Design of a Three-Phase Flux Reversal Machine

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Electric Machines &Power Systems

ISSN: 0731-356X (Print) (Online) Journal homepage: www.tandfonline.com/journals/uemp19

Design of a Three-Phase Flux Reversal Machine

Ion Boldea, Congxiao Wang, Syed A. Nasar

To cite this article: Ion Boldea, Congxiao Wang, Syed A. Nasar (1999) Design of a Three-
Phase Flux Reversal Machine, Electric Machines &Power Systems, 27:8, 849-863, DOI:
10.1080/073135699268885

To link to this article: https://doi.org/10.1080/073135699268885

Published online: 29 Oct 2010.

Submit your article to this journal

Article views: 143

View related articles

Citing articles: 4 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=uemp20
Electric Machines and Power Systems, 27:849–863, 1999
Copyright ° c 1999 Taylor & Francis, Inc.
0731-356X / 99 $12.00 + .00

Design of a Three-Phase Flux Reversal M achine

ION BOLDEA
University Politehnica
Timisoara, Romania

CONGXIAO WANG
SYED A. NASAR
University of Kentucky
453, Anderson Hall
Lexington KY40506 USA

A comprehensive design of a novel doubly-salient, stator-permanent magnet


(PM) three-phase ux reversal machine (FRM) is presented in this paper.
FRM has a robust and easy-to-build structure and is suitable for high-speed
applications. Detailed design procedures and formulae are given. All of the di-
mensions of the FRM are derived. Basic parameters are calculated. The most
important aspect of this design is to consider the wide operating speed range.
The work presented may serve as a guideline for the general principle of design
on three-phase FRM. Furthermore, the methodology and formulae introduced
in this paper are also applicable to other types of PM and/or reluctance ma-
chine.

I Introduction
FRM is a doubly-salient machine with PMs on the stator. PM ux linkage in the
stator phase concentrated coils reverses polarity with the traveling of the rotor.
Although the machine operating on this principle was introduced as early as 1955
[1], it was only in a single-phase conŽguration. The poor usage of rotor volume, the
stator vibration, and di culties in manufacturing the stator are major obstacles to
its widespread application. In an eŒort to improve the torque density, reduce stator
vibrations, and simplify the manufacturing process, an improved single-phase FRM
was introduced in [2] and thoroughly analyzed using FEA in [3]. It is to be noted
that other single-phase [4] or three-phase [5] stator-PM doubly salient machines
have been proposed. However, the ux linkage in any stator coil varies from a
minimum to a maximum value without changing polarity. Flux reversal tends to
produce faster ux variation with rotor position than ux pulsation, and thus larger
torque for a given stator MMF is expected. In an eŒort to achieve a maximum
utilization of materials and to obtain a reasonably low torque pulsation and a high
e ciency for both generating and motoring modes of operation, comprehensive
design of an FRM in its three-phase conŽguration is presented in this paper.

Manuscript received in Žnal form on September 16, 1998.


Address correspondence to Syed A. Nasar

849
850 I. Boldea et al.

II Design Procedure
A Basic ConŽguration
A general rule regarding the stator and rotor number of poles N s , N r , and the
number of phases m for the multiphase FRM is given by
Ns m
= . (1)
Nr m+ 1
Therefore, N s =N r = 3/ 4, 6/ 8, 12/ 16 are some candidates for a three-phase FRM.
From experience, we choose 6/ 8 conŽguration, as shown in Figure 1(a). The coil
terminal connections are shown in Figure 1(b).
The eight rotor poles and interpoles are equally spaced. Thus, their span ¿ is
given by
¼D r
¿= , (2)
16

Figure 1. (a) Cross section of the three-phase FRM. (b) Interconnection of stator
coils.
Design of a Three-Phase Flux Reversal Machine 851

Table 1
Design speciŽcations

Generator Motor

42V (DC output) Speed range: 180–18,000 rpm


1.5 kW @ 1800 rpm DC line voltage: 500V DC
3.0 kW @ 9000 to 18,000 rpm 10 kW output @ 18,000 rpm
Liquid-cooled

where D r is the outer diameter of the rotor. The stator has six salient poles with
a span of 2¿ each. Interpole openings in the stator are about 0.66 ¿ (circumferen-
tially). Each stator pole supports two PMs of alternate polarities. As may be seen
from Figure 1, the polarities of the adjacent PMs on diŒerent poles are the same.
Consequently, the PM ux fringing between adjacent poles is considerably reduced.
Each stator pole is wound with a concentrated coil. Two coils in series make one
phase, or two coils may be connected in parallel to constitute one phase, as shown
in Figure 1(b).

B Design SpeciŽcations
The FRM to be designed should be able to operate as both generator and motor.
The main design speciŽcations are listed in Table 1.

C Induced EM F
Energy conversion in the FRM occurs by virtue of interaction between the PM
Želd and the stator coil magnetomotive forces (MMFs). The EMF per phase e is
given by
d¸P M dx
e = 2N c , (3)
ds dt
where dx = dt is the tip (or peripheral) speed and ¸P M is the ux linkage variation
per stator pole (without armature reaction).
Ideally, ¸P M varies linearly with the mover position, which can be expressed as
³ ´
x
¸P Mi » B gP M i ¿l ¿ < x < ¿, (4)
¿
where B gP Mi is the ideal PM ux density in the airgap and l is the stack length.
However, in reality, owing to fringing, ¸P M departs from being linear. Then, we
have
d ¸P M
» 2B gP Mi lk f r , (5)
dx
where k f r is a fringing factor and the coe cient 2 is included to account for 2
coil-sides per coil. We assume
kf r » 0.5 ~ 0 .65 . (6)
To determine k f r with a greater accuracy, Žnite element analysis (FEA) is necessary.
This approach is also necessary for design optimization. In any event, equations (3),
(5), and (6) yield the induced EMF.
852 I. Boldea et al.

D Electromagnetic Power at Base Speed


Assuming an engine idling speed of 900 rpm, with a 1/ 2 belt-transmission ratio, a
speed of 1800 rpm may be achieved. This speed is considered as the base speed at
which the torque is maximum. The e ciency ´b at base speed is approximately 0.6
to 0.70. These are initial values to be veriŽed at the end of the design process.
As a generator, the electromagnetic power P e b is given by
Pb 1500
Pe b = = = 2307 .7 W . (7)
´b 0 .65
Next, we choose a tangential force density of f x = 1 .8 to 2.5 N = cm 2 . Also, we let
® = l = D r = 1 .1 to 1.8 so that equation (7) may be rewritten as

Dr ¼ 3
P e b = f x ( ¼D r l ) W r = f x ®D r W r, (8)
2 2
where W r is the rotor speed in rad/ sec. Substituting numerical values in equation
(8) yields
³ ´1/ 3
2 × 2307 .7
Dr = = 0.06763m = 6 .763cm, (9)
2 .1 × 104 × ¼ × 1 .2 × 2¼ × 30

where we have assumed f x = 2 .1 × 104 N = m 2 and ® = 1 .2. Consequently, from


equation (9) we have D r = 0 .07m (for safety). The stack length is thus

l = ®D r = 1 .2 × 0 .6763 = 0 .0812m . (10)

With a stacking factor k s of 0.95, the actual stack length l s becomes

l s = l = k s = 0 .0812= 0 .95 = 0 .0855m . (11)

It should be noted that 0.1-mm thick laminations with very thin insulation are
recommended to achieve 0.95 stacking factor.
Corresponding to the base electromagnetic power, the base electromagnetic
torque is
Pe b 2307 .7
Te b = = = 12 .24N m . (12)
W rb 2¼30

E PM Airgap Flux Density


The PM ideal ux path is shown in Figure 2, and ideally the ux density at the
surface of the PM (on the airgap side) B m is the same as the airgap ux density
B g i . Also, with a linear demagnetization ( B m , H c ) curve for the PM, we have

Br
Bb = " # = B gi, (13)
¹r e (1 + k st ) g
1+
¹o hm

where ¹r e is the recoil permeability, ¹ o = 4¼10 7 H/ m, the permeability of free


space, h m is the PM radial thickness, g is the airgap, and k st is a globla saturation
factor.
To Žnd the magnet thickness, we proceed as follows. First, to obtain a low
inductance and hence a low time-constant, a thick magnet would be desirable.
Design of a Three-Phase Flux Reversal Machine 853

Figure 2. Ideal path of the PM uxes.

However, a thick PM would imply a higher cost. From experience, we observe that
the ratio of h pr to h m should be about 2.5. Thus, with D r = 0 .07 m, we choose
h pr = 17 mm and h m = 6 mm. Magnequench MQ3-F38H Nd Fe B magnets are used.
The properties of this magnet are given in Table 2.
At an operating temperature of 25° c, we have
" #
kB r
Br = Bro 1+ ( T 25) = 1.213T , (14)
100
" #
kH c
Hc = Hco 1 + ( T 25) = 0.652MA = m, (15)
100
and the new recoil permeability becomes
Br 1.213 6 6
¹r e = = × 10 = 1.86 × 10 = 1 .48¹ o . (16)
Hc 0.652
Finally, we choose airgap g = 0.5m. Consequently, from equation (13), we obtain
1 .213
B m = B gi = " # = 1 .074T , (17)
1 .48¹ o (1 + 0 .05)0 .5
1+
¹o 6

Table 2
Properties of Magnequench MW3-F38H

Residual ux density at 25° C B r o = 1 .27 T


Coercive force at 25° C H c o = 0 .93 MA/ m
Recoil Permeability at 25° C ¹r e = 1 .07 ¹ o H/ m
Temperature coe cient for B r k B r = 0 .09%= ° C
Temperature coe cient for H c k H c = 0 .6%= ° C
854 I. Boldea et al.

Figure 3. EMFs, uxes, and ideal currents.

where we let k st = 0 .05. Now, let k f r = 0 .65 so that equation (5) yields
d ¸P M
= 2 × 1.074 × 0 .0812 × 0 .65 = 0.11337W b= m . (18)
dx
In terms of the rotor displacement angle µr , we have
d ¸P M 1 d¸ P M 0.07
= Dr = 0 .11337 = 3 .98mW b= rad . (19)
dµr 2 dx 2
Finally, with two coils in series, the induced EMF becomes
d ¸P M dµr d ¸P M
e = 2N c = 2N c (2¼n ) , (20)
dµr dt dµr
where n is the rotor speed in rad/ sec.

F M M F per Coil at 1800 rpm


With two phases working at a time (Fig. 3) and E¯ and I¯ in phase with each other,
the electromagnetic power is
P e = 2E I . (21)
Notice from Figure 3 that the actual EMF is trapezoidal. Also, at high speeds,
overlapping during commutation will make the currents like trapezoids. At 1800 rpm
(or 240 Hz) nonsimultaneous conduction of the rectiŽer diodes (Fig. 4) is present.
At high speeds, this phenomenon is severe and inuences the choice of the Žlter
capacitor, C f (Fig. 4).
At 1800 rpm, let the current be constant during a 120° (electrical) conduc-
tion interval, and the current is still in phase with the EMF (Fig. 3). Thus, from
equations (20) and (21) we obtain
d¸ P M
P e b = 2 × 2N c (2¼n ) Ib. (22)
dµr
Design of a Three-Phase Flux Reversal Machine 855

Figure 4. Equivalent circuit with diode rectiŽer and capacitor Žlter.

A simple manipulation of equation (22) yields

N c I b = 770A t =coil, (23)

which is rather large.

G Stator Slot Geometry


Stator slot/ pole geometry is shown in Figure 5. The slots are semiclosed with a slot
opening w s o = 2¿= 3.
To calculate the slot area, we Žrst determine the slot Žll factor. Because space
is required for axial air circulation, we choose a low Žll factor k f il l = 0 .38. To choose
the core ux density, generator design would be diŒerent from motor design owing to
the mode of cooling in the two cases. Finally, the choice of current density is a matter
of compromise. A high current density reduces the machine size and machine time
constant, but also decreases the machine e ciency. Low time constant improves the
commutation process. Furthermore, at high speeds such as 18,000 rpm (or 2.4 kHz),
skin eŒect is not negligible and results in a signiŽcant increase in copper losses. We
have two choices here: either we use stranded wire or increase the slot area, hence
the stator diameter and the machine mass. Litz wire may be appropriate for the
motor primarily operating in extreme high speed. For generator, we choose the base
current density
2
J c b = 12A = m m . (24)

Figure 5. Stator slot/ pole geometry.


856 I. Boldea et al.

Figure 6. ReŽned slot geometry.

Hence the slot area becomes


2N c I b 2 × 770
A sl ot = = = 338m 2 . (25)
k f il l J c b 0 .38 × 12
The precise geometry of a slot is shown in Figure 6.
The maximum ux per pole is given by
1 1
¸P M max =
(1 + k f r ) l¿ps B max = (1 + 0 .65) l¿ 1 .0739 = 0 .886l ¿ . (26)
2 2
Choosing a pole ux density B ps = 1 .2T , the pole width wt s is given by
¸P M max 0 .886l¿ps
wt s = = = 0 .738¿ps . (27)
lB p s l × 1 .2
The pole pitch ¿ps is obtained from
¼ ¼
¿=
( D + 2g ) = (71) = 13 .94mm . (28)
16 r 16
Hence, from equations (27) and (28) we obtain
w t s = 0 .738 × 13 .94 = 10 .287m m . (29)
The slot opening l w t is given by
¼D r ¼× 93
lw i = wt s = 10 .287 = 38 .38m m . (30)
6 6
Because A sl ot = 338mm 2 , from equation (25), the useful slot height h ps becomes
10 mm. Finally, the external pole diameter is given by
D pe = D r + 2h ps = 93 + 2 × 10 = 113m m (31)
The stator back-iron thickness is obtained from
k f il l B gP Mi ¿ 0 .65 × 1 .0739 × 13 .94
bc s = = = 6 .08 » 8m m . (32)
2B c s 2 × 0 .8
Consequently, the stator external diameter D se becomes
D se = D pe + 2bc s = 113 + 2 × 8 = 129mm . (33)
Design of a Three-Phase Flux Reversal Machine 857

Figure 7. Rotor core geometry.

H Rotor Core Geometry


The rotor core geometry is shown in Figure 7. For the purpose of reducing windage
loss, the interpole areas should be Žlled with lightweight nonmagnetic material.
The rotor has 8 poles and 8 interpoles. The motor back core is designated by
bc r and the shaft by D sh . The angles subtended by the pole ® and by interpoles ¯
are such that ® = ¯ = ¼= 8 = 22 .5o . The interpolar space is rounded oŒnear the
back iron to reduce fringing. The approximate (average) ux density in the rotor
pole is given by
1 1
B pr » (1 + k f r ) B gP Mi = (1 + 0 .65)1 .07 = 0 .88T . (34)
2 2
For mechanical integrity, it is estimated that

D sh = 20m m . (35)

Also, we assume that the rotor back iron thickness bc r could be made the same as
that of the stator bc s . Thus
bc r = bc s = 8m m, (36)

and the pole height h pr becomes


1 1
h pr = (D D sh 2bcr ) = (70 20 2 × 8) = 17m m . (37)
2 r 2
The average ux density in the rotor back iron is given by
1 ¿ B gP M i 13 .92 1.07
B cr = k = × 0 .65 × = 0 .6T . (38)
2 f r bc r 2 8

III Parameter Calculations


A M achine Parameters R s and L s
The stator coil geometry is shown in Figure 8.
858 I. Boldea et al.

Figure 8. Stator coil geometry.

With two coils in series, the stator resistance per phase R s is given by
2
l c 2N c ½c o
Rs = = k R N c2 , (39)
( N c I b) = J c b
where
2l c ½c o
kR = . (40)
( N c I b) = J c b

In equation (40), ½c o = 2 .1 × 10 8 W m , the resistivity of copper; ( N c I b) = 770, as


given by equation (23); J c o = 12 × 106 A = m 2 , the base current density; and l c is the
mean length per turn, as given by

l c = 2 × 88 + 2 × 1 .3( w t s + l w ) = 0 .2696m . (41)

Thus, equation (40) yields

2 × 0 .2696 × 12 × 106 × 2.1 × 10 8


4
kR = = 1 .761 × 10 . (42)
770
The phase inductance L s consists of the magnetizing inductance L m and the leakage
inductance L l such that
L s = Ll + Lm , (43)

where
2 6 2
L m = 2N c [¹ o =[g + h m = ( ¹r e =¹ o )]l¿ (1 + k f r )] = 0 .759 × 10 Nc . (44)

The leakage inductance L l has two major components: the slot leakage inductance
and the end connection leakage inductance. Referring to Figure 9, the slot-speciŽc
permeance c is
h ps o 2h ps 1 2h ps
¸s = + + = 0.43 . (45)
wo lw i + wso 3( l w e + l w i )
Design of a Three-Phase Flux Reversal Machine 859

Figure 9. Stator slot dimensions.

We assume the end-connection–speciŽc permeance ¸e c is half of ¸ s , i.e.,


1
¸e c = ¸ = 0 .215 . (46)
2 s
For simplicity, the leakage-speciŽc permeance may be approximately taken as the
average of ¸ s and ¸e c , i.e., ¸sec = 0.323.
Therefore, the leakage inductance becomes
2 7 2
L l = 2N c ¹ o ¸sec l c = 2 .19 × 10 Nc . (47)
Finally, the total inductance per phase is obtained from equations (43), (44), and
(47). Hence
7 2 7 2 2
L s = (7 .59 + 2 .19) × 10 N c = 9 .7 × 10 N c = kL N c . (48)
The time constant is given by
T s = L s = R s = k L = k R = 5m sec . (49)
Due to the skin eŒect, the eŒective resistance R s may be larger than the value
calculated in equation (42). On the downside, this results in an increase in copper
loss. However, the time constant would be decreased, which is desirable in reducing
the size of the Žlter capacitor.
Finally, with design data obtained thus far, the induced EMF, as given by
equations (3), (5), (19), and (20), is
d ¸P M 3 2
e = 2N c (2¼n ) = 2(2¼)3 .98 × 10 N c n = 5 × 10 Ncn. (50)
dµr

B Copper and Core Losses


Copper losses at the base speed (1800 rpm) with two phases conducting at any
time is
2 2 4
P c o = 2I b R s = 2k R ( N c I b ) = 2 × 1 .761 × 10 × (770) 2 = 209W . (51)
Total mass of copper is
6N c I b 0 .2696 × 6 × 770 × 8900
G c o = lc °c o = = 0.924k g . (52)
Jco 12 × 106
860 I. Boldea et al.

To evaluate the core loss, we Žrst Žnd the masses of the various parts. The stator
pole mass is
" ³ ´#
2¿ + wt s
G ps = 6 w t s h t s + 2¿ h ps o + h p s 1 l° ir on = 0 .797k g . (53)
2

Mass of the stator yoke (or back-iron) is


¼
Gcs = ( D 2se 2
D pe ) l°ir on = 1 .876k g . (54)
4
Rotor pole mass is
1 ¼ 2
G pr = × [D (D r 2h ps ) 2 ]l° ir on = 0 .873 (55)
2 4 r
Rotor back-iron mass is
¼
G cr = [( D r 2h pr ) 2 2
D st ]l° ir on = 0 .124k g . (56)
4
Now, to a Žrst approximation the core loss is given by
1. 7 2 3
P c or e = 96k e f B i W =m , (57)

where k e » 0 .11 for 0.5-min thick M-19 and k e » 0.0076 for 0.1-mm thick lamina-
tions. The frequency f is 240 Hz at 1800 rpm. Choosing k e » 0 .04, core losses at
1800 rpm become

2401. 7
( P c or e ) 1800 = 96 × 0 .05 × [B 2ps ~ G ps + B c2s ~ G c s + B 2pr ~ G pr + B c2r ~ G c r ]
° ir on
2401. 7
= 96 × 0 .05 × [0 .62 × 0 .797 + 0.82 × 1 .876 + 0.882 × 0 .873 + 0 .82 × 0 .124]
7600
= 12 .6W . (58)

At 18000 rpm, the core loss becomes

( P c or e ) 18, 000 = ( P c or e ) 18, 00 × 101. 7 = 631 .5W . (59)

C E ciency VeriŽcation
At the base speed, we assumed a 0.65 e ciency. This implies that the total losses
are
X Pb 1500
Pl oss = Pb = 1500 = 807 .7W . (60)
´b 0.65
As calculated earlier, P c o = 209 .6W and P c or e = 12 .6W . Thus, mechanical losses
could be as high as

P m e c h = 807 .7 209 .6 12 .6 = 585 .5W . (61)

Because mechanical losses could not be so high, we may reŽne the design with
a higher value of e ciency. For fabricating the prototype, we retain the present
design.
Design of a Three-Phase Flux Reversal Machine 861

D Number of Turns per Coil


First, we recall that the rectiŽer output voltage is V o = 42V DC. Including the ef-
fects of the resistive voltage drop and the inductive voltage drop due to overlapping,
we have
3
(2E ) 1800 2R s I b !b L s I b = V o . (63)
¼

Substituting the numerical values,


2 4
2 × 5 × 10 N c × 30 2 × 1 .761 × 10 (N c I b)N c
3 6
(2¼ × 240)0 .992 × 10 ( N c I b ) N c = 42 . (64)
¼

Solving for N c yields N c = 26 turns/ coil.


The active power circulation is given by

(2E 2R s I b) I b = V o I o , (65)

where
NcIb 770
Ib = = = 30A (66)
Nc 26
or
(77 .7 0 .2717 × 26) × 30 = V o I o = 2100, (67)

which is much greater than the initially assumed value of 1500 W. The DC current is
2100
Io = = 50A . (68)
42

E W ire Size
At 1800 rpm at 1.5 kW, approximate phase current = 30A. With the assumed base
current density,
30
A w ir e = = 2 .5mm 2 (69)
12
Stranded wire with a total cross-sectional area of 2.5 mm2 and class F insulation
is recommended for the prototype machine.

IV Design Summary
Geometry and dimensions of the stator and rotor dimensions (in millimeters) are
shown in Figure 10. Other data are as follows:
i) Stator and rotor stack length: 86mm.
ii) Lamination thickness: 0.1mm (with very thin insulation layers).
iii) Magnets: Magnequench MQ3 assembled in four pieces along the stack
length.
The geometry of the coils are shown in Figure 11, each having 26 turns of
stranded wire of 2.5 mm2 cross-section with class F insulation.
862 I. Boldea et al.

Figure 10. Stator and rotor core Žnal dimensions.

Figure 11. Coil geometry and overall dimensions.


Design of a Three-Phase Flux Reversal Machine 863

V Conclusion
A comprehensive design of a novel three-phase FRM is presented in this paper.
The design procedure is illustrated in detail. The formulae involved in the design
and parameter calculation are presented based on the classic design method. All
the dimensions of the FRM are given. Basic parameters are calculated. The high-
speed application requirement is considered in the whole process of design. The
work presented may serve as a guideline for the general principle of the design of
three-phase FRM. Furthermore, the methodology and formulae introduced in this
paper may also be applicable to other types of PM and/ or reluctance machine.

A cknowledgment
This work was, in part, supported by NSF Grant No. ECS-98-02454.

References
[1] Rauch, S. E., Johnson, L. J., 1955, “ Design principles of ux switch alternator,” AIEE
Trans. Vol. 74III, pp. 1261–1268.
[2] Boldea, I., Serban, E., Basiu, R., 1996, “ Flux reversal stator-PM single-phase generator
with controlled d.c. output,” Record of OPTIM-96, Brasou, Romania, pp. 1123–1137.
[3] Deodhar, R., Anderson, S., Boldea, I., Miller, T.J.E., 1997, “ The ux-reversal machine:
A new brushless doubly-salient permanent magnet machine,” IEEE Trans. Ind. Appl.,
Vol. 33, No. 4, pp. 925–934.
[4] Luo, X., Qiu, D., Lipo, T. A., “ A novel two phase double salient permanent magnet
motor,” Record of IEEE-IAS-1996 Annual Meeting, Vol. A, pp. 808–815.
[5] Liao, Y., Liang, F., Lipo, T.A., 1995, “ A novel permanent-magnet motor with doubly-
salient structure,” IEEE Trans. Ind. Appl., Vol. 31, pp. 1069–1078.

You might also like