Professional Documents
Culture Documents
Abstract—This paper proposes a simplified analytical machine the machine’s parameters/design. System designers have used
sizing procedure for a three-phase surface mounted permanent simple scaling laws for components and machines [1] while
magnet synchronous machines appropriate for both system level integrated machine and drive design for weight optimisation
analysis and preliminary machine design. For system-level analy-
sis, the proposed method can generate candidate machine models, has been considered in [2] and [3].
be used to check if proposed performance values for a machine in In this paper, an analytical machine sizing procedure is
a system level design problem are feasible or be integrated into a presented. This procedure bridges the gap between system
system-level optimization considering power electronics, passives, level and machine level design. The optimisation problem
and machine. In machine design, the advantage of an analytical
process over FEA is its computational efficient nature, capable defined by the design procedure can be solved using simple
of quickly generating numerous candidate designs across a wide iterative methods or more advanced heuristic search methods.
search-space. Selected designs can be further refined using more [4] used an iterative sizing method that progressively increases
computational intensive methods such as FEA, CFD, and machine the machine dimensions until the desired design targets and
design software. Outputs of the sizing procedure include, mass, specifications are met. [5] investigated population-based op-
volume, efficiency, slot and rotor geometry, winding dimensions,
stator resistance, stator inductance, magnet flux linkage, air gap timisation design methods of permanent-magnet synchronous
width, magnet thickness, and steady state thermal results. The machines using genetic algorithm and particle swarm opti-
sizing procedure is validated using 2D electromagnetic solver misation. Detailed discussion of these methods is beyond the
FEMM, and machine design software Motor-CAD. scope of this paper. The presented design procedure was tested
Index Terms—Analytical machine sizing, surface mounted using an iterative method within a user defined search space
permanent magnet synchronous machines, SPMSM.
and resolution.
I. I NTRODUCTION An outline of the design procedure is presented in section II.
The input specifications for the design procedure are outlined
Surface mounted permanent magnet machines have gained in section III. The system level machine sizing process is
popularity due to their high power-to-weight ratio and low presented in section IV. This is followed by the detailed design
power loss, making them suitable for servo and traction appli- in section V where the air gap and magnets are sized. Next,
cations for industry, transport, domestic items, and aerospace. the thermal modelling is presented in section VI. The full
The main objective in machine design is to meet mechanical design procedure is used to design three machines of different
operating requirements, typically defined by nominal and configurations and power ratings, validated using FEMM and
peak torque, speed, and duty cycle, for a given operating Motor-CAD, and presented in section VII.
environment while minimising cost, weight, and losses.
For system level designers who don’t have prior machine
Machine design techniques can be split into analytical and
design experience, typical values for various applications
finite element analysis (FEA) techniques with the prior typi-
are provided to aid preliminary design. For preliminary ma-
cally being used for preliminary machine design and sizing,
chine design, this analytical method can be incorporated into
and the latter to analyse and refine said preliminary designs.
computational-efficient design methods.
Analytical machine sizing requires less computational effort
The following assumptions about the candidate machine
compared to finite element based techniques and, for that
designs are made:
reason, can be integrated into a machine sizing optimisation
process or a system level optimisation. If desired, these designs • A three-phase, outer stator, inner rotor design is consid-
can be later analysed and refined using FEA. ered.
Machine, drive, and system design are typically done sep- • Only external cooling, such as natural air convection,
arately. An example would by a system designer selecting a forced air convection, is used.
machine with fixed parameters in a catalogue/database that • The permeability of the stator and rotor iron compared
delivers the mechanical requirement for the available DC link, to that of the air is considered infinite when determining
from which the inverter, filters and controllers are designed. stator inductance and magnet operating point.
However, the system designer does not have any influence on • Magnetic saturation is not considered.
752
Fig. 3. Skew angle.
yQ − n r π
kp1 = sin , (7)
yQ 2
π
2 sin( 2m )
kd1 = , (8)
Nsl
mp sin( Nπpsl )
Fig. 2. Machine dimensions.
sin(θsq π2 )
ksq1 = , (9)
TABLE I θsq π2
T YPICAL VALUES OF TORQUE PER ROTOR VOLUME AND CURRENT
DENSITIES ( CONTINUOUS OPERATION ) [6]. where yQ is the full pitch slot count equal to Nsl /Np , nr is
the pitch shortening in number of slots, and θsq is the skew
angle, as shown in Fig. 3, typically set at one slot such that
Class of machine TRV (kNm/m3 )
Small totally-enclosed motor (Ferrite magnets) 7 - 14
θsq = 2π/Nsl . Typical values for kw1 is 0.85 - 0.95 [6] and
Totally-enclosed motors (sintered rare earth or NdFeB) 14 - 42 can be used for system level analysis.
Integral-hp industrial motors 7 - 30 For a given value of TRV and stack length Lstk = xD, rotor
High-performance servomotors 15 - 50
Aerospace machines 30 - 75
volume Vr , and rotor diameter D are calculated. The average
Condition J (A/mm2 ) magnetic loading B is related to the peak tooth magnetic flux
Totally-enclosed 1.5 - 5 Bt(pk) as follows,
Air-over, fan-cooled 5 - 10
Liquid cooled 10 - 30 B = 2τ Bt(pk) /π (10)
753
The CSA of a single copper conductor Acu(wire) is deter-
mined,
Acu(wire) = Acu(total) /(6Nph ) (15)
The diameter of the conductor in a round wire dcu(wire) can
determined from this,
q
dcu(wire) = 2 Acu(wire) /π. (16)
Fig. 4. Continuous and peak operating points on the dq-current plot.
The maximum diameter of wire consisting of insulation and
conductor assuming square packing pattern dwire is deter-
mined as follows. This value is the minimum value of the Voltage and current limits Vmax , Imax , expressed as follows,
slot opening at the stator bore wss . need to be considered during operation,
dcu(wire)
r
π
dwire = (17) Vdc q
2 Fslot Vmax = √ ≥ vd2 + vq2 , (28)
3
Total length of wire for each phase lph is determined as follows
where lend is the per-turn mean end winding length for semi- 2Tpk q
circular end winding path and nr is the pitch shortening in Imax = ≥ i2d + i2q . (29)
3pλm
number of slots [8],
The considered continuous and peak torque operating limits
lph = Nph (lend + 2Lstk ), (18)
2 are shown in Fig. 4. Portions of the voltage limit ellipses
π nr Np crossing the continuous and peak limit operating points are
lend = 1− (rra + rrad ). (19)
2Nsl Nsl shown by the dashed lines and a portion of the current limit
circle is represented by the solid line arc. For these operating
The stator resistance Rs can now be determined knowing the
points, no field-weakening is present, thus id = 0.
conductor’s total length and wire copper cross-sectional area,
If Ls , λm , and iq are known, these values can be sub-
lph stituted into (21), (22), and (28) to check if the voltage
Rs = , (20)
σcu Acu(wire) limit is respected for an operating electrical angular speed
where σcu is the copper electrical conductivity. ωe and respective mechanical angular speed ωm . Rs has
been determined during the winding, conductor, and resistance
C. Operating point, stator inductance, and magnetic flux calculations for a given TRV, and winding configuration. Ls
linkage and λm are determined as follows.
The steady state dq voltage equations in the synchronous To determine a suitable value of Ls for desired current ripple
reference frame for a three-phase non-salient pole SPMSM c, the PWM waveform is approximated to a square wave with
are as follows where id and iq are the d- and q- axis currents, duty cycle of 0.5 and peak-to-peak amplitude of Vdc . The first
vd and vq are the d- and q- voltages, λd and λq are the d- and harmonic of the square wave is represented as a sin-wave of
q- flux linkages, λm is the permanent magnet flux linkage, and amplitude (4/π)Vdc at switching frequency fsw . For a desired
ωe is the electrical frequency. For an SPMSM, it is assumed peak to peak current ripple ∆I at steady state condition, a
that d- and q-axis stator inductances are equal such that Ls = value c can be determined as c = ∆I/(2iq ) as shown. Stator
Ld = Lq . inductance Ls can be determined as follows,
vd = Rs id − ωe Ls iq , (21)
Vdc
Ls = √ . (30)
vq = Rs iq + ωe Ls id + ωe λm . (22) 2π 2 ciq fsw
λd = Ls id + λm (23) Equation (27) can be solved for λm to get an expression with
λq = Ls iq (24) respect to iq and Tem ,
q
|λdq | = λ2d + λ2q (25) 2Tem
λm = . (31)
3piq
The relationship between electrical and mechanical angular
speed ωe and ωm is as follows where p is the pole pairs, Equations (30) and (31) are substituted into (21) and (22) for
id = 0, and then into voltage limit equation (28) as a function
ωe = ωm p, (26) of iq to give,
The electromechanical torque equation is as follows, s 2 2
3p Vdc Tsw 2Tem
Tem = λm iq . (27) |vdq | = − ωe 2 + Rs iq + ωe . (32)
2 π ciq 3piq
754
The operating electrical angular speed ωe at the voltage and Kc is the Carter’s coefficient approximated as follows [9],
limit for given current operating point iq for id = 0 can be 5 + wss /da
determined using the following equation, Kc = . (38)
5 + wss /da − (wss /da )2 (da /λ)
ωe = The slot leakage inductance is calculated assuming a rect-
angular slot shape [9]. The slot permanence per unit length
q
2
(Vmax − |idq |2 Rs2 )|λdq |2 + i2q Rs2 λ2m − iq Rs λm (33)
. is,
|λdq |2 1 hs hsh
P = µ0 + . (39)
This equation can be used to determine the maximum op- 3 ws wss
erating electrical angular speed at peak torque, where iq = The slots in each phase belt Nb and no of coils per turn Nc
2Tpk /(3pλm ) assuming no saturation. are as follows,
Nsl
Nb = , (40)
D. Power Loss, Efficiency, Mass 2pm
Copper losses Pcu can be calculated from Rs and nominal Nph
Nc = . (41)
current Iph as follows, 2pNb
2
Pcu = 3Iph Rs . (34) The slot leakage inductance is calculated as follows,
La = 2pLstk P Nc2 (4Nb − nr ). (42)
Power factor (PF) for the nominal and peak operating point
are calculated from their respective operating dq voltages and The total synchronous inductance Ls is the sum of air-gap and
currents, leakage inductances,
P F = cos(arctan(vq /vd ) − arctan(iq /id )). (35) Ls = Ls1 + La . (43)
755
TABLE II
M ACHINE DESIGNS .
Parameter A B C
Slots 12 24 24
Poles 4 2 4
Nominal torque (Nm) 2 5 20
Peak torque (Nm) 3 7.5 30
Fig. 5. Magnet’s open-circuit equivalent reluctance network [10]. Speed (RPM) 2500 5000 2500
Mass (kg) 5.65 12.62 25.11
Volume (litre) 1.46 3.07 6.74
Nominal Line Current (Arms ) 4.4 22 44
C. Air gap and magnet length sizing Current Density (A/mm2 ) 3.9 5 4.3
The air gap da and magnet length dm need to be calculated. TRV (kNm/m3 ) 14.6 13.4 26.8
First, calculate the slot leakage inductance La using (39)-(42). Stator Resistance (Ω) 0.32 0.047 0.0207
Stator Inductance (mH) 5.92 0.83 0.207
Then determine the desired air-gap inductance Ls1 using (43). Magnet Flux Linkage (Wb) 0.107 0.107 0.107
Equations (37), (38) are substituted into (36) to give an Rotor Diameter (mm) 47 62 83
expression of Ls1 as a function of da and dm , Stator Diameter (mm) 94 124 163
Stack Length (mm) 79 103 138
Ls1 =
2 2
6µ0 rra Lstk Nph 2µrec kw1 (5λda + λwss − wss )
.
πp2 (λµrec (da wss + 5d2a ) + dm (5λda + λwss − wss2 )) Substituting (52) into (53) and solving for Iarm gives the
(49) magnitude of the demagnetisation current limit. This value
This can be solved for dm with respect to Ls1 and da , must not be greater than λm /Ls such that the full current
2
6µ0 rra Lstk Nph 2
µrec kw1 2
(5λda − λwss − wss ) operating range is reachable. Thus all designs must be checked
dm = for their demagnetisation current limit.
πp2 Ls1 (5λda + λwss − wss2 )
(50)
λµrec (da wss + 5d2a ) VI. T HERMAL M ODELLING
− 2
.
5λda + λwss − wss The candidate machine’s steady state temperatures and tran-
A function for magnet flux-linkage λm1 with respect to sient temperature profiles can be determined using a lumped
magnet depth dm can be created using (44)-(48). A value of parameter thermal network (LPTN) [11]. All machine compo-
da can be found numerically such that the calculated value nents are approximated by a general cylindrical shape whereby
of dm (da , Ls1 ) from (50) when substituted into λm1 (dm ) axial and radial heat flows are assumed to be independent
matches the desired magnet flux λm . of each other and no circumferential heat flow is present.
Given the calculated components’ dimensions, masses, and
D. Demagnetisation limit their respective thermal conductivities, the equivalent thermal
During field-weakening operation, the negative id current network for each component can be generated.
is injected into the machine to reduce the λd and increase Frame to ambient thermal resistance as a function of frame
the operating range at the voltage limit. However, excess and ambient temperature is needed to accurately determine
negative id can demagnetise the magnets. The operating point both transient and steady state machine temperatures [12].
of the magnet should never go below the knee point of the The equivalent frame to ambient thermal resistance for natural
linear portion of the demagnetisation curve (BD , HcJ ) where free convection and radiation cooling calculation procedure is
HcJ is the magnet’s intrinsic coercive force. This point is presented in [13] where the machine is approximated as a
temperature dependent with magnets being more sensitive horizontal isothermal cylinder.
to demagnetisation at higher temperatures. The maximum
magnitude of negative id is determined as follows. VII. R ESULTS
The demagnetisation point BD is determined as follows, The design and specifications of three machines sized by
the presented design procedure are shown in Table II. The
BD = µrec µ0 (−HcJ ) + Br . (51)
magnet material is NdFeB 37 and the natural free convection
The magnetic flux φm for this operating point is calculated is considered. The ambient temperature is 50◦ C. The TRV and
by setting Bm as BD such that φm = BD Am . The armature current density search space is selected from Table I as 14 -
MMF Farm for this condition is as follows, 42 kNm/m3 and 1.5 - 10 A/mm2 respectively.
Rm0
The designs were validated using 2D electromagnetic solver
Farm = φm RL − (RL + Rg ) φm + (φm − φr ) (52) FEMM [14] and machine design software Motor-CAD. The
RL
flux density plots generated from FEMM are shown in Fig. 6.
The relationship between the peak current Iarm and the Tool, FEMM, and Motor-CAD calculated values for torque,
armature MMF is, stator inductance, tooth flux density, air gap flux density,
pπFarm losses, winding temperature, stator resistance, efficiency, and
Iarm = (53) power factor are shown in Table III. Absolute values are
3kw1 Nph
756
TABLE III
C OMPARISON OF TOOL , FEMM, AND M OTOR -CAD CALCULATED
VALUES .
757