Research ArticleAn Analytical Solution to the ElectromagneticPerformance of an Ultra-High-Speed PM Motor
Wenjie Cheng 1 Zhikai Deng1 Ling Xiao1 Bin Zhong2 and Yanhua Sun3
1College of Science Xirsquoan University of Science and Technology No 58 Yanta Road Xirsquoan China2School of Mechanical Engineering Xirsquoan University of Science and Technology No 58 Yanta Road Xirsquoan China3State Key Laboratory for Strength and Vibration of Mechanical Structures Xirsquoan Jiaotong University No 28Xianning West Road Xirsquoan China
Correspondence should be addressed to Wenjie Cheng cwj20070807163com
Received 21 February 2019 Revised 23 July 2019 Accepted 18 August 2019 Published 12 September 2019
Academic Editor Gorazd Stumberger
Copyright copy 2019Wenjie Cheng et al-is is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
Bipolar rotors with cylindrical permanent magnets (PMs) which are magnetized in parallel are generally used in ultra-high-speedpermanent magnet synchronous motors (PMSMs) Aiming for this type of PMSM two methods namely the Poynting theoremand the Maxwell stress tensor were used to deduce the analytic expression of the electromagnetic driving torque A locked-rotorsimulation by the finite element (FE) method verifies the analytical solution In addition the cogging torque back-electromotiveforce (emf) armature reaction inductance etc were calculated compared and discussed-e no-load back-emf was checked by a60000 rpm free coast experiment -e proposed analytical solution is very convenient to guide the design of the ultra-high-speedPMSM and check its performance
1 Introduction
With advancements in new materials electrical and elec-tronics control technology etc the speed limits of rotatingmachines are being pushed higher and ultra-high-speedmotors are increasingly used in industry applications [1]Particularly in recent times the high-speed PMSM hasattracted more attention due to its high efficiency simplestructure high power density and small size [2 3] forexample it can be used as the driving motor for the fuel cellair blower [4 5] as an air centrifugal compressor [6] or asthe generator of a micro-gas turbine [7]
It is very significant to calculate the electromagnetictorque because it can be used to check the motor powerestimate noise and vibration and provide further guidancefor the motor control It is well known that the electro-magnetic torque is generated from the interaction of thePM field and the armature reaction field therefore how todetermine the above two magnetic fields becomes keyPMSMs often adopt a surface-mounted PM rotor and thiskind of rotor as seen in most relevant literatures [8 9] has
a common characteristic which is that the magnetic polearc is less than a pole pitch (the pole embrace is less than1) which means the air space between the magnets isoccupied by unmagnetized materials such as air or alu-minum alloy Rotors of the ultra-high-speed PMSM arealways subject to extreme conditions such as high speedand high temperature To improve the bending stiffnessand strength of the rotor ultra-high-speed PMSMsgenerally use a special rotor construction in which acylindrical PM magnetized in parallel is interference-fitted into a metallic sleeve therefore the rotor has singlepole pairs and each half of the magnet has a 180deg electricpole arc (the pole embrace is equal to 1) as shown inFigure 1 As mentioned in the literature [10] existinganalytical methods are successful only in certain cir-cumstances and are likely to fail in others for examplewhen the magnets have a 180deg electric pole arc In theliterature [11] study an open-circuit magnetostatic fieldsolution to a 2-pole rotor with a diametrically magnetizedsintered NdFeB magnet was given for a 20000 rpm 3-phase brushless PM dc motor Recently the literature [12]
HindawiJournal of Electrical and Computer EngineeringVolume 2019 Article ID 1469637 11 pageshttpsdoiorg10115520191469637
gave analytical solutions to the PM field of bipolar rotorswith ring or cylindrical PMs that are magnetized inparallel Nevertheless most relevant literature studies onanalytical models for predicting the armature reactionfield distribution are successful for the motor owing tothe high permeability shaft [13] however few studiesinvolve the hollow stator bore as shown in Figure 2(b)
For ultra-high-speed PMSMs including a stator withoverlapping windings and 2-pole parallel magnetizedcylindrical PMs analytical solutions to the magnetic fieldand corresponding electromagnetic torque are rarely re-ported at present which has resulted in a prolonged motordesign cycle to evaluating the motor performance Due tothe need of cooling and reduction of pulse torque theultra-high-speed PMSM generally adopts a large air gaphereupon the slot effect is very weak -erefore the as-sumption of ldquoslotless stator + equivalent current sheetrdquocan be introduced and some convenient formulas ofestimating motor performance are expected to beobtained
Hence this paper will first deduce the magnetic fieldthen obtain the electromagnetic driving torque by using thePoynting theorem andMaxwell stress tensor and then verifythe correctness by using the FE method In addition thecogging torque the back-electromotive force (emf) andarmature reaction inductance are calculated compared anddiscussed A 10 kW 120000 rpm prototype is designed andmanufactured and the no-load back-emf is checked by a60000 rpm free coast experiment
2 Analytical Model
-e cross section of the ultra-high-speed PMSMs is shown inFigure 1 in which slotted stator and distribution winding areused and the rotor is composed by installing a cylindricalparallel polarized PM into the metallic sleeve
For the sintered NdFeB magnet the demagnetizationcurve is approximately a straight line and the relativerecoil permeability μr is approximate to that of air μr waschosen from 1029ndash1100 in most literatures To simplifythe field calculation the following assumptions aremade
(1) -e PM has a linear second-quadrant de-magnetization curve and the permeability of thestator iron is infinite
(2) -e winding end effect and rotor eddy current areneglected
(3) -e PM field is calculated with a 2D solution byassuming a smooth stator surface that is the slottingis neglected as shown in Figure 2(a)
(4) Conductors in the stator slot can be equivalent to thecurrent sheet distributing evenly along the slotopening as shown in Figure 2(b)
In the calculation of the PM field the slot-opening effectis neglected therefore the calculation model of the PM fieldis shown in Figure 2(a) where R1 and R2 are the outer radius
and the inner radius of the PM and I and II represent thePM and the air region respectively
-e sleeve applies to a low permeability alloy which canbe physically equivalent to the air therefore in the calcu-lation model of the armature reaction field the stator boreseems to be full of air as shown in Figure 2(b) where b0 isthe width of the slot opening
-e loaded magnetic field is the superposition of theopen-circuit PM and the armature reaction field and theloaded air-gap flux density is defined as follows
BloadIIr BW
IIr + BPMIIr
BloadIIθ BW
IIθ + BPMIIθ
⎧⎨
⎩ (1)
In equation (1) load PM and W represent the loadedPM and winding magnetic fields respectively r and θrepresent the radial and circumferential directionsrespectively
21 Open-Circuit Magnetic Field -e literature [12] hasdeduced the solution to the PM field shown in Figure 2(a)which is given as follows
BPMIIr
minus μ0MrR21 1 + R2
2rminus 2( 1113857
R21 μr minus 1( 1113857 minus R2
2 μr + 1( 1113857cos θ
BPMIIθ
μ0MrR21 minus 1 + R2
2rminus 2( 1113857
R21 μr minus 1( 1113857 minus R2
2 μr + 1( 1113857sin θ
(2)
where Mr is the amplitude of the residual magnetization ofthe PM μr is the relative recoil permeability of the PM μ0 isthe vacuum permeability and r and θ represent the radialand circumferential positions respectively
22 Armature Reaction Field -e governing equation of theair-gap region is as follows
zϕ2
z2r+
zϕrzr
+zϕ2
r2zθ2 0 (3)
where ϕ is a scalar magnetic potential and the relation of ϕand H (magnetic field intensity) is expressed as follows
Hθ minuszϕ
r zθ
Hr minuszϕzr
⎧⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎩
(4)
In the inner radius of the stator
Hθ1113868111386811138681113868 rR2
minus J (5)
where J is the Fourier series expansion of the current sheetAt the center of the stator the H should meet the following
Hθ1113868111386811138681113868 r⟶0neinfin
Hr
1113868111386811138681113868 r⟶0neinfin(6)
Taking into account the symmetry of the magneticfield the single conductor field can be solved according toequations (3)ndash(6) -e method to solve the armature
2 Journal of Electrical and Computer Engineering
reaction field is as follows the single conductor field issolved first and then the single-turn coil field can beobtained by adding two single conductor fields at a polardistance apart from each other Whereupon the single-phase winding field is superpositioned onto fields that areexcited by the multiple single-turn coils and finally thearmature reaction field can be calculated by super-imposing the fields of three-phase winding
For distributed winding multiple single-turn coil fieldsthat are a pitch apart from each other can be equivalent to aconcentrated winding by taking into account the distribu-tion coefficient and the pitch-shortening coefficient -eabove-mentioned treatment will lightly change higherharmonic components of the armature reaction field butvastly simplify the solution of the single-phase winding fieldwith good accuracy In addition it is assumed that in-stantaneous three-phase currents (ia ib ic) only involvefundamental components and they can be written asfollows
ia Im cos(Pωt + β)
ib Im cos Pωt + β minus2π3
1113874 1113875
ic Im cos Pωt + β minus4π3
1113874 1113875
⎧⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎪⎪⎪⎪⎩
(7)
where Im is the current amplitude of a single conductor P isthe number of pole pairs ω is the synchronous angularvelocity and β is the current phase
-e armature reaction field in Figure 2(b) is expressed asfollows
BWIIr
minus 2qNsμ0π
1113944
infin
v11113896Kdpv(2v minus 1)Ksov(2v minus 1)
middot Fo(2v minus 1 r) middot (minus 1)v
middot 1113890ia cos[(2v minus 1)(α minus c)]
+ ib cos (2v minus 1) α minus c minus2π3
1113874 11138751113876 1113877
+ ic cos (2v minus 1) α minus c minus4π3
1113874 11138751113876 111387711138911113897
BWIIα
minus 2qNsμ0π
1113944
infin
v11113896Kdpv(2v minus 1)Ksov(2v minus 1)
middot Fo(2v minus 1 r) middot (minus 1)vminus 1
middot 1113890ia sin[(2v minus 1)(α minus c)]
+ ib sin (2v minus 1) α minus c minus2π3
1113874 11138751113876 1113877
+ ic sin (2v minus 1) α minus c minus4π3
1113874 11138751113876 111387711138911113897
(8)
where Ksov(v) sin(vb0(2R2))(vb0(2R2)) is the slot-opening coefficient which is a function of the spatial har-monic order vFo(r v) rvminus 1Rv
2 reflects the influence of theair-gap thickness on the radialcircumferential flux densityof the v order harmonic wave α is the angle relative to thehorizontal direction as shown in Figure 2(b) and is positive
Mrθ
Mr
Mrr
R1
R2
θ
I (PM)
II (Air gap)
(a)
Stator bore
II
Air b0Slot
α = 0
Current sheetR2
(b)
Figure 2 Models of the electromagnetic field of the PMSMs (a) Open-circuit PM field (b) Armature reaction field
Stator iron
Winding
Sleeve
Magnet
Magnetizationdirection
Figure 1 Cross section of the ultra-high-speed PMSM
Journal of Electrical and Computer Engineering 3
in the counter clockwise direction c is an angle between theaxis of winding A and the d-axis q is the number of slots perphase per pole and Ns is the number of conductors per slot
23 Electromagnetic Torque andCoggingTorque In this parttwo methods namely the Poynting theorem and theMaxwell stress tensor were used to deduce the analyticexpression of the electromagnetic torque For the 2D fieldthe electric field strength vector E in the air-gap region onlyincludes the axial component Ez therefore the Poyntingvector is given by the following
1113937r
minus EzHθ
1113937θ
EzHr
1113937z
0
⎧⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎩
(9)
where Ez μ0 1113938(zHθzt)dr -e electromagnetic powertransmitting from the stator to the rotor is mainly dependenton the radial component of the Poynting vector Πr whichcan be calculated from equation (10) as follows
PM Temω lef 11139462π
01113945
r
r dθ (10)
where lef is the length of the stator iron Taking the hori-zontal axis in Figure 2(b) as the reference position the PMfield is changed as follows
BPMIIr minus
μ0MrR21 1 + R2
2rminus 2( 1113857
R21 μr minus 1( 1113857 minus R2
2 μr + 1( 1113857middot cos(φ)
BPMIIθ
μ0MrR21 minus 1 + R2
2rminus 2( 1113857
R21 μr minus 1( 1113857 minus R2
2 μr + 1( 1113857middot sin(φ)
(11)
where φ θ minus π2 α It is assumed that the electromag-netic power is mainly transmitted by the fundamentalmagnetic field of armature reaction field When the numberof pole pairs equals 1 by substituting equation (7) intoequation (8) the armature reaction field will become asfollows using the properties of trigonometric functions
BWIIr μ0G1 cos(ωt minus φ + δ)
BWIIθ μ0G1 sin(ωt minus φ + δ)
(12)
where δ β + c represents the angle between the axis of ar-mature reaction field and the axis of PM field andG1 (3qNsKdp1Kso1Im)(πR2) Equation (12) shows that thearmature reaction field is a travelling wave but the PM field is astanding wave shown by equation (11) To make the PM fieldrotate counterclockwise synchronously relative to the armaturereaction field equation (11) can written as the following
BPMIIr μ0 minus G2r
minus 2minus G31113872 1113873 middot cos(φ minus ωt)
BPMIIθ μ0 G2r
minus 2minus G31113872 1113873 middot sin(φ minus ωt)
(13)
where G2 MrR21R
22[R2
1(μr minus 1) minus R22(μr + 1)] and G3
MrR21[R2
1(μr minus 1) minus R22(μr + 1)] -erefore the loaded gap
magnetic intensity can be obtained as follows
HIIr G1 cos(ωt minus φ + δ) + minus G2rminus 2
minus G31113872 1113873 middot cos(φ minus ωt)1113960 1113961
HIIθ G1 sin(ωt minus φ + δ) + G2rminus 2
minus G31113872 1113873 middot sin(φ minus ωt)1113960 1113961
(14)
Furthermore Ez can be obtained byHIIθ and is expressedas the following
Ez μ0ωr G1 cos(ωt minus φ + δ) + G2rminus 2
+ G31113872 1113873cos(φ minus ωt)1113960 1113961
(15)
By substituting equation (15) into equation (9)Πrwill becalculated and according to equation (10) the electro-magnetic torque can be obtained as follows
Tem minus 2πμ0lefG1G2 sin δ δ isin [0 π] (16)
In the 2D electromagnetic field fθ BrBθμ0 and fθ isthe circumferential electromagnetic force density acting onthe stator or rotor -erefore the electromagnetic torque ofthe rotor which is shown in Figure 1 can be written as thefollowing
Tem 2μ0lef 1113946π
0HIIrHIIθr
2dθ (17)
By substituting equation (14) into equation (17) theexpression of the electromagnetic torque Tem is as follows
Tem minus 2πμ0lefG1G3r2sin δ δ isin [0 π]
asymp 6πμ0qKdp1Kso1NsIm
2πR2
R1
R21113888 1113889
2
Mrlefr2sin δ
(18)
In equation (18) when rR2 equation (18) is equal toequation (16) which indicates that the torque acting on therotor surface generated by the Maxwell stress equals thetorque acting on the current sheet generated by the ampereforce -e results show that the electromagnetic torque isproportional to the length of stator iron lef the ampere turnsper unit length of the stator inner surface NsIm2πR2 andthe magnetization Mr while it is proportional to the squareof the ratio (R1R2)
2 (PM radius to stator inner radius)-e cogging torque arises from the interaction of the
magnetomotive force (mmf) harmonics and the airgappermeance harmonics hence it exists in almost all types ofmotors in which the airgap permeance is not constant -ecogging torque will cause undesirable speed pulsationsvibrations possible resonance and acoustic noise [10] Forthe motor shown in Figure (1) because the radial andtangential components of the magnetic flux density vectoras presented in equation (11) are pure sinusoidal the mmfharmonics only contain the fundamental component andthe cogging torque is caused by stator slots In fact thecogging torque can also be calculated by equation (17)however HIIr and HIIθ should be replaced by HPM
IIr and HPMIIθ
respectively In the paper HPMIIr and HPM
IIθ are obtained by thehypothesis of a smooth stator surface hereupon they cannotreveal the slot effects However slot effects can be consideredanalytically by the subdomain technique Works in theliterature [10] have proposed a simple method to predict the
4 Journal of Electrical and Computer Engineering
cogging torque based on the analytical calculation of the air-gap field distribution and the net lateral force acting on theteeth
24 Back-Electromotive Force -e flux through a phasewinding excited by PM and by winding respectively can bewritten as follows
Φpm 2 BPM
IIr( 1113857maxπ
τplef
ΦW 2 BW
IIr( 1113857maxπ
τplef
⎧⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎩
(19)
where τp is the pole pitch and lef is the length of stator iron-e back-emf amplitude in each phase winding assuming a120-degree coil span is then obtained as follows
E 444fkdpvNΦload (20)
where f is the angular frequency of the rotor kdpv is thewinding coefficient N is the number of turns per phase andΦload Φpm +ΦW When ΦW 0 it corresponds to anopen-circuit situation and the back-emf becomes the no-load back-emf E0
25 Armature Reaction Inductance -e constant-powertransformation from the three-phase coordinates to the d-qcoordinate is given by the following
id
iq
i0
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
23
1113970
cos θ cos θ minus2π3
1113874 1113875 cos θ +2π3
1113874 1113875
minus sin θ minus sin θ minus2π3
1113874 1113875 minus sin θ +2π3
1113874 1113875
12
1113970 12
1113970 12
1113970
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
(21)
where θ 1113938ω dt represents the rotor position which is anangle between the d-axis and winding axis of phase A Whenthe above two axes coincide θ 0Whereupon according toequation (21) there are
id
iq
⎡⎣ ⎤⎦ [C]
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (22a)
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ [C]
Tid
iq1113890 1113891 (22b)
where [C] 23
radic 1 minus 12 120
3
radic2 minus
3
radic21113890 1113891 and [C]T and [C]minus 1
are transposed matrix and inverse matrix of [C] re-spectively Notably there is [C]T [C]minus 1 According toequation (22a) we have
id
23
1113970
iA minus12
iB minus12
iC1113874 1113875
iq
23
1113970 3
radic
2iB minus
3
radic
2iC1113888 1113889
⎧⎪⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎪⎩
(23)
θ 0 which means Pωt 0 in equation (7)Substituting (7) into equation (23) yields
id 3
radicIRmscos(β)
iq 3
radicIRmssin(β)
⎧⎨
⎩ (24)
where IRms is the RMS of phase current IRms Im2
radic and β
denotes the electrical angle of the phase current vectorbeyond the d-axis Let β 0 by substituting equation (24)into equation (22b) to get the three-phase AC [iA iB iC]when the three-phase AC flows into the stator windings itwill generate the d-axis armature reaction field Obviouslythe stator iron saturation is most likely to occur when thedirection of the armature reaction field is in line with that ofthe PM field accordingly we only consider this situation
Denoting the inductance matrix in the abc coordinatesystem by
LABC1113858 1113859
LAA LAB LAC
LBA LBB LBC
LCA LCB LCC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (25)
-e flux linkage in the abc coordinate system can berepresented as
λA
λB
λC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ LABC
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (26)
-e flux linkage in the dq coordinate system is alsoapplied to equation (22a) that is
λd
λq
⎡⎣ ⎤⎦ [C]
λA
λB
λC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (27)
Substituting equation (26) and equation (22b) intoequation (27) yields the following
λd
λq
⎡⎣ ⎤⎦ [C] LABC1113858 1113859[C]T
id
iq
⎡⎣ ⎤⎦ Ldq1113960 1113961id
iq
⎡⎣ ⎤⎦ (28)
where [Ldq] Ld Lmd
Lmq Lq1113890 1113891 [C][LABC][C]T it represents
the inductance matrix in the dq coordinate system If the endleakage is not considered and other leakage factors such ashigher harmonics and slot are also ignored then Ld and Lq
will represent the d-axis and q-axis armature reaction in-ductances namely Lad and Laq respectively From the FEM[LABC] will be obtained and then it will turn into [Ldq] byusing the above transformation
Lad and Laq can also be calculated by analytical methodand the process is as follows
Journal of Electrical and Computer Engineering 5
-e load flux linkage of phase A is calculated as
λA NR2lef 1113946αy2
minus αy2BloadIIr R2 θ( 1113857dθ (29)
where N is the number of turns per coil of one phasewinding and αy is the coil pitch angle -e load flux linkageof phase B and C is expressed as
λB λAcos2π3
λC λAcos4π3
⎧⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎩
(30)
According to the above-mentioned [λA λB λC] fromequation (27) we will get [λd λq] And then there is
Ld λd
id (31)
Similarly neglecting some leakage factors Ld calculatedby equation (31) corresponds to Lad
3 Example Validation
To verify the analytic solution to the electromagnetic torquea locked-rotor simulation is carried out by the FEMmethod-e motor parameters are listed in Table 1
In the simulation the windings are fed by a symmetricalthree-phase alternating current to produce a rotating ar-mature reaction field but the rotor speed is set to zero thatis the rotor is locked When the armature reaction fieldrotates relative to the rotor with the angular velocity ω thelocked-rotor torque will generate and vary periodically withδ (an angle between two axes of the armature reaction fieldand PM field and it changes periodically from 0 to 360degrees) -e locked-rotor phenomenon can be applied toverify the electromagnetic torque of equation (16) andequation (18)
When the motor is fed with currents as shown inequation (7) the locked-rotor torques are as shown inFigure 3 -ree kinds of input currents of 2000Hz withdifferent amplitudes such as 10A 30A and 50A areconsidered Figure 3(a) shows the three-phase currents witha 10A peak value in FEM -e locked-rotor torque curvesobtained by FEM are shown in Figure 3(b) which shows thatthe locked-rotor torque is also sinusoidal with time and itspeak value is increased with the increasing amplitude of thephase current A comparison of the results of FEM and theanalytical model is illustrated in Figure 3(c) in which it canbe found that the analytical solution is in good agreementwith the FEM solution
Peak torques obtained by the FE model are comparedwith those from the derived analytical equations-e resultswhich are listed in Table 2 show that the errors are less than5 which proves the correctness of equation (16) andequation (18)
-emagnetic field results that were calculated by FEM areshown in Figure 4 where red and blue conductors indicate the
current flows out of and into the cross section respectively Inaddition the PMhas a parallel magnetization (the direction ofthe magnetization is vertically upward)
Because the PM field is much larger than the armaturereaction field the magnetic fields of the motor in Figure 4 atdifferent times are approximately the same At 02ms thelocked-rotor torque in Figure 3(b) is negative and thecorresponding motor magnetic field is shown in Figure 4(a)in which the armature reaction field lags behind the PM fieldby approximately 135 degrees which just causes a negativetorque Moreover the armature reaction field weakens thePM field the maximum flux density occurs in the statortooth and its value is 138 T At 04ms the locked-rotortorque equals zero as shown in Figure 3(b) and the cor-responding motor magnetic field is shown in Figure 4(b)where the armature reaction field coincides with the PMfield which verifies the null torque At 04ms the maximumflux density of the stator tooth is 170 T At 05ms thelocked-rotor torque approaches the maximum as shown inFigure 3(b) and the corresponding motor magnetic field isshown in Figure 4(c) It can be seen that the armature re-action field direction is left (90 degrees ahead of the PMfield) which verifies the maximum torque At 05ms themaximum flux density of the stator tooth is 160 T
-ree methods namely the FM method the air-gappermeance method [10] and the subdomain method wereused to estimate the cogging torque and the results areshown in Figure 5 -e air-gap permeance method is basedon the assumption that the flux crosses the magnet and airgap in a straight line wherever a magnet faces a toothHowever the above assumption will no longer be suitablefor the motor mentioned in the paper because the flux willbend in the air gap as shown in Figure 5(a) the coggingtorque calculated by the air-gap permeance method is smallrelative to that from FEM and the subdomain method asshown in Figure 5(b) -e cogging torque peak is approx-imately 370 μNm and is 005 of the rated torque (08Nm)which is too small to be ignored In fact some degrees ofharmonic components will exist in the polarized PM thusthe cogging torque peak will increase
-e stator tooth tips in relation to the width of the slotopenings also influence the back-emf waveform When theheight of the tooth tips is small magnetic saturation existsand the emf waveform tends to be a trapezoid Howeverwhen the tooth tip height is too large the emf waveform willdeteriorate [11] -e motor consists of a stator having 12
Table 1 Parameters of the high-speed PMSM
Symbol Quantity ValuesR1 PM radius 12times10minus 3mR2 Inner radius of stator 174times10minus 3mQ Slot numbers 12lef Core length 55times10minus 3mM Magnetization 820000Amω Angular velocity 12560 radsNs Conductors per slot 20μr Relative recoil permeability 1038P Number of pole pairs 1
6 Journal of Electrical and Computer Engineering
teeth which carry overlapping winding -e no-load back-emf waveform obtained from the FEM is displayed inFigure 6(a) and the emf waveform is almost purely sinu-soidal which is ideal for ultra-high-speed PMSMs As shownin Figure 5(a) a stator tooth has the maximum flux density1542 T which does not reach the saturation region of thematerial-e comparison of the no-load back-emf calculatedby equation (20) and that of FEM is shown in Figure 6(b)which manifests a good agreement of both methodsFigure 6(c) shows the prototype of the PMSM in which therotor is supported by gas foil bearings and the experimentalresults are shown in Figure 6(d) A coast-down test under60000 rmin is carried out that is the frequency converterwill shut down when the rotor is stable at 60000 rmin and
then the rotor speed will decrease to zero gradually byovercoming air friction In this process the stator has a back-emf but no current thereupon the rotor is not subjected tothe electromagnetic force Hereupon the curve of a no-loadback-emf vs speed can be achieved -e experimental anddesign values of the no-load back-emf are 570V and 596 Vat 746Hz respectively and the difference between them iswithin 46
-e results of Lad and Laq are shown in Figure 7 It canbe found that Lad and Laq are same when current is smalland both of them decrease with the increase of the currentBecause the d-axial magnetic circuit is more saturated Laddescends faster -e magnetic saturation is neglected inthe analytical method whereas Lad not varies with thecurrent
When IRms 10A and 100A the errors of the twomethods are 9 and 2 respectively It shows that theresults of the analytical method and FEM are very close Inthe analytical method only the fundamental harmonic isused to calculate the flux linkage However in the FEMexcept for the fundamental harmonic high-order harmonics
009 020 040 060 080 100 120ndash1000
1000
XY Plot 3torque_Cal_10kw120krpm
ndash750
750500
ndash500ndash250
250000
Curr
ent (
A)
Input current (phase A)
Input current (phase C)Input current (phase B)
Time (ms)
(a)
XY Plot 1torque_Cal_10kw120krpm
Torque Im = lsquo10ArsquoTorque_1 Im = lsquo30ArsquoTorque_2 Im = lsquo50Arsquo
Time (ms)
ndash150
150
ndash100
100
ndash050
050
000
020 040 060 080 120100 130000
Y1 (N
ewto
nMet
er)
(b)
0 02 04 06 08 1 12ndash15
ndash05
0
05
1
15
ndash1
2
t (ms)
Torq
ue T
emN
m
FE 10AFE 30AFE 50A
Analytic 10AAnalytic 30AAnalytic 50A
(c)
Figure 3-e results of the locked-rotor (a)-e three-phase alternating currents (b)-e torque vs time (c) Result comparison of FEM andthe analytical model
Table 2 Peak torque comparison between FE and the theory
Im (A) FEM (Nmiddotm) Analytical (Nmiddotm) Error ()10 0266 0271 18730 0802 0815 16250 1333 1357 176
Journal of Electrical and Computer Engineering 7
B (tesla)13842e + 000
12112e + 00011247e + 000
95165e ndash 00186514e ndash 001
69212e ndash 00160561e ndash 00151910e ndash 00143259e ndash 00134608e ndash 00125957e ndash 001
86550e ndash 00240351e ndash 005
17306e ndash 001
77863e ndash 001
10382e + 000
12977e + 000
Time = 00002sSpeed = 0000000rpmPosition = 0000000deg
0 40 80 (mm)
(a)
B (tesla)17001e + 000
13813e + 000
10626e + 000
12540e ndash 00410637e ndash 00121262e ndash 00131887e ndash 00142512e ndash 00153136e ndash 00163761e ndash 00174386e ndash 00185011e ndash 00195936e ndash 001
11689e + 00012751e + 000
14876e + 00015938e + 000
Speed = 0000000rpmPosition = 0000000deg
Time = 00004s0 40 80 (mm)
(b)B (tesla)
16056e + 000
90322e ndash 001
14749e ndash 00410049e ndash 00120083e ndash 00130117e ndash 00140152e ndash 00150186e ndash 00160220e ndash 00170254e ndash 00180288e ndash 001
10036e + 00011039e + 00012043e + 00013046e + 00014049e + 00015053e + 000
Position = 0000000deg
Time = 0000500000000000001sSpeed = 0000000rpm
40 80 (mm)0
(c)
Figure 4 Flux density vector field and current vector field of the FE model (a) Fields at 00002 s (b) Fields at 00004 s (c) Fields at 00005 s
B (tesla)15428e + 00014464e + 00013500e + 000
11572e + 00010608e + 000
96871e ndash 00219326e ndash 00128966e ndash 00138605e ndash 00148245e ndash 00157884e ndash 00167523e ndash 001
47663e ndash 004
77163e ndash 00186802e ndash 00196442e ndash 001
12536e + 000
(a)
10 20155 25 300Angle (degmech)
ndash400
400
ndash200
200
ndash100
100
0
ndash300
300
Cog
ging
torq
ue (μ
Nm
)
FEPermeanceSubdomain
(b)
Figure 5 PM field and cogging torque (a) PM field (b) Comparison of the cogging torque of FEM and the analytical model
8 Journal of Electrical and Computer Engineering
are also taken into account in the flux linkage -erefore theresult of the analytical method is smaller than that of FEM
It is worth noting that the stator tooth has the mostserious saturation as shown in Figure 8(a) and the fluxdensities reach 2086 T and 1880 T when IRms 100A(28IN) and 50A (14IN) respectively IN denotes the ratedcurrent of the motor -e working points of stator toothcorresponding to the both working condition are illustratedin Figure 8(b) It shows that the working point is located atthe end of the linear segment of the B-H curve whenIRms 14IN It indicates that the magnetic saturation of thestator iron is not prominent when the prototype worksunder rated working conditions and it applies to the pro-posed analytical method in this paper
In addition in order to enhance the cooling of the rotorthe air gap should be chosen as large as possible so thatmagnetic saturation in the tooth tips rarely occurs Fur-thermore the teeth in a 12-slot stator are diametricallysymmetrical the air-gap field distribution is symmetrical
000 020 040 060 080 100Time (ms)
ndash25000
ndash12500
000
12500
25000
Y1 (V
)XY Plot 1
Curve info rms_1 maxInduced voltage (phase A) 1560441 2188502Induced voltage (phase B) 1563378 2186613Induced voltage (phase C) 1573086 2184715
(a)
0 01 02 03 04 05Time (ms)
FE (phase A)Analytic
Vol
tage
(V)
ndash300
ndash200
ndash100
0
100
200
300
(b)
(c)
Time (s)0 50 100 150 200
0
500
1000
1500Fr
eque
ncy
(Hz)
0
50
100
150
Vol
tage
RM
SV
FrequencyVoltage RMS
(d)
Figure 6 Results of the no-load back-emf (a) No-load back-emf of FEM at 120000 rpm (b) Comparison of no-load back-emf at120000 rpm (c) Prototype of the ultra-high-speed PMSM (d) Experimental curves of the phase voltage and frequency
0 20 40 60 80 1000098
01
0102
0104
0106
0108
011
Phase current IRMSA
Lad FEMLaq FEMLad Analytical
d- an
d q-
axis
indu
ctan
ces (
mH
)
Figure 7 Curves of Lad and Laq vs the current
Journal of Electrical and Computer Engineering 9
and no unbalanced magnetic pull will exist if the rotor andstator axes are coincident Although the rotor rotates ec-centrically with respect to the stator in fact the eccentricityis too tiny due to the constraint of bearings to cause a notableunbalanced magnetic pull
4 Conclusion
For the ultra-high-speed PMSM including a stator withoverlapping windings and the 2-pole parallel magnetizedcylindrical PM electromagnetic torques are rarely reportedat present which leads to a prolonged motor design cycle toevaluating motor performance -erefore the work of thispaper is as follows
(1) -is paper deduces the analytical solution to theelectromagnetic driving torque of ultra-high-speedPMSMs which is verified by the FE methodaccording to locked-rotor simulation -e resultsshow that the electromagnetic torque is pro-portional to the length of stator iron the ampereturns per unit length of the stator inner surface andthe magnetization while it is proportional to thesquare of the ratio ie PM radius to stator innerradius
(2) -e cogging torque back-emf armature reactioninductance and so on are calculated compared anddiscussed For the PMSM studied in the paper theair-gap permeance method was shown not to beapplicable to the estimation of the cogging torquedue to the fact that the flux bends in the air gap andthus a high-accuracy method (eg the subdomainmethod) should be adopted
(3) A prototype is designed and manufactured and theno-load back-emf is checked by a 60000 rpm freecoast experiment -e experimental and designvalues of the no-load back-emf are 570 V and 596V
at 746Hz respectively and the difference betweenthem is within 46
Data Availability
-e data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
-e authors declare that they have no conflicts of interest
Acknowledgments
-is work is supported by National Natural ScienceFoundation of China (Grant nos 51705413 51705416 and11502196) and China Postdoctoral Science Foundation(Grant no 2017M613291XB)
References
[1] A Tenconi S Vaschetto and A Vigliani ldquoElectrical machinesfor high-speed applications design considerations andtradeoffsrdquo IEEE Transactions on Industrial Electronics vol 61no 6 pp 3022ndash3029 2014
[2] J H Ahn J Y Choi C H Park C Han C W Kim andT G Yoon ldquoCorrelation between rotor vibration and me-chanical stress in ultra-high-speed permanent magnet syn-chronous motorsrdquo IEEE Transactions on Magnetics vol 53no 11 Article ID 8209906 2017
[3] M-S Lim S-H Chai J-S Yang and J-P Hong ldquoDesign andverification of 150 krpm PMSM based on experiment resultsof prototyperdquo IEEE Transactions on Industrial Electronicsvol 62 no 12 pp 7827ndash7836 2015
[4] D-K Hong B-C Woo J-Y Lee and D-H Koo ldquoUltra highspeed motor supported by air foil bearings for air blowercooling fuel cellsrdquo IEEE Transactions on Magnetics vol 48no 2 pp 871ndash874 2012
20864e + 00019560e + 00018256e + 00016953e + 00015649e + 00014345e + 00013041e + 00011737e + 00010433e + 00091289e ndash 00178250e ndash 00165210e ndash 00152171e ndash 00139132e ndash 00126092e ndash 00113053e ndash 00113431e ndash 004
B (tesla)
(a)
375
250
125
000000E + 000 500E + 005
H (A_per_meter)
B (te
sla)
Working point 28IN
Working point 14IN
(b)
Figure 8 Magnetic saturation condition of the stator iron (a) Distribution of flux density for IRms 100A (b) Working points of the statortooth
10 Journal of Electrical and Computer Engineering
[5] D-K Hong B-C Woo Y-H Jeong D-H Koo andC-W Ahn ldquoDevelopment of an ultra high speed permanentmagnet synchronous motorrdquo International Journal of Pre-cision Engineering and Manufacturing vol 14 no 3pp 493ndash499 2013
[6] S-M Jang H-W Cho and S-K Choi ldquoDesign and analysisof a high-speed brushless DC motor for centrifugal com-pressorrdquo IEEE Transactions on Magnetics vol 43 no 6pp 2573ndash2575 2007
[7] D-K Hong D-S Joo B-CWoo D-H Koo and C-W AhnldquoUnbalance response analysis and experimental validation ofan ultra high speed motor-generator for microturbine gen-erators considering balancingrdquo Sensors vol 14 no 9pp 16117ndash16127 2014
[8] A B Proca A Keyhani A EL-Antably W Lu and M DaildquoAnalytical model for permanent magnet motors with surfacemounted magnetsrdquo IEEE Transactions on Energy Conversionvol 18 no 3 pp 386ndash391 2003
[9] F Dubas C Espanet and A Miraoui ldquoAn original analyticalexpression of the maximum magnet thickness in surfacemounted permanent magnet motorsrdquo De European PhysicalJournal Applied Physics vol 38 no 2 pp 169ndash176 2007
[10] Z Q Zhu and D Howe ldquoAnalytical prediction of the coggingtorque in radial-field permanent magnet brushless motorsrdquoIEEE Transactions on Magnetics vol 28 no 2 pp 1371ndash13741992
[11] Z Q Zhu K Ng and D Howe ldquoDesign and analysis of high-speed brushless permanent magnet motorsrdquo in Proceedings ofthe International Conference on Electrical Machines amp Drivespp 381ndash385 IET Cambridge UK September 1997
[12] W Cheng L Yu L Huang Y Lv L Li and Y Sun ldquoAn-alytical solution to magnetic field distribution of a parallelmagnetised rotor with cylindrical or ring-type permanentmagnetrdquo IET Electric Power Applications vol 9 no 6pp 429ndash437 2015
[13] Z Q Zhu and D Howe ldquoInstantaneous magnetic field dis-tribution in brushless permanent magnet DC motors IIArmature-reaction fieldrdquo IEEE Transactions on Magneticsvol 29 no 1 pp 136ndash142 1993
Journal of Electrical and Computer Engineering 11
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
gave analytical solutions to the PM field of bipolar rotorswith ring or cylindrical PMs that are magnetized inparallel Nevertheless most relevant literature studies onanalytical models for predicting the armature reactionfield distribution are successful for the motor owing tothe high permeability shaft [13] however few studiesinvolve the hollow stator bore as shown in Figure 2(b)
For ultra-high-speed PMSMs including a stator withoverlapping windings and 2-pole parallel magnetizedcylindrical PMs analytical solutions to the magnetic fieldand corresponding electromagnetic torque are rarely re-ported at present which has resulted in a prolonged motordesign cycle to evaluating the motor performance Due tothe need of cooling and reduction of pulse torque theultra-high-speed PMSM generally adopts a large air gaphereupon the slot effect is very weak -erefore the as-sumption of ldquoslotless stator + equivalent current sheetrdquocan be introduced and some convenient formulas ofestimating motor performance are expected to beobtained
Hence this paper will first deduce the magnetic fieldthen obtain the electromagnetic driving torque by using thePoynting theorem andMaxwell stress tensor and then verifythe correctness by using the FE method In addition thecogging torque the back-electromotive force (emf) andarmature reaction inductance are calculated compared anddiscussed A 10 kW 120000 rpm prototype is designed andmanufactured and the no-load back-emf is checked by a60000 rpm free coast experiment
2 Analytical Model
-e cross section of the ultra-high-speed PMSMs is shown inFigure 1 in which slotted stator and distribution winding areused and the rotor is composed by installing a cylindricalparallel polarized PM into the metallic sleeve
For the sintered NdFeB magnet the demagnetizationcurve is approximately a straight line and the relativerecoil permeability μr is approximate to that of air μr waschosen from 1029ndash1100 in most literatures To simplifythe field calculation the following assumptions aremade
(1) -e PM has a linear second-quadrant de-magnetization curve and the permeability of thestator iron is infinite
(2) -e winding end effect and rotor eddy current areneglected
(3) -e PM field is calculated with a 2D solution byassuming a smooth stator surface that is the slottingis neglected as shown in Figure 2(a)
(4) Conductors in the stator slot can be equivalent to thecurrent sheet distributing evenly along the slotopening as shown in Figure 2(b)
In the calculation of the PM field the slot-opening effectis neglected therefore the calculation model of the PM fieldis shown in Figure 2(a) where R1 and R2 are the outer radius
and the inner radius of the PM and I and II represent thePM and the air region respectively
-e sleeve applies to a low permeability alloy which canbe physically equivalent to the air therefore in the calcu-lation model of the armature reaction field the stator boreseems to be full of air as shown in Figure 2(b) where b0 isthe width of the slot opening
-e loaded magnetic field is the superposition of theopen-circuit PM and the armature reaction field and theloaded air-gap flux density is defined as follows
BloadIIr BW
IIr + BPMIIr
BloadIIθ BW
IIθ + BPMIIθ
⎧⎨
⎩ (1)
In equation (1) load PM and W represent the loadedPM and winding magnetic fields respectively r and θrepresent the radial and circumferential directionsrespectively
21 Open-Circuit Magnetic Field -e literature [12] hasdeduced the solution to the PM field shown in Figure 2(a)which is given as follows
BPMIIr
minus μ0MrR21 1 + R2
2rminus 2( 1113857
R21 μr minus 1( 1113857 minus R2
2 μr + 1( 1113857cos θ
BPMIIθ
μ0MrR21 minus 1 + R2
2rminus 2( 1113857
R21 μr minus 1( 1113857 minus R2
2 μr + 1( 1113857sin θ
(2)
where Mr is the amplitude of the residual magnetization ofthe PM μr is the relative recoil permeability of the PM μ0 isthe vacuum permeability and r and θ represent the radialand circumferential positions respectively
22 Armature Reaction Field -e governing equation of theair-gap region is as follows
zϕ2
z2r+
zϕrzr
+zϕ2
r2zθ2 0 (3)
where ϕ is a scalar magnetic potential and the relation of ϕand H (magnetic field intensity) is expressed as follows
Hθ minuszϕ
r zθ
Hr minuszϕzr
⎧⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎩
(4)
In the inner radius of the stator
Hθ1113868111386811138681113868 rR2
minus J (5)
where J is the Fourier series expansion of the current sheetAt the center of the stator the H should meet the following
Hθ1113868111386811138681113868 r⟶0neinfin
Hr
1113868111386811138681113868 r⟶0neinfin(6)
Taking into account the symmetry of the magneticfield the single conductor field can be solved according toequations (3)ndash(6) -e method to solve the armature
2 Journal of Electrical and Computer Engineering
reaction field is as follows the single conductor field issolved first and then the single-turn coil field can beobtained by adding two single conductor fields at a polardistance apart from each other Whereupon the single-phase winding field is superpositioned onto fields that areexcited by the multiple single-turn coils and finally thearmature reaction field can be calculated by super-imposing the fields of three-phase winding
For distributed winding multiple single-turn coil fieldsthat are a pitch apart from each other can be equivalent to aconcentrated winding by taking into account the distribu-tion coefficient and the pitch-shortening coefficient -eabove-mentioned treatment will lightly change higherharmonic components of the armature reaction field butvastly simplify the solution of the single-phase winding fieldwith good accuracy In addition it is assumed that in-stantaneous three-phase currents (ia ib ic) only involvefundamental components and they can be written asfollows
ia Im cos(Pωt + β)
ib Im cos Pωt + β minus2π3
1113874 1113875
ic Im cos Pωt + β minus4π3
1113874 1113875
⎧⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎪⎪⎪⎪⎩
(7)
where Im is the current amplitude of a single conductor P isthe number of pole pairs ω is the synchronous angularvelocity and β is the current phase
-e armature reaction field in Figure 2(b) is expressed asfollows
BWIIr
minus 2qNsμ0π
1113944
infin
v11113896Kdpv(2v minus 1)Ksov(2v minus 1)
middot Fo(2v minus 1 r) middot (minus 1)v
middot 1113890ia cos[(2v minus 1)(α minus c)]
+ ib cos (2v minus 1) α minus c minus2π3
1113874 11138751113876 1113877
+ ic cos (2v minus 1) α minus c minus4π3
1113874 11138751113876 111387711138911113897
BWIIα
minus 2qNsμ0π
1113944
infin
v11113896Kdpv(2v minus 1)Ksov(2v minus 1)
middot Fo(2v minus 1 r) middot (minus 1)vminus 1
middot 1113890ia sin[(2v minus 1)(α minus c)]
+ ib sin (2v minus 1) α minus c minus2π3
1113874 11138751113876 1113877
+ ic sin (2v minus 1) α minus c minus4π3
1113874 11138751113876 111387711138911113897
(8)
where Ksov(v) sin(vb0(2R2))(vb0(2R2)) is the slot-opening coefficient which is a function of the spatial har-monic order vFo(r v) rvminus 1Rv
2 reflects the influence of theair-gap thickness on the radialcircumferential flux densityof the v order harmonic wave α is the angle relative to thehorizontal direction as shown in Figure 2(b) and is positive
Mrθ
Mr
Mrr
R1
R2
θ
I (PM)
II (Air gap)
(a)
Stator bore
II
Air b0Slot
α = 0
Current sheetR2
(b)
Figure 2 Models of the electromagnetic field of the PMSMs (a) Open-circuit PM field (b) Armature reaction field
Stator iron
Winding
Sleeve
Magnet
Magnetizationdirection
Figure 1 Cross section of the ultra-high-speed PMSM
Journal of Electrical and Computer Engineering 3
in the counter clockwise direction c is an angle between theaxis of winding A and the d-axis q is the number of slots perphase per pole and Ns is the number of conductors per slot
23 Electromagnetic Torque andCoggingTorque In this parttwo methods namely the Poynting theorem and theMaxwell stress tensor were used to deduce the analyticexpression of the electromagnetic torque For the 2D fieldthe electric field strength vector E in the air-gap region onlyincludes the axial component Ez therefore the Poyntingvector is given by the following
1113937r
minus EzHθ
1113937θ
EzHr
1113937z
0
⎧⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎩
(9)
where Ez μ0 1113938(zHθzt)dr -e electromagnetic powertransmitting from the stator to the rotor is mainly dependenton the radial component of the Poynting vector Πr whichcan be calculated from equation (10) as follows
PM Temω lef 11139462π
01113945
r
r dθ (10)
where lef is the length of the stator iron Taking the hori-zontal axis in Figure 2(b) as the reference position the PMfield is changed as follows
BPMIIr minus
μ0MrR21 1 + R2
2rminus 2( 1113857
R21 μr minus 1( 1113857 minus R2
2 μr + 1( 1113857middot cos(φ)
BPMIIθ
μ0MrR21 minus 1 + R2
2rminus 2( 1113857
R21 μr minus 1( 1113857 minus R2
2 μr + 1( 1113857middot sin(φ)
(11)
where φ θ minus π2 α It is assumed that the electromag-netic power is mainly transmitted by the fundamentalmagnetic field of armature reaction field When the numberof pole pairs equals 1 by substituting equation (7) intoequation (8) the armature reaction field will become asfollows using the properties of trigonometric functions
BWIIr μ0G1 cos(ωt minus φ + δ)
BWIIθ μ0G1 sin(ωt minus φ + δ)
(12)
where δ β + c represents the angle between the axis of ar-mature reaction field and the axis of PM field andG1 (3qNsKdp1Kso1Im)(πR2) Equation (12) shows that thearmature reaction field is a travelling wave but the PM field is astanding wave shown by equation (11) To make the PM fieldrotate counterclockwise synchronously relative to the armaturereaction field equation (11) can written as the following
BPMIIr μ0 minus G2r
minus 2minus G31113872 1113873 middot cos(φ minus ωt)
BPMIIθ μ0 G2r
minus 2minus G31113872 1113873 middot sin(φ minus ωt)
(13)
where G2 MrR21R
22[R2
1(μr minus 1) minus R22(μr + 1)] and G3
MrR21[R2
1(μr minus 1) minus R22(μr + 1)] -erefore the loaded gap
magnetic intensity can be obtained as follows
HIIr G1 cos(ωt minus φ + δ) + minus G2rminus 2
minus G31113872 1113873 middot cos(φ minus ωt)1113960 1113961
HIIθ G1 sin(ωt minus φ + δ) + G2rminus 2
minus G31113872 1113873 middot sin(φ minus ωt)1113960 1113961
(14)
Furthermore Ez can be obtained byHIIθ and is expressedas the following
Ez μ0ωr G1 cos(ωt minus φ + δ) + G2rminus 2
+ G31113872 1113873cos(φ minus ωt)1113960 1113961
(15)
By substituting equation (15) into equation (9)Πrwill becalculated and according to equation (10) the electro-magnetic torque can be obtained as follows
Tem minus 2πμ0lefG1G2 sin δ δ isin [0 π] (16)
In the 2D electromagnetic field fθ BrBθμ0 and fθ isthe circumferential electromagnetic force density acting onthe stator or rotor -erefore the electromagnetic torque ofthe rotor which is shown in Figure 1 can be written as thefollowing
Tem 2μ0lef 1113946π
0HIIrHIIθr
2dθ (17)
By substituting equation (14) into equation (17) theexpression of the electromagnetic torque Tem is as follows
Tem minus 2πμ0lefG1G3r2sin δ δ isin [0 π]
asymp 6πμ0qKdp1Kso1NsIm
2πR2
R1
R21113888 1113889
2
Mrlefr2sin δ
(18)
In equation (18) when rR2 equation (18) is equal toequation (16) which indicates that the torque acting on therotor surface generated by the Maxwell stress equals thetorque acting on the current sheet generated by the ampereforce -e results show that the electromagnetic torque isproportional to the length of stator iron lef the ampere turnsper unit length of the stator inner surface NsIm2πR2 andthe magnetization Mr while it is proportional to the squareof the ratio (R1R2)
2 (PM radius to stator inner radius)-e cogging torque arises from the interaction of the
magnetomotive force (mmf) harmonics and the airgappermeance harmonics hence it exists in almost all types ofmotors in which the airgap permeance is not constant -ecogging torque will cause undesirable speed pulsationsvibrations possible resonance and acoustic noise [10] Forthe motor shown in Figure (1) because the radial andtangential components of the magnetic flux density vectoras presented in equation (11) are pure sinusoidal the mmfharmonics only contain the fundamental component andthe cogging torque is caused by stator slots In fact thecogging torque can also be calculated by equation (17)however HIIr and HIIθ should be replaced by HPM
IIr and HPMIIθ
respectively In the paper HPMIIr and HPM
IIθ are obtained by thehypothesis of a smooth stator surface hereupon they cannotreveal the slot effects However slot effects can be consideredanalytically by the subdomain technique Works in theliterature [10] have proposed a simple method to predict the
4 Journal of Electrical and Computer Engineering
cogging torque based on the analytical calculation of the air-gap field distribution and the net lateral force acting on theteeth
24 Back-Electromotive Force -e flux through a phasewinding excited by PM and by winding respectively can bewritten as follows
Φpm 2 BPM
IIr( 1113857maxπ
τplef
ΦW 2 BW
IIr( 1113857maxπ
τplef
⎧⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎩
(19)
where τp is the pole pitch and lef is the length of stator iron-e back-emf amplitude in each phase winding assuming a120-degree coil span is then obtained as follows
E 444fkdpvNΦload (20)
where f is the angular frequency of the rotor kdpv is thewinding coefficient N is the number of turns per phase andΦload Φpm +ΦW When ΦW 0 it corresponds to anopen-circuit situation and the back-emf becomes the no-load back-emf E0
25 Armature Reaction Inductance -e constant-powertransformation from the three-phase coordinates to the d-qcoordinate is given by the following
id
iq
i0
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
23
1113970
cos θ cos θ minus2π3
1113874 1113875 cos θ +2π3
1113874 1113875
minus sin θ minus sin θ minus2π3
1113874 1113875 minus sin θ +2π3
1113874 1113875
12
1113970 12
1113970 12
1113970
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
(21)
where θ 1113938ω dt represents the rotor position which is anangle between the d-axis and winding axis of phase A Whenthe above two axes coincide θ 0Whereupon according toequation (21) there are
id
iq
⎡⎣ ⎤⎦ [C]
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (22a)
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ [C]
Tid
iq1113890 1113891 (22b)
where [C] 23
radic 1 minus 12 120
3
radic2 minus
3
radic21113890 1113891 and [C]T and [C]minus 1
are transposed matrix and inverse matrix of [C] re-spectively Notably there is [C]T [C]minus 1 According toequation (22a) we have
id
23
1113970
iA minus12
iB minus12
iC1113874 1113875
iq
23
1113970 3
radic
2iB minus
3
radic
2iC1113888 1113889
⎧⎪⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎪⎩
(23)
θ 0 which means Pωt 0 in equation (7)Substituting (7) into equation (23) yields
id 3
radicIRmscos(β)
iq 3
radicIRmssin(β)
⎧⎨
⎩ (24)
where IRms is the RMS of phase current IRms Im2
radic and β
denotes the electrical angle of the phase current vectorbeyond the d-axis Let β 0 by substituting equation (24)into equation (22b) to get the three-phase AC [iA iB iC]when the three-phase AC flows into the stator windings itwill generate the d-axis armature reaction field Obviouslythe stator iron saturation is most likely to occur when thedirection of the armature reaction field is in line with that ofthe PM field accordingly we only consider this situation
Denoting the inductance matrix in the abc coordinatesystem by
LABC1113858 1113859
LAA LAB LAC
LBA LBB LBC
LCA LCB LCC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (25)
-e flux linkage in the abc coordinate system can berepresented as
λA
λB
λC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ LABC
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (26)
-e flux linkage in the dq coordinate system is alsoapplied to equation (22a) that is
λd
λq
⎡⎣ ⎤⎦ [C]
λA
λB
λC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (27)
Substituting equation (26) and equation (22b) intoequation (27) yields the following
λd
λq
⎡⎣ ⎤⎦ [C] LABC1113858 1113859[C]T
id
iq
⎡⎣ ⎤⎦ Ldq1113960 1113961id
iq
⎡⎣ ⎤⎦ (28)
where [Ldq] Ld Lmd
Lmq Lq1113890 1113891 [C][LABC][C]T it represents
the inductance matrix in the dq coordinate system If the endleakage is not considered and other leakage factors such ashigher harmonics and slot are also ignored then Ld and Lq
will represent the d-axis and q-axis armature reaction in-ductances namely Lad and Laq respectively From the FEM[LABC] will be obtained and then it will turn into [Ldq] byusing the above transformation
Lad and Laq can also be calculated by analytical methodand the process is as follows
Journal of Electrical and Computer Engineering 5
-e load flux linkage of phase A is calculated as
λA NR2lef 1113946αy2
minus αy2BloadIIr R2 θ( 1113857dθ (29)
where N is the number of turns per coil of one phasewinding and αy is the coil pitch angle -e load flux linkageof phase B and C is expressed as
λB λAcos2π3
λC λAcos4π3
⎧⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎩
(30)
According to the above-mentioned [λA λB λC] fromequation (27) we will get [λd λq] And then there is
Ld λd
id (31)
Similarly neglecting some leakage factors Ld calculatedby equation (31) corresponds to Lad
3 Example Validation
To verify the analytic solution to the electromagnetic torquea locked-rotor simulation is carried out by the FEMmethod-e motor parameters are listed in Table 1
In the simulation the windings are fed by a symmetricalthree-phase alternating current to produce a rotating ar-mature reaction field but the rotor speed is set to zero thatis the rotor is locked When the armature reaction fieldrotates relative to the rotor with the angular velocity ω thelocked-rotor torque will generate and vary periodically withδ (an angle between two axes of the armature reaction fieldand PM field and it changes periodically from 0 to 360degrees) -e locked-rotor phenomenon can be applied toverify the electromagnetic torque of equation (16) andequation (18)
When the motor is fed with currents as shown inequation (7) the locked-rotor torques are as shown inFigure 3 -ree kinds of input currents of 2000Hz withdifferent amplitudes such as 10A 30A and 50A areconsidered Figure 3(a) shows the three-phase currents witha 10A peak value in FEM -e locked-rotor torque curvesobtained by FEM are shown in Figure 3(b) which shows thatthe locked-rotor torque is also sinusoidal with time and itspeak value is increased with the increasing amplitude of thephase current A comparison of the results of FEM and theanalytical model is illustrated in Figure 3(c) in which it canbe found that the analytical solution is in good agreementwith the FEM solution
Peak torques obtained by the FE model are comparedwith those from the derived analytical equations-e resultswhich are listed in Table 2 show that the errors are less than5 which proves the correctness of equation (16) andequation (18)
-emagnetic field results that were calculated by FEM areshown in Figure 4 where red and blue conductors indicate the
current flows out of and into the cross section respectively Inaddition the PMhas a parallel magnetization (the direction ofthe magnetization is vertically upward)
Because the PM field is much larger than the armaturereaction field the magnetic fields of the motor in Figure 4 atdifferent times are approximately the same At 02ms thelocked-rotor torque in Figure 3(b) is negative and thecorresponding motor magnetic field is shown in Figure 4(a)in which the armature reaction field lags behind the PM fieldby approximately 135 degrees which just causes a negativetorque Moreover the armature reaction field weakens thePM field the maximum flux density occurs in the statortooth and its value is 138 T At 04ms the locked-rotortorque equals zero as shown in Figure 3(b) and the cor-responding motor magnetic field is shown in Figure 4(b)where the armature reaction field coincides with the PMfield which verifies the null torque At 04ms the maximumflux density of the stator tooth is 170 T At 05ms thelocked-rotor torque approaches the maximum as shown inFigure 3(b) and the corresponding motor magnetic field isshown in Figure 4(c) It can be seen that the armature re-action field direction is left (90 degrees ahead of the PMfield) which verifies the maximum torque At 05ms themaximum flux density of the stator tooth is 160 T
-ree methods namely the FM method the air-gappermeance method [10] and the subdomain method wereused to estimate the cogging torque and the results areshown in Figure 5 -e air-gap permeance method is basedon the assumption that the flux crosses the magnet and airgap in a straight line wherever a magnet faces a toothHowever the above assumption will no longer be suitablefor the motor mentioned in the paper because the flux willbend in the air gap as shown in Figure 5(a) the coggingtorque calculated by the air-gap permeance method is smallrelative to that from FEM and the subdomain method asshown in Figure 5(b) -e cogging torque peak is approx-imately 370 μNm and is 005 of the rated torque (08Nm)which is too small to be ignored In fact some degrees ofharmonic components will exist in the polarized PM thusthe cogging torque peak will increase
-e stator tooth tips in relation to the width of the slotopenings also influence the back-emf waveform When theheight of the tooth tips is small magnetic saturation existsand the emf waveform tends to be a trapezoid Howeverwhen the tooth tip height is too large the emf waveform willdeteriorate [11] -e motor consists of a stator having 12
Table 1 Parameters of the high-speed PMSM
Symbol Quantity ValuesR1 PM radius 12times10minus 3mR2 Inner radius of stator 174times10minus 3mQ Slot numbers 12lef Core length 55times10minus 3mM Magnetization 820000Amω Angular velocity 12560 radsNs Conductors per slot 20μr Relative recoil permeability 1038P Number of pole pairs 1
6 Journal of Electrical and Computer Engineering
teeth which carry overlapping winding -e no-load back-emf waveform obtained from the FEM is displayed inFigure 6(a) and the emf waveform is almost purely sinu-soidal which is ideal for ultra-high-speed PMSMs As shownin Figure 5(a) a stator tooth has the maximum flux density1542 T which does not reach the saturation region of thematerial-e comparison of the no-load back-emf calculatedby equation (20) and that of FEM is shown in Figure 6(b)which manifests a good agreement of both methodsFigure 6(c) shows the prototype of the PMSM in which therotor is supported by gas foil bearings and the experimentalresults are shown in Figure 6(d) A coast-down test under60000 rmin is carried out that is the frequency converterwill shut down when the rotor is stable at 60000 rmin and
then the rotor speed will decrease to zero gradually byovercoming air friction In this process the stator has a back-emf but no current thereupon the rotor is not subjected tothe electromagnetic force Hereupon the curve of a no-loadback-emf vs speed can be achieved -e experimental anddesign values of the no-load back-emf are 570V and 596 Vat 746Hz respectively and the difference between them iswithin 46
-e results of Lad and Laq are shown in Figure 7 It canbe found that Lad and Laq are same when current is smalland both of them decrease with the increase of the currentBecause the d-axial magnetic circuit is more saturated Laddescends faster -e magnetic saturation is neglected inthe analytical method whereas Lad not varies with thecurrent
When IRms 10A and 100A the errors of the twomethods are 9 and 2 respectively It shows that theresults of the analytical method and FEM are very close Inthe analytical method only the fundamental harmonic isused to calculate the flux linkage However in the FEMexcept for the fundamental harmonic high-order harmonics
009 020 040 060 080 100 120ndash1000
1000
XY Plot 3torque_Cal_10kw120krpm
ndash750
750500
ndash500ndash250
250000
Curr
ent (
A)
Input current (phase A)
Input current (phase C)Input current (phase B)
Time (ms)
(a)
XY Plot 1torque_Cal_10kw120krpm
Torque Im = lsquo10ArsquoTorque_1 Im = lsquo30ArsquoTorque_2 Im = lsquo50Arsquo
Time (ms)
ndash150
150
ndash100
100
ndash050
050
000
020 040 060 080 120100 130000
Y1 (N
ewto
nMet
er)
(b)
0 02 04 06 08 1 12ndash15
ndash05
0
05
1
15
ndash1
2
t (ms)
Torq
ue T
emN
m
FE 10AFE 30AFE 50A
Analytic 10AAnalytic 30AAnalytic 50A
(c)
Figure 3-e results of the locked-rotor (a)-e three-phase alternating currents (b)-e torque vs time (c) Result comparison of FEM andthe analytical model
Table 2 Peak torque comparison between FE and the theory
Im (A) FEM (Nmiddotm) Analytical (Nmiddotm) Error ()10 0266 0271 18730 0802 0815 16250 1333 1357 176
Journal of Electrical and Computer Engineering 7
B (tesla)13842e + 000
12112e + 00011247e + 000
95165e ndash 00186514e ndash 001
69212e ndash 00160561e ndash 00151910e ndash 00143259e ndash 00134608e ndash 00125957e ndash 001
86550e ndash 00240351e ndash 005
17306e ndash 001
77863e ndash 001
10382e + 000
12977e + 000
Time = 00002sSpeed = 0000000rpmPosition = 0000000deg
0 40 80 (mm)
(a)
B (tesla)17001e + 000
13813e + 000
10626e + 000
12540e ndash 00410637e ndash 00121262e ndash 00131887e ndash 00142512e ndash 00153136e ndash 00163761e ndash 00174386e ndash 00185011e ndash 00195936e ndash 001
11689e + 00012751e + 000
14876e + 00015938e + 000
Speed = 0000000rpmPosition = 0000000deg
Time = 00004s0 40 80 (mm)
(b)B (tesla)
16056e + 000
90322e ndash 001
14749e ndash 00410049e ndash 00120083e ndash 00130117e ndash 00140152e ndash 00150186e ndash 00160220e ndash 00170254e ndash 00180288e ndash 001
10036e + 00011039e + 00012043e + 00013046e + 00014049e + 00015053e + 000
Position = 0000000deg
Time = 0000500000000000001sSpeed = 0000000rpm
40 80 (mm)0
(c)
Figure 4 Flux density vector field and current vector field of the FE model (a) Fields at 00002 s (b) Fields at 00004 s (c) Fields at 00005 s
B (tesla)15428e + 00014464e + 00013500e + 000
11572e + 00010608e + 000
96871e ndash 00219326e ndash 00128966e ndash 00138605e ndash 00148245e ndash 00157884e ndash 00167523e ndash 001
47663e ndash 004
77163e ndash 00186802e ndash 00196442e ndash 001
12536e + 000
(a)
10 20155 25 300Angle (degmech)
ndash400
400
ndash200
200
ndash100
100
0
ndash300
300
Cog
ging
torq
ue (μ
Nm
)
FEPermeanceSubdomain
(b)
Figure 5 PM field and cogging torque (a) PM field (b) Comparison of the cogging torque of FEM and the analytical model
8 Journal of Electrical and Computer Engineering
are also taken into account in the flux linkage -erefore theresult of the analytical method is smaller than that of FEM
It is worth noting that the stator tooth has the mostserious saturation as shown in Figure 8(a) and the fluxdensities reach 2086 T and 1880 T when IRms 100A(28IN) and 50A (14IN) respectively IN denotes the ratedcurrent of the motor -e working points of stator toothcorresponding to the both working condition are illustratedin Figure 8(b) It shows that the working point is located atthe end of the linear segment of the B-H curve whenIRms 14IN It indicates that the magnetic saturation of thestator iron is not prominent when the prototype worksunder rated working conditions and it applies to the pro-posed analytical method in this paper
In addition in order to enhance the cooling of the rotorthe air gap should be chosen as large as possible so thatmagnetic saturation in the tooth tips rarely occurs Fur-thermore the teeth in a 12-slot stator are diametricallysymmetrical the air-gap field distribution is symmetrical
000 020 040 060 080 100Time (ms)
ndash25000
ndash12500
000
12500
25000
Y1 (V
)XY Plot 1
Curve info rms_1 maxInduced voltage (phase A) 1560441 2188502Induced voltage (phase B) 1563378 2186613Induced voltage (phase C) 1573086 2184715
(a)
0 01 02 03 04 05Time (ms)
FE (phase A)Analytic
Vol
tage
(V)
ndash300
ndash200
ndash100
0
100
200
300
(b)
(c)
Time (s)0 50 100 150 200
0
500
1000
1500Fr
eque
ncy
(Hz)
0
50
100
150
Vol
tage
RM
SV
FrequencyVoltage RMS
(d)
Figure 6 Results of the no-load back-emf (a) No-load back-emf of FEM at 120000 rpm (b) Comparison of no-load back-emf at120000 rpm (c) Prototype of the ultra-high-speed PMSM (d) Experimental curves of the phase voltage and frequency
0 20 40 60 80 1000098
01
0102
0104
0106
0108
011
Phase current IRMSA
Lad FEMLaq FEMLad Analytical
d- an
d q-
axis
indu
ctan
ces (
mH
)
Figure 7 Curves of Lad and Laq vs the current
Journal of Electrical and Computer Engineering 9
and no unbalanced magnetic pull will exist if the rotor andstator axes are coincident Although the rotor rotates ec-centrically with respect to the stator in fact the eccentricityis too tiny due to the constraint of bearings to cause a notableunbalanced magnetic pull
4 Conclusion
For the ultra-high-speed PMSM including a stator withoverlapping windings and the 2-pole parallel magnetizedcylindrical PM electromagnetic torques are rarely reportedat present which leads to a prolonged motor design cycle toevaluating motor performance -erefore the work of thispaper is as follows
(1) -is paper deduces the analytical solution to theelectromagnetic driving torque of ultra-high-speedPMSMs which is verified by the FE methodaccording to locked-rotor simulation -e resultsshow that the electromagnetic torque is pro-portional to the length of stator iron the ampereturns per unit length of the stator inner surface andthe magnetization while it is proportional to thesquare of the ratio ie PM radius to stator innerradius
(2) -e cogging torque back-emf armature reactioninductance and so on are calculated compared anddiscussed For the PMSM studied in the paper theair-gap permeance method was shown not to beapplicable to the estimation of the cogging torquedue to the fact that the flux bends in the air gap andthus a high-accuracy method (eg the subdomainmethod) should be adopted
(3) A prototype is designed and manufactured and theno-load back-emf is checked by a 60000 rpm freecoast experiment -e experimental and designvalues of the no-load back-emf are 570 V and 596V
at 746Hz respectively and the difference betweenthem is within 46
Data Availability
-e data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
-e authors declare that they have no conflicts of interest
Acknowledgments
-is work is supported by National Natural ScienceFoundation of China (Grant nos 51705413 51705416 and11502196) and China Postdoctoral Science Foundation(Grant no 2017M613291XB)
References
[1] A Tenconi S Vaschetto and A Vigliani ldquoElectrical machinesfor high-speed applications design considerations andtradeoffsrdquo IEEE Transactions on Industrial Electronics vol 61no 6 pp 3022ndash3029 2014
[2] J H Ahn J Y Choi C H Park C Han C W Kim andT G Yoon ldquoCorrelation between rotor vibration and me-chanical stress in ultra-high-speed permanent magnet syn-chronous motorsrdquo IEEE Transactions on Magnetics vol 53no 11 Article ID 8209906 2017
[3] M-S Lim S-H Chai J-S Yang and J-P Hong ldquoDesign andverification of 150 krpm PMSM based on experiment resultsof prototyperdquo IEEE Transactions on Industrial Electronicsvol 62 no 12 pp 7827ndash7836 2015
[4] D-K Hong B-C Woo J-Y Lee and D-H Koo ldquoUltra highspeed motor supported by air foil bearings for air blowercooling fuel cellsrdquo IEEE Transactions on Magnetics vol 48no 2 pp 871ndash874 2012
20864e + 00019560e + 00018256e + 00016953e + 00015649e + 00014345e + 00013041e + 00011737e + 00010433e + 00091289e ndash 00178250e ndash 00165210e ndash 00152171e ndash 00139132e ndash 00126092e ndash 00113053e ndash 00113431e ndash 004
B (tesla)
(a)
375
250
125
000000E + 000 500E + 005
H (A_per_meter)
B (te
sla)
Working point 28IN
Working point 14IN
(b)
Figure 8 Magnetic saturation condition of the stator iron (a) Distribution of flux density for IRms 100A (b) Working points of the statortooth
10 Journal of Electrical and Computer Engineering
[5] D-K Hong B-C Woo Y-H Jeong D-H Koo andC-W Ahn ldquoDevelopment of an ultra high speed permanentmagnet synchronous motorrdquo International Journal of Pre-cision Engineering and Manufacturing vol 14 no 3pp 493ndash499 2013
[6] S-M Jang H-W Cho and S-K Choi ldquoDesign and analysisof a high-speed brushless DC motor for centrifugal com-pressorrdquo IEEE Transactions on Magnetics vol 43 no 6pp 2573ndash2575 2007
[7] D-K Hong D-S Joo B-CWoo D-H Koo and C-W AhnldquoUnbalance response analysis and experimental validation ofan ultra high speed motor-generator for microturbine gen-erators considering balancingrdquo Sensors vol 14 no 9pp 16117ndash16127 2014
[8] A B Proca A Keyhani A EL-Antably W Lu and M DaildquoAnalytical model for permanent magnet motors with surfacemounted magnetsrdquo IEEE Transactions on Energy Conversionvol 18 no 3 pp 386ndash391 2003
[9] F Dubas C Espanet and A Miraoui ldquoAn original analyticalexpression of the maximum magnet thickness in surfacemounted permanent magnet motorsrdquo De European PhysicalJournal Applied Physics vol 38 no 2 pp 169ndash176 2007
[10] Z Q Zhu and D Howe ldquoAnalytical prediction of the coggingtorque in radial-field permanent magnet brushless motorsrdquoIEEE Transactions on Magnetics vol 28 no 2 pp 1371ndash13741992
[11] Z Q Zhu K Ng and D Howe ldquoDesign and analysis of high-speed brushless permanent magnet motorsrdquo in Proceedings ofthe International Conference on Electrical Machines amp Drivespp 381ndash385 IET Cambridge UK September 1997
[12] W Cheng L Yu L Huang Y Lv L Li and Y Sun ldquoAn-alytical solution to magnetic field distribution of a parallelmagnetised rotor with cylindrical or ring-type permanentmagnetrdquo IET Electric Power Applications vol 9 no 6pp 429ndash437 2015
[13] Z Q Zhu and D Howe ldquoInstantaneous magnetic field dis-tribution in brushless permanent magnet DC motors IIArmature-reaction fieldrdquo IEEE Transactions on Magneticsvol 29 no 1 pp 136ndash142 1993
Journal of Electrical and Computer Engineering 11
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
reaction field is as follows the single conductor field issolved first and then the single-turn coil field can beobtained by adding two single conductor fields at a polardistance apart from each other Whereupon the single-phase winding field is superpositioned onto fields that areexcited by the multiple single-turn coils and finally thearmature reaction field can be calculated by super-imposing the fields of three-phase winding
For distributed winding multiple single-turn coil fieldsthat are a pitch apart from each other can be equivalent to aconcentrated winding by taking into account the distribu-tion coefficient and the pitch-shortening coefficient -eabove-mentioned treatment will lightly change higherharmonic components of the armature reaction field butvastly simplify the solution of the single-phase winding fieldwith good accuracy In addition it is assumed that in-stantaneous three-phase currents (ia ib ic) only involvefundamental components and they can be written asfollows
ia Im cos(Pωt + β)
ib Im cos Pωt + β minus2π3
1113874 1113875
ic Im cos Pωt + β minus4π3
1113874 1113875
⎧⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎪⎪⎪⎪⎩
(7)
where Im is the current amplitude of a single conductor P isthe number of pole pairs ω is the synchronous angularvelocity and β is the current phase
-e armature reaction field in Figure 2(b) is expressed asfollows
BWIIr
minus 2qNsμ0π
1113944
infin
v11113896Kdpv(2v minus 1)Ksov(2v minus 1)
middot Fo(2v minus 1 r) middot (minus 1)v
middot 1113890ia cos[(2v minus 1)(α minus c)]
+ ib cos (2v minus 1) α minus c minus2π3
1113874 11138751113876 1113877
+ ic cos (2v minus 1) α minus c minus4π3
1113874 11138751113876 111387711138911113897
BWIIα
minus 2qNsμ0π
1113944
infin
v11113896Kdpv(2v minus 1)Ksov(2v minus 1)
middot Fo(2v minus 1 r) middot (minus 1)vminus 1
middot 1113890ia sin[(2v minus 1)(α minus c)]
+ ib sin (2v minus 1) α minus c minus2π3
1113874 11138751113876 1113877
+ ic sin (2v minus 1) α minus c minus4π3
1113874 11138751113876 111387711138911113897
(8)
where Ksov(v) sin(vb0(2R2))(vb0(2R2)) is the slot-opening coefficient which is a function of the spatial har-monic order vFo(r v) rvminus 1Rv
2 reflects the influence of theair-gap thickness on the radialcircumferential flux densityof the v order harmonic wave α is the angle relative to thehorizontal direction as shown in Figure 2(b) and is positive
Mrθ
Mr
Mrr
R1
R2
θ
I (PM)
II (Air gap)
(a)
Stator bore
II
Air b0Slot
α = 0
Current sheetR2
(b)
Figure 2 Models of the electromagnetic field of the PMSMs (a) Open-circuit PM field (b) Armature reaction field
Stator iron
Winding
Sleeve
Magnet
Magnetizationdirection
Figure 1 Cross section of the ultra-high-speed PMSM
Journal of Electrical and Computer Engineering 3
in the counter clockwise direction c is an angle between theaxis of winding A and the d-axis q is the number of slots perphase per pole and Ns is the number of conductors per slot
23 Electromagnetic Torque andCoggingTorque In this parttwo methods namely the Poynting theorem and theMaxwell stress tensor were used to deduce the analyticexpression of the electromagnetic torque For the 2D fieldthe electric field strength vector E in the air-gap region onlyincludes the axial component Ez therefore the Poyntingvector is given by the following
1113937r
minus EzHθ
1113937θ
EzHr
1113937z
0
⎧⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎩
(9)
where Ez μ0 1113938(zHθzt)dr -e electromagnetic powertransmitting from the stator to the rotor is mainly dependenton the radial component of the Poynting vector Πr whichcan be calculated from equation (10) as follows
PM Temω lef 11139462π
01113945
r
r dθ (10)
where lef is the length of the stator iron Taking the hori-zontal axis in Figure 2(b) as the reference position the PMfield is changed as follows
BPMIIr minus
μ0MrR21 1 + R2
2rminus 2( 1113857
R21 μr minus 1( 1113857 minus R2
2 μr + 1( 1113857middot cos(φ)
BPMIIθ
μ0MrR21 minus 1 + R2
2rminus 2( 1113857
R21 μr minus 1( 1113857 minus R2
2 μr + 1( 1113857middot sin(φ)
(11)
where φ θ minus π2 α It is assumed that the electromag-netic power is mainly transmitted by the fundamentalmagnetic field of armature reaction field When the numberof pole pairs equals 1 by substituting equation (7) intoequation (8) the armature reaction field will become asfollows using the properties of trigonometric functions
BWIIr μ0G1 cos(ωt minus φ + δ)
BWIIθ μ0G1 sin(ωt minus φ + δ)
(12)
where δ β + c represents the angle between the axis of ar-mature reaction field and the axis of PM field andG1 (3qNsKdp1Kso1Im)(πR2) Equation (12) shows that thearmature reaction field is a travelling wave but the PM field is astanding wave shown by equation (11) To make the PM fieldrotate counterclockwise synchronously relative to the armaturereaction field equation (11) can written as the following
BPMIIr μ0 minus G2r
minus 2minus G31113872 1113873 middot cos(φ minus ωt)
BPMIIθ μ0 G2r
minus 2minus G31113872 1113873 middot sin(φ minus ωt)
(13)
where G2 MrR21R
22[R2
1(μr minus 1) minus R22(μr + 1)] and G3
MrR21[R2
1(μr minus 1) minus R22(μr + 1)] -erefore the loaded gap
magnetic intensity can be obtained as follows
HIIr G1 cos(ωt minus φ + δ) + minus G2rminus 2
minus G31113872 1113873 middot cos(φ minus ωt)1113960 1113961
HIIθ G1 sin(ωt minus φ + δ) + G2rminus 2
minus G31113872 1113873 middot sin(φ minus ωt)1113960 1113961
(14)
Furthermore Ez can be obtained byHIIθ and is expressedas the following
Ez μ0ωr G1 cos(ωt minus φ + δ) + G2rminus 2
+ G31113872 1113873cos(φ minus ωt)1113960 1113961
(15)
By substituting equation (15) into equation (9)Πrwill becalculated and according to equation (10) the electro-magnetic torque can be obtained as follows
Tem minus 2πμ0lefG1G2 sin δ δ isin [0 π] (16)
In the 2D electromagnetic field fθ BrBθμ0 and fθ isthe circumferential electromagnetic force density acting onthe stator or rotor -erefore the electromagnetic torque ofthe rotor which is shown in Figure 1 can be written as thefollowing
Tem 2μ0lef 1113946π
0HIIrHIIθr
2dθ (17)
By substituting equation (14) into equation (17) theexpression of the electromagnetic torque Tem is as follows
Tem minus 2πμ0lefG1G3r2sin δ δ isin [0 π]
asymp 6πμ0qKdp1Kso1NsIm
2πR2
R1
R21113888 1113889
2
Mrlefr2sin δ
(18)
In equation (18) when rR2 equation (18) is equal toequation (16) which indicates that the torque acting on therotor surface generated by the Maxwell stress equals thetorque acting on the current sheet generated by the ampereforce -e results show that the electromagnetic torque isproportional to the length of stator iron lef the ampere turnsper unit length of the stator inner surface NsIm2πR2 andthe magnetization Mr while it is proportional to the squareof the ratio (R1R2)
2 (PM radius to stator inner radius)-e cogging torque arises from the interaction of the
magnetomotive force (mmf) harmonics and the airgappermeance harmonics hence it exists in almost all types ofmotors in which the airgap permeance is not constant -ecogging torque will cause undesirable speed pulsationsvibrations possible resonance and acoustic noise [10] Forthe motor shown in Figure (1) because the radial andtangential components of the magnetic flux density vectoras presented in equation (11) are pure sinusoidal the mmfharmonics only contain the fundamental component andthe cogging torque is caused by stator slots In fact thecogging torque can also be calculated by equation (17)however HIIr and HIIθ should be replaced by HPM
IIr and HPMIIθ
respectively In the paper HPMIIr and HPM
IIθ are obtained by thehypothesis of a smooth stator surface hereupon they cannotreveal the slot effects However slot effects can be consideredanalytically by the subdomain technique Works in theliterature [10] have proposed a simple method to predict the
4 Journal of Electrical and Computer Engineering
cogging torque based on the analytical calculation of the air-gap field distribution and the net lateral force acting on theteeth
24 Back-Electromotive Force -e flux through a phasewinding excited by PM and by winding respectively can bewritten as follows
Φpm 2 BPM
IIr( 1113857maxπ
τplef
ΦW 2 BW
IIr( 1113857maxπ
τplef
⎧⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎩
(19)
where τp is the pole pitch and lef is the length of stator iron-e back-emf amplitude in each phase winding assuming a120-degree coil span is then obtained as follows
E 444fkdpvNΦload (20)
where f is the angular frequency of the rotor kdpv is thewinding coefficient N is the number of turns per phase andΦload Φpm +ΦW When ΦW 0 it corresponds to anopen-circuit situation and the back-emf becomes the no-load back-emf E0
25 Armature Reaction Inductance -e constant-powertransformation from the three-phase coordinates to the d-qcoordinate is given by the following
id
iq
i0
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
23
1113970
cos θ cos θ minus2π3
1113874 1113875 cos θ +2π3
1113874 1113875
minus sin θ minus sin θ minus2π3
1113874 1113875 minus sin θ +2π3
1113874 1113875
12
1113970 12
1113970 12
1113970
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
(21)
where θ 1113938ω dt represents the rotor position which is anangle between the d-axis and winding axis of phase A Whenthe above two axes coincide θ 0Whereupon according toequation (21) there are
id
iq
⎡⎣ ⎤⎦ [C]
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (22a)
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ [C]
Tid
iq1113890 1113891 (22b)
where [C] 23
radic 1 minus 12 120
3
radic2 minus
3
radic21113890 1113891 and [C]T and [C]minus 1
are transposed matrix and inverse matrix of [C] re-spectively Notably there is [C]T [C]minus 1 According toequation (22a) we have
id
23
1113970
iA minus12
iB minus12
iC1113874 1113875
iq
23
1113970 3
radic
2iB minus
3
radic
2iC1113888 1113889
⎧⎪⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎪⎩
(23)
θ 0 which means Pωt 0 in equation (7)Substituting (7) into equation (23) yields
id 3
radicIRmscos(β)
iq 3
radicIRmssin(β)
⎧⎨
⎩ (24)
where IRms is the RMS of phase current IRms Im2
radic and β
denotes the electrical angle of the phase current vectorbeyond the d-axis Let β 0 by substituting equation (24)into equation (22b) to get the three-phase AC [iA iB iC]when the three-phase AC flows into the stator windings itwill generate the d-axis armature reaction field Obviouslythe stator iron saturation is most likely to occur when thedirection of the armature reaction field is in line with that ofthe PM field accordingly we only consider this situation
Denoting the inductance matrix in the abc coordinatesystem by
LABC1113858 1113859
LAA LAB LAC
LBA LBB LBC
LCA LCB LCC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (25)
-e flux linkage in the abc coordinate system can berepresented as
λA
λB
λC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ LABC
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (26)
-e flux linkage in the dq coordinate system is alsoapplied to equation (22a) that is
λd
λq
⎡⎣ ⎤⎦ [C]
λA
λB
λC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (27)
Substituting equation (26) and equation (22b) intoequation (27) yields the following
λd
λq
⎡⎣ ⎤⎦ [C] LABC1113858 1113859[C]T
id
iq
⎡⎣ ⎤⎦ Ldq1113960 1113961id
iq
⎡⎣ ⎤⎦ (28)
where [Ldq] Ld Lmd
Lmq Lq1113890 1113891 [C][LABC][C]T it represents
the inductance matrix in the dq coordinate system If the endleakage is not considered and other leakage factors such ashigher harmonics and slot are also ignored then Ld and Lq
will represent the d-axis and q-axis armature reaction in-ductances namely Lad and Laq respectively From the FEM[LABC] will be obtained and then it will turn into [Ldq] byusing the above transformation
Lad and Laq can also be calculated by analytical methodand the process is as follows
Journal of Electrical and Computer Engineering 5
-e load flux linkage of phase A is calculated as
λA NR2lef 1113946αy2
minus αy2BloadIIr R2 θ( 1113857dθ (29)
where N is the number of turns per coil of one phasewinding and αy is the coil pitch angle -e load flux linkageof phase B and C is expressed as
λB λAcos2π3
λC λAcos4π3
⎧⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎩
(30)
According to the above-mentioned [λA λB λC] fromequation (27) we will get [λd λq] And then there is
Ld λd
id (31)
Similarly neglecting some leakage factors Ld calculatedby equation (31) corresponds to Lad
3 Example Validation
To verify the analytic solution to the electromagnetic torquea locked-rotor simulation is carried out by the FEMmethod-e motor parameters are listed in Table 1
In the simulation the windings are fed by a symmetricalthree-phase alternating current to produce a rotating ar-mature reaction field but the rotor speed is set to zero thatis the rotor is locked When the armature reaction fieldrotates relative to the rotor with the angular velocity ω thelocked-rotor torque will generate and vary periodically withδ (an angle between two axes of the armature reaction fieldand PM field and it changes periodically from 0 to 360degrees) -e locked-rotor phenomenon can be applied toverify the electromagnetic torque of equation (16) andequation (18)
When the motor is fed with currents as shown inequation (7) the locked-rotor torques are as shown inFigure 3 -ree kinds of input currents of 2000Hz withdifferent amplitudes such as 10A 30A and 50A areconsidered Figure 3(a) shows the three-phase currents witha 10A peak value in FEM -e locked-rotor torque curvesobtained by FEM are shown in Figure 3(b) which shows thatthe locked-rotor torque is also sinusoidal with time and itspeak value is increased with the increasing amplitude of thephase current A comparison of the results of FEM and theanalytical model is illustrated in Figure 3(c) in which it canbe found that the analytical solution is in good agreementwith the FEM solution
Peak torques obtained by the FE model are comparedwith those from the derived analytical equations-e resultswhich are listed in Table 2 show that the errors are less than5 which proves the correctness of equation (16) andequation (18)
-emagnetic field results that were calculated by FEM areshown in Figure 4 where red and blue conductors indicate the
current flows out of and into the cross section respectively Inaddition the PMhas a parallel magnetization (the direction ofthe magnetization is vertically upward)
Because the PM field is much larger than the armaturereaction field the magnetic fields of the motor in Figure 4 atdifferent times are approximately the same At 02ms thelocked-rotor torque in Figure 3(b) is negative and thecorresponding motor magnetic field is shown in Figure 4(a)in which the armature reaction field lags behind the PM fieldby approximately 135 degrees which just causes a negativetorque Moreover the armature reaction field weakens thePM field the maximum flux density occurs in the statortooth and its value is 138 T At 04ms the locked-rotortorque equals zero as shown in Figure 3(b) and the cor-responding motor magnetic field is shown in Figure 4(b)where the armature reaction field coincides with the PMfield which verifies the null torque At 04ms the maximumflux density of the stator tooth is 170 T At 05ms thelocked-rotor torque approaches the maximum as shown inFigure 3(b) and the corresponding motor magnetic field isshown in Figure 4(c) It can be seen that the armature re-action field direction is left (90 degrees ahead of the PMfield) which verifies the maximum torque At 05ms themaximum flux density of the stator tooth is 160 T
-ree methods namely the FM method the air-gappermeance method [10] and the subdomain method wereused to estimate the cogging torque and the results areshown in Figure 5 -e air-gap permeance method is basedon the assumption that the flux crosses the magnet and airgap in a straight line wherever a magnet faces a toothHowever the above assumption will no longer be suitablefor the motor mentioned in the paper because the flux willbend in the air gap as shown in Figure 5(a) the coggingtorque calculated by the air-gap permeance method is smallrelative to that from FEM and the subdomain method asshown in Figure 5(b) -e cogging torque peak is approx-imately 370 μNm and is 005 of the rated torque (08Nm)which is too small to be ignored In fact some degrees ofharmonic components will exist in the polarized PM thusthe cogging torque peak will increase
-e stator tooth tips in relation to the width of the slotopenings also influence the back-emf waveform When theheight of the tooth tips is small magnetic saturation existsand the emf waveform tends to be a trapezoid Howeverwhen the tooth tip height is too large the emf waveform willdeteriorate [11] -e motor consists of a stator having 12
Table 1 Parameters of the high-speed PMSM
Symbol Quantity ValuesR1 PM radius 12times10minus 3mR2 Inner radius of stator 174times10minus 3mQ Slot numbers 12lef Core length 55times10minus 3mM Magnetization 820000Amω Angular velocity 12560 radsNs Conductors per slot 20μr Relative recoil permeability 1038P Number of pole pairs 1
6 Journal of Electrical and Computer Engineering
teeth which carry overlapping winding -e no-load back-emf waveform obtained from the FEM is displayed inFigure 6(a) and the emf waveform is almost purely sinu-soidal which is ideal for ultra-high-speed PMSMs As shownin Figure 5(a) a stator tooth has the maximum flux density1542 T which does not reach the saturation region of thematerial-e comparison of the no-load back-emf calculatedby equation (20) and that of FEM is shown in Figure 6(b)which manifests a good agreement of both methodsFigure 6(c) shows the prototype of the PMSM in which therotor is supported by gas foil bearings and the experimentalresults are shown in Figure 6(d) A coast-down test under60000 rmin is carried out that is the frequency converterwill shut down when the rotor is stable at 60000 rmin and
then the rotor speed will decrease to zero gradually byovercoming air friction In this process the stator has a back-emf but no current thereupon the rotor is not subjected tothe electromagnetic force Hereupon the curve of a no-loadback-emf vs speed can be achieved -e experimental anddesign values of the no-load back-emf are 570V and 596 Vat 746Hz respectively and the difference between them iswithin 46
-e results of Lad and Laq are shown in Figure 7 It canbe found that Lad and Laq are same when current is smalland both of them decrease with the increase of the currentBecause the d-axial magnetic circuit is more saturated Laddescends faster -e magnetic saturation is neglected inthe analytical method whereas Lad not varies with thecurrent
When IRms 10A and 100A the errors of the twomethods are 9 and 2 respectively It shows that theresults of the analytical method and FEM are very close Inthe analytical method only the fundamental harmonic isused to calculate the flux linkage However in the FEMexcept for the fundamental harmonic high-order harmonics
009 020 040 060 080 100 120ndash1000
1000
XY Plot 3torque_Cal_10kw120krpm
ndash750
750500
ndash500ndash250
250000
Curr
ent (
A)
Input current (phase A)
Input current (phase C)Input current (phase B)
Time (ms)
(a)
XY Plot 1torque_Cal_10kw120krpm
Torque Im = lsquo10ArsquoTorque_1 Im = lsquo30ArsquoTorque_2 Im = lsquo50Arsquo
Time (ms)
ndash150
150
ndash100
100
ndash050
050
000
020 040 060 080 120100 130000
Y1 (N
ewto
nMet
er)
(b)
0 02 04 06 08 1 12ndash15
ndash05
0
05
1
15
ndash1
2
t (ms)
Torq
ue T
emN
m
FE 10AFE 30AFE 50A
Analytic 10AAnalytic 30AAnalytic 50A
(c)
Figure 3-e results of the locked-rotor (a)-e three-phase alternating currents (b)-e torque vs time (c) Result comparison of FEM andthe analytical model
Table 2 Peak torque comparison between FE and the theory
Im (A) FEM (Nmiddotm) Analytical (Nmiddotm) Error ()10 0266 0271 18730 0802 0815 16250 1333 1357 176
Journal of Electrical and Computer Engineering 7
B (tesla)13842e + 000
12112e + 00011247e + 000
95165e ndash 00186514e ndash 001
69212e ndash 00160561e ndash 00151910e ndash 00143259e ndash 00134608e ndash 00125957e ndash 001
86550e ndash 00240351e ndash 005
17306e ndash 001
77863e ndash 001
10382e + 000
12977e + 000
Time = 00002sSpeed = 0000000rpmPosition = 0000000deg
0 40 80 (mm)
(a)
B (tesla)17001e + 000
13813e + 000
10626e + 000
12540e ndash 00410637e ndash 00121262e ndash 00131887e ndash 00142512e ndash 00153136e ndash 00163761e ndash 00174386e ndash 00185011e ndash 00195936e ndash 001
11689e + 00012751e + 000
14876e + 00015938e + 000
Speed = 0000000rpmPosition = 0000000deg
Time = 00004s0 40 80 (mm)
(b)B (tesla)
16056e + 000
90322e ndash 001
14749e ndash 00410049e ndash 00120083e ndash 00130117e ndash 00140152e ndash 00150186e ndash 00160220e ndash 00170254e ndash 00180288e ndash 001
10036e + 00011039e + 00012043e + 00013046e + 00014049e + 00015053e + 000
Position = 0000000deg
Time = 0000500000000000001sSpeed = 0000000rpm
40 80 (mm)0
(c)
Figure 4 Flux density vector field and current vector field of the FE model (a) Fields at 00002 s (b) Fields at 00004 s (c) Fields at 00005 s
B (tesla)15428e + 00014464e + 00013500e + 000
11572e + 00010608e + 000
96871e ndash 00219326e ndash 00128966e ndash 00138605e ndash 00148245e ndash 00157884e ndash 00167523e ndash 001
47663e ndash 004
77163e ndash 00186802e ndash 00196442e ndash 001
12536e + 000
(a)
10 20155 25 300Angle (degmech)
ndash400
400
ndash200
200
ndash100
100
0
ndash300
300
Cog
ging
torq
ue (μ
Nm
)
FEPermeanceSubdomain
(b)
Figure 5 PM field and cogging torque (a) PM field (b) Comparison of the cogging torque of FEM and the analytical model
8 Journal of Electrical and Computer Engineering
are also taken into account in the flux linkage -erefore theresult of the analytical method is smaller than that of FEM
It is worth noting that the stator tooth has the mostserious saturation as shown in Figure 8(a) and the fluxdensities reach 2086 T and 1880 T when IRms 100A(28IN) and 50A (14IN) respectively IN denotes the ratedcurrent of the motor -e working points of stator toothcorresponding to the both working condition are illustratedin Figure 8(b) It shows that the working point is located atthe end of the linear segment of the B-H curve whenIRms 14IN It indicates that the magnetic saturation of thestator iron is not prominent when the prototype worksunder rated working conditions and it applies to the pro-posed analytical method in this paper
In addition in order to enhance the cooling of the rotorthe air gap should be chosen as large as possible so thatmagnetic saturation in the tooth tips rarely occurs Fur-thermore the teeth in a 12-slot stator are diametricallysymmetrical the air-gap field distribution is symmetrical
000 020 040 060 080 100Time (ms)
ndash25000
ndash12500
000
12500
25000
Y1 (V
)XY Plot 1
Curve info rms_1 maxInduced voltage (phase A) 1560441 2188502Induced voltage (phase B) 1563378 2186613Induced voltage (phase C) 1573086 2184715
(a)
0 01 02 03 04 05Time (ms)
FE (phase A)Analytic
Vol
tage
(V)
ndash300
ndash200
ndash100
0
100
200
300
(b)
(c)
Time (s)0 50 100 150 200
0
500
1000
1500Fr
eque
ncy
(Hz)
0
50
100
150
Vol
tage
RM
SV
FrequencyVoltage RMS
(d)
Figure 6 Results of the no-load back-emf (a) No-load back-emf of FEM at 120000 rpm (b) Comparison of no-load back-emf at120000 rpm (c) Prototype of the ultra-high-speed PMSM (d) Experimental curves of the phase voltage and frequency
0 20 40 60 80 1000098
01
0102
0104
0106
0108
011
Phase current IRMSA
Lad FEMLaq FEMLad Analytical
d- an
d q-
axis
indu
ctan
ces (
mH
)
Figure 7 Curves of Lad and Laq vs the current
Journal of Electrical and Computer Engineering 9
and no unbalanced magnetic pull will exist if the rotor andstator axes are coincident Although the rotor rotates ec-centrically with respect to the stator in fact the eccentricityis too tiny due to the constraint of bearings to cause a notableunbalanced magnetic pull
4 Conclusion
For the ultra-high-speed PMSM including a stator withoverlapping windings and the 2-pole parallel magnetizedcylindrical PM electromagnetic torques are rarely reportedat present which leads to a prolonged motor design cycle toevaluating motor performance -erefore the work of thispaper is as follows
(1) -is paper deduces the analytical solution to theelectromagnetic driving torque of ultra-high-speedPMSMs which is verified by the FE methodaccording to locked-rotor simulation -e resultsshow that the electromagnetic torque is pro-portional to the length of stator iron the ampereturns per unit length of the stator inner surface andthe magnetization while it is proportional to thesquare of the ratio ie PM radius to stator innerradius
(2) -e cogging torque back-emf armature reactioninductance and so on are calculated compared anddiscussed For the PMSM studied in the paper theair-gap permeance method was shown not to beapplicable to the estimation of the cogging torquedue to the fact that the flux bends in the air gap andthus a high-accuracy method (eg the subdomainmethod) should be adopted
(3) A prototype is designed and manufactured and theno-load back-emf is checked by a 60000 rpm freecoast experiment -e experimental and designvalues of the no-load back-emf are 570 V and 596V
at 746Hz respectively and the difference betweenthem is within 46
Data Availability
-e data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
-e authors declare that they have no conflicts of interest
Acknowledgments
-is work is supported by National Natural ScienceFoundation of China (Grant nos 51705413 51705416 and11502196) and China Postdoctoral Science Foundation(Grant no 2017M613291XB)
References
[1] A Tenconi S Vaschetto and A Vigliani ldquoElectrical machinesfor high-speed applications design considerations andtradeoffsrdquo IEEE Transactions on Industrial Electronics vol 61no 6 pp 3022ndash3029 2014
[2] J H Ahn J Y Choi C H Park C Han C W Kim andT G Yoon ldquoCorrelation between rotor vibration and me-chanical stress in ultra-high-speed permanent magnet syn-chronous motorsrdquo IEEE Transactions on Magnetics vol 53no 11 Article ID 8209906 2017
[3] M-S Lim S-H Chai J-S Yang and J-P Hong ldquoDesign andverification of 150 krpm PMSM based on experiment resultsof prototyperdquo IEEE Transactions on Industrial Electronicsvol 62 no 12 pp 7827ndash7836 2015
[4] D-K Hong B-C Woo J-Y Lee and D-H Koo ldquoUltra highspeed motor supported by air foil bearings for air blowercooling fuel cellsrdquo IEEE Transactions on Magnetics vol 48no 2 pp 871ndash874 2012
20864e + 00019560e + 00018256e + 00016953e + 00015649e + 00014345e + 00013041e + 00011737e + 00010433e + 00091289e ndash 00178250e ndash 00165210e ndash 00152171e ndash 00139132e ndash 00126092e ndash 00113053e ndash 00113431e ndash 004
B (tesla)
(a)
375
250
125
000000E + 000 500E + 005
H (A_per_meter)
B (te
sla)
Working point 28IN
Working point 14IN
(b)
Figure 8 Magnetic saturation condition of the stator iron (a) Distribution of flux density for IRms 100A (b) Working points of the statortooth
10 Journal of Electrical and Computer Engineering
[5] D-K Hong B-C Woo Y-H Jeong D-H Koo andC-W Ahn ldquoDevelopment of an ultra high speed permanentmagnet synchronous motorrdquo International Journal of Pre-cision Engineering and Manufacturing vol 14 no 3pp 493ndash499 2013
[6] S-M Jang H-W Cho and S-K Choi ldquoDesign and analysisof a high-speed brushless DC motor for centrifugal com-pressorrdquo IEEE Transactions on Magnetics vol 43 no 6pp 2573ndash2575 2007
[7] D-K Hong D-S Joo B-CWoo D-H Koo and C-W AhnldquoUnbalance response analysis and experimental validation ofan ultra high speed motor-generator for microturbine gen-erators considering balancingrdquo Sensors vol 14 no 9pp 16117ndash16127 2014
[8] A B Proca A Keyhani A EL-Antably W Lu and M DaildquoAnalytical model for permanent magnet motors with surfacemounted magnetsrdquo IEEE Transactions on Energy Conversionvol 18 no 3 pp 386ndash391 2003
[9] F Dubas C Espanet and A Miraoui ldquoAn original analyticalexpression of the maximum magnet thickness in surfacemounted permanent magnet motorsrdquo De European PhysicalJournal Applied Physics vol 38 no 2 pp 169ndash176 2007
[10] Z Q Zhu and D Howe ldquoAnalytical prediction of the coggingtorque in radial-field permanent magnet brushless motorsrdquoIEEE Transactions on Magnetics vol 28 no 2 pp 1371ndash13741992
[11] Z Q Zhu K Ng and D Howe ldquoDesign and analysis of high-speed brushless permanent magnet motorsrdquo in Proceedings ofthe International Conference on Electrical Machines amp Drivespp 381ndash385 IET Cambridge UK September 1997
[12] W Cheng L Yu L Huang Y Lv L Li and Y Sun ldquoAn-alytical solution to magnetic field distribution of a parallelmagnetised rotor with cylindrical or ring-type permanentmagnetrdquo IET Electric Power Applications vol 9 no 6pp 429ndash437 2015
[13] Z Q Zhu and D Howe ldquoInstantaneous magnetic field dis-tribution in brushless permanent magnet DC motors IIArmature-reaction fieldrdquo IEEE Transactions on Magneticsvol 29 no 1 pp 136ndash142 1993
Journal of Electrical and Computer Engineering 11
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
in the counter clockwise direction c is an angle between theaxis of winding A and the d-axis q is the number of slots perphase per pole and Ns is the number of conductors per slot
23 Electromagnetic Torque andCoggingTorque In this parttwo methods namely the Poynting theorem and theMaxwell stress tensor were used to deduce the analyticexpression of the electromagnetic torque For the 2D fieldthe electric field strength vector E in the air-gap region onlyincludes the axial component Ez therefore the Poyntingvector is given by the following
1113937r
minus EzHθ
1113937θ
EzHr
1113937z
0
⎧⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎩
(9)
where Ez μ0 1113938(zHθzt)dr -e electromagnetic powertransmitting from the stator to the rotor is mainly dependenton the radial component of the Poynting vector Πr whichcan be calculated from equation (10) as follows
PM Temω lef 11139462π
01113945
r
r dθ (10)
where lef is the length of the stator iron Taking the hori-zontal axis in Figure 2(b) as the reference position the PMfield is changed as follows
BPMIIr minus
μ0MrR21 1 + R2
2rminus 2( 1113857
R21 μr minus 1( 1113857 minus R2
2 μr + 1( 1113857middot cos(φ)
BPMIIθ
μ0MrR21 minus 1 + R2
2rminus 2( 1113857
R21 μr minus 1( 1113857 minus R2
2 μr + 1( 1113857middot sin(φ)
(11)
where φ θ minus π2 α It is assumed that the electromag-netic power is mainly transmitted by the fundamentalmagnetic field of armature reaction field When the numberof pole pairs equals 1 by substituting equation (7) intoequation (8) the armature reaction field will become asfollows using the properties of trigonometric functions
BWIIr μ0G1 cos(ωt minus φ + δ)
BWIIθ μ0G1 sin(ωt minus φ + δ)
(12)
where δ β + c represents the angle between the axis of ar-mature reaction field and the axis of PM field andG1 (3qNsKdp1Kso1Im)(πR2) Equation (12) shows that thearmature reaction field is a travelling wave but the PM field is astanding wave shown by equation (11) To make the PM fieldrotate counterclockwise synchronously relative to the armaturereaction field equation (11) can written as the following
BPMIIr μ0 minus G2r
minus 2minus G31113872 1113873 middot cos(φ minus ωt)
BPMIIθ μ0 G2r
minus 2minus G31113872 1113873 middot sin(φ minus ωt)
(13)
where G2 MrR21R
22[R2
1(μr minus 1) minus R22(μr + 1)] and G3
MrR21[R2
1(μr minus 1) minus R22(μr + 1)] -erefore the loaded gap
magnetic intensity can be obtained as follows
HIIr G1 cos(ωt minus φ + δ) + minus G2rminus 2
minus G31113872 1113873 middot cos(φ minus ωt)1113960 1113961
HIIθ G1 sin(ωt minus φ + δ) + G2rminus 2
minus G31113872 1113873 middot sin(φ minus ωt)1113960 1113961
(14)
Furthermore Ez can be obtained byHIIθ and is expressedas the following
Ez μ0ωr G1 cos(ωt minus φ + δ) + G2rminus 2
+ G31113872 1113873cos(φ minus ωt)1113960 1113961
(15)
By substituting equation (15) into equation (9)Πrwill becalculated and according to equation (10) the electro-magnetic torque can be obtained as follows
Tem minus 2πμ0lefG1G2 sin δ δ isin [0 π] (16)
In the 2D electromagnetic field fθ BrBθμ0 and fθ isthe circumferential electromagnetic force density acting onthe stator or rotor -erefore the electromagnetic torque ofthe rotor which is shown in Figure 1 can be written as thefollowing
Tem 2μ0lef 1113946π
0HIIrHIIθr
2dθ (17)
By substituting equation (14) into equation (17) theexpression of the electromagnetic torque Tem is as follows
Tem minus 2πμ0lefG1G3r2sin δ δ isin [0 π]
asymp 6πμ0qKdp1Kso1NsIm
2πR2
R1
R21113888 1113889
2
Mrlefr2sin δ
(18)
In equation (18) when rR2 equation (18) is equal toequation (16) which indicates that the torque acting on therotor surface generated by the Maxwell stress equals thetorque acting on the current sheet generated by the ampereforce -e results show that the electromagnetic torque isproportional to the length of stator iron lef the ampere turnsper unit length of the stator inner surface NsIm2πR2 andthe magnetization Mr while it is proportional to the squareof the ratio (R1R2)
2 (PM radius to stator inner radius)-e cogging torque arises from the interaction of the
magnetomotive force (mmf) harmonics and the airgappermeance harmonics hence it exists in almost all types ofmotors in which the airgap permeance is not constant -ecogging torque will cause undesirable speed pulsationsvibrations possible resonance and acoustic noise [10] Forthe motor shown in Figure (1) because the radial andtangential components of the magnetic flux density vectoras presented in equation (11) are pure sinusoidal the mmfharmonics only contain the fundamental component andthe cogging torque is caused by stator slots In fact thecogging torque can also be calculated by equation (17)however HIIr and HIIθ should be replaced by HPM
IIr and HPMIIθ
respectively In the paper HPMIIr and HPM
IIθ are obtained by thehypothesis of a smooth stator surface hereupon they cannotreveal the slot effects However slot effects can be consideredanalytically by the subdomain technique Works in theliterature [10] have proposed a simple method to predict the
4 Journal of Electrical and Computer Engineering
cogging torque based on the analytical calculation of the air-gap field distribution and the net lateral force acting on theteeth
24 Back-Electromotive Force -e flux through a phasewinding excited by PM and by winding respectively can bewritten as follows
Φpm 2 BPM
IIr( 1113857maxπ
τplef
ΦW 2 BW
IIr( 1113857maxπ
τplef
⎧⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎩
(19)
where τp is the pole pitch and lef is the length of stator iron-e back-emf amplitude in each phase winding assuming a120-degree coil span is then obtained as follows
E 444fkdpvNΦload (20)
where f is the angular frequency of the rotor kdpv is thewinding coefficient N is the number of turns per phase andΦload Φpm +ΦW When ΦW 0 it corresponds to anopen-circuit situation and the back-emf becomes the no-load back-emf E0
25 Armature Reaction Inductance -e constant-powertransformation from the three-phase coordinates to the d-qcoordinate is given by the following
id
iq
i0
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
23
1113970
cos θ cos θ minus2π3
1113874 1113875 cos θ +2π3
1113874 1113875
minus sin θ minus sin θ minus2π3
1113874 1113875 minus sin θ +2π3
1113874 1113875
12
1113970 12
1113970 12
1113970
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
(21)
where θ 1113938ω dt represents the rotor position which is anangle between the d-axis and winding axis of phase A Whenthe above two axes coincide θ 0Whereupon according toequation (21) there are
id
iq
⎡⎣ ⎤⎦ [C]
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (22a)
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ [C]
Tid
iq1113890 1113891 (22b)
where [C] 23
radic 1 minus 12 120
3
radic2 minus
3
radic21113890 1113891 and [C]T and [C]minus 1
are transposed matrix and inverse matrix of [C] re-spectively Notably there is [C]T [C]minus 1 According toequation (22a) we have
id
23
1113970
iA minus12
iB minus12
iC1113874 1113875
iq
23
1113970 3
radic
2iB minus
3
radic
2iC1113888 1113889
⎧⎪⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎪⎩
(23)
θ 0 which means Pωt 0 in equation (7)Substituting (7) into equation (23) yields
id 3
radicIRmscos(β)
iq 3
radicIRmssin(β)
⎧⎨
⎩ (24)
where IRms is the RMS of phase current IRms Im2
radic and β
denotes the electrical angle of the phase current vectorbeyond the d-axis Let β 0 by substituting equation (24)into equation (22b) to get the three-phase AC [iA iB iC]when the three-phase AC flows into the stator windings itwill generate the d-axis armature reaction field Obviouslythe stator iron saturation is most likely to occur when thedirection of the armature reaction field is in line with that ofthe PM field accordingly we only consider this situation
Denoting the inductance matrix in the abc coordinatesystem by
LABC1113858 1113859
LAA LAB LAC
LBA LBB LBC
LCA LCB LCC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (25)
-e flux linkage in the abc coordinate system can berepresented as
λA
λB
λC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ LABC
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (26)
-e flux linkage in the dq coordinate system is alsoapplied to equation (22a) that is
λd
λq
⎡⎣ ⎤⎦ [C]
λA
λB
λC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (27)
Substituting equation (26) and equation (22b) intoequation (27) yields the following
λd
λq
⎡⎣ ⎤⎦ [C] LABC1113858 1113859[C]T
id
iq
⎡⎣ ⎤⎦ Ldq1113960 1113961id
iq
⎡⎣ ⎤⎦ (28)
where [Ldq] Ld Lmd
Lmq Lq1113890 1113891 [C][LABC][C]T it represents
the inductance matrix in the dq coordinate system If the endleakage is not considered and other leakage factors such ashigher harmonics and slot are also ignored then Ld and Lq
will represent the d-axis and q-axis armature reaction in-ductances namely Lad and Laq respectively From the FEM[LABC] will be obtained and then it will turn into [Ldq] byusing the above transformation
Lad and Laq can also be calculated by analytical methodand the process is as follows
Journal of Electrical and Computer Engineering 5
-e load flux linkage of phase A is calculated as
λA NR2lef 1113946αy2
minus αy2BloadIIr R2 θ( 1113857dθ (29)
where N is the number of turns per coil of one phasewinding and αy is the coil pitch angle -e load flux linkageof phase B and C is expressed as
λB λAcos2π3
λC λAcos4π3
⎧⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎩
(30)
According to the above-mentioned [λA λB λC] fromequation (27) we will get [λd λq] And then there is
Ld λd
id (31)
Similarly neglecting some leakage factors Ld calculatedby equation (31) corresponds to Lad
3 Example Validation
To verify the analytic solution to the electromagnetic torquea locked-rotor simulation is carried out by the FEMmethod-e motor parameters are listed in Table 1
In the simulation the windings are fed by a symmetricalthree-phase alternating current to produce a rotating ar-mature reaction field but the rotor speed is set to zero thatis the rotor is locked When the armature reaction fieldrotates relative to the rotor with the angular velocity ω thelocked-rotor torque will generate and vary periodically withδ (an angle between two axes of the armature reaction fieldand PM field and it changes periodically from 0 to 360degrees) -e locked-rotor phenomenon can be applied toverify the electromagnetic torque of equation (16) andequation (18)
When the motor is fed with currents as shown inequation (7) the locked-rotor torques are as shown inFigure 3 -ree kinds of input currents of 2000Hz withdifferent amplitudes such as 10A 30A and 50A areconsidered Figure 3(a) shows the three-phase currents witha 10A peak value in FEM -e locked-rotor torque curvesobtained by FEM are shown in Figure 3(b) which shows thatthe locked-rotor torque is also sinusoidal with time and itspeak value is increased with the increasing amplitude of thephase current A comparison of the results of FEM and theanalytical model is illustrated in Figure 3(c) in which it canbe found that the analytical solution is in good agreementwith the FEM solution
Peak torques obtained by the FE model are comparedwith those from the derived analytical equations-e resultswhich are listed in Table 2 show that the errors are less than5 which proves the correctness of equation (16) andequation (18)
-emagnetic field results that were calculated by FEM areshown in Figure 4 where red and blue conductors indicate the
current flows out of and into the cross section respectively Inaddition the PMhas a parallel magnetization (the direction ofthe magnetization is vertically upward)
Because the PM field is much larger than the armaturereaction field the magnetic fields of the motor in Figure 4 atdifferent times are approximately the same At 02ms thelocked-rotor torque in Figure 3(b) is negative and thecorresponding motor magnetic field is shown in Figure 4(a)in which the armature reaction field lags behind the PM fieldby approximately 135 degrees which just causes a negativetorque Moreover the armature reaction field weakens thePM field the maximum flux density occurs in the statortooth and its value is 138 T At 04ms the locked-rotortorque equals zero as shown in Figure 3(b) and the cor-responding motor magnetic field is shown in Figure 4(b)where the armature reaction field coincides with the PMfield which verifies the null torque At 04ms the maximumflux density of the stator tooth is 170 T At 05ms thelocked-rotor torque approaches the maximum as shown inFigure 3(b) and the corresponding motor magnetic field isshown in Figure 4(c) It can be seen that the armature re-action field direction is left (90 degrees ahead of the PMfield) which verifies the maximum torque At 05ms themaximum flux density of the stator tooth is 160 T
-ree methods namely the FM method the air-gappermeance method [10] and the subdomain method wereused to estimate the cogging torque and the results areshown in Figure 5 -e air-gap permeance method is basedon the assumption that the flux crosses the magnet and airgap in a straight line wherever a magnet faces a toothHowever the above assumption will no longer be suitablefor the motor mentioned in the paper because the flux willbend in the air gap as shown in Figure 5(a) the coggingtorque calculated by the air-gap permeance method is smallrelative to that from FEM and the subdomain method asshown in Figure 5(b) -e cogging torque peak is approx-imately 370 μNm and is 005 of the rated torque (08Nm)which is too small to be ignored In fact some degrees ofharmonic components will exist in the polarized PM thusthe cogging torque peak will increase
-e stator tooth tips in relation to the width of the slotopenings also influence the back-emf waveform When theheight of the tooth tips is small magnetic saturation existsand the emf waveform tends to be a trapezoid Howeverwhen the tooth tip height is too large the emf waveform willdeteriorate [11] -e motor consists of a stator having 12
Table 1 Parameters of the high-speed PMSM
Symbol Quantity ValuesR1 PM radius 12times10minus 3mR2 Inner radius of stator 174times10minus 3mQ Slot numbers 12lef Core length 55times10minus 3mM Magnetization 820000Amω Angular velocity 12560 radsNs Conductors per slot 20μr Relative recoil permeability 1038P Number of pole pairs 1
6 Journal of Electrical and Computer Engineering
teeth which carry overlapping winding -e no-load back-emf waveform obtained from the FEM is displayed inFigure 6(a) and the emf waveform is almost purely sinu-soidal which is ideal for ultra-high-speed PMSMs As shownin Figure 5(a) a stator tooth has the maximum flux density1542 T which does not reach the saturation region of thematerial-e comparison of the no-load back-emf calculatedby equation (20) and that of FEM is shown in Figure 6(b)which manifests a good agreement of both methodsFigure 6(c) shows the prototype of the PMSM in which therotor is supported by gas foil bearings and the experimentalresults are shown in Figure 6(d) A coast-down test under60000 rmin is carried out that is the frequency converterwill shut down when the rotor is stable at 60000 rmin and
then the rotor speed will decrease to zero gradually byovercoming air friction In this process the stator has a back-emf but no current thereupon the rotor is not subjected tothe electromagnetic force Hereupon the curve of a no-loadback-emf vs speed can be achieved -e experimental anddesign values of the no-load back-emf are 570V and 596 Vat 746Hz respectively and the difference between them iswithin 46
-e results of Lad and Laq are shown in Figure 7 It canbe found that Lad and Laq are same when current is smalland both of them decrease with the increase of the currentBecause the d-axial magnetic circuit is more saturated Laddescends faster -e magnetic saturation is neglected inthe analytical method whereas Lad not varies with thecurrent
When IRms 10A and 100A the errors of the twomethods are 9 and 2 respectively It shows that theresults of the analytical method and FEM are very close Inthe analytical method only the fundamental harmonic isused to calculate the flux linkage However in the FEMexcept for the fundamental harmonic high-order harmonics
009 020 040 060 080 100 120ndash1000
1000
XY Plot 3torque_Cal_10kw120krpm
ndash750
750500
ndash500ndash250
250000
Curr
ent (
A)
Input current (phase A)
Input current (phase C)Input current (phase B)
Time (ms)
(a)
XY Plot 1torque_Cal_10kw120krpm
Torque Im = lsquo10ArsquoTorque_1 Im = lsquo30ArsquoTorque_2 Im = lsquo50Arsquo
Time (ms)
ndash150
150
ndash100
100
ndash050
050
000
020 040 060 080 120100 130000
Y1 (N
ewto
nMet
er)
(b)
0 02 04 06 08 1 12ndash15
ndash05
0
05
1
15
ndash1
2
t (ms)
Torq
ue T
emN
m
FE 10AFE 30AFE 50A
Analytic 10AAnalytic 30AAnalytic 50A
(c)
Figure 3-e results of the locked-rotor (a)-e three-phase alternating currents (b)-e torque vs time (c) Result comparison of FEM andthe analytical model
Table 2 Peak torque comparison between FE and the theory
Im (A) FEM (Nmiddotm) Analytical (Nmiddotm) Error ()10 0266 0271 18730 0802 0815 16250 1333 1357 176
Journal of Electrical and Computer Engineering 7
B (tesla)13842e + 000
12112e + 00011247e + 000
95165e ndash 00186514e ndash 001
69212e ndash 00160561e ndash 00151910e ndash 00143259e ndash 00134608e ndash 00125957e ndash 001
86550e ndash 00240351e ndash 005
17306e ndash 001
77863e ndash 001
10382e + 000
12977e + 000
Time = 00002sSpeed = 0000000rpmPosition = 0000000deg
0 40 80 (mm)
(a)
B (tesla)17001e + 000
13813e + 000
10626e + 000
12540e ndash 00410637e ndash 00121262e ndash 00131887e ndash 00142512e ndash 00153136e ndash 00163761e ndash 00174386e ndash 00185011e ndash 00195936e ndash 001
11689e + 00012751e + 000
14876e + 00015938e + 000
Speed = 0000000rpmPosition = 0000000deg
Time = 00004s0 40 80 (mm)
(b)B (tesla)
16056e + 000
90322e ndash 001
14749e ndash 00410049e ndash 00120083e ndash 00130117e ndash 00140152e ndash 00150186e ndash 00160220e ndash 00170254e ndash 00180288e ndash 001
10036e + 00011039e + 00012043e + 00013046e + 00014049e + 00015053e + 000
Position = 0000000deg
Time = 0000500000000000001sSpeed = 0000000rpm
40 80 (mm)0
(c)
Figure 4 Flux density vector field and current vector field of the FE model (a) Fields at 00002 s (b) Fields at 00004 s (c) Fields at 00005 s
B (tesla)15428e + 00014464e + 00013500e + 000
11572e + 00010608e + 000
96871e ndash 00219326e ndash 00128966e ndash 00138605e ndash 00148245e ndash 00157884e ndash 00167523e ndash 001
47663e ndash 004
77163e ndash 00186802e ndash 00196442e ndash 001
12536e + 000
(a)
10 20155 25 300Angle (degmech)
ndash400
400
ndash200
200
ndash100
100
0
ndash300
300
Cog
ging
torq
ue (μ
Nm
)
FEPermeanceSubdomain
(b)
Figure 5 PM field and cogging torque (a) PM field (b) Comparison of the cogging torque of FEM and the analytical model
8 Journal of Electrical and Computer Engineering
are also taken into account in the flux linkage -erefore theresult of the analytical method is smaller than that of FEM
It is worth noting that the stator tooth has the mostserious saturation as shown in Figure 8(a) and the fluxdensities reach 2086 T and 1880 T when IRms 100A(28IN) and 50A (14IN) respectively IN denotes the ratedcurrent of the motor -e working points of stator toothcorresponding to the both working condition are illustratedin Figure 8(b) It shows that the working point is located atthe end of the linear segment of the B-H curve whenIRms 14IN It indicates that the magnetic saturation of thestator iron is not prominent when the prototype worksunder rated working conditions and it applies to the pro-posed analytical method in this paper
In addition in order to enhance the cooling of the rotorthe air gap should be chosen as large as possible so thatmagnetic saturation in the tooth tips rarely occurs Fur-thermore the teeth in a 12-slot stator are diametricallysymmetrical the air-gap field distribution is symmetrical
000 020 040 060 080 100Time (ms)
ndash25000
ndash12500
000
12500
25000
Y1 (V
)XY Plot 1
Curve info rms_1 maxInduced voltage (phase A) 1560441 2188502Induced voltage (phase B) 1563378 2186613Induced voltage (phase C) 1573086 2184715
(a)
0 01 02 03 04 05Time (ms)
FE (phase A)Analytic
Vol
tage
(V)
ndash300
ndash200
ndash100
0
100
200
300
(b)
(c)
Time (s)0 50 100 150 200
0
500
1000
1500Fr
eque
ncy
(Hz)
0
50
100
150
Vol
tage
RM
SV
FrequencyVoltage RMS
(d)
Figure 6 Results of the no-load back-emf (a) No-load back-emf of FEM at 120000 rpm (b) Comparison of no-load back-emf at120000 rpm (c) Prototype of the ultra-high-speed PMSM (d) Experimental curves of the phase voltage and frequency
0 20 40 60 80 1000098
01
0102
0104
0106
0108
011
Phase current IRMSA
Lad FEMLaq FEMLad Analytical
d- an
d q-
axis
indu
ctan
ces (
mH
)
Figure 7 Curves of Lad and Laq vs the current
Journal of Electrical and Computer Engineering 9
and no unbalanced magnetic pull will exist if the rotor andstator axes are coincident Although the rotor rotates ec-centrically with respect to the stator in fact the eccentricityis too tiny due to the constraint of bearings to cause a notableunbalanced magnetic pull
4 Conclusion
For the ultra-high-speed PMSM including a stator withoverlapping windings and the 2-pole parallel magnetizedcylindrical PM electromagnetic torques are rarely reportedat present which leads to a prolonged motor design cycle toevaluating motor performance -erefore the work of thispaper is as follows
(1) -is paper deduces the analytical solution to theelectromagnetic driving torque of ultra-high-speedPMSMs which is verified by the FE methodaccording to locked-rotor simulation -e resultsshow that the electromagnetic torque is pro-portional to the length of stator iron the ampereturns per unit length of the stator inner surface andthe magnetization while it is proportional to thesquare of the ratio ie PM radius to stator innerradius
(2) -e cogging torque back-emf armature reactioninductance and so on are calculated compared anddiscussed For the PMSM studied in the paper theair-gap permeance method was shown not to beapplicable to the estimation of the cogging torquedue to the fact that the flux bends in the air gap andthus a high-accuracy method (eg the subdomainmethod) should be adopted
(3) A prototype is designed and manufactured and theno-load back-emf is checked by a 60000 rpm freecoast experiment -e experimental and designvalues of the no-load back-emf are 570 V and 596V
at 746Hz respectively and the difference betweenthem is within 46
Data Availability
-e data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
-e authors declare that they have no conflicts of interest
Acknowledgments
-is work is supported by National Natural ScienceFoundation of China (Grant nos 51705413 51705416 and11502196) and China Postdoctoral Science Foundation(Grant no 2017M613291XB)
References
[1] A Tenconi S Vaschetto and A Vigliani ldquoElectrical machinesfor high-speed applications design considerations andtradeoffsrdquo IEEE Transactions on Industrial Electronics vol 61no 6 pp 3022ndash3029 2014
[2] J H Ahn J Y Choi C H Park C Han C W Kim andT G Yoon ldquoCorrelation between rotor vibration and me-chanical stress in ultra-high-speed permanent magnet syn-chronous motorsrdquo IEEE Transactions on Magnetics vol 53no 11 Article ID 8209906 2017
[3] M-S Lim S-H Chai J-S Yang and J-P Hong ldquoDesign andverification of 150 krpm PMSM based on experiment resultsof prototyperdquo IEEE Transactions on Industrial Electronicsvol 62 no 12 pp 7827ndash7836 2015
[4] D-K Hong B-C Woo J-Y Lee and D-H Koo ldquoUltra highspeed motor supported by air foil bearings for air blowercooling fuel cellsrdquo IEEE Transactions on Magnetics vol 48no 2 pp 871ndash874 2012
20864e + 00019560e + 00018256e + 00016953e + 00015649e + 00014345e + 00013041e + 00011737e + 00010433e + 00091289e ndash 00178250e ndash 00165210e ndash 00152171e ndash 00139132e ndash 00126092e ndash 00113053e ndash 00113431e ndash 004
B (tesla)
(a)
375
250
125
000000E + 000 500E + 005
H (A_per_meter)
B (te
sla)
Working point 28IN
Working point 14IN
(b)
Figure 8 Magnetic saturation condition of the stator iron (a) Distribution of flux density for IRms 100A (b) Working points of the statortooth
10 Journal of Electrical and Computer Engineering
[5] D-K Hong B-C Woo Y-H Jeong D-H Koo andC-W Ahn ldquoDevelopment of an ultra high speed permanentmagnet synchronous motorrdquo International Journal of Pre-cision Engineering and Manufacturing vol 14 no 3pp 493ndash499 2013
[6] S-M Jang H-W Cho and S-K Choi ldquoDesign and analysisof a high-speed brushless DC motor for centrifugal com-pressorrdquo IEEE Transactions on Magnetics vol 43 no 6pp 2573ndash2575 2007
[7] D-K Hong D-S Joo B-CWoo D-H Koo and C-W AhnldquoUnbalance response analysis and experimental validation ofan ultra high speed motor-generator for microturbine gen-erators considering balancingrdquo Sensors vol 14 no 9pp 16117ndash16127 2014
[8] A B Proca A Keyhani A EL-Antably W Lu and M DaildquoAnalytical model for permanent magnet motors with surfacemounted magnetsrdquo IEEE Transactions on Energy Conversionvol 18 no 3 pp 386ndash391 2003
[9] F Dubas C Espanet and A Miraoui ldquoAn original analyticalexpression of the maximum magnet thickness in surfacemounted permanent magnet motorsrdquo De European PhysicalJournal Applied Physics vol 38 no 2 pp 169ndash176 2007
[10] Z Q Zhu and D Howe ldquoAnalytical prediction of the coggingtorque in radial-field permanent magnet brushless motorsrdquoIEEE Transactions on Magnetics vol 28 no 2 pp 1371ndash13741992
[11] Z Q Zhu K Ng and D Howe ldquoDesign and analysis of high-speed brushless permanent magnet motorsrdquo in Proceedings ofthe International Conference on Electrical Machines amp Drivespp 381ndash385 IET Cambridge UK September 1997
[12] W Cheng L Yu L Huang Y Lv L Li and Y Sun ldquoAn-alytical solution to magnetic field distribution of a parallelmagnetised rotor with cylindrical or ring-type permanentmagnetrdquo IET Electric Power Applications vol 9 no 6pp 429ndash437 2015
[13] Z Q Zhu and D Howe ldquoInstantaneous magnetic field dis-tribution in brushless permanent magnet DC motors IIArmature-reaction fieldrdquo IEEE Transactions on Magneticsvol 29 no 1 pp 136ndash142 1993
Journal of Electrical and Computer Engineering 11
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
cogging torque based on the analytical calculation of the air-gap field distribution and the net lateral force acting on theteeth
24 Back-Electromotive Force -e flux through a phasewinding excited by PM and by winding respectively can bewritten as follows
Φpm 2 BPM
IIr( 1113857maxπ
τplef
ΦW 2 BW
IIr( 1113857maxπ
τplef
⎧⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎩
(19)
where τp is the pole pitch and lef is the length of stator iron-e back-emf amplitude in each phase winding assuming a120-degree coil span is then obtained as follows
E 444fkdpvNΦload (20)
where f is the angular frequency of the rotor kdpv is thewinding coefficient N is the number of turns per phase andΦload Φpm +ΦW When ΦW 0 it corresponds to anopen-circuit situation and the back-emf becomes the no-load back-emf E0
25 Armature Reaction Inductance -e constant-powertransformation from the three-phase coordinates to the d-qcoordinate is given by the following
id
iq
i0
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
23
1113970
cos θ cos θ minus2π3
1113874 1113875 cos θ +2π3
1113874 1113875
minus sin θ minus sin θ minus2π3
1113874 1113875 minus sin θ +2π3
1113874 1113875
12
1113970 12
1113970 12
1113970
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
(21)
where θ 1113938ω dt represents the rotor position which is anangle between the d-axis and winding axis of phase A Whenthe above two axes coincide θ 0Whereupon according toequation (21) there are
id
iq
⎡⎣ ⎤⎦ [C]
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (22a)
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ [C]
Tid
iq1113890 1113891 (22b)
where [C] 23
radic 1 minus 12 120
3
radic2 minus
3
radic21113890 1113891 and [C]T and [C]minus 1
are transposed matrix and inverse matrix of [C] re-spectively Notably there is [C]T [C]minus 1 According toequation (22a) we have
id
23
1113970
iA minus12
iB minus12
iC1113874 1113875
iq
23
1113970 3
radic
2iB minus
3
radic
2iC1113888 1113889
⎧⎪⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎪⎩
(23)
θ 0 which means Pωt 0 in equation (7)Substituting (7) into equation (23) yields
id 3
radicIRmscos(β)
iq 3
radicIRmssin(β)
⎧⎨
⎩ (24)
where IRms is the RMS of phase current IRms Im2
radic and β
denotes the electrical angle of the phase current vectorbeyond the d-axis Let β 0 by substituting equation (24)into equation (22b) to get the three-phase AC [iA iB iC]when the three-phase AC flows into the stator windings itwill generate the d-axis armature reaction field Obviouslythe stator iron saturation is most likely to occur when thedirection of the armature reaction field is in line with that ofthe PM field accordingly we only consider this situation
Denoting the inductance matrix in the abc coordinatesystem by
LABC1113858 1113859
LAA LAB LAC
LBA LBB LBC
LCA LCB LCC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (25)
-e flux linkage in the abc coordinate system can berepresented as
λA
λB
λC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ LABC
iA
iB
iC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (26)
-e flux linkage in the dq coordinate system is alsoapplied to equation (22a) that is
λd
λq
⎡⎣ ⎤⎦ [C]
λA
λB
λC
⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (27)
Substituting equation (26) and equation (22b) intoequation (27) yields the following
λd
λq
⎡⎣ ⎤⎦ [C] LABC1113858 1113859[C]T
id
iq
⎡⎣ ⎤⎦ Ldq1113960 1113961id
iq
⎡⎣ ⎤⎦ (28)
where [Ldq] Ld Lmd
Lmq Lq1113890 1113891 [C][LABC][C]T it represents
the inductance matrix in the dq coordinate system If the endleakage is not considered and other leakage factors such ashigher harmonics and slot are also ignored then Ld and Lq
will represent the d-axis and q-axis armature reaction in-ductances namely Lad and Laq respectively From the FEM[LABC] will be obtained and then it will turn into [Ldq] byusing the above transformation
Lad and Laq can also be calculated by analytical methodand the process is as follows
Journal of Electrical and Computer Engineering 5
-e load flux linkage of phase A is calculated as
λA NR2lef 1113946αy2
minus αy2BloadIIr R2 θ( 1113857dθ (29)
where N is the number of turns per coil of one phasewinding and αy is the coil pitch angle -e load flux linkageof phase B and C is expressed as
λB λAcos2π3
λC λAcos4π3
⎧⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎩
(30)
According to the above-mentioned [λA λB λC] fromequation (27) we will get [λd λq] And then there is
Ld λd
id (31)
Similarly neglecting some leakage factors Ld calculatedby equation (31) corresponds to Lad
3 Example Validation
To verify the analytic solution to the electromagnetic torquea locked-rotor simulation is carried out by the FEMmethod-e motor parameters are listed in Table 1
In the simulation the windings are fed by a symmetricalthree-phase alternating current to produce a rotating ar-mature reaction field but the rotor speed is set to zero thatis the rotor is locked When the armature reaction fieldrotates relative to the rotor with the angular velocity ω thelocked-rotor torque will generate and vary periodically withδ (an angle between two axes of the armature reaction fieldand PM field and it changes periodically from 0 to 360degrees) -e locked-rotor phenomenon can be applied toverify the electromagnetic torque of equation (16) andequation (18)
When the motor is fed with currents as shown inequation (7) the locked-rotor torques are as shown inFigure 3 -ree kinds of input currents of 2000Hz withdifferent amplitudes such as 10A 30A and 50A areconsidered Figure 3(a) shows the three-phase currents witha 10A peak value in FEM -e locked-rotor torque curvesobtained by FEM are shown in Figure 3(b) which shows thatthe locked-rotor torque is also sinusoidal with time and itspeak value is increased with the increasing amplitude of thephase current A comparison of the results of FEM and theanalytical model is illustrated in Figure 3(c) in which it canbe found that the analytical solution is in good agreementwith the FEM solution
Peak torques obtained by the FE model are comparedwith those from the derived analytical equations-e resultswhich are listed in Table 2 show that the errors are less than5 which proves the correctness of equation (16) andequation (18)
-emagnetic field results that were calculated by FEM areshown in Figure 4 where red and blue conductors indicate the
current flows out of and into the cross section respectively Inaddition the PMhas a parallel magnetization (the direction ofthe magnetization is vertically upward)
Because the PM field is much larger than the armaturereaction field the magnetic fields of the motor in Figure 4 atdifferent times are approximately the same At 02ms thelocked-rotor torque in Figure 3(b) is negative and thecorresponding motor magnetic field is shown in Figure 4(a)in which the armature reaction field lags behind the PM fieldby approximately 135 degrees which just causes a negativetorque Moreover the armature reaction field weakens thePM field the maximum flux density occurs in the statortooth and its value is 138 T At 04ms the locked-rotortorque equals zero as shown in Figure 3(b) and the cor-responding motor magnetic field is shown in Figure 4(b)where the armature reaction field coincides with the PMfield which verifies the null torque At 04ms the maximumflux density of the stator tooth is 170 T At 05ms thelocked-rotor torque approaches the maximum as shown inFigure 3(b) and the corresponding motor magnetic field isshown in Figure 4(c) It can be seen that the armature re-action field direction is left (90 degrees ahead of the PMfield) which verifies the maximum torque At 05ms themaximum flux density of the stator tooth is 160 T
-ree methods namely the FM method the air-gappermeance method [10] and the subdomain method wereused to estimate the cogging torque and the results areshown in Figure 5 -e air-gap permeance method is basedon the assumption that the flux crosses the magnet and airgap in a straight line wherever a magnet faces a toothHowever the above assumption will no longer be suitablefor the motor mentioned in the paper because the flux willbend in the air gap as shown in Figure 5(a) the coggingtorque calculated by the air-gap permeance method is smallrelative to that from FEM and the subdomain method asshown in Figure 5(b) -e cogging torque peak is approx-imately 370 μNm and is 005 of the rated torque (08Nm)which is too small to be ignored In fact some degrees ofharmonic components will exist in the polarized PM thusthe cogging torque peak will increase
-e stator tooth tips in relation to the width of the slotopenings also influence the back-emf waveform When theheight of the tooth tips is small magnetic saturation existsand the emf waveform tends to be a trapezoid Howeverwhen the tooth tip height is too large the emf waveform willdeteriorate [11] -e motor consists of a stator having 12
Table 1 Parameters of the high-speed PMSM
Symbol Quantity ValuesR1 PM radius 12times10minus 3mR2 Inner radius of stator 174times10minus 3mQ Slot numbers 12lef Core length 55times10minus 3mM Magnetization 820000Amω Angular velocity 12560 radsNs Conductors per slot 20μr Relative recoil permeability 1038P Number of pole pairs 1
6 Journal of Electrical and Computer Engineering
teeth which carry overlapping winding -e no-load back-emf waveform obtained from the FEM is displayed inFigure 6(a) and the emf waveform is almost purely sinu-soidal which is ideal for ultra-high-speed PMSMs As shownin Figure 5(a) a stator tooth has the maximum flux density1542 T which does not reach the saturation region of thematerial-e comparison of the no-load back-emf calculatedby equation (20) and that of FEM is shown in Figure 6(b)which manifests a good agreement of both methodsFigure 6(c) shows the prototype of the PMSM in which therotor is supported by gas foil bearings and the experimentalresults are shown in Figure 6(d) A coast-down test under60000 rmin is carried out that is the frequency converterwill shut down when the rotor is stable at 60000 rmin and
then the rotor speed will decrease to zero gradually byovercoming air friction In this process the stator has a back-emf but no current thereupon the rotor is not subjected tothe electromagnetic force Hereupon the curve of a no-loadback-emf vs speed can be achieved -e experimental anddesign values of the no-load back-emf are 570V and 596 Vat 746Hz respectively and the difference between them iswithin 46
-e results of Lad and Laq are shown in Figure 7 It canbe found that Lad and Laq are same when current is smalland both of them decrease with the increase of the currentBecause the d-axial magnetic circuit is more saturated Laddescends faster -e magnetic saturation is neglected inthe analytical method whereas Lad not varies with thecurrent
When IRms 10A and 100A the errors of the twomethods are 9 and 2 respectively It shows that theresults of the analytical method and FEM are very close Inthe analytical method only the fundamental harmonic isused to calculate the flux linkage However in the FEMexcept for the fundamental harmonic high-order harmonics
009 020 040 060 080 100 120ndash1000
1000
XY Plot 3torque_Cal_10kw120krpm
ndash750
750500
ndash500ndash250
250000
Curr
ent (
A)
Input current (phase A)
Input current (phase C)Input current (phase B)
Time (ms)
(a)
XY Plot 1torque_Cal_10kw120krpm
Torque Im = lsquo10ArsquoTorque_1 Im = lsquo30ArsquoTorque_2 Im = lsquo50Arsquo
Time (ms)
ndash150
150
ndash100
100
ndash050
050
000
020 040 060 080 120100 130000
Y1 (N
ewto
nMet
er)
(b)
0 02 04 06 08 1 12ndash15
ndash05
0
05
1
15
ndash1
2
t (ms)
Torq
ue T
emN
m
FE 10AFE 30AFE 50A
Analytic 10AAnalytic 30AAnalytic 50A
(c)
Figure 3-e results of the locked-rotor (a)-e three-phase alternating currents (b)-e torque vs time (c) Result comparison of FEM andthe analytical model
Table 2 Peak torque comparison between FE and the theory
Im (A) FEM (Nmiddotm) Analytical (Nmiddotm) Error ()10 0266 0271 18730 0802 0815 16250 1333 1357 176
Journal of Electrical and Computer Engineering 7
B (tesla)13842e + 000
12112e + 00011247e + 000
95165e ndash 00186514e ndash 001
69212e ndash 00160561e ndash 00151910e ndash 00143259e ndash 00134608e ndash 00125957e ndash 001
86550e ndash 00240351e ndash 005
17306e ndash 001
77863e ndash 001
10382e + 000
12977e + 000
Time = 00002sSpeed = 0000000rpmPosition = 0000000deg
0 40 80 (mm)
(a)
B (tesla)17001e + 000
13813e + 000
10626e + 000
12540e ndash 00410637e ndash 00121262e ndash 00131887e ndash 00142512e ndash 00153136e ndash 00163761e ndash 00174386e ndash 00185011e ndash 00195936e ndash 001
11689e + 00012751e + 000
14876e + 00015938e + 000
Speed = 0000000rpmPosition = 0000000deg
Time = 00004s0 40 80 (mm)
(b)B (tesla)
16056e + 000
90322e ndash 001
14749e ndash 00410049e ndash 00120083e ndash 00130117e ndash 00140152e ndash 00150186e ndash 00160220e ndash 00170254e ndash 00180288e ndash 001
10036e + 00011039e + 00012043e + 00013046e + 00014049e + 00015053e + 000
Position = 0000000deg
Time = 0000500000000000001sSpeed = 0000000rpm
40 80 (mm)0
(c)
Figure 4 Flux density vector field and current vector field of the FE model (a) Fields at 00002 s (b) Fields at 00004 s (c) Fields at 00005 s
B (tesla)15428e + 00014464e + 00013500e + 000
11572e + 00010608e + 000
96871e ndash 00219326e ndash 00128966e ndash 00138605e ndash 00148245e ndash 00157884e ndash 00167523e ndash 001
47663e ndash 004
77163e ndash 00186802e ndash 00196442e ndash 001
12536e + 000
(a)
10 20155 25 300Angle (degmech)
ndash400
400
ndash200
200
ndash100
100
0
ndash300
300
Cog
ging
torq
ue (μ
Nm
)
FEPermeanceSubdomain
(b)
Figure 5 PM field and cogging torque (a) PM field (b) Comparison of the cogging torque of FEM and the analytical model
8 Journal of Electrical and Computer Engineering
are also taken into account in the flux linkage -erefore theresult of the analytical method is smaller than that of FEM
It is worth noting that the stator tooth has the mostserious saturation as shown in Figure 8(a) and the fluxdensities reach 2086 T and 1880 T when IRms 100A(28IN) and 50A (14IN) respectively IN denotes the ratedcurrent of the motor -e working points of stator toothcorresponding to the both working condition are illustratedin Figure 8(b) It shows that the working point is located atthe end of the linear segment of the B-H curve whenIRms 14IN It indicates that the magnetic saturation of thestator iron is not prominent when the prototype worksunder rated working conditions and it applies to the pro-posed analytical method in this paper
In addition in order to enhance the cooling of the rotorthe air gap should be chosen as large as possible so thatmagnetic saturation in the tooth tips rarely occurs Fur-thermore the teeth in a 12-slot stator are diametricallysymmetrical the air-gap field distribution is symmetrical
000 020 040 060 080 100Time (ms)
ndash25000
ndash12500
000
12500
25000
Y1 (V
)XY Plot 1
Curve info rms_1 maxInduced voltage (phase A) 1560441 2188502Induced voltage (phase B) 1563378 2186613Induced voltage (phase C) 1573086 2184715
(a)
0 01 02 03 04 05Time (ms)
FE (phase A)Analytic
Vol
tage
(V)
ndash300
ndash200
ndash100
0
100
200
300
(b)
(c)
Time (s)0 50 100 150 200
0
500
1000
1500Fr
eque
ncy
(Hz)
0
50
100
150
Vol
tage
RM
SV
FrequencyVoltage RMS
(d)
Figure 6 Results of the no-load back-emf (a) No-load back-emf of FEM at 120000 rpm (b) Comparison of no-load back-emf at120000 rpm (c) Prototype of the ultra-high-speed PMSM (d) Experimental curves of the phase voltage and frequency
0 20 40 60 80 1000098
01
0102
0104
0106
0108
011
Phase current IRMSA
Lad FEMLaq FEMLad Analytical
d- an
d q-
axis
indu
ctan
ces (
mH
)
Figure 7 Curves of Lad and Laq vs the current
Journal of Electrical and Computer Engineering 9
and no unbalanced magnetic pull will exist if the rotor andstator axes are coincident Although the rotor rotates ec-centrically with respect to the stator in fact the eccentricityis too tiny due to the constraint of bearings to cause a notableunbalanced magnetic pull
4 Conclusion
For the ultra-high-speed PMSM including a stator withoverlapping windings and the 2-pole parallel magnetizedcylindrical PM electromagnetic torques are rarely reportedat present which leads to a prolonged motor design cycle toevaluating motor performance -erefore the work of thispaper is as follows
(1) -is paper deduces the analytical solution to theelectromagnetic driving torque of ultra-high-speedPMSMs which is verified by the FE methodaccording to locked-rotor simulation -e resultsshow that the electromagnetic torque is pro-portional to the length of stator iron the ampereturns per unit length of the stator inner surface andthe magnetization while it is proportional to thesquare of the ratio ie PM radius to stator innerradius
(2) -e cogging torque back-emf armature reactioninductance and so on are calculated compared anddiscussed For the PMSM studied in the paper theair-gap permeance method was shown not to beapplicable to the estimation of the cogging torquedue to the fact that the flux bends in the air gap andthus a high-accuracy method (eg the subdomainmethod) should be adopted
(3) A prototype is designed and manufactured and theno-load back-emf is checked by a 60000 rpm freecoast experiment -e experimental and designvalues of the no-load back-emf are 570 V and 596V
at 746Hz respectively and the difference betweenthem is within 46
Data Availability
-e data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
-e authors declare that they have no conflicts of interest
Acknowledgments
-is work is supported by National Natural ScienceFoundation of China (Grant nos 51705413 51705416 and11502196) and China Postdoctoral Science Foundation(Grant no 2017M613291XB)
References
[1] A Tenconi S Vaschetto and A Vigliani ldquoElectrical machinesfor high-speed applications design considerations andtradeoffsrdquo IEEE Transactions on Industrial Electronics vol 61no 6 pp 3022ndash3029 2014
[2] J H Ahn J Y Choi C H Park C Han C W Kim andT G Yoon ldquoCorrelation between rotor vibration and me-chanical stress in ultra-high-speed permanent magnet syn-chronous motorsrdquo IEEE Transactions on Magnetics vol 53no 11 Article ID 8209906 2017
[3] M-S Lim S-H Chai J-S Yang and J-P Hong ldquoDesign andverification of 150 krpm PMSM based on experiment resultsof prototyperdquo IEEE Transactions on Industrial Electronicsvol 62 no 12 pp 7827ndash7836 2015
[4] D-K Hong B-C Woo J-Y Lee and D-H Koo ldquoUltra highspeed motor supported by air foil bearings for air blowercooling fuel cellsrdquo IEEE Transactions on Magnetics vol 48no 2 pp 871ndash874 2012
20864e + 00019560e + 00018256e + 00016953e + 00015649e + 00014345e + 00013041e + 00011737e + 00010433e + 00091289e ndash 00178250e ndash 00165210e ndash 00152171e ndash 00139132e ndash 00126092e ndash 00113053e ndash 00113431e ndash 004
B (tesla)
(a)
375
250
125
000000E + 000 500E + 005
H (A_per_meter)
B (te
sla)
Working point 28IN
Working point 14IN
(b)
Figure 8 Magnetic saturation condition of the stator iron (a) Distribution of flux density for IRms 100A (b) Working points of the statortooth
10 Journal of Electrical and Computer Engineering
[5] D-K Hong B-C Woo Y-H Jeong D-H Koo andC-W Ahn ldquoDevelopment of an ultra high speed permanentmagnet synchronous motorrdquo International Journal of Pre-cision Engineering and Manufacturing vol 14 no 3pp 493ndash499 2013
[6] S-M Jang H-W Cho and S-K Choi ldquoDesign and analysisof a high-speed brushless DC motor for centrifugal com-pressorrdquo IEEE Transactions on Magnetics vol 43 no 6pp 2573ndash2575 2007
[7] D-K Hong D-S Joo B-CWoo D-H Koo and C-W AhnldquoUnbalance response analysis and experimental validation ofan ultra high speed motor-generator for microturbine gen-erators considering balancingrdquo Sensors vol 14 no 9pp 16117ndash16127 2014
[8] A B Proca A Keyhani A EL-Antably W Lu and M DaildquoAnalytical model for permanent magnet motors with surfacemounted magnetsrdquo IEEE Transactions on Energy Conversionvol 18 no 3 pp 386ndash391 2003
[9] F Dubas C Espanet and A Miraoui ldquoAn original analyticalexpression of the maximum magnet thickness in surfacemounted permanent magnet motorsrdquo De European PhysicalJournal Applied Physics vol 38 no 2 pp 169ndash176 2007
[10] Z Q Zhu and D Howe ldquoAnalytical prediction of the coggingtorque in radial-field permanent magnet brushless motorsrdquoIEEE Transactions on Magnetics vol 28 no 2 pp 1371ndash13741992
[11] Z Q Zhu K Ng and D Howe ldquoDesign and analysis of high-speed brushless permanent magnet motorsrdquo in Proceedings ofthe International Conference on Electrical Machines amp Drivespp 381ndash385 IET Cambridge UK September 1997
[12] W Cheng L Yu L Huang Y Lv L Li and Y Sun ldquoAn-alytical solution to magnetic field distribution of a parallelmagnetised rotor with cylindrical or ring-type permanentmagnetrdquo IET Electric Power Applications vol 9 no 6pp 429ndash437 2015
[13] Z Q Zhu and D Howe ldquoInstantaneous magnetic field dis-tribution in brushless permanent magnet DC motors IIArmature-reaction fieldrdquo IEEE Transactions on Magneticsvol 29 no 1 pp 136ndash142 1993
Journal of Electrical and Computer Engineering 11
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
-e load flux linkage of phase A is calculated as
λA NR2lef 1113946αy2
minus αy2BloadIIr R2 θ( 1113857dθ (29)
where N is the number of turns per coil of one phasewinding and αy is the coil pitch angle -e load flux linkageof phase B and C is expressed as
λB λAcos2π3
λC λAcos4π3
⎧⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎩
(30)
According to the above-mentioned [λA λB λC] fromequation (27) we will get [λd λq] And then there is
Ld λd
id (31)
Similarly neglecting some leakage factors Ld calculatedby equation (31) corresponds to Lad
3 Example Validation
To verify the analytic solution to the electromagnetic torquea locked-rotor simulation is carried out by the FEMmethod-e motor parameters are listed in Table 1
In the simulation the windings are fed by a symmetricalthree-phase alternating current to produce a rotating ar-mature reaction field but the rotor speed is set to zero thatis the rotor is locked When the armature reaction fieldrotates relative to the rotor with the angular velocity ω thelocked-rotor torque will generate and vary periodically withδ (an angle between two axes of the armature reaction fieldand PM field and it changes periodically from 0 to 360degrees) -e locked-rotor phenomenon can be applied toverify the electromagnetic torque of equation (16) andequation (18)
When the motor is fed with currents as shown inequation (7) the locked-rotor torques are as shown inFigure 3 -ree kinds of input currents of 2000Hz withdifferent amplitudes such as 10A 30A and 50A areconsidered Figure 3(a) shows the three-phase currents witha 10A peak value in FEM -e locked-rotor torque curvesobtained by FEM are shown in Figure 3(b) which shows thatthe locked-rotor torque is also sinusoidal with time and itspeak value is increased with the increasing amplitude of thephase current A comparison of the results of FEM and theanalytical model is illustrated in Figure 3(c) in which it canbe found that the analytical solution is in good agreementwith the FEM solution
Peak torques obtained by the FE model are comparedwith those from the derived analytical equations-e resultswhich are listed in Table 2 show that the errors are less than5 which proves the correctness of equation (16) andequation (18)
-emagnetic field results that were calculated by FEM areshown in Figure 4 where red and blue conductors indicate the
current flows out of and into the cross section respectively Inaddition the PMhas a parallel magnetization (the direction ofthe magnetization is vertically upward)
Because the PM field is much larger than the armaturereaction field the magnetic fields of the motor in Figure 4 atdifferent times are approximately the same At 02ms thelocked-rotor torque in Figure 3(b) is negative and thecorresponding motor magnetic field is shown in Figure 4(a)in which the armature reaction field lags behind the PM fieldby approximately 135 degrees which just causes a negativetorque Moreover the armature reaction field weakens thePM field the maximum flux density occurs in the statortooth and its value is 138 T At 04ms the locked-rotortorque equals zero as shown in Figure 3(b) and the cor-responding motor magnetic field is shown in Figure 4(b)where the armature reaction field coincides with the PMfield which verifies the null torque At 04ms the maximumflux density of the stator tooth is 170 T At 05ms thelocked-rotor torque approaches the maximum as shown inFigure 3(b) and the corresponding motor magnetic field isshown in Figure 4(c) It can be seen that the armature re-action field direction is left (90 degrees ahead of the PMfield) which verifies the maximum torque At 05ms themaximum flux density of the stator tooth is 160 T
-ree methods namely the FM method the air-gappermeance method [10] and the subdomain method wereused to estimate the cogging torque and the results areshown in Figure 5 -e air-gap permeance method is basedon the assumption that the flux crosses the magnet and airgap in a straight line wherever a magnet faces a toothHowever the above assumption will no longer be suitablefor the motor mentioned in the paper because the flux willbend in the air gap as shown in Figure 5(a) the coggingtorque calculated by the air-gap permeance method is smallrelative to that from FEM and the subdomain method asshown in Figure 5(b) -e cogging torque peak is approx-imately 370 μNm and is 005 of the rated torque (08Nm)which is too small to be ignored In fact some degrees ofharmonic components will exist in the polarized PM thusthe cogging torque peak will increase
-e stator tooth tips in relation to the width of the slotopenings also influence the back-emf waveform When theheight of the tooth tips is small magnetic saturation existsand the emf waveform tends to be a trapezoid Howeverwhen the tooth tip height is too large the emf waveform willdeteriorate [11] -e motor consists of a stator having 12
Table 1 Parameters of the high-speed PMSM
Symbol Quantity ValuesR1 PM radius 12times10minus 3mR2 Inner radius of stator 174times10minus 3mQ Slot numbers 12lef Core length 55times10minus 3mM Magnetization 820000Amω Angular velocity 12560 radsNs Conductors per slot 20μr Relative recoil permeability 1038P Number of pole pairs 1
6 Journal of Electrical and Computer Engineering
teeth which carry overlapping winding -e no-load back-emf waveform obtained from the FEM is displayed inFigure 6(a) and the emf waveform is almost purely sinu-soidal which is ideal for ultra-high-speed PMSMs As shownin Figure 5(a) a stator tooth has the maximum flux density1542 T which does not reach the saturation region of thematerial-e comparison of the no-load back-emf calculatedby equation (20) and that of FEM is shown in Figure 6(b)which manifests a good agreement of both methodsFigure 6(c) shows the prototype of the PMSM in which therotor is supported by gas foil bearings and the experimentalresults are shown in Figure 6(d) A coast-down test under60000 rmin is carried out that is the frequency converterwill shut down when the rotor is stable at 60000 rmin and
then the rotor speed will decrease to zero gradually byovercoming air friction In this process the stator has a back-emf but no current thereupon the rotor is not subjected tothe electromagnetic force Hereupon the curve of a no-loadback-emf vs speed can be achieved -e experimental anddesign values of the no-load back-emf are 570V and 596 Vat 746Hz respectively and the difference between them iswithin 46
-e results of Lad and Laq are shown in Figure 7 It canbe found that Lad and Laq are same when current is smalland both of them decrease with the increase of the currentBecause the d-axial magnetic circuit is more saturated Laddescends faster -e magnetic saturation is neglected inthe analytical method whereas Lad not varies with thecurrent
When IRms 10A and 100A the errors of the twomethods are 9 and 2 respectively It shows that theresults of the analytical method and FEM are very close Inthe analytical method only the fundamental harmonic isused to calculate the flux linkage However in the FEMexcept for the fundamental harmonic high-order harmonics
009 020 040 060 080 100 120ndash1000
1000
XY Plot 3torque_Cal_10kw120krpm
ndash750
750500
ndash500ndash250
250000
Curr
ent (
A)
Input current (phase A)
Input current (phase C)Input current (phase B)
Time (ms)
(a)
XY Plot 1torque_Cal_10kw120krpm
Torque Im = lsquo10ArsquoTorque_1 Im = lsquo30ArsquoTorque_2 Im = lsquo50Arsquo
Time (ms)
ndash150
150
ndash100
100
ndash050
050
000
020 040 060 080 120100 130000
Y1 (N
ewto
nMet
er)
(b)
0 02 04 06 08 1 12ndash15
ndash05
0
05
1
15
ndash1
2
t (ms)
Torq
ue T
emN
m
FE 10AFE 30AFE 50A
Analytic 10AAnalytic 30AAnalytic 50A
(c)
Figure 3-e results of the locked-rotor (a)-e three-phase alternating currents (b)-e torque vs time (c) Result comparison of FEM andthe analytical model
Table 2 Peak torque comparison between FE and the theory
Im (A) FEM (Nmiddotm) Analytical (Nmiddotm) Error ()10 0266 0271 18730 0802 0815 16250 1333 1357 176
Journal of Electrical and Computer Engineering 7
B (tesla)13842e + 000
12112e + 00011247e + 000
95165e ndash 00186514e ndash 001
69212e ndash 00160561e ndash 00151910e ndash 00143259e ndash 00134608e ndash 00125957e ndash 001
86550e ndash 00240351e ndash 005
17306e ndash 001
77863e ndash 001
10382e + 000
12977e + 000
Time = 00002sSpeed = 0000000rpmPosition = 0000000deg
0 40 80 (mm)
(a)
B (tesla)17001e + 000
13813e + 000
10626e + 000
12540e ndash 00410637e ndash 00121262e ndash 00131887e ndash 00142512e ndash 00153136e ndash 00163761e ndash 00174386e ndash 00185011e ndash 00195936e ndash 001
11689e + 00012751e + 000
14876e + 00015938e + 000
Speed = 0000000rpmPosition = 0000000deg
Time = 00004s0 40 80 (mm)
(b)B (tesla)
16056e + 000
90322e ndash 001
14749e ndash 00410049e ndash 00120083e ndash 00130117e ndash 00140152e ndash 00150186e ndash 00160220e ndash 00170254e ndash 00180288e ndash 001
10036e + 00011039e + 00012043e + 00013046e + 00014049e + 00015053e + 000
Position = 0000000deg
Time = 0000500000000000001sSpeed = 0000000rpm
40 80 (mm)0
(c)
Figure 4 Flux density vector field and current vector field of the FE model (a) Fields at 00002 s (b) Fields at 00004 s (c) Fields at 00005 s
B (tesla)15428e + 00014464e + 00013500e + 000
11572e + 00010608e + 000
96871e ndash 00219326e ndash 00128966e ndash 00138605e ndash 00148245e ndash 00157884e ndash 00167523e ndash 001
47663e ndash 004
77163e ndash 00186802e ndash 00196442e ndash 001
12536e + 000
(a)
10 20155 25 300Angle (degmech)
ndash400
400
ndash200
200
ndash100
100
0
ndash300
300
Cog
ging
torq
ue (μ
Nm
)
FEPermeanceSubdomain
(b)
Figure 5 PM field and cogging torque (a) PM field (b) Comparison of the cogging torque of FEM and the analytical model
8 Journal of Electrical and Computer Engineering
are also taken into account in the flux linkage -erefore theresult of the analytical method is smaller than that of FEM
It is worth noting that the stator tooth has the mostserious saturation as shown in Figure 8(a) and the fluxdensities reach 2086 T and 1880 T when IRms 100A(28IN) and 50A (14IN) respectively IN denotes the ratedcurrent of the motor -e working points of stator toothcorresponding to the both working condition are illustratedin Figure 8(b) It shows that the working point is located atthe end of the linear segment of the B-H curve whenIRms 14IN It indicates that the magnetic saturation of thestator iron is not prominent when the prototype worksunder rated working conditions and it applies to the pro-posed analytical method in this paper
In addition in order to enhance the cooling of the rotorthe air gap should be chosen as large as possible so thatmagnetic saturation in the tooth tips rarely occurs Fur-thermore the teeth in a 12-slot stator are diametricallysymmetrical the air-gap field distribution is symmetrical
000 020 040 060 080 100Time (ms)
ndash25000
ndash12500
000
12500
25000
Y1 (V
)XY Plot 1
Curve info rms_1 maxInduced voltage (phase A) 1560441 2188502Induced voltage (phase B) 1563378 2186613Induced voltage (phase C) 1573086 2184715
(a)
0 01 02 03 04 05Time (ms)
FE (phase A)Analytic
Vol
tage
(V)
ndash300
ndash200
ndash100
0
100
200
300
(b)
(c)
Time (s)0 50 100 150 200
0
500
1000
1500Fr
eque
ncy
(Hz)
0
50
100
150
Vol
tage
RM
SV
FrequencyVoltage RMS
(d)
Figure 6 Results of the no-load back-emf (a) No-load back-emf of FEM at 120000 rpm (b) Comparison of no-load back-emf at120000 rpm (c) Prototype of the ultra-high-speed PMSM (d) Experimental curves of the phase voltage and frequency
0 20 40 60 80 1000098
01
0102
0104
0106
0108
011
Phase current IRMSA
Lad FEMLaq FEMLad Analytical
d- an
d q-
axis
indu
ctan
ces (
mH
)
Figure 7 Curves of Lad and Laq vs the current
Journal of Electrical and Computer Engineering 9
and no unbalanced magnetic pull will exist if the rotor andstator axes are coincident Although the rotor rotates ec-centrically with respect to the stator in fact the eccentricityis too tiny due to the constraint of bearings to cause a notableunbalanced magnetic pull
4 Conclusion
For the ultra-high-speed PMSM including a stator withoverlapping windings and the 2-pole parallel magnetizedcylindrical PM electromagnetic torques are rarely reportedat present which leads to a prolonged motor design cycle toevaluating motor performance -erefore the work of thispaper is as follows
(1) -is paper deduces the analytical solution to theelectromagnetic driving torque of ultra-high-speedPMSMs which is verified by the FE methodaccording to locked-rotor simulation -e resultsshow that the electromagnetic torque is pro-portional to the length of stator iron the ampereturns per unit length of the stator inner surface andthe magnetization while it is proportional to thesquare of the ratio ie PM radius to stator innerradius
(2) -e cogging torque back-emf armature reactioninductance and so on are calculated compared anddiscussed For the PMSM studied in the paper theair-gap permeance method was shown not to beapplicable to the estimation of the cogging torquedue to the fact that the flux bends in the air gap andthus a high-accuracy method (eg the subdomainmethod) should be adopted
(3) A prototype is designed and manufactured and theno-load back-emf is checked by a 60000 rpm freecoast experiment -e experimental and designvalues of the no-load back-emf are 570 V and 596V
at 746Hz respectively and the difference betweenthem is within 46
Data Availability
-e data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
-e authors declare that they have no conflicts of interest
Acknowledgments
-is work is supported by National Natural ScienceFoundation of China (Grant nos 51705413 51705416 and11502196) and China Postdoctoral Science Foundation(Grant no 2017M613291XB)
References
[1] A Tenconi S Vaschetto and A Vigliani ldquoElectrical machinesfor high-speed applications design considerations andtradeoffsrdquo IEEE Transactions on Industrial Electronics vol 61no 6 pp 3022ndash3029 2014
[2] J H Ahn J Y Choi C H Park C Han C W Kim andT G Yoon ldquoCorrelation between rotor vibration and me-chanical stress in ultra-high-speed permanent magnet syn-chronous motorsrdquo IEEE Transactions on Magnetics vol 53no 11 Article ID 8209906 2017
[3] M-S Lim S-H Chai J-S Yang and J-P Hong ldquoDesign andverification of 150 krpm PMSM based on experiment resultsof prototyperdquo IEEE Transactions on Industrial Electronicsvol 62 no 12 pp 7827ndash7836 2015
[4] D-K Hong B-C Woo J-Y Lee and D-H Koo ldquoUltra highspeed motor supported by air foil bearings for air blowercooling fuel cellsrdquo IEEE Transactions on Magnetics vol 48no 2 pp 871ndash874 2012
20864e + 00019560e + 00018256e + 00016953e + 00015649e + 00014345e + 00013041e + 00011737e + 00010433e + 00091289e ndash 00178250e ndash 00165210e ndash 00152171e ndash 00139132e ndash 00126092e ndash 00113053e ndash 00113431e ndash 004
B (tesla)
(a)
375
250
125
000000E + 000 500E + 005
H (A_per_meter)
B (te
sla)
Working point 28IN
Working point 14IN
(b)
Figure 8 Magnetic saturation condition of the stator iron (a) Distribution of flux density for IRms 100A (b) Working points of the statortooth
10 Journal of Electrical and Computer Engineering
[5] D-K Hong B-C Woo Y-H Jeong D-H Koo andC-W Ahn ldquoDevelopment of an ultra high speed permanentmagnet synchronous motorrdquo International Journal of Pre-cision Engineering and Manufacturing vol 14 no 3pp 493ndash499 2013
[6] S-M Jang H-W Cho and S-K Choi ldquoDesign and analysisof a high-speed brushless DC motor for centrifugal com-pressorrdquo IEEE Transactions on Magnetics vol 43 no 6pp 2573ndash2575 2007
[7] D-K Hong D-S Joo B-CWoo D-H Koo and C-W AhnldquoUnbalance response analysis and experimental validation ofan ultra high speed motor-generator for microturbine gen-erators considering balancingrdquo Sensors vol 14 no 9pp 16117ndash16127 2014
[8] A B Proca A Keyhani A EL-Antably W Lu and M DaildquoAnalytical model for permanent magnet motors with surfacemounted magnetsrdquo IEEE Transactions on Energy Conversionvol 18 no 3 pp 386ndash391 2003
[9] F Dubas C Espanet and A Miraoui ldquoAn original analyticalexpression of the maximum magnet thickness in surfacemounted permanent magnet motorsrdquo De European PhysicalJournal Applied Physics vol 38 no 2 pp 169ndash176 2007
[10] Z Q Zhu and D Howe ldquoAnalytical prediction of the coggingtorque in radial-field permanent magnet brushless motorsrdquoIEEE Transactions on Magnetics vol 28 no 2 pp 1371ndash13741992
[11] Z Q Zhu K Ng and D Howe ldquoDesign and analysis of high-speed brushless permanent magnet motorsrdquo in Proceedings ofthe International Conference on Electrical Machines amp Drivespp 381ndash385 IET Cambridge UK September 1997
[12] W Cheng L Yu L Huang Y Lv L Li and Y Sun ldquoAn-alytical solution to magnetic field distribution of a parallelmagnetised rotor with cylindrical or ring-type permanentmagnetrdquo IET Electric Power Applications vol 9 no 6pp 429ndash437 2015
[13] Z Q Zhu and D Howe ldquoInstantaneous magnetic field dis-tribution in brushless permanent magnet DC motors IIArmature-reaction fieldrdquo IEEE Transactions on Magneticsvol 29 no 1 pp 136ndash142 1993
Journal of Electrical and Computer Engineering 11
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
teeth which carry overlapping winding -e no-load back-emf waveform obtained from the FEM is displayed inFigure 6(a) and the emf waveform is almost purely sinu-soidal which is ideal for ultra-high-speed PMSMs As shownin Figure 5(a) a stator tooth has the maximum flux density1542 T which does not reach the saturation region of thematerial-e comparison of the no-load back-emf calculatedby equation (20) and that of FEM is shown in Figure 6(b)which manifests a good agreement of both methodsFigure 6(c) shows the prototype of the PMSM in which therotor is supported by gas foil bearings and the experimentalresults are shown in Figure 6(d) A coast-down test under60000 rmin is carried out that is the frequency converterwill shut down when the rotor is stable at 60000 rmin and
then the rotor speed will decrease to zero gradually byovercoming air friction In this process the stator has a back-emf but no current thereupon the rotor is not subjected tothe electromagnetic force Hereupon the curve of a no-loadback-emf vs speed can be achieved -e experimental anddesign values of the no-load back-emf are 570V and 596 Vat 746Hz respectively and the difference between them iswithin 46
-e results of Lad and Laq are shown in Figure 7 It canbe found that Lad and Laq are same when current is smalland both of them decrease with the increase of the currentBecause the d-axial magnetic circuit is more saturated Laddescends faster -e magnetic saturation is neglected inthe analytical method whereas Lad not varies with thecurrent
When IRms 10A and 100A the errors of the twomethods are 9 and 2 respectively It shows that theresults of the analytical method and FEM are very close Inthe analytical method only the fundamental harmonic isused to calculate the flux linkage However in the FEMexcept for the fundamental harmonic high-order harmonics
009 020 040 060 080 100 120ndash1000
1000
XY Plot 3torque_Cal_10kw120krpm
ndash750
750500
ndash500ndash250
250000
Curr
ent (
A)
Input current (phase A)
Input current (phase C)Input current (phase B)
Time (ms)
(a)
XY Plot 1torque_Cal_10kw120krpm
Torque Im = lsquo10ArsquoTorque_1 Im = lsquo30ArsquoTorque_2 Im = lsquo50Arsquo
Time (ms)
ndash150
150
ndash100
100
ndash050
050
000
020 040 060 080 120100 130000
Y1 (N
ewto
nMet
er)
(b)
0 02 04 06 08 1 12ndash15
ndash05
0
05
1
15
ndash1
2
t (ms)
Torq
ue T
emN
m
FE 10AFE 30AFE 50A
Analytic 10AAnalytic 30AAnalytic 50A
(c)
Figure 3-e results of the locked-rotor (a)-e three-phase alternating currents (b)-e torque vs time (c) Result comparison of FEM andthe analytical model
Table 2 Peak torque comparison between FE and the theory
Im (A) FEM (Nmiddotm) Analytical (Nmiddotm) Error ()10 0266 0271 18730 0802 0815 16250 1333 1357 176
Journal of Electrical and Computer Engineering 7
B (tesla)13842e + 000
12112e + 00011247e + 000
95165e ndash 00186514e ndash 001
69212e ndash 00160561e ndash 00151910e ndash 00143259e ndash 00134608e ndash 00125957e ndash 001
86550e ndash 00240351e ndash 005
17306e ndash 001
77863e ndash 001
10382e + 000
12977e + 000
Time = 00002sSpeed = 0000000rpmPosition = 0000000deg
0 40 80 (mm)
(a)
B (tesla)17001e + 000
13813e + 000
10626e + 000
12540e ndash 00410637e ndash 00121262e ndash 00131887e ndash 00142512e ndash 00153136e ndash 00163761e ndash 00174386e ndash 00185011e ndash 00195936e ndash 001
11689e + 00012751e + 000
14876e + 00015938e + 000
Speed = 0000000rpmPosition = 0000000deg
Time = 00004s0 40 80 (mm)
(b)B (tesla)
16056e + 000
90322e ndash 001
14749e ndash 00410049e ndash 00120083e ndash 00130117e ndash 00140152e ndash 00150186e ndash 00160220e ndash 00170254e ndash 00180288e ndash 001
10036e + 00011039e + 00012043e + 00013046e + 00014049e + 00015053e + 000
Position = 0000000deg
Time = 0000500000000000001sSpeed = 0000000rpm
40 80 (mm)0
(c)
Figure 4 Flux density vector field and current vector field of the FE model (a) Fields at 00002 s (b) Fields at 00004 s (c) Fields at 00005 s
B (tesla)15428e + 00014464e + 00013500e + 000
11572e + 00010608e + 000
96871e ndash 00219326e ndash 00128966e ndash 00138605e ndash 00148245e ndash 00157884e ndash 00167523e ndash 001
47663e ndash 004
77163e ndash 00186802e ndash 00196442e ndash 001
12536e + 000
(a)
10 20155 25 300Angle (degmech)
ndash400
400
ndash200
200
ndash100
100
0
ndash300
300
Cog
ging
torq
ue (μ
Nm
)
FEPermeanceSubdomain
(b)
Figure 5 PM field and cogging torque (a) PM field (b) Comparison of the cogging torque of FEM and the analytical model
8 Journal of Electrical and Computer Engineering
are also taken into account in the flux linkage -erefore theresult of the analytical method is smaller than that of FEM
It is worth noting that the stator tooth has the mostserious saturation as shown in Figure 8(a) and the fluxdensities reach 2086 T and 1880 T when IRms 100A(28IN) and 50A (14IN) respectively IN denotes the ratedcurrent of the motor -e working points of stator toothcorresponding to the both working condition are illustratedin Figure 8(b) It shows that the working point is located atthe end of the linear segment of the B-H curve whenIRms 14IN It indicates that the magnetic saturation of thestator iron is not prominent when the prototype worksunder rated working conditions and it applies to the pro-posed analytical method in this paper
In addition in order to enhance the cooling of the rotorthe air gap should be chosen as large as possible so thatmagnetic saturation in the tooth tips rarely occurs Fur-thermore the teeth in a 12-slot stator are diametricallysymmetrical the air-gap field distribution is symmetrical
000 020 040 060 080 100Time (ms)
ndash25000
ndash12500
000
12500
25000
Y1 (V
)XY Plot 1
Curve info rms_1 maxInduced voltage (phase A) 1560441 2188502Induced voltage (phase B) 1563378 2186613Induced voltage (phase C) 1573086 2184715
(a)
0 01 02 03 04 05Time (ms)
FE (phase A)Analytic
Vol
tage
(V)
ndash300
ndash200
ndash100
0
100
200
300
(b)
(c)
Time (s)0 50 100 150 200
0
500
1000
1500Fr
eque
ncy
(Hz)
0
50
100
150
Vol
tage
RM
SV
FrequencyVoltage RMS
(d)
Figure 6 Results of the no-load back-emf (a) No-load back-emf of FEM at 120000 rpm (b) Comparison of no-load back-emf at120000 rpm (c) Prototype of the ultra-high-speed PMSM (d) Experimental curves of the phase voltage and frequency
0 20 40 60 80 1000098
01
0102
0104
0106
0108
011
Phase current IRMSA
Lad FEMLaq FEMLad Analytical
d- an
d q-
axis
indu
ctan
ces (
mH
)
Figure 7 Curves of Lad and Laq vs the current
Journal of Electrical and Computer Engineering 9
and no unbalanced magnetic pull will exist if the rotor andstator axes are coincident Although the rotor rotates ec-centrically with respect to the stator in fact the eccentricityis too tiny due to the constraint of bearings to cause a notableunbalanced magnetic pull
4 Conclusion
For the ultra-high-speed PMSM including a stator withoverlapping windings and the 2-pole parallel magnetizedcylindrical PM electromagnetic torques are rarely reportedat present which leads to a prolonged motor design cycle toevaluating motor performance -erefore the work of thispaper is as follows
(1) -is paper deduces the analytical solution to theelectromagnetic driving torque of ultra-high-speedPMSMs which is verified by the FE methodaccording to locked-rotor simulation -e resultsshow that the electromagnetic torque is pro-portional to the length of stator iron the ampereturns per unit length of the stator inner surface andthe magnetization while it is proportional to thesquare of the ratio ie PM radius to stator innerradius
(2) -e cogging torque back-emf armature reactioninductance and so on are calculated compared anddiscussed For the PMSM studied in the paper theair-gap permeance method was shown not to beapplicable to the estimation of the cogging torquedue to the fact that the flux bends in the air gap andthus a high-accuracy method (eg the subdomainmethod) should be adopted
(3) A prototype is designed and manufactured and theno-load back-emf is checked by a 60000 rpm freecoast experiment -e experimental and designvalues of the no-load back-emf are 570 V and 596V
at 746Hz respectively and the difference betweenthem is within 46
Data Availability
-e data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
-e authors declare that they have no conflicts of interest
Acknowledgments
-is work is supported by National Natural ScienceFoundation of China (Grant nos 51705413 51705416 and11502196) and China Postdoctoral Science Foundation(Grant no 2017M613291XB)
References
[1] A Tenconi S Vaschetto and A Vigliani ldquoElectrical machinesfor high-speed applications design considerations andtradeoffsrdquo IEEE Transactions on Industrial Electronics vol 61no 6 pp 3022ndash3029 2014
[2] J H Ahn J Y Choi C H Park C Han C W Kim andT G Yoon ldquoCorrelation between rotor vibration and me-chanical stress in ultra-high-speed permanent magnet syn-chronous motorsrdquo IEEE Transactions on Magnetics vol 53no 11 Article ID 8209906 2017
[3] M-S Lim S-H Chai J-S Yang and J-P Hong ldquoDesign andverification of 150 krpm PMSM based on experiment resultsof prototyperdquo IEEE Transactions on Industrial Electronicsvol 62 no 12 pp 7827ndash7836 2015
[4] D-K Hong B-C Woo J-Y Lee and D-H Koo ldquoUltra highspeed motor supported by air foil bearings for air blowercooling fuel cellsrdquo IEEE Transactions on Magnetics vol 48no 2 pp 871ndash874 2012
20864e + 00019560e + 00018256e + 00016953e + 00015649e + 00014345e + 00013041e + 00011737e + 00010433e + 00091289e ndash 00178250e ndash 00165210e ndash 00152171e ndash 00139132e ndash 00126092e ndash 00113053e ndash 00113431e ndash 004
B (tesla)
(a)
375
250
125
000000E + 000 500E + 005
H (A_per_meter)
B (te
sla)
Working point 28IN
Working point 14IN
(b)
Figure 8 Magnetic saturation condition of the stator iron (a) Distribution of flux density for IRms 100A (b) Working points of the statortooth
10 Journal of Electrical and Computer Engineering
[5] D-K Hong B-C Woo Y-H Jeong D-H Koo andC-W Ahn ldquoDevelopment of an ultra high speed permanentmagnet synchronous motorrdquo International Journal of Pre-cision Engineering and Manufacturing vol 14 no 3pp 493ndash499 2013
[6] S-M Jang H-W Cho and S-K Choi ldquoDesign and analysisof a high-speed brushless DC motor for centrifugal com-pressorrdquo IEEE Transactions on Magnetics vol 43 no 6pp 2573ndash2575 2007
[7] D-K Hong D-S Joo B-CWoo D-H Koo and C-W AhnldquoUnbalance response analysis and experimental validation ofan ultra high speed motor-generator for microturbine gen-erators considering balancingrdquo Sensors vol 14 no 9pp 16117ndash16127 2014
[8] A B Proca A Keyhani A EL-Antably W Lu and M DaildquoAnalytical model for permanent magnet motors with surfacemounted magnetsrdquo IEEE Transactions on Energy Conversionvol 18 no 3 pp 386ndash391 2003
[9] F Dubas C Espanet and A Miraoui ldquoAn original analyticalexpression of the maximum magnet thickness in surfacemounted permanent magnet motorsrdquo De European PhysicalJournal Applied Physics vol 38 no 2 pp 169ndash176 2007
[10] Z Q Zhu and D Howe ldquoAnalytical prediction of the coggingtorque in radial-field permanent magnet brushless motorsrdquoIEEE Transactions on Magnetics vol 28 no 2 pp 1371ndash13741992
[11] Z Q Zhu K Ng and D Howe ldquoDesign and analysis of high-speed brushless permanent magnet motorsrdquo in Proceedings ofthe International Conference on Electrical Machines amp Drivespp 381ndash385 IET Cambridge UK September 1997
[12] W Cheng L Yu L Huang Y Lv L Li and Y Sun ldquoAn-alytical solution to magnetic field distribution of a parallelmagnetised rotor with cylindrical or ring-type permanentmagnetrdquo IET Electric Power Applications vol 9 no 6pp 429ndash437 2015
[13] Z Q Zhu and D Howe ldquoInstantaneous magnetic field dis-tribution in brushless permanent magnet DC motors IIArmature-reaction fieldrdquo IEEE Transactions on Magneticsvol 29 no 1 pp 136ndash142 1993
Journal of Electrical and Computer Engineering 11
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
B (tesla)13842e + 000
12112e + 00011247e + 000
95165e ndash 00186514e ndash 001
69212e ndash 00160561e ndash 00151910e ndash 00143259e ndash 00134608e ndash 00125957e ndash 001
86550e ndash 00240351e ndash 005
17306e ndash 001
77863e ndash 001
10382e + 000
12977e + 000
Time = 00002sSpeed = 0000000rpmPosition = 0000000deg
0 40 80 (mm)
(a)
B (tesla)17001e + 000
13813e + 000
10626e + 000
12540e ndash 00410637e ndash 00121262e ndash 00131887e ndash 00142512e ndash 00153136e ndash 00163761e ndash 00174386e ndash 00185011e ndash 00195936e ndash 001
11689e + 00012751e + 000
14876e + 00015938e + 000
Speed = 0000000rpmPosition = 0000000deg
Time = 00004s0 40 80 (mm)
(b)B (tesla)
16056e + 000
90322e ndash 001
14749e ndash 00410049e ndash 00120083e ndash 00130117e ndash 00140152e ndash 00150186e ndash 00160220e ndash 00170254e ndash 00180288e ndash 001
10036e + 00011039e + 00012043e + 00013046e + 00014049e + 00015053e + 000
Position = 0000000deg
Time = 0000500000000000001sSpeed = 0000000rpm
40 80 (mm)0
(c)
Figure 4 Flux density vector field and current vector field of the FE model (a) Fields at 00002 s (b) Fields at 00004 s (c) Fields at 00005 s
B (tesla)15428e + 00014464e + 00013500e + 000
11572e + 00010608e + 000
96871e ndash 00219326e ndash 00128966e ndash 00138605e ndash 00148245e ndash 00157884e ndash 00167523e ndash 001
47663e ndash 004
77163e ndash 00186802e ndash 00196442e ndash 001
12536e + 000
(a)
10 20155 25 300Angle (degmech)
ndash400
400
ndash200
200
ndash100
100
0
ndash300
300
Cog
ging
torq
ue (μ
Nm
)
FEPermeanceSubdomain
(b)
Figure 5 PM field and cogging torque (a) PM field (b) Comparison of the cogging torque of FEM and the analytical model
8 Journal of Electrical and Computer Engineering
are also taken into account in the flux linkage -erefore theresult of the analytical method is smaller than that of FEM
It is worth noting that the stator tooth has the mostserious saturation as shown in Figure 8(a) and the fluxdensities reach 2086 T and 1880 T when IRms 100A(28IN) and 50A (14IN) respectively IN denotes the ratedcurrent of the motor -e working points of stator toothcorresponding to the both working condition are illustratedin Figure 8(b) It shows that the working point is located atthe end of the linear segment of the B-H curve whenIRms 14IN It indicates that the magnetic saturation of thestator iron is not prominent when the prototype worksunder rated working conditions and it applies to the pro-posed analytical method in this paper
In addition in order to enhance the cooling of the rotorthe air gap should be chosen as large as possible so thatmagnetic saturation in the tooth tips rarely occurs Fur-thermore the teeth in a 12-slot stator are diametricallysymmetrical the air-gap field distribution is symmetrical
000 020 040 060 080 100Time (ms)
ndash25000
ndash12500
000
12500
25000
Y1 (V
)XY Plot 1
Curve info rms_1 maxInduced voltage (phase A) 1560441 2188502Induced voltage (phase B) 1563378 2186613Induced voltage (phase C) 1573086 2184715
(a)
0 01 02 03 04 05Time (ms)
FE (phase A)Analytic
Vol
tage
(V)
ndash300
ndash200
ndash100
0
100
200
300
(b)
(c)
Time (s)0 50 100 150 200
0
500
1000
1500Fr
eque
ncy
(Hz)
0
50
100
150
Vol
tage
RM
SV
FrequencyVoltage RMS
(d)
Figure 6 Results of the no-load back-emf (a) No-load back-emf of FEM at 120000 rpm (b) Comparison of no-load back-emf at120000 rpm (c) Prototype of the ultra-high-speed PMSM (d) Experimental curves of the phase voltage and frequency
0 20 40 60 80 1000098
01
0102
0104
0106
0108
011
Phase current IRMSA
Lad FEMLaq FEMLad Analytical
d- an
d q-
axis
indu
ctan
ces (
mH
)
Figure 7 Curves of Lad and Laq vs the current
Journal of Electrical and Computer Engineering 9
and no unbalanced magnetic pull will exist if the rotor andstator axes are coincident Although the rotor rotates ec-centrically with respect to the stator in fact the eccentricityis too tiny due to the constraint of bearings to cause a notableunbalanced magnetic pull
4 Conclusion
For the ultra-high-speed PMSM including a stator withoverlapping windings and the 2-pole parallel magnetizedcylindrical PM electromagnetic torques are rarely reportedat present which leads to a prolonged motor design cycle toevaluating motor performance -erefore the work of thispaper is as follows
(1) -is paper deduces the analytical solution to theelectromagnetic driving torque of ultra-high-speedPMSMs which is verified by the FE methodaccording to locked-rotor simulation -e resultsshow that the electromagnetic torque is pro-portional to the length of stator iron the ampereturns per unit length of the stator inner surface andthe magnetization while it is proportional to thesquare of the ratio ie PM radius to stator innerradius
(2) -e cogging torque back-emf armature reactioninductance and so on are calculated compared anddiscussed For the PMSM studied in the paper theair-gap permeance method was shown not to beapplicable to the estimation of the cogging torquedue to the fact that the flux bends in the air gap andthus a high-accuracy method (eg the subdomainmethod) should be adopted
(3) A prototype is designed and manufactured and theno-load back-emf is checked by a 60000 rpm freecoast experiment -e experimental and designvalues of the no-load back-emf are 570 V and 596V
at 746Hz respectively and the difference betweenthem is within 46
Data Availability
-e data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
-e authors declare that they have no conflicts of interest
Acknowledgments
-is work is supported by National Natural ScienceFoundation of China (Grant nos 51705413 51705416 and11502196) and China Postdoctoral Science Foundation(Grant no 2017M613291XB)
References
[1] A Tenconi S Vaschetto and A Vigliani ldquoElectrical machinesfor high-speed applications design considerations andtradeoffsrdquo IEEE Transactions on Industrial Electronics vol 61no 6 pp 3022ndash3029 2014
[2] J H Ahn J Y Choi C H Park C Han C W Kim andT G Yoon ldquoCorrelation between rotor vibration and me-chanical stress in ultra-high-speed permanent magnet syn-chronous motorsrdquo IEEE Transactions on Magnetics vol 53no 11 Article ID 8209906 2017
[3] M-S Lim S-H Chai J-S Yang and J-P Hong ldquoDesign andverification of 150 krpm PMSM based on experiment resultsof prototyperdquo IEEE Transactions on Industrial Electronicsvol 62 no 12 pp 7827ndash7836 2015
[4] D-K Hong B-C Woo J-Y Lee and D-H Koo ldquoUltra highspeed motor supported by air foil bearings for air blowercooling fuel cellsrdquo IEEE Transactions on Magnetics vol 48no 2 pp 871ndash874 2012
20864e + 00019560e + 00018256e + 00016953e + 00015649e + 00014345e + 00013041e + 00011737e + 00010433e + 00091289e ndash 00178250e ndash 00165210e ndash 00152171e ndash 00139132e ndash 00126092e ndash 00113053e ndash 00113431e ndash 004
B (tesla)
(a)
375
250
125
000000E + 000 500E + 005
H (A_per_meter)
B (te
sla)
Working point 28IN
Working point 14IN
(b)
Figure 8 Magnetic saturation condition of the stator iron (a) Distribution of flux density for IRms 100A (b) Working points of the statortooth
10 Journal of Electrical and Computer Engineering
[5] D-K Hong B-C Woo Y-H Jeong D-H Koo andC-W Ahn ldquoDevelopment of an ultra high speed permanentmagnet synchronous motorrdquo International Journal of Pre-cision Engineering and Manufacturing vol 14 no 3pp 493ndash499 2013
[6] S-M Jang H-W Cho and S-K Choi ldquoDesign and analysisof a high-speed brushless DC motor for centrifugal com-pressorrdquo IEEE Transactions on Magnetics vol 43 no 6pp 2573ndash2575 2007
[7] D-K Hong D-S Joo B-CWoo D-H Koo and C-W AhnldquoUnbalance response analysis and experimental validation ofan ultra high speed motor-generator for microturbine gen-erators considering balancingrdquo Sensors vol 14 no 9pp 16117ndash16127 2014
[8] A B Proca A Keyhani A EL-Antably W Lu and M DaildquoAnalytical model for permanent magnet motors with surfacemounted magnetsrdquo IEEE Transactions on Energy Conversionvol 18 no 3 pp 386ndash391 2003
[9] F Dubas C Espanet and A Miraoui ldquoAn original analyticalexpression of the maximum magnet thickness in surfacemounted permanent magnet motorsrdquo De European PhysicalJournal Applied Physics vol 38 no 2 pp 169ndash176 2007
[10] Z Q Zhu and D Howe ldquoAnalytical prediction of the coggingtorque in radial-field permanent magnet brushless motorsrdquoIEEE Transactions on Magnetics vol 28 no 2 pp 1371ndash13741992
[11] Z Q Zhu K Ng and D Howe ldquoDesign and analysis of high-speed brushless permanent magnet motorsrdquo in Proceedings ofthe International Conference on Electrical Machines amp Drivespp 381ndash385 IET Cambridge UK September 1997
[12] W Cheng L Yu L Huang Y Lv L Li and Y Sun ldquoAn-alytical solution to magnetic field distribution of a parallelmagnetised rotor with cylindrical or ring-type permanentmagnetrdquo IET Electric Power Applications vol 9 no 6pp 429ndash437 2015
[13] Z Q Zhu and D Howe ldquoInstantaneous magnetic field dis-tribution in brushless permanent magnet DC motors IIArmature-reaction fieldrdquo IEEE Transactions on Magneticsvol 29 no 1 pp 136ndash142 1993
Journal of Electrical and Computer Engineering 11
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
are also taken into account in the flux linkage -erefore theresult of the analytical method is smaller than that of FEM
It is worth noting that the stator tooth has the mostserious saturation as shown in Figure 8(a) and the fluxdensities reach 2086 T and 1880 T when IRms 100A(28IN) and 50A (14IN) respectively IN denotes the ratedcurrent of the motor -e working points of stator toothcorresponding to the both working condition are illustratedin Figure 8(b) It shows that the working point is located atthe end of the linear segment of the B-H curve whenIRms 14IN It indicates that the magnetic saturation of thestator iron is not prominent when the prototype worksunder rated working conditions and it applies to the pro-posed analytical method in this paper
In addition in order to enhance the cooling of the rotorthe air gap should be chosen as large as possible so thatmagnetic saturation in the tooth tips rarely occurs Fur-thermore the teeth in a 12-slot stator are diametricallysymmetrical the air-gap field distribution is symmetrical
000 020 040 060 080 100Time (ms)
ndash25000
ndash12500
000
12500
25000
Y1 (V
)XY Plot 1
Curve info rms_1 maxInduced voltage (phase A) 1560441 2188502Induced voltage (phase B) 1563378 2186613Induced voltage (phase C) 1573086 2184715
(a)
0 01 02 03 04 05Time (ms)
FE (phase A)Analytic
Vol
tage
(V)
ndash300
ndash200
ndash100
0
100
200
300
(b)
(c)
Time (s)0 50 100 150 200
0
500
1000
1500Fr
eque
ncy
(Hz)
0
50
100
150
Vol
tage
RM
SV
FrequencyVoltage RMS
(d)
Figure 6 Results of the no-load back-emf (a) No-load back-emf of FEM at 120000 rpm (b) Comparison of no-load back-emf at120000 rpm (c) Prototype of the ultra-high-speed PMSM (d) Experimental curves of the phase voltage and frequency
0 20 40 60 80 1000098
01
0102
0104
0106
0108
011
Phase current IRMSA
Lad FEMLaq FEMLad Analytical
d- an
d q-
axis
indu
ctan
ces (
mH
)
Figure 7 Curves of Lad and Laq vs the current
Journal of Electrical and Computer Engineering 9
and no unbalanced magnetic pull will exist if the rotor andstator axes are coincident Although the rotor rotates ec-centrically with respect to the stator in fact the eccentricityis too tiny due to the constraint of bearings to cause a notableunbalanced magnetic pull
4 Conclusion
For the ultra-high-speed PMSM including a stator withoverlapping windings and the 2-pole parallel magnetizedcylindrical PM electromagnetic torques are rarely reportedat present which leads to a prolonged motor design cycle toevaluating motor performance -erefore the work of thispaper is as follows
(1) -is paper deduces the analytical solution to theelectromagnetic driving torque of ultra-high-speedPMSMs which is verified by the FE methodaccording to locked-rotor simulation -e resultsshow that the electromagnetic torque is pro-portional to the length of stator iron the ampereturns per unit length of the stator inner surface andthe magnetization while it is proportional to thesquare of the ratio ie PM radius to stator innerradius
(2) -e cogging torque back-emf armature reactioninductance and so on are calculated compared anddiscussed For the PMSM studied in the paper theair-gap permeance method was shown not to beapplicable to the estimation of the cogging torquedue to the fact that the flux bends in the air gap andthus a high-accuracy method (eg the subdomainmethod) should be adopted
(3) A prototype is designed and manufactured and theno-load back-emf is checked by a 60000 rpm freecoast experiment -e experimental and designvalues of the no-load back-emf are 570 V and 596V
at 746Hz respectively and the difference betweenthem is within 46
Data Availability
-e data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
-e authors declare that they have no conflicts of interest
Acknowledgments
-is work is supported by National Natural ScienceFoundation of China (Grant nos 51705413 51705416 and11502196) and China Postdoctoral Science Foundation(Grant no 2017M613291XB)
References
[1] A Tenconi S Vaschetto and A Vigliani ldquoElectrical machinesfor high-speed applications design considerations andtradeoffsrdquo IEEE Transactions on Industrial Electronics vol 61no 6 pp 3022ndash3029 2014
[2] J H Ahn J Y Choi C H Park C Han C W Kim andT G Yoon ldquoCorrelation between rotor vibration and me-chanical stress in ultra-high-speed permanent magnet syn-chronous motorsrdquo IEEE Transactions on Magnetics vol 53no 11 Article ID 8209906 2017
[3] M-S Lim S-H Chai J-S Yang and J-P Hong ldquoDesign andverification of 150 krpm PMSM based on experiment resultsof prototyperdquo IEEE Transactions on Industrial Electronicsvol 62 no 12 pp 7827ndash7836 2015
[4] D-K Hong B-C Woo J-Y Lee and D-H Koo ldquoUltra highspeed motor supported by air foil bearings for air blowercooling fuel cellsrdquo IEEE Transactions on Magnetics vol 48no 2 pp 871ndash874 2012
20864e + 00019560e + 00018256e + 00016953e + 00015649e + 00014345e + 00013041e + 00011737e + 00010433e + 00091289e ndash 00178250e ndash 00165210e ndash 00152171e ndash 00139132e ndash 00126092e ndash 00113053e ndash 00113431e ndash 004
B (tesla)
(a)
375
250
125
000000E + 000 500E + 005
H (A_per_meter)
B (te
sla)
Working point 28IN
Working point 14IN
(b)
Figure 8 Magnetic saturation condition of the stator iron (a) Distribution of flux density for IRms 100A (b) Working points of the statortooth
10 Journal of Electrical and Computer Engineering
[5] D-K Hong B-C Woo Y-H Jeong D-H Koo andC-W Ahn ldquoDevelopment of an ultra high speed permanentmagnet synchronous motorrdquo International Journal of Pre-cision Engineering and Manufacturing vol 14 no 3pp 493ndash499 2013
[6] S-M Jang H-W Cho and S-K Choi ldquoDesign and analysisof a high-speed brushless DC motor for centrifugal com-pressorrdquo IEEE Transactions on Magnetics vol 43 no 6pp 2573ndash2575 2007
[7] D-K Hong D-S Joo B-CWoo D-H Koo and C-W AhnldquoUnbalance response analysis and experimental validation ofan ultra high speed motor-generator for microturbine gen-erators considering balancingrdquo Sensors vol 14 no 9pp 16117ndash16127 2014
[8] A B Proca A Keyhani A EL-Antably W Lu and M DaildquoAnalytical model for permanent magnet motors with surfacemounted magnetsrdquo IEEE Transactions on Energy Conversionvol 18 no 3 pp 386ndash391 2003
[9] F Dubas C Espanet and A Miraoui ldquoAn original analyticalexpression of the maximum magnet thickness in surfacemounted permanent magnet motorsrdquo De European PhysicalJournal Applied Physics vol 38 no 2 pp 169ndash176 2007
[10] Z Q Zhu and D Howe ldquoAnalytical prediction of the coggingtorque in radial-field permanent magnet brushless motorsrdquoIEEE Transactions on Magnetics vol 28 no 2 pp 1371ndash13741992
[11] Z Q Zhu K Ng and D Howe ldquoDesign and analysis of high-speed brushless permanent magnet motorsrdquo in Proceedings ofthe International Conference on Electrical Machines amp Drivespp 381ndash385 IET Cambridge UK September 1997
[12] W Cheng L Yu L Huang Y Lv L Li and Y Sun ldquoAn-alytical solution to magnetic field distribution of a parallelmagnetised rotor with cylindrical or ring-type permanentmagnetrdquo IET Electric Power Applications vol 9 no 6pp 429ndash437 2015
[13] Z Q Zhu and D Howe ldquoInstantaneous magnetic field dis-tribution in brushless permanent magnet DC motors IIArmature-reaction fieldrdquo IEEE Transactions on Magneticsvol 29 no 1 pp 136ndash142 1993
Journal of Electrical and Computer Engineering 11
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
and no unbalanced magnetic pull will exist if the rotor andstator axes are coincident Although the rotor rotates ec-centrically with respect to the stator in fact the eccentricityis too tiny due to the constraint of bearings to cause a notableunbalanced magnetic pull
4 Conclusion
For the ultra-high-speed PMSM including a stator withoverlapping windings and the 2-pole parallel magnetizedcylindrical PM electromagnetic torques are rarely reportedat present which leads to a prolonged motor design cycle toevaluating motor performance -erefore the work of thispaper is as follows
(1) -is paper deduces the analytical solution to theelectromagnetic driving torque of ultra-high-speedPMSMs which is verified by the FE methodaccording to locked-rotor simulation -e resultsshow that the electromagnetic torque is pro-portional to the length of stator iron the ampereturns per unit length of the stator inner surface andthe magnetization while it is proportional to thesquare of the ratio ie PM radius to stator innerradius
(2) -e cogging torque back-emf armature reactioninductance and so on are calculated compared anddiscussed For the PMSM studied in the paper theair-gap permeance method was shown not to beapplicable to the estimation of the cogging torquedue to the fact that the flux bends in the air gap andthus a high-accuracy method (eg the subdomainmethod) should be adopted
(3) A prototype is designed and manufactured and theno-load back-emf is checked by a 60000 rpm freecoast experiment -e experimental and designvalues of the no-load back-emf are 570 V and 596V
at 746Hz respectively and the difference betweenthem is within 46
Data Availability
-e data used to support the findings of this study areavailable from the corresponding author upon request
Conflicts of Interest
-e authors declare that they have no conflicts of interest
Acknowledgments
-is work is supported by National Natural ScienceFoundation of China (Grant nos 51705413 51705416 and11502196) and China Postdoctoral Science Foundation(Grant no 2017M613291XB)
References
[1] A Tenconi S Vaschetto and A Vigliani ldquoElectrical machinesfor high-speed applications design considerations andtradeoffsrdquo IEEE Transactions on Industrial Electronics vol 61no 6 pp 3022ndash3029 2014
[2] J H Ahn J Y Choi C H Park C Han C W Kim andT G Yoon ldquoCorrelation between rotor vibration and me-chanical stress in ultra-high-speed permanent magnet syn-chronous motorsrdquo IEEE Transactions on Magnetics vol 53no 11 Article ID 8209906 2017
[3] M-S Lim S-H Chai J-S Yang and J-P Hong ldquoDesign andverification of 150 krpm PMSM based on experiment resultsof prototyperdquo IEEE Transactions on Industrial Electronicsvol 62 no 12 pp 7827ndash7836 2015
[4] D-K Hong B-C Woo J-Y Lee and D-H Koo ldquoUltra highspeed motor supported by air foil bearings for air blowercooling fuel cellsrdquo IEEE Transactions on Magnetics vol 48no 2 pp 871ndash874 2012
20864e + 00019560e + 00018256e + 00016953e + 00015649e + 00014345e + 00013041e + 00011737e + 00010433e + 00091289e ndash 00178250e ndash 00165210e ndash 00152171e ndash 00139132e ndash 00126092e ndash 00113053e ndash 00113431e ndash 004
B (tesla)
(a)
375
250
125
000000E + 000 500E + 005
H (A_per_meter)
B (te
sla)
Working point 28IN
Working point 14IN
(b)
Figure 8 Magnetic saturation condition of the stator iron (a) Distribution of flux density for IRms 100A (b) Working points of the statortooth
10 Journal of Electrical and Computer Engineering
[5] D-K Hong B-C Woo Y-H Jeong D-H Koo andC-W Ahn ldquoDevelopment of an ultra high speed permanentmagnet synchronous motorrdquo International Journal of Pre-cision Engineering and Manufacturing vol 14 no 3pp 493ndash499 2013
[6] S-M Jang H-W Cho and S-K Choi ldquoDesign and analysisof a high-speed brushless DC motor for centrifugal com-pressorrdquo IEEE Transactions on Magnetics vol 43 no 6pp 2573ndash2575 2007
[7] D-K Hong D-S Joo B-CWoo D-H Koo and C-W AhnldquoUnbalance response analysis and experimental validation ofan ultra high speed motor-generator for microturbine gen-erators considering balancingrdquo Sensors vol 14 no 9pp 16117ndash16127 2014
[8] A B Proca A Keyhani A EL-Antably W Lu and M DaildquoAnalytical model for permanent magnet motors with surfacemounted magnetsrdquo IEEE Transactions on Energy Conversionvol 18 no 3 pp 386ndash391 2003
[9] F Dubas C Espanet and A Miraoui ldquoAn original analyticalexpression of the maximum magnet thickness in surfacemounted permanent magnet motorsrdquo De European PhysicalJournal Applied Physics vol 38 no 2 pp 169ndash176 2007
[10] Z Q Zhu and D Howe ldquoAnalytical prediction of the coggingtorque in radial-field permanent magnet brushless motorsrdquoIEEE Transactions on Magnetics vol 28 no 2 pp 1371ndash13741992
[11] Z Q Zhu K Ng and D Howe ldquoDesign and analysis of high-speed brushless permanent magnet motorsrdquo in Proceedings ofthe International Conference on Electrical Machines amp Drivespp 381ndash385 IET Cambridge UK September 1997
[12] W Cheng L Yu L Huang Y Lv L Li and Y Sun ldquoAn-alytical solution to magnetic field distribution of a parallelmagnetised rotor with cylindrical or ring-type permanentmagnetrdquo IET Electric Power Applications vol 9 no 6pp 429ndash437 2015
[13] Z Q Zhu and D Howe ldquoInstantaneous magnetic field dis-tribution in brushless permanent magnet DC motors IIArmature-reaction fieldrdquo IEEE Transactions on Magneticsvol 29 no 1 pp 136ndash142 1993
Journal of Electrical and Computer Engineering 11
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
[5] D-K Hong B-C Woo Y-H Jeong D-H Koo andC-W Ahn ldquoDevelopment of an ultra high speed permanentmagnet synchronous motorrdquo International Journal of Pre-cision Engineering and Manufacturing vol 14 no 3pp 493ndash499 2013
[6] S-M Jang H-W Cho and S-K Choi ldquoDesign and analysisof a high-speed brushless DC motor for centrifugal com-pressorrdquo IEEE Transactions on Magnetics vol 43 no 6pp 2573ndash2575 2007
[7] D-K Hong D-S Joo B-CWoo D-H Koo and C-W AhnldquoUnbalance response analysis and experimental validation ofan ultra high speed motor-generator for microturbine gen-erators considering balancingrdquo Sensors vol 14 no 9pp 16117ndash16127 2014
[8] A B Proca A Keyhani A EL-Antably W Lu and M DaildquoAnalytical model for permanent magnet motors with surfacemounted magnetsrdquo IEEE Transactions on Energy Conversionvol 18 no 3 pp 386ndash391 2003
[9] F Dubas C Espanet and A Miraoui ldquoAn original analyticalexpression of the maximum magnet thickness in surfacemounted permanent magnet motorsrdquo De European PhysicalJournal Applied Physics vol 38 no 2 pp 169ndash176 2007
[10] Z Q Zhu and D Howe ldquoAnalytical prediction of the coggingtorque in radial-field permanent magnet brushless motorsrdquoIEEE Transactions on Magnetics vol 28 no 2 pp 1371ndash13741992
[11] Z Q Zhu K Ng and D Howe ldquoDesign and analysis of high-speed brushless permanent magnet motorsrdquo in Proceedings ofthe International Conference on Electrical Machines amp Drivespp 381ndash385 IET Cambridge UK September 1997
[12] W Cheng L Yu L Huang Y Lv L Li and Y Sun ldquoAn-alytical solution to magnetic field distribution of a parallelmagnetised rotor with cylindrical or ring-type permanentmagnetrdquo IET Electric Power Applications vol 9 no 6pp 429ndash437 2015
[13] Z Q Zhu and D Howe ldquoInstantaneous magnetic field dis-tribution in brushless permanent magnet DC motors IIArmature-reaction fieldrdquo IEEE Transactions on Magneticsvol 29 no 1 pp 136ndash142 1993
Journal of Electrical and Computer Engineering 11
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Top Related