turns per coil (phase)
p1 is the pole pairs:
3 3
Te = ⋅ p ⋅ (ψ d ⋅ I q − ψ q ⋅ I d ) = ⋅ p1 ⋅ ψ PM ⋅ I q
2 1 2
For steady state dtd = 0, and the vector diagram for generating is as shown in Figure 11.13a and
Figure 11.13b for pure Iq control (Id = 0) and for unity power factor (with diode rectifier). Pure Iq
control for Ls = constant corresponds to maximum torque per current, but the machine requires
reactive power for magnetization. A controlled front rectifier is required.
The performance is acceptable, but the force density is not large enough.
For the diode rectifier, the current always contains an Id component (Figure 11.13b), and thus, the
torque is produced with more losses. However, the reactive power required is zero, and the diode rectifier
is less expensive. If constant direct current (DC) link voltage operation at variable speed is required,
either an active front or PWM converter is provided on the machine side or a diode rectifier plus a

ωr
jq jq ωr
ΨPM d Id ΨPM
d
Lsj I q L s Id
δV = ϕ1 < 0
δ V jI Ψs Ls jI q
is = jIq q
Vs Is
Vs
−jω rΨs Es −Rs Is −jω rΨs

−Rs Is
Es
−jω r Ls Is −jω r Ls Is
(a) (b)
FIGURE 11.13 The transverse flux machine (TFM) vector diagram for steady state: (a) pure Iq control and (b) with
power factor control (with diode rectifier).
DC–DC boost converter is required. The core losses might be considered, mainly for steady state, as
follows:
3 ω ⋅ψ
2
piron = ⋅ r s (11.43)
2 Riron
In Equation 11.43, the core losses are considered to be produced by the resultant flux in the machine,
in the presence of stator current. The core losses piron might be determined from tests that segregate the
iron losses, or they may be calculated from analytical or FEM models, as shown in Chapter 10.
The efficiency may be defined as follows:
P2el
η= (11.44)
P2el + pmec + piron + pcopper + protor + pAl
The rotor losses protor refer to the eddy current losses in the PMs, which may be neglected only for
f1 ≤ 100 Hz, in general. The strayload losses are included in piron and pcopper , but the eddy current losses in
the aluminum carriers of the U- and I-cores are individualized as pAl and may be calculated only through
3D FEM eddy current models [7].
When the fundamental frequency goes up, above 100 Hz, piron and protor become important and have
to be dealt with using great care (see Chapter 10 for more details).
Example 11.2
Consider the TFM design in Example 11.1, and calculate the number of turns per phase for a line
voltage VLn = 584 V (RMS) — star connection — at rated speed for E, I in phase, and for the same
number of turns W1 and unity power factor, and calculate the performance for the same current as above.
Solution
We simply make use of the vector diagram (Figure 11.13a), where yPM is
E(rms) ⋅ 2 53.61 ⋅ W1 ⋅ 2
ψ PM = = = 0.1203 ⋅ W1
ωr 2 ⋅ π ⋅100
(11.45)
VLn (rms) ⋅ 2 584 ⋅ 2
Vsn = = = 476 V
3 3

From Figure 11.13a,
Es = E(rms) ⋅ 2 = Vs ⋅ cosϕ1 + Rs ⋅ I s (11.46)
or
75.59 ⋅W1 = 476 ⋅ 0.933 + 0.454 ⋅10−3 ⋅W12 ⋅ I1 ⋅ 2 (11.47)
With W1I1 = 3885 Aturns/coil, we obtain W1 ≈ 6 turns/coil. As the core and mechanical losses have
not been considered, the efficiency will reflect only the presence of the already calculated copper losses:
P1elm − pcon P2 628 − 13.70 614.33 kW

η= = = = = 0.978 (11.48)
P1elm P1elm 628 628 kW
Alternatively,
3 ⋅VLn ⋅ I1 ⋅ cosϕ1
ηen = (11.49)
P1elm
with
W1 ⋅ I1 3885
I1 = = = 647.5 A (11.50)
W1 6
3 ⋅ 584 ⋅ 647 ⋅ 0.933 609

ηen = = = 0.971 (11.51)
628 ⋅103 628
The difference in efficiency from the two formulas above is only due to calculation errors, as the
model is the same.
Now, for the second case, we have to notice that if we consider the same phase current (RMS) I1
and voltage VLn, the vector diagram in Figure 11.13b provides the following:
(Vs + Rs ⋅ I1)2 + ω12 ⋅ L2s ⋅ I12 = Es2 (11.52)
with W1 = 6 turns, I1 = 647.5 A, we can calculate, from Equation 11.52, the terminal voltage:
Vs′ = Es2 − ω12 ⋅ L2s ⋅ I s2 − Rs ⋅ I s
(11.53)
= (75.59 ⋅ 6)2 − (2 ⋅ π ⋅100 ⋅1.5044 ⋅10−5 )2 ⋅ 64 ⋅ 647.52
− 0.3026 ⋅10−3 ⋅ W12 ⋅ I1 ⋅ 2 = 396.48 − 9.945 = 386.63 V
Vs′⋅ 3 3
VL′ = = 386.63 ⋅ = 473.40 V < 584 V (11.54)
2 2
The line terminal voltage for the same current decreased notably for unity power factor.
Now, the output power is only
(P2′)cos ϕ =1 = 3 ⋅ 473.40 ⋅ 647.5 = 530.29 kW (11.55)

1

So, for the same number of turns W1 = 6 and the same conductor cross-section and current and speed,
with a diode rectifier, the generator will produce 20% less power. The additional reactive power provided
through the active front PWM converter, when E and I are in phase, allows for delivering power in better
conditions (higher voltage and power).
Note on the Control of TFM

Once a three-phase machine is considered, the control may be implemented with sinusoidal current (for
almost sinusoidal emf) or with trapezoidal current (for trapezoidal emf), with third harmonic tapping,
if a diode rectifier is used. The control system, with or without motion sensors, is practically the same
as that for standard PMSGs, as detailed in Chapter 10. This is why we do not elaborate here on TFM
control.
The methods used to reduce the cogging torque are similar to those for standard PMSGs.
It is to be seen if the TFM will make it to the markets soon, at least as a low-speed high-torque direct-
driven generator.
11.4 The Three-Phase Flux Reversal Permanent Magnet

Generator: Magnetic and Electric Circuit Design
As mentioned in the section on TFMs, the large PM flux fringing and, consequently, the low power factor
are the main obstacles in the way for producing very good performance, even at very low speeds. The
FRMs face similar problems, but they are easier to manufacture [8–11].
We already introduced the PM flux concentration schemes for FRM with stator (Figure 11.6) and rotor
PMs (Figure 11.7). We will, however, treat in detail here only the surface stator PM configuration, as it is
similar to the TFM investigated in the previous paragraph and is characterized by a smaller inductance (in
P.U.), though PM flux fringing is large. We will first present a conceptual design and complete FEM
investigations of a 200 Nm, 128 rpm (60 Hz) FRM. Then, through a numerical example, a FRM preliminary
design methodology is presented for 200 kNm, 30 rpm, 100 Hz system, as in Example 11.1 and Example 11.2.
We will refer to very low speeds, as we feel that for speeds above 500 to 600 rpm, PMSGs with distributed
or fractional (with tooth-wound coils) windings are already well established (Figure 11.14).
The low-speed operation at a frequency of 50 to 60 Hz (to secure moderate core loss but still make good
use of iron) implies a large number of “poles” or “pole pairs” or “periods” of the flux reversal along the
machine periphery. In the FRM with Nr rotor salient poles, the speed n and frequency f1 are related by the
following:
f1 = n ⋅ N r (11.56)
The two-pole pitch angle corresponds to two PM poles of alternate polarities on the stator, that is,
π ⋅ Dr
2 ⋅τ PM = 2 ⋅τ rot = (11.57)
Nr
where Dr is the rotor diameter. A stator pole may accommodate 2 · np PMs. The electrical angle of the
space between two neighboring stator pole coils should be 120 electrical degrees or 120°/Nr geometrical
degrees or 2 · τPM /3. Thus,
 2 ⋅τ PM 
N s ⋅  2 ⋅ np ⋅τ PM + = π ⋅(Dr + 2 ⋅ g ) ≈ 2 ⋅τ PM ⋅ N r (11.58)
 3 

2/3tPM
tPM
hPM
N
S
S
2tPM
N Dis
N
S
S
N
FIGURE 11.14 Surface permanent magnet (PM) flux reversal machine (FRM) geometry details (Ns = 12, np = 2,
Nr = 28).
where Ns is the number of stator poles, and g is the mechanical airgap. Eliminating τPM from Equation
11.58 yields the following:
 1
N s ⋅  np +  = N r (11.59)
 3
The typical configuration for Ns = 12 is shown in Figure 11.4 and Figure 11.15. The PMs on the stator
pole shoes are very close to the airgap. We may say it has pole PMs.
The FRM for low-speed drives has the following distinct features:
• It uses conventional stamped laminations both on the stator and on the rotor.
• It has no PMs or windings on the rotor.
• It has PMs on the stator, where their temperature can be easily monitored and controlled.
• The stator has concentrated coils that are easy to manufacture.
• The lowest pole pitch τPM is to be larger than (PM + airgap) thickness [8–11] to limit flux fringing
in PM utilization, as is the case with the transverse PM flux machine that has the PMs placed on
the rotor.
• The higher the rotor diameter (or torque), the larger the maximum number of pole pairs (2 · τPM),
and thus, the lower the speed at 50 (60) Hz.
• The pole PM FRM has lower inductances than the inset PM FRM (Figure 11.15), as expected.
Also, the cogging torque of the latter is intrinsically smaller.
• For the inset PM FRM, the airgap should be as small as mechanically feasible, and the PM thickness
hPM > 6 · g. Also, the stator and rotor teeth should be equal, while the rotor interpole will span
the rest of the double-pitch τPM:
τ PM = bt1 + hPM ; 2 ⋅ τ PM = bt 2 + bs 2 ; bt 2 = bt1 (11.60)

FIGURE 11.15 Low-speed flux reversal machine (FRM) with inset permanent magnets (PMs) on stator.
Moreover, bt1/hPM > 2/3 provides enough saliency. Thus, the inset PM FRM with an airgap of 0.2 mm,
hPM = 1.5 mm, would allow for bt1min = 5 mm or a pole pitch τPM = 5 + 1.5 = 6.5 mm, which is
65% of the one considered for the pole PM configuration. Further reductions of the airgap and
magnet thickness may lead to even smaller pole pitches. Thus, at 60 Hz, the speed is accordingly
reduced to a value depending on the rotor diameter.
• The inset PM FRM has the PMs parallel to the stator magnet flux lines and is much more difficult
to demagnetize. As a bonus, the flux in the magnets varies less (especially under load). Thus, the
eddy currents induced with PMs are notably smaller than those for the pole PM configuration,
that has PMs that directly experience the stator current additional field.
• The much lower total (magnetic) airgap of the inset PM FRM leads to much larger inductance,
which is limited by heavy saturation of the core for high currents (overload). Therefore, the stator
current limits are mainly governed by stator temperature and magnetic oversaturation in the inset
PM configuration, and stator temperature and PM demagnetization limit the currents for the pole
PM FRM.
11.4.1 Preliminary Geometry for 200 Nm at 128 rpm

via Conceptual Design
Consider the pole PM configuration only, with Nr = 28 rotor poles, Ns = 12 stator poles, with 2np = 4 PMs
per stator pole. The designated primary frequency is f1 = 60 Hz. It corresponds to the rated speed nn =
(f1/Nr) · 60 = 128.5 rpm. We start the preliminary design from the specific tangential force ft to calculate
the torque Te:
Dr
Te = f t ⋅ π ⋅ Dr ⋅ lstack ⋅ ; f tn = 1.5 − 4 N /cm 2 (11.61)
2
where Dr is the rotor diameter, and lstack is the stack length. A stack-length-to-rotor-diameter Dr ratio λ
is defined as follows:
lstack
λ= = 0.2 − 1.5 (11.62)
Dr

The lower values of λ are justified for a large number of stator poles. With Ns = 12, λ = 1.055, Te = 200 Nm,
and ftn = 2 N/cm2 (continuous duty), from Equation 11.62, we obtain
2 ⋅Ten 2 ⋅ 200
Dr = 3 = 3 = 0.182 m (11.63)
π ⋅ f tn ⋅ λ π ⋅ 2.0 ⋅104 ⋅1.055
We may choose Dr = 0.180 m to obtain lstack = 1.055 · 0.18 = 0.19 m. The PM flux per stator pole ΦPM is
as follows:
lstack ⋅ BPMi ⋅τ PM ⋅ npp

φ PM = (11.64)
1 + k fringe
BPMi is the ideal flux density in the airgap as given by
hPM
BPMi = Br ⋅ (11.65)
hPM + g
where hPM = 2.5 mm, and g = 0.5 mm. At 75°C, Br = 1.21 T, Hc = 651 · 103 A/m for neodymium–iron–boron
(NdFeB) PMs. kfringe is a flux fringing coefficient to be determined by FEM. We choose here an initial safe
value kfringe = 2.33. The flux is to vary almost sinusoidally with the rotor position θr:
φ PM (θr ) = φ PM ⋅ sin(N r ⋅θr ) (11.66)
Therefore, the PM flux per stator pole derivative with θr is
dφ PM (θr ) lstack ⋅ BPMi ⋅τ PM ⋅ npp ⋅ N r ⋅ cos(N r ⋅θr )

= (11.67)
dθr 1+ k fringe
There are Ns/3 stator poles per phase and, with all coils per phase in series, according to Equation
11.59, the emf amplitude per phase Em is
Ns 2 ⋅ π ⋅ n ⋅ lstack ⋅ π ⋅ Dr ⋅ npp ⋅ BPMi

Em = ⋅ nc ⋅ (11.68)
3 1 + k fringe
where nc is the number of turns per coil. For a given sinusoidal current I (RMS), in phase with the emf
(Iq current control), the torque Te is, again,
3 2
Te = ⋅ Em ⋅ I ⋅ . (11.69)
2 2 ⋅π ⋅n
In our case, we find ncIn 2 :
2 2 ⋅π ⋅n 2 1 1
nc ⋅ In ⋅ 2 = nc ⋅ ⋅Te ⋅ = ⋅ 200 ⋅ ⋅
3 Em 3 4 ⋅ 0.19 ⋅ π 0.18 ⋅ 0.3 ⋅1.21 (11.70)
≈ 0.855 ⋅ 103 Aturns/coil.
For Em, we used Equation 11.68.

Choosing a rated current density jcn = 3.5 A/mm2 and a slot fill factor kfill = 0.4, and noticing that there
are two coils per slot, the slot area Aslot is as follows:
2 ⋅ nc ⋅ I 2 ⋅ 0.855 ⋅103
Aslot = = = 0.863 ⋅10−3 m 2 (11.71)
jcn ⋅ k fill 3.5 ⋅106 ⋅ 0.44
The slot detailed geometry may now be calculated if the stator pole (tooth) width is first found. As
the current loading in such low-speed machines tends to be large, and the coupling between the PM and
the coils is not so strong, an analytical dimensioning of its magnetic core is hardly practical, although
apparently standard methods could be used. The rather low rated current density is a safety precaution
for the torque capability (above 200 Nm) exploration by FEM.
11.4.2 FEM Analysis of Pole-PM FRM at No Load

The geometry with PM height hPM = 2.5 mm and airgap g = 0.5 mm was, in fact, obtained after many
FEM calculations with thinner and thicker PMs and eventually larger airgaps. The maximum PM flux,
calculated in the middle of the coil, was found for hPM = 2.5 mm and g = 0.5 mm (PM pitch is equal to
10 mm). The PM flux per stator pole with straight rotor poles is given in Figure 11.16. It should be noted
that the waveforms are symmetric and almost sinusoidal [4].
However, the cogging torque, although not large (less than 10% rated torque), for general purposes, may
be further reduced. Skewing the rotor by various angles leads to cogging torque reduction (Figure 11.17a).
For 1.8° mechanical skewing, the best result is obtained (Figure 11.17b).
A 7% loss in the maximum PM flux per phase is encountered with 1.80 mechanical degrees of skewing
(Figure 11.17c). We consider this to be acceptable, as the cogging torque was reduced below 1.5%, which
qualifies FRM for high performance. The skewing has the additional effect of shifting the phase PM flux
(as expected). This effect is important when designing the control system.
After filtering the phase PM fluxes obtained from FEM, their derivation with rotor position was
performed. As demonstrated in Figure 11.18a and Figure 11.18b, there are serious grounds to use vector
(sinusoidal) control, as the emfs are proportional to dΦ/dθ, and they are almost sinusoidal.
×10−3
1.5
PM flux per pole in phase A
PM flux per pole in phase B
PM flux per pole in phase C
1
PM flux per pole (Wb)
0.5
−0.5
−1
−1.5
0 2 4 6 8 10 12 14
Rotor position (mech. deg.)
FIGURE 11.16 Permanent magnet (PM) flux per pole vs. rotor position.

40
+ Cogging torque (N.m, without skewing)
Cogging torque (N.m, skewing 0.9 mech. deg.)
30 ∗ Cogging torque (N.m, skewing 1.8 mech. deg.)
++++++
20 +++++++ ++ ++
+
Cogging torque (Nm)

++ + +
+++ ++ ++
+++ + + +
10 + + + +
+ + +
+ + +
+ + + +
0 ∗ ∗ ∗+ ∗ ∗ + ∗ ∗ ∗+ ∗ ∗ ∗
+ +
+ + + +
+ + +
−10 + + +
+ ++
+ ++
+ ++ + +++
++ ++
−20 + ++
++ ++ +++++++
++++
−30
−1 0 1 2 3 4 5
(a)
0.25
Rated torque 200 N.m
0.2
Cogging torque (per unit)
0.15
0.1
0.05
0
−1 0 1 2 3 4 5
(b)
1.05
Peak PM flux in phase A (per unit)
0.95
0.9
0.85
0.8
−0.5 0 0.5 1 1.5 2 2.5 3 3.5
Skewing angle (mech. deg.)
(c)
FIGURE 11.17 Cogging torque for different skewing angles: (a) cogging torque vs. position, (b) cogging torque vs.
skewing angle, and (c) peak permanent magnet (PM) flux in a phase coil vs. skewing angle.

0.05
Derivative of PM flux per pole in phase A
Derivative of PM flux per pole in phase B
0.04 Derivative of PM flux per pole in phase C
0.03
Derivative of PM flux per pole

0.02
0.01
−0.01
−0.02
−0.03
−0.04
−2 0 2 4 6 8 10 12 14
Rotor position(mech. deg.)
(a)
ω 1LendI
Va
Φa
Rsl
Ea Iq
ΦPM d
(b)
FIGURE 11.18 Waveforms of d ΦPM/dθr per pole in phases A, B, and C: (a) fundamental and third harmonic and
(b) vector diagram with pure Iq control.
11.4.3 FEM Analysis at Steady State on Load

We may simulate steady-state load conditions by FEM if we let sinusoidal currents flow in the three
phases. Therefore, in fact, by gearing the currents to rotor position such that the current in phase A is
maximum when the PM flux in phase A is zero, we produce pure q axis control. Figure 11.19a shows
the phase currents vs. position. They are “in phase” with the flux derivatives (Figure 11.18a).
For various rotor positions, the flux per pole of phases A, B, and C with all currents present with
instantaneous values, which are position dependent, FEM flux distribution was calculated. The flux per
pole derivatives with θr for the three phases are determined and shown in Figure 11.19b. It should be
noted that, as expected, the on-load total flux derivatives are not in phase with the currents, as they, in

1500
Phase A
Phase B
Phase C
Coil MMF (peak value, ampere turns)

1000
500
−500
−1000
−1500
−2 0 2 4 6 8 10 12 14
(a)
0.25
Derivative of the total flux per pole in phase A
Derivative of the total flux (with current and PM)
Derivative of the total flux per pole in phase B

0.2 Derivative of the total flux per pole in phase C
0.15
0.1
0.05
−0.05
−0.1
−0.15
−0.2
−2 0 2 4 6 8 10 12 14
(b)
FIGURE 11.19 (a) Phase currents and (b) total flux derivatives with respect to rotor position θr .
fact, represent the total derivatives, and because current varies, also. That is to say, the total E(A, B, C)
under steady state is proportional to dΦA, B, C /dθr:
− dφ A,B ,C (θr ) N s
E At ,B ,C (θr ) = ⋅ ⋅ n ⋅ 2 ⋅ π ⋅ n V (rms) (11.72)
dθr 3 c
Taking the maximum value of EAt(θr) and dividing it by 2, we may add the voltage phasor RIA (Figure 11.17c)
and get the phase voltage VA:
( E (θ ) ) + R ⋅I
2
VA ≈ t
A r max
+ Rs ⋅ I ⋅ cos γ s
2
⋅ sin 2 γ (11.73)

800
700
Torque (N.m) for various peak MMF

600
500
400
300
200 Torque (N.m) at 1500 ampere turns

Torque (N.m) at 2500 ampere turns
Torque (N.m) at 3500 ampere turns
100 Torque (N.m) at 5000 ampere turns
0
−2 0 2 4 6 8 10 12 14
(a)
700
600
Average torque(N.m)
500
400
300
Average torque(N.m)
200
Pulsation torque (N.m)
100
0
1000 2000 3000 4000 5000 6000
Peak MMF(Nc)(ampere turns)
(b)
0.2
0.15
Tave/Nc
0.1
0.05
0
1000 2000 3000 4000 5000 6000
Peak MMF (Ncl)(ampere turns)
(c)
FIGURE 11.20 Curves representing current-related torque vs. position for various coil peak magnetomotive force
(mmf) (sinusoidal currents in all phases): (a) and (b) average and pulsation torque vs. mmf, (c) average torque/pole
mmf, and (d) cogging torque vs. position.

1.5
Cogging torque (N.m)

0.5
−0.5
−1
−1.5
−1 0 1 2 3 4 5
(d)
FIGURE 11.20 (Continued).
The only component missing is the end-connection leakage inductance component Lend. This inductance
produces an additional voltage drop ω1 · Lend · I (ω1 = 2 · π ·n · Nr = 2 · π ·f1, Figure 11.17b). This way, the
steady-state characteristics may be calculated for various current levels in the machine. The flux per pole in
phase A for a coil peak mmf of 1500, 2500, 3500, and 5000 Aturns demonstrates the effect of saturation on
performance. This is more evident in instantaneous torque Te(t):
Ns  dφ t (θ ) dφ t (θ ) dφ t (θ ) 
Te (t ) = ⋅ nc ⋅  A r ⋅ i A (θr ) + B r ⋅ iB (θr ) + C r ⋅ iC (θr ) (11.74)
3  dθr dθr dθr 
The results are shown in Figure 11.20a through Figure 11.20d. A few remarks are in order.
• Tangential force densities up to 7.108 N/cm2 are practical, but at a low power factor (less than 0.3
in our case).
• The current-related torque pulsations increase with current due to saturation and the presence of
small reluctance torque (Figure 11.20a).
• The torque/pole ampere turns (Figure 11.20c) decreases with current, especially for mmfs above
3500 ampere turns/pole.
• Even for 5000 ampere turns/pole, the PMs are not demagnetized (Figure 11.21).
• The current-related torque pulsations are 3% of the average torque at 3500 ampere turns/pole.
• If the cogging torque peak value is considered, the total torque pulsation/amplitude is 3.01% at
3500 ampere turns. This is not a large value and may be further reduced by introducing a small
harmonic in the current reference waveform to compensate for it.
As the PM height hPM = 2.5 mm, and the airgap g = 0.5 mm, the question arises if, for 5000 ampere
turns, the PMs do not get demagnetized. Through FEM, the flux density distributions in the PM (half
North Pole and half South Pole) lower and upper borders were calculated (Figure 11.21). The PMs do
not get demagnetized mainly because of the large fringing effects in a large magnetic gap structure. In
other words, the not so good coupling between the PMs and the stator coils allows for a high current
loading.

Flux density of PM (telsa) (coil MMF

0.5
500 ampere turns)

S N
0
N S
−0.5
−1
−1.5
−8 −6 −4 −2 0 2 4 6 8
Position (mech. deg.) (rotor position at 0 mech. deg.)
FIGURE 11.21 Flux density on the permanent magnet (PM) upper and lower border vs. position for 5000 A turns/
pole and θr = 0.
While the torque (tangential force) density is important, so is the ratio between stator losses and torque
(W/Nm). For our case, and rated current (jcon = 3.5 A/mm2), the coil loss is as follows:
lcoil ⋅ jcon N s 2 2
Pcoil = 3 ⋅ Rs ⋅ In2 = 3 ⋅ ρco ⋅ ⋅ ⋅ nc ⋅ In = 3 ⋅ kR ⋅ nc2 ⋅ In2
nc ⋅ In 3
(11.75)
855
= 3 ⋅ 2.1 ⋅10−8 ⋅ 0.46 ⋅ 3.5 ⋅106 ⋅ 4 ⋅ = 244.95 W
2
and
lcoil ≈ 2 ⋅ (lstack + 2 ⋅ np ⋅ τ PM ) = 2 ⋅ (0.19 + 2 ⋅ 2 ⋅ 0.1) = 0.46 m (11.76)
The total torque Ten is considered to be 200 Nm, and, thus, the ratio Pcoil /Ten = 244.95/200 = 1.225 W/Nm.
This is a very good result, but the force density is moderate:
2 ⋅Ten 2 ⋅ 200
f tn = = = 2.1⋅104 N /m 2 = 2.1 N /cm 2 (11.77)
π ⋅ Dr2 ⋅ lstack π ⋅ 0.1792 ⋅ 0.1892
This situation corresponds to a pole peak value of mmf of 885 ampere turns. For 5000 ampere turns/
pole, the torque is 677 Nm (7.1085 N/cm2), so the torque increases by 3.385 times, but the losses increase
(5000/855)2 = 34.2 times. Also, the current density for peak torque is 16.95 A/mm2. Such a current density
may be sustained for continuous operation only with liquid stator cooling.
For the peak torque, the efficiency with core and PM eddy current losses neglected, as they are relatively
small, is as follows:
Tek ⋅ Ωn 677 ⋅ 2 ⋅ π ⋅
128
ηk ≈ = 60
= 0.52 (11.78)
Tek ⋅ Ω + pcoil 677 ⋅ 2 ⋅ π ⋅
1228
+ 244.95 ⋅ 23.47
60
For rated conditions, however,
Ten ⋅ Ωn 200 ⋅ 2 ⋅ π ⋅ 128

ηn ≈ = 60
= 0.91 (11.79)
Ten ⋅ Ω + pcoil 200 ⋅ 2 ⋅ π ⋅
1228
60
+ 244.95

The ideal power factor angle ϕ1 is (for nc I 2 = 855 Aturns, from Equation 11.5),
ω1 ⋅ Ls ⋅ I ⋅ 2
ϕ1 = tan −1 = tan −1 2.2 = 65°
E peak
or cosϕ1 = 0.41. Considering the low speed of 128 rpm, the efficiencies are reasonable. Note that the 0.52
efficiency at very heavy current load implies a large voltage drop along stator resistance, to be considered
when calculating the number of turns/coil (nc) for a given inverter voltage value. The power factor is
already low, as in general, the magnetic airgap is small, and LsIs is still large.
11.4.4 FEM Computation of Inductances

As already mentioned, the inductance needed for the circuit model is the transient inductance:
 ∆φ   ∆φ 
LAAt =  A  ; LABt =  B  ; θr , iB , iC = const . (11.80)
 ∆i A   ∆i A 
Therefore, only the current amplitude in phase A was modified by a relatively small value, while the
values of currents in the other phases remain unchanged. When the rotor position is varied, the instan-
taneous values of the currents in all phases vary accordingly. The self-inductance and mutual transient
inductance, LAAt and LABt , respectively, are shown in Figure 11.22a and Figure 11.22b for two-phase current
amplitudes (corresponding to 2500 and 2600 peak mmf/pole). Two remarks are appropriate [4]:
• The variation of transient inductance with θr may be neglected.
• The mutual transient inductance is almost ten times smaller than the self-inductance.
Repeating the FEM computation for a low value of current in the machine, the average self-inductance
and mutual inductance are shown to vary notably with current due to magnetic saturation (Figure 11.23).
To check the necessity of considering all currents present when calculating the transient inductances, the
computation has also been done with current in phase A only. Figure 11.23 shows clearly that a correct
estimation of transient inductance requires all phases to be energized. The same is true when calculating
the torque from emfs and currents.
For the circuit model, we may use either the phase coordinate or the d–q model. Let us first define
the phase coordinate model:
di A,B ,C
i A,B ,C ⋅ Rs − VA,B ,C = E A,B ,C (θ s , is ) − Lt (is ) ⋅ (11.81)
dt
 dφ (θ , i )  N
E A,B ,C = −  A,B ,C r s  ⋅ s ⋅ nc ⋅ 2 ⋅ π ⋅n (11.82)
 d θ r i 3
s
11.4.5 Inductances and the Circuit Model of FRM

First, the synchronous inductance is
Lt (is ) = Lend + [LAAt (is ) − LABt (is )] (11.83)
2 
⋅ i (t ) + iB (t ) ⋅ e j⋅(2⋅π 3) + iC (t ) ⋅ e − j⋅(2⋅π 3)  ⋅ e r r
− j ⋅N ⋅θ
is = (11.84)
3 A

×10−6
3.5
Laa (H, Nc = 1) at 2500 ampere turns

3
2.5
1.5
0.5 Fitted laa (H, Nc = 1) at 2500 ampere turns

Laa (H, Nc = 1) at 2500 ampere turns
0
−2 0 2 4 6 8 10 12 14
(a)
×10−6
1
Lab(H, Nc = 1 per phase)
0.5
−0.5
−1
−2 0 2 4 6 8 10 12 14
(b)
FIGURE 11.22 Self-inductance and mutual inductance with all currents in phases A, B, and C considered: (a) self-
inductance and (b) mutual inductance.
We use here the current space vector absolute value to account for saturation with the A, B, C model.
Also, analytical approximations in total emf and inductance dependence on θr and is are required:
dφ(is ,θr )
dθr
(
= a0 + a1 ⋅ nc ⋅ I s − a2 ⋅ nc2 ⋅ I s2 ⋅[cos(N r ⋅θrA,B ,C ) + a3 ⋅ cos(3 ⋅ N r ⋅θrA,B ,C )] ) (11.85)
where
2 ⋅π
θrA,B ,C = θr + γ 0 (nc ⋅ I s ) − (k − 1) ⋅ , k = 1, 2, 3
3
and
γ 0 (nc ⋅ I s ) = γ θ 0 + c1 ⋅ nc ⋅ I s + c 2 ⋅ nc2 ⋅ I s2
(11.86)
Lt (is ) ≈  Lend + b1 − b2 ⋅ nc2 ⋅ I s2  ⋅ nc2

×10−6
3
2.5
Lt(H, Ncl = 1) 2
1.5
Lt = 2.81e–6–8.89e–11∗Ncl–2.56e–14∗Ncl∗Ncl (H)
1 Lend + Lt (H, Nc = 1)
Lend(H, Nc = 1)
0.5
0
−1000 0 1000 2000 3000 4000 5000 6000
Coil MMF(ampere turns, peak value)
FIGURE 11.23 Average transient inductance dependence on current.
The torque equation Te is as follows:

1
Te = (E A ⋅ i A + EB ⋅ iB + EC ⋅ iC ) ⋅ (11.87)
2 ⋅π ⋅n
The motion equations complete the model:
dn dθr
J ⋅ 2 ⋅π ⋅ = Te − Tload ; =n (11.88)
dt dt
11.4.6 The d–q Model of FRM

The study of transients may be performed in A, B, C coordinates, but for control purposes, the d–q model is
required. To simplify the d–q model, we will neglect the third harmonic in the emfs, as it is below 5%. When
implementing the control for very low torque pulsation, the current iq may be supplied an additional compo-
nent to cancel the already small torque pulsations. The d–q model, in stator coordinates, is simply,
dψ s dψ s di s
is ⋅ Rs − V s = − − j ⋅ωr ⋅ψ s ; = Lt ⋅ (11.89)
dt dt dt
ψ s = ψ d + j ⋅ ψ q ; is = id + j ⋅ iq ;
(11.90)
V s = Vd + j ⋅ Vq
is = id2 + iq2 (11.91)
  2 ⋅π   2 ⋅π  
cos(−θr ) cos  −θr +  cos  −θr −  
2  3   3  
P(θr ) =  
3  2 ⋅π   2 ⋅π  
 sin(−θr ) sin  −θr + 3  sin  −θr − 3  
 

 E (θ , i )
ω r ⋅ ψ d (is )  A r s 
⋅  = P(θr ) ⋅  EB (θr , is )  (11.92)
ω r ⋅ ψ q (is )   
 EC (θr , is )
ψ d (is ) = Ls (is ) ⋅ id + ψ PM
(11.93)
ψ q (is ) = Ls (is ) ⋅ iq
2
Te = ⋅ N ⋅ (ψ d ⋅ iq − ψ q ⋅ id ) (11.94)
3 r
dn dθ
J ⋅ 2 ⋅π ⋅ = Te − TL ; r = 2 ⋅ π ⋅ n ⋅ p (11.95)
dt dt
Notice that Ls is the average normal steady-state inductance. Ls and the transient inductance Lt both
depend on is due to saturation (Figure 11.23):
Vs =
2
3
(
⋅ VA (t ) + VB (t ) ⋅ e ( ) + VC (t ) ⋅ e − j⋅(2⋅π 3)
j ⋅ 2⋅π 3
) (11.96)
As expected, with sinusoidal E(θr), the d–q model will not exhibit θr other than in the Park transformation.
The magnetic saturation is considered according to Equation 11.83 through Equation 11.86.
Example 11.3
Consider the preliminary design of a FRM as generator for a torque Ten = 200 kNm at a speed
nn = 30 rpm for a frequency fn ≈ 100 Hz.
Solution
Making use of Equation 11.63, with ftn = 2.66 N/cm2 with λ = 0.3, we obtain directly the interior
diameter, Dir:
2 ⋅Ten 2 ⋅ 200 ⋅103

Dir = 3 = 3 = 2.5158 m
π ⋅ f tn ⋅ λ π ⋅ 2.66 ⋅104 ⋅ 0.3
The stack length lstack = λ · Dir = 0.3 · 2.5158 = 0.7545 m. This is the same as in Example 11.1 for
the three-phase TFM.
For 100 Hz, the number of rotor salient poles Nr is as follows (Equation 11.56):
fni 100
N ri = ≈ 30 = 200
nn 60
We are choosing the same PM material, NdFeB, with Br = 1.3 T and µrem = 1.05 at 100°C.
The number of stator large poles (coils) Ns and the number of PM poles npp per stator large pole
are as follows:
 2
N s ⋅  2 ⋅ npp +  = 2 ⋅ N r
 3

and thus, with Ns = 6 · k and npp = 4, we obtain Ns = 48 for Nr = 208. Consequently, the PM pole
width τPM is as follows:
π ⋅ Dir π ⋅ 2.515
τ PM = = = 0.01898 m
2 ⋅ Nr 2 ⋅ 208
and the frequency fn = nn · Nr = 30/60 · 208 = 104 Hz.

As the configuration is more rugged (it resembles the SRM topology), we may adopt a slightly
smaller airgap of g = 3 mm.
We then assign the PM radial height hPM a value, which is about three times the airgap hPM = 3 · g =
9 mm, to provide for high enough ideal maximum PM flux density in the airgap:
hPM 9
BgPMaxi = Br ⋅ = 1.3 ⋅ = 0.975 T
hPM + g 9+3
The actual maximum PM flux density in the airgap BgPMax is as follows:
1
BgPMax = BgPMaxi ⋅ (11.97)
(1 + k fringe ) ⋅(1 + ks )
with the fringing coefficient kfringe = 2.33 and the magnetic core contribution ks = 0.1, BgPMax becomes
1
BgPMax = 0.975 ⋅ = 0.266 T
(1 + 2.33) ⋅ (1 + 0.1)
Note that this is about the same as for TFM in Example 11.1.
This seems a small value, but at this small pole pitch, it may be justified, given the large gap imposed
by the large rotor diameter.
The peak value of the phase emf, with Ns coils in series, and each having nc turns, is as follows
(Equation 11.68):
Ns π ⋅ Dr
Em = ⋅ n ⋅ 2 ⋅ π ⋅ n ⋅ lstack ⋅ ⋅ npp ⋅ BgPMax
3 c 2
48 30 2.515
= ⋅ 2 ⋅ π ⋅ ⋅ 0.7545 ⋅ π ⋅ ⋅ 4 ⋅ 0.266 ⋅ nc = 159.2 ⋅ nc
3 60 2
The RMS ampere turns per coil nc In for rated torque are as follows (Equation 11.70):
2 nc 2 ⋅ π ⋅ n 2 ⋅ 200 ⋅103 ⋅ 2 ⋅ π ⋅1 2
nc ⋅ In = ⋅ ⋅Ten ⋅ = = 1865 Aturns(rms)/coil (11.98)
3 2 Em 3 ⋅ 2 ⋅159.2
The area required in the slot to accommodate two coils Aslot is
2 ⋅ nc ⋅ In 2 ⋅1865
Aslot = = = 1.036 ⋅10−3 m 2 = 1036 mm 2
jcon ⋅ k fill 9 ⋅106 ⋅ 0.4
where jcon = 9 A/mm2 and kfill = 0.4.

bps ª 3t PM = 54 mm
ª12.52 mm
hys = 30 mm ª 1.5t PM
t PM = 18 mm
hsu
240°
408
hos ª t PM/4 ª 5 mm R 1244.5 mm
ª18.98 mm
R 1257.5 mm
R 1324.5 mm
R 1275.5 mm
FIGURE 11.24 Flux reversal machine (FRM) stator pole geometry.
The slot bottom width b1 is, approximately,
2 ⋅ π ⋅ R1 2 ⋅ π ⋅1284.5
b1 = − bps = − 54 = 144 mm
Ns 48
The pole body width bps is approximated having in mind the low useful PM flux density in the
airgap. The same is valid for the stator yoke height hys, which is chosen as 30 mm for mechanical
rather than magnetization constraints (Figure 11.24).
The rotor poles (Figure 11.14) should be tall enough to create enough saliency for the rather large
magnetic airgap (hPM + g = 12 mm).
This only shows that even a slot height of hsu = 10 mm would provide enough room for the two
coils that require 1036 mm2 in all (Figure 11.25).
The machine resistance per phase Rs is
Ns l ⋅ n2 48 2.3 ⋅10−8 ⋅1.77 ⋅ nc2

Rs = ⋅ ρco ⋅ coiln ⋅I c = ⋅ 1865 = 3.14 ⋅10−3 ⋅ nc2
3 c 3
j
con
6
9⋅10
where the coil average length lcoil is
2 ⋅ npp ⋅ τ PM − bps
lcoil = 2 ⋅ lstack + 2 ⋅ bps + π ⋅
2
π
= 2 ⋅ 0.7545 + 2 ⋅ 0.054 + ⋅ (2 ⋅ 4 ⋅ 0.019 − 0.054) = 1.77 m
2

2 t PM
>(4 − 5) (hPM + g)
(0.7 – 1)t PM
hyr = hys
FIGURE 11.25 Rotor salient-pole geometry.
The copper losses for rated current (torque) are as follows:
pcon = 3 ⋅ Rs ⋅ In2 = 3 ⋅ 3.14 ⋅10−3 ⋅ (nc ⋅ In )2 = 9.42 ⋅10−3 ⋅18652 = 32.76 kW
Note that the simplicity of the manufacturing process in FRM is paid for dearly in copper losses,
which increase from 13.70 kW in the TFM to 32.76 kW in the FRM.
As there are no aluminum carriers to hold the stator core, their eddy current losses are also absent
in FRM, but still the copper losses of FRM are larger in a machine with the same size and torque.
The machine inductance Ls is again made up of two components: the airgap inductance Lm and the
leakage inductance Lsl. The leakage inductance is moderate, as the slot aspect ratio hsu/b1 is small,
and thus, we concentrate on the airgap inductance:
Ns τ PM ⋅ npp 1 + k f
Lm ≈ ⋅ µ0 ⋅ nc2 ⋅ ⋅ ⋅l (11.99)
3 g + hPM 1 + ks stack
kf is a fringing effect coefficient kf < 0.2, which accounts for the large airgap above the rotor interpole
contribution to the coil self-flux linkage:
48 0.01898 ⋅ 4 1 + 0.2
Lm = ⋅1.256 ⋅10−6 ⋅ ⋅ ⋅ 0.7545 ⋅ nc2 = 0.1046 ⋅10−3 ⋅ nc2
3 (3 + 9) ⋅10−3 1 + 0.1
Considering that the leakage inductance represents 10% of Lm, Ls = (1 + 0.1) · 0.115 · 103 · nc2. The
power factor angle for pure Iq control is, again (Equation 11.5),
 ω ⋅L ⋅I ⋅ 2   2 ⋅ π ⋅104 ⋅ 0.115 ⋅10−3 ⋅ n2 ⋅ I ⋅ 2 

ϕ1 = tan −1  1 s −1
 = tan 
c −1
 = tan 1.2375 = 51°
 Em   159.6 ⋅ nc 
cos ϕ1 = 0.6285
Remember that the power factor for the same output conditions was 0.933 for the TFM. We also
have to point out that the PM flux fringing ratio of 0.33 (that is, 33% of ideal PM flux reaches
the coils) is a bit too severe for the FRM, because the rotor poles are one pole pitch apart
tangentially, while for the TFM, the axial distance between neighboring U- and I-shape cores is

lPM
hPM
hsu
Wos = lPM bsu

2t
Wpr
Wts
2t
W
lif = 2 τ + os
3 2
FIGURE 11.26 Flux reversal generator (FRG) with stator permanent magnet (PM) flux concentration.
two to three times smaller. Finally, the PM and airgap height were reduced by 25% for FRM. In-
depth FEM studies are required to document the best solution of the two. However, with four
airgaps per coil, TFM is expected to show smaller inductance.
Example 11.4: FRM with Stator PM-Flux Concentration

Let us consider again a direct-driven wind generator system operating again at Ten = 200 kNm
and n = 30 rpm, fn ≈ 100 Hz. The design of the magnetic circuit, losses, and power factor angle for
pure Iq control are required.
Solution
The machine geometry (Figure 11.6) is represented with one pole only (Figure 11.26).
Making use of the PM flux concentration, we adopted a larger force density ftn than in previous
design examples to offset the effect of the inductance increase due to the fringing effect: ftn = 6 N/cm2.
To yield the same diameter as before, λ = lstack/Dis is decreased accordingly to λ = 0.133:
2 ⋅Ten 2 ⋅ 200 ⋅103

Dis = 3 = 3 = 2.515 m
π ⋅ f tn ⋅ λ π ⋅ 6 ⋅104 ⋅ 0.133
The stack length is, however,
lstack = λ ⋅ Dis = 0.133 ⋅ 2.515 = 0.3345 m

The number of rotor salient poles for around 100 Hz should be around Nr = 200, calculated as
follows (Figure 11.26):
 2 ⋅τ 
2 ⋅ N r ⋅τ = 6 ⋅ k ⋅  4 ⋅τ + = 28 ⋅ k ⋅τ (11.100)
 3 
with Nr = 186; k = 14; f1 = Nr · n = 186 · 30/60 = 93 Hz.

The airgap g = 2.0 mm, and the pole pitch τ is
π ⋅ Dis π ⋅ 2.515
τ= = = 0.02014 m
2 ⋅ Nr 2 ⋅186
The slot opening, Wos, equal to the PM thickness, lPM, is considered to be 8 mm. So, the stator tooth Wts
= τ − Wos = 20 − 8 = 12 mm. The rotor salient pole Wpr may be as wide as the pole pitch or smaller:
Wpr = (0.8 ÷ 1.0) ⋅ τ (11.101)
The ideal PM flux density in the airgap BgPMax is now as follows:
Br
BgPMaxi = µrec
(11.102)
+ ⋅ τ2⋅g
Wts
hPM µ0 PM
This expression is derived from
Bm ⋅ hPM = Bg ⋅ Wts
Bq
Hm ⋅ lPM + ⋅2⋅ g = 0 (11.103)
µ0
Bm = Br + µrec ⋅ Hm
with Br = 1.3 T and µrec = 1.05 · µ0.

The fringing effect will reduce the ideal PM airgap flux density to a smaller value, as in previous cases:
BgPMaxi
BgPMax = (11.104)
(1 + k fringe ) ⋅(1 + ks )
The fringing coefficient depends again on the airgap g/Wts ratio, lPM/g ratio, and on the degree of
saturation. With kfringe = 1.5, ks = 0.2, we set the actual airgap flux density BgPMax at a reasonable
value, say BgPMax = 0.7 T.
In this case, from Equation 11.104,
BgPMaxi = 0.7 ⋅ (1 + 1.5) ⋅ (1 + 0.2) = 2.1 T
This is a high value, but remember that it is only a theoretical one. From Equation 11.102, we may
size the PM height hPM:
1.3
2.1 = 1.05⋅µ0 2⋅20
; hPM = 120 mm
+ ⋅
12
hPM µ0 8

The PM maximum flux per coil turn, ΦPMaxc , is as follows:
φ PMaxc = BgPMax ⋅ Wts ⋅ lstack = 0.7 ⋅ 0.012 ⋅ 0.3345 = 2.81 ⋅10−3 Wb (11.105)
The maximum flux linkage ψPMphase is
ψ PMphase = 2 ⋅ k ⋅φ PMaxc ⋅ nc = 2 ⋅14 ⋅ 2.81⋅10−3 ⋅ nc = 0.007867⋅nc (11.106)
The emf per phase (peak value) Em is
Em = 2 ⋅ π ⋅ f1 ⋅ψ PMphase = 2 ⋅ π ⋅ 93 ⋅ 0.07867 ⋅ nc = 45.949 ⋅ nc (11.107)
The ampere turns per slot ncIn may be calculated from Equation 11.98:
2 ⋅ nc 2 ⋅ π ⋅ n 2 ⋅ nc 200 ⋅103 ⋅ 2 ⋅ π ⋅ 30/60

nc ⋅ In = ⋅Ten ⋅ = ⋅ = 6.4589 ⋅103 Aturns/coil
3⋅ 2 Em 3⋅ 2 45.949 ⋅ nc
The slot width is about equal to the pole pitch (Figure 11.26). Wsu = (1 to 1.2) · τPM ≈ 1.1 · 20 = 22
mm. This way, the stator tooth average width is around half the pole pitch τPM.
As the total slot height hsu ≈ hPM – 2/3 · Wts – 0.005 = 103 · 10−3 – 2/3 · 12 · 10−3 – 5 · 10−3 = 90 · 10−3 m,
the current density required to host the coil jcon is as follows:
nc ⋅ In 6458.9
jcon = = = 6.524 A/mm 2
hsu ⋅ Wsu ⋅ k fill 90 ⋅ 22 ⋅ 0.5
The stator coil resistance is
l 2.3 ⋅10−8 ⋅ 0.818 2

Rsc = ρco ⋅ ncoil
⋅I
⋅ nc
2
= 6458.9
⋅ nc = 1.9 ⋅10−5 ⋅ nc2
c n
jcon 6.524⋅106 (11.108)
lcoil = 2 ⋅ lstacck + 4 ⋅ τ PM + π ⋅ Wsu = 2 ⋅ 0.3345 + 4 ⋅ 0.020 + π ⋅ 0.022 = 0.8818 m
For the entire phase,
Rs = 2 ⋅ k ⋅ Rsc = 2 ⋅14 ⋅1.9 ⋅10−5 ⋅ nc2 = 0.532 ⋅10−3 ⋅ nc2
The stator copper losses are then
Pcon = 3 ⋅ Rs ⋅ In2 = 3 ⋅ 0.532 ⋅10−3 ⋅ 6458.92 = 66.58 kW
A few remarks are in order:

• The FRM with stator PM flux concentration cannot appropriately take full advantage of the
principle of PM flux magnification.
• The machine size was reduced (the stack length was 0.7545 m for the surface PM stator FRM). This
reduction in size is paid for by larger copper losses (66.58 kW for an input power of 628 kW). There
is notably more copper in this machine.
• It is possible to redo the design for a smaller force density, but for larger stack length, the same
interior stator diameter Dis is needed to notably reduce the copper losses. Still, the inductance
seems to be higher as each coil “entertains” two airgaps.
• The FRM with stator PM flux concentration seems to be restricted to a smaller number of poles
per rotor and small torque machines, where the fabrication costs may be reduced due to its
relatively more rugged topology.

11.4.7 Notes on Flux Reversal Generator (FRG) Control

With its easy rotor skewing, an FRM can produce a rather sinusoidal emf and is thus eligible for field orientation
or direct power (torque) control as any PM synchronous generator (see Chapter 10 for details and results).
Example 11.5
Let us now consider the FRG with rotor PM flux concentration (Figure 11.27) illustrated here with
only a few poles (Figure 11.27). For the same data as in Example 11.3, prepare an adequate design.
Solution
The two stator cores and the rotor core are all made up of standard stamped laminations. Only the
external rotor is provided with windings to save room in the rotor (Figure 11.7) and thus make it
more rugged mechanically. Each stator large pole now contains 2 · np + 1 poles and 2 · np interpoles.
It is clearly visible that for maximum flux per large pole, all the PMs in the rotor are active. This
is a better PM utilization in addition to PM flux concentration.
First, let us keep ftn = 6 N/cm2 and λ = 0.133. Then, Dis (stator interior diameter) stays the same as
in Example 11.3; Dis = 2.515 m.
External stator
A'
C'
3tPM
tPM
g = 1.5 mm S
SS NN
>5g NN
SS
A
NN
SS
2/3tPM g = 1.5 mm
NN
B >5g
SS
2 tPM
N
SS N
2 tPM
2/3tPM
2tPM
S NN
Interior stator
Rotor
tPM hPM
lPM
Wtr
lPM
≤ 0.4
tPM
(1−1.2) mm bridge
FIGURE 11.27 Flux reversal machine (FRM) with rotor permanent magnet (PM) flux concentration and dual stator.

The number of stator poles 6 · k is now as follows:
 2
6 ⋅ k ⋅  2 ⋅ npp + 1 + 2 ⋅ npp +  = 2 ⋅ N r ≈ 400 (11.109)
 3
with npp = 2, k = 7, and 2 · Nr = 406 poles.

So, the PM pole pitch τPM = π · Dis /2 · Nr = π · 2.515/406 = 0.01945 m.
The ideal PM flux density in the airgap BgPMaxi is as follows:
Br
BgPMaxi = Wtr µrec 2⋅g
; Br = 1.3 T ; µrec = 1.05 ⋅ µ0 (11.110)
2⋅hPM
+ µ0
⋅l
PM
In contrast to Equation 11.102, the factor 2 · hPM instead of hPM is used in Equation 11.110, because
two PM magnets cooperate in the rotor tooth Wtr .
With the actual airgap PM flux density BgPMax = 0.9 T, the ideal airgap PM flux density BgPMaxi is as
follows:
BgPMaxi = BgPMax ⋅ (1 + k fringe ) ⋅ (1 + ks ) = 0.9 ⋅ (1 + 1.5) ⋅ (1 + 0.2) = 2.7 T (11.111)
With this value in Equation 11.108, the ratio hPM/Wtr is obtained (the PM thickness lPM = 0.4 · τPM =
0.4· 20 = 8 mm):
1.3
2.7 = 1.05⋅µ0
(11.112)
+ ⋅ 83
Wtr
2⋅hPM µ0
It follows that 2 · hPM /Wtr = 11.4.

Consequently, the PM radial height hPM = 11.4 · Wtr /2 = 5.7 · 12 = 68.40 mm.
This is a notable reduction in PM weight for a better effect in comparison with Example 13.3. The
maximum PM flux linkage per phase is obtained by adapting Equation 11.105 and Equation 11.106
for the case in point:
ψ PMphase = 2 ⋅ k ⋅ Wts ⋅ BgPMax ⋅ lstack ⋅ (2 ⋅ npp + 1) ⋅ nc

(11.113)
= 2 ⋅ 7 ⋅ 0.012 ⋅ 0.9 ⋅ 0.3345 ⋅ (2 ⋅ 2 + 1) ⋅ nc = 0.2528 ⋅ nc
Consequently, the peak value of emf per phase Em is now calculated from Equation 11.106:
30
Em = 2 ⋅ π ⋅ f1 ⋅φ PMphase = 2 ⋅ π ⋅ ⋅ 203 ⋅ 0.2528 ⋅ nc = 161.19 ⋅nc (11.114)
60
By now, we have all the data necessary to calculate the rated mmf per coil ncIn (Equation 11.98):
2 2 ⋅π ⋅n 2 200 ⋅103 ⋅ 2 ⋅ π ⋅ 30/60

nc ⋅ In = ⋅ nc ⋅Ten ⋅ = ⋅ nc ⋅ = 1842.1 Aturns
3⋅ 2 Em 3⋅ 2 161.19 ⋅ nc
A current density jcon = 6 A/mm2 may be adopted, as the slot useful area Aslot is
2 ⋅ nc ⋅ In 2 ⋅1842.1
Aslot = = = 1.535 ⋅10−3 m 2 = 1535 mm 2
jcon ⋅ k fill 6 ⋅106 ⋅ 0.4
There is plenty of room to locate such a slot with low height and, thus, with low slot leakage
inductance contribution.

The phase resistance Rs (Equation 11.108) is as follows:
l 2 ⋅ 7 ⋅ 2.3 ⋅10−8 ⋅1.1045 2

Rs = 2 ⋅ k ⋅ ρco ⋅ ncoil
⋅I
⋅ nc
2
= 1842.1
⋅ nc = 1.1684 ⋅10−3 ⋅ nc2
c n
jcon 6⋅106
π ⋅ Dis 2.515
lcoil ≈ 2 ⋅ lstack + 2 ⋅ = 2 ⋅ 0.3345 + 2 ⋅ π ⋅ = 1.045 m
6⋅k 6 ⋅7
The copper losses pcon are
pcon = 3 ⋅ Rs ⋅ In2 = 3 ⋅1.1684 ⋅10−3 ⋅1842.12 = 11.7925 kW
This displays less copper losses than for the TFM (13.7 kW), while the machine axial total length
is about half in the TFM for the same diameter. There is still one more problem: the power factor.
For maximum Iq current control, the PM flux is zero in that phase; thus, the airgap inductance is
rather large, as the PMs “do not stay in the way” of coil mmf flux.
Consequently,
Wtr ⋅ lstack
Lm ≈ µ0 ⋅ 2 ⋅ k ⋅ nc2 ⋅ (2 ⋅ npp + 1) ⋅
2 ⋅ g ⋅ (1 + ks )
0.012 ⋅ 3345
= 1.256 ⋅10−6 ⋅ 2 ⋅ 7 ⋅ nc2 ⋅ (2 ⋅ 2 + 1) ⋅ = 0.98 ⋅10−4 ⋅ nc2 (11.115)
2 ⋅1.5 ⋅10−3 ⋅ (1 + 0.2)
Ls = (1 + 0.1) ⋅ Lm = 1.078 ⋅10−4 ⋅ nc2
Finally, the power factor angle ϕ1 is as follows (with ncIn = 1842.1 Aturns):
 ω ⋅L ⋅I ⋅ 2   203 1.078 ⋅10 ⋅ nc ⋅ In ⋅ 2 

−4 2
ϕ1 = tan −1  1 s n −1
 = tan  2 ⋅ π ⋅ ⋅ −1
 = tan 1.107 = 48°
 Em   2 161.2 ⋅ nc 
cosϕ1 = 0.67
Let us try to reduce copper losses further by reducing the number of stator coils with k = 2 (four
coils per phase). We obtain np = 8 and 2 · N′r = 6 · k (4 · np + 1 + 2/3) = 404. The PM pole pitch
τPM remains about the same at 0.01947 m.
Repeating the design, the PM flux per phase (Equation 11.113) is
(2 ⋅ npp
′ + 1) ⋅ k ′ (2 ⋅ 8 + 1) ⋅ 2
ψ PMphase = 0.2528 ⋅ nc ⋅ = 0.2528 ⋅ nc ⋅ = 0.2455 ⋅ nc
(2 ⋅ npp + 1) ⋅ k (2 ⋅ 2 + 1) ⋅ 7
and the emf (peak value) E′m (Equation 11.114) is as follows:
202
Em′ = 2 ⋅ π ⋅ N r ⋅ n ⋅ φ PMphase = 2 ⋅ π ⋅ ⋅ 0.2455 ⋅ nc = 155.76 ⋅nc
2
So, the coil mmf (ncIn)′ is
161.19
(nc ⋅ In )′ = 1842.1 ⋅ = 1906.26 Aturns(rms)
155.76

With the same jcon = 6 A/mm2, the stator resistance Rs′ becomes
′ ⋅ nc2
lcoil 2 ⋅ 2 ⋅ 2.3 ⋅10−8 ⋅ nc2 ⋅1.985
Rs′ = 2 ⋅ k ⋅ ρco ⋅ (nc ⋅In )'
= 1906.26
= 0.575 ⋅10−3 ⋅ nc2
jcon 6⋅106
2 ⋅ π ⋅ Dis 2 ⋅ π ⋅ 2.515
′ ≈ 2 ⋅ lstack +
lcoil = 2 ⋅ 0.3345 + = 1.985 m
6⋅k 12
The copper losses p′con are now
′ = 3 ⋅ Rs′ ⋅ (nc ⋅ In )′ 2 = 3 ⋅ 0.575 ⋅10−3 ⋅1906.262 = 6.266 kW (about 1%)

pcon
Note that the copper losses were reduced, by half, at the price of thicker stator and iron yokes. Unfor-
tunately, the inductance stays about the same, so the power factor stays about the same, around 0.67.
The following are some final remarks:
• In comparison with the FRM with stator surface PMs, the machine axial length was reduced by
half, at the same interior and outer diameter, smaller copper losses, but same power factor.
• The PM weight is only slightly larger than in the TFM with rotor PM flux concentration.
• In comparison with the TFM, the machine length is half at about the same interior and outer
stator diameter, the copper losses are only 30% less, but the power factor is 0.67 in comparison
with the 0.933 of TFM.
• Reducing the number of stator coils could be another way to further improve FRM performance.
• The FRM is considered more manufacturable than the TFM.
• It could be argued that the double-sided TFM with flux concentration can produce even better
results. This is true, but in a less manufacturable topology.
• Though in the numerical examples of this chapter the fringing flux coefficients were chosen
conservatively low (0.3 to 0.4), full FEM studies are still required to validate the performance
claims to precision, on a case-by-case basis.
11.5 Summary
• This chapter investigates two special PM brushless generators with large numbers of poles and
nonoverlapping stator coils.
• One is called the TFM, and it uses circular single coils per phase. The PMs are on the rotor or on
the stator, with or without PM flux concentration. The flux paths are generally three dimensional.
• The other one is called the FRM, and it makes use of the switched reluctance machine standard
laminated core. The PMs may be placed on top of stator poles in a large even-numbered 2 · np
with alternate polarity.
• In both, TFM and FRM, the PM flux linkage in the stator coils reverses polarity, but, generally,
for the same diameter and pole pitch (number of poles per periphery), the TFM coil embraces
more alternate PM poles and is destined for better torque magnification.
• As the PM pole pitch decreases with a large number of poles, the PM fringing flux increases to
the point that the latter becomes overwhelming. In essence, τPM > g + hPM, where g is the mechanical
gap, and hPM is the PM radial thickness to secure a less than 66% reduction of flux in coil due to
fringing. Reductions of only 35 to 40% are reported in Reference 2 for a small machine.
• The large fringing translates into poor usage of PM and core material and high current loading
for good torque density, higher copper losses, and lower power factor.
• The torque in Newton meters per watts of copper losses and the power factor cos ϕ1 are defined
as performance indexes, which are independent of speed (frequency), to characterize TFM and
FRM. Also, the total cost has to be considered, as these machines use less copper but more iron
and PM materials than standard machines.

• The chapter develops preliminary electromagnetic models for TFM and FRM and then uses the
same specifications: a 200 kNm, 30 rpm, 100 Hz generator in four different designs — a TFM
with surface PMs and three FRM designs (one with surface PM stator, one with stator PM flux
concentration, and one with rotor PM flux concentration).
• All four topologies could be designed for the specifications, but the TFM, for the same volume,
was slightly better than FRM with the surface PM stator in power factor. However, a reduction in
volume by one half, for about the same copper losses and lower power factor, was obtained with
the PM rotor flux concentration FRM. It may be argued that the TFM could also be built with
PM flux concentration. This is true, but the already low manufacturability of TFM is further
reduced this way.
• It is too early to discriminate between TFM and FRM, as one seems slightly better in torque/
copper losses for given volume, but the other is notably more manufacturable, for a very high
number of poles.
• In terms of control, the FRM is easily capable (through skewing) of controlling sinusoidal emfs
and is thus directly eligible for standard field orientation or direct power (torque) control (see
Chapter 10 on this issue).
• New PM generator/motor configurations that depart from the standard PM synchronous generators/
motors (Chapter 10) are still being proposed in the search for better very low speed direct-driven
generator systems with full power electronics control.
References
1. H. Weh, and H. May, Achievable force densities, Record of ICEM-1996, München, Germany, vol. 3,
1996, pp. 1101–1111.
2. J. Luo, S. Huang, S.Chen, and T.A. Lipo, Design and experiments of a novel axial circumferential
current permanent magnet machine (AFCCM) with radial airgap, Record of IEEE–IAS, 2001,
Annual Meeting, 2001.
3. G. Henneberger, and I.A. Viorel, Variable Reluctance Electrical Machines, Shaker-Verlag, Aachen,
2001, chap. 6.
4. I. Boldea, J. Zhang, and S.A. Nasar, Theoretical characterization of flux reversal machine in low
speed drives — the pole-PM configuration, IEEE Trans., IA-38, 6, 2002, pp. 1549–1557.
5. S.E. Rauch, and L.J. Johnson, Design principles of flux switch alternator, AIEE Trans., part III,
1955, pp. 1261–1268.
6. A. Njeh, A. Masmoudi, and A. El Antably, 3D FEA based investigation of cogging torque of claw
pole transverse flux PM machine, Record of IEEE–IEMDC, vol. I, 2003, pp. 319–324.
7. M. Bork, R. Blissenbach, and G. Henneberger, Identification of the loss distribution in a transverse
flux machine, Record of ICEM, vol. 3, Istanbul, Turkey, 1998, pp. 1826–1831.
8. I. Boldea, E. Serban, and R. Babau, Flux-reversal stator-PM single phase generator with controlled
DC output, Record of OPTIM, vol. 4, Poiana Brasov, Romania, 1996, pp. 1124–1134.
9. R. Deodhar, S. Andersson, I. Boldea, and T.J.E. Miller, The flux reversal machine: a new brushless
doubly-salient PM machine, IEEE Trans., IA-33, 4, pp. 925–934.
10. I. Boldea, C.X. Wang, and S.A. Nasar, Design of a three-phase flux reversal machine, Electr. Mach.
Power Syst., 27, 1999, pp. 849–863.
11. C.X. Wang, S.A. Nasar, and I. Boldea, Three-phase flux reversal machine, Proc. IEE, EPA-146, 2,
1999, pp. 139–146.

5715ch11 PDF

Uploaded by

Copyright:

Available Formats

5715ch11 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

5715ch11 PDF

Uploaded by

Copyright:

Available Formats

11

Transverse Flux and

11.1 Introduction ..................................................................... 11-1

© 2006 by Taylor & Francis Group, LLC

© 2006 by Taylor & Francis Group, LLC

Es ≈ −W1 ⋅ p12 ⋅ Bg τ ⋅ lstack ⋅ 2 ⋅ π ⋅ n = −W1 ⋅ Bg ⋅ τ ⋅ p1 ⋅ lstack ⋅ ω1 ⋅ sin pθr (11.2)

© 2006 by Taylor & Francis Group, LLC

© 2006 by Taylor & Francis Group, LLC

A typical three-phase FRM is shown in Figure 11.5a and Figure 11.5b.

© 2006 by Taylor & Francis Group, LLC

11.2 The Three-Phase Transverse Flux Machine (TFM): Magnetic

© 2006 by Taylor & Francis Group, LLC

© 2006 by Taylor & Francis Group, LLC

2 Ryr Ryi + Ryu

2 ⋅(hslot + hyi ) lstack + lslot

© 2006 by Taylor & Francis Group, LLC

φ PMax = b yu ⋅ lstack ⋅ Bcu = b yu ⋅ hyu ⋅ B yu = hyi ⋅ bi ⋅ B yi

µ(1) = µi (0) + c(µi (1) − µi (0)); c = 0.2 − 0.3 (11.12)

such as to speed up convergence.

FPM (qer) 1 (bPM + bu ) p =q

© 2006 by Taylor & Francis Group, LLC

© 2006 by Taylor & Francis Group, LLC

Es ⋅ I1 ⋅ cos γ 1 W1 ⋅ p 2 ⋅ φ PMax ⋅ I1 ⋅ cos γ 1

φ PM (θr ) = φ PMax ⋅ cos( p ⋅θr ) (11.17)

11.2.1 The Phase Inductance Ls

11.2.2 Phase Resistance and Slot Area

π ⋅(Dis + 2 ⋅ hyi + hs ) ⋅W12

© 2006 by Taylor & Francis Group, LLC

In our case, with λ = 0.05, ft = 2.66 N/cm2:

The stack length is as follows:

© 2006 by Taylor & Francis Group, LLC

Ten ⋅ 2 200 ⋅103 ⋅ 2

This is a reasonable value.

lslot = 2(g + hPM) = 32 mm (11.32)

the slot height hs is

2.3 ⋅10−8 ⋅ π ⋅(2.515 + 2 ⋅ 0.083 + 0.03372)

Pcon = 3 ⋅ Rs ⋅ I12 = 3 ⋅ 0.3206 ⋅10−3 ⋅ 38852 = 13.70 kW (11.35)

© 2006 by Taylor & Francis Group, LLC

The electromagnetic power, Pelm, is

Also, from Equation 11.20, Lse, the leakage inductance is as follows:

The emf E (rms value) is as follows:

Ten ⋅ 2 ⋅ π ⋅ n 628 ⋅103 209.3 ⋅103

The ideal power factor angle ϕ 1 (for γ = 0) is

ω1 ⋅ Ls ⋅ I1  2 ⋅ π ⋅100 ⋅ 0.844 ⋅10−5 

11.3 TFM — the d–q Model and Steady State

© 2006 by Taylor & Francis Group, LLC

TABLE 11.1 Example Design Summary

© 2006 by Taylor & Francis Group, LLC

−jω rΨs Es −Rs Is −jω rΨs

© 2006 by Taylor & Francis Group, LLC

From Figure 11.13a,

Es = E(rms) ⋅ 2 = Vs ⋅ cosϕ1 + Rs ⋅ I s (11.46)

75.59 ⋅W1 = 476 ⋅ 0.933 + 0.454 ⋅10−3 ⋅W12 ⋅ I1 ⋅ 2 (11.47)

P1elm − pcon P2 628 − 13.70 614.33 kW

3 ⋅ 584 ⋅ 647 ⋅ 0.933 609

(Vs + Rs ⋅ I1)2 + ω12 ⋅ L2s ⋅ I12 = Es2 (11.52)

Vs′ = Es2 − ω12 ⋅ L2s ⋅ I s2 − Rs ⋅ I s

− 0.3026 ⋅10−3 ⋅ W12 ⋅ I1 ⋅ 2 = 396.48 − 9.945 = 386.63 V

(P2′)cos ϕ =1 = 3 ⋅ 473.40 ⋅ 647.5 = 530.29 kW (11.55)