Filippo Milanesi Tesi

University of Bologna
DEPARTMENT OF ELECTRICAL ENGINEERING

FACULTY OF ENGINEERING
Ph. D. in Electrical Engineering
XXI year
POWER ELECTRONICS, MACHINES AND DRIVES (ING-IND/32)
Design optimization and

control strategies for PM
Multiphase Tubular Linear
Actuators
Ph. D. Thesis:
Filippo Milanesi
Tutor:
Prof. Giovanni Serra
Ph. D. Coordinator:
Prof. Francesco Negrini
Final Dissertation in 2009
TomyparentsandtoSilvia
DesignoptimizationandcontrolstrategiesforPMMultiphaseTubularLinearActuators
TABLEOFCONTENTS
TABLEOFCONTENTS
CHAPTER1
Page9
BasicofTubularLinearMachines
1.1. Introduction....................................................................................................................
1.2. Structureoflinearactuators.................................................................................
A. Geometries...................................................................................................................
B. Forcesandfundamentalprincipleofoperations........................................
C. Applications................................................................................................................
1.3. Briefcomparisonamongdifferentlinearactuators................................
A. Linearinductionactuators....................................................................................
B. Linearpermanentmagnetsynchronousactuators....................................
C. Linearreluctancesynchronousactuators......................................................
D. Linearswitchedreluctanceandstepperactuators....................................
1.4. Technicalcharacteristicsoftubularlineardrives...................................
A. Lineartubularactuator..........................................................................................
B. Lineartubularactuatorcompetitors................................................................
1.5. IroncorePMtubularlinearactuator...............................................................
A. Permanentmagnets.................................................................................................
B. Magneticcircuit.........................................................................................................
C. Windings.......................................................................................................................
1.6. Conclusions......................................................................................................................
CHAPTER2
Page11
Page11
Page11
Page12
Page13
Page15
Page15
Page15
Page16
Page17
Page17
Page17
Page20
Page20
Page21
Page22
Page25
Page26
BasicofMultiphaseDrives
2.1. Introduction....................................................................................................................
2.2. Multiphasespacevectorrepresentation........................................................
2.3. MathematicalmodelofmultiphasePMsynchronousmachines.......
2.4. Mainfeatureofmultiphasedrives.....................................................................
2.5. Multiphasedrivesfortorquecapabilityenhancement.........................
2.6. Faulttolerantdrives..................................................................................................
2.7. MultiMotordrives......................................................................................................
A. MultiMotordriveconcept....................................................................................
B. Machineconnectivity...............................................................................................
Page27
Page27
Page29
Page30
Page31
Page33
Page34
Page37
Page39
7|P a g e
Tableofcontents
C. Control...........................................................................................................................
2.8. Conclusion........................................................................................................................
CHAPTER3
Page41
Page42
BasicofMultiphaseDrives
3.1. Introduction....................................................................................................................
3.2. Reviewofthedesignoptimizationmethodologies..................................
3.3. Electromagneticmodel.............................................................................................
A. Preliminarydesign...................................................................................................
B. Thrustforcecontributions....................................................................................
C. Fundamentalhypothesis.......................................................................................
D. Designofthemagnets.............................................................................................
E. Sliderandstatordesign.........................................................................................
F. Thrustforcecalculations.......................................................................................
3.4. Thermalmodel..............................................................................................................
3.5. Mechanicalmodeloftheslider............................................................................
A. Staticequilibriumanalysis....................................................................................
B. Maximumsliderdisplacement............................................................................
C. Reactionforcesonthebearings.........................................................................
3.6. Optimizeddesignalgorithm..................................................................................
A. Inputparameteranddesignconstraints........................................................
B. Preliminarydesign:solutionprocedureoftheintegratedmodel
equations......................................................................................................................
C. Thermalmodelsolution.........................................................................................
D. Finiteelementanalysis...........................................................................................
E. Numberofwiresperslotoptimization...........................................................
3.7. Conclusion........................................................................................................................
CHAPTER4
Page43
Page43
Page44
Page44
Page44
Page45
Page46
Page47
Page48
Page50
Page52
Page52
Page55
Page56
Page57
Page57
Page59
Page59
Page60
Page61
Page63
OptimizeddesignofaPMtubularlinear
actuatorandexperimentalresults
4.1. Introduction....................................................................................................................
4.2. Definitionofthedesignconstraints..................................................................
4.3. Analysisoftheinfluenceoftheairgaplengthonthemotor
performance...................................................................................................................
4.4. Optimizeddesignofatubularlinearactuator............................................
A. Preliminarydesign...................................................................................................
B. FEMrefining................................................................................................................
C. Optimizationofthenumberofwiresperslot..............................................
4.5. Experimentalresults.................................................................................................
8|P a g e
Page65
Page65
Page67
Page69
Page69
Page70
Page72
Page74
A. Thrustforce.................................................................................................................
B. Thermalbehaviour...................................................................................................
4.6. Conclusion........................................................................................................................
CHAPTER5
Page74
Page77
Page77
Five phasedualmotorseriesconnected
drive:simulationsandexperimentalresults
5.1. Introduction....................................................................................................................
5.2. Descriptionofthefivephasedualmotorseiresconnecteddrive..
A. Motorconnections....................................................................................................
B. Invertervoltagelimit...............................................................................................
C. Magneticdesignofthefivephasetubularmotor.......................................
D. Controlscheme...........................................................................................................
E. Tuningoftheregulators.........................................................................................
5.3. Experimentalresults..................................................................................................
A. Startuptest.................................................................................................................
B. Positionreferencevariation.................................................................................
C. Velocityvariation......................................................................................................
C. Independentcontrotest.........................................................................................
5.4. Conclusion........................................................................................................................
Page79
Page79
Page79
Page81
Page84
Page88
Page89
Page90
Page91
Page92
Page92
Page93
Page93
Conclusions
Page95
Bibliography
Page97
9|P a g e
CHAPTER
1
BasicsofTubularLinear
Machines
1.1 Introduction
Thedemandforlinearelectricmachine,speciallyforcontrolledmotion,hasregistered
acontinuousgrowthinrecentyears,sinceitsintegrationinindustrialapplicationsleadsto
important advantages [1]. Furthermore, the tubular structure seems to be attractive for
industrial purposes due to both its closed form and the inherently absence of attractive
forcebetweenthestatorandthemover.
Inthischapterthefundamentalcharacteristicsofthelinearmachineswillbeillustrated
inordertohighlinetheirmainadvantagesanddrawbacks.Particularattentionwillbepaidto
the technologic developments in soft and hard magnetic materials, which enable the
industrialrealizationofmotorwithhighforcedensity,improveddynamicbehaviourwithina
relativelylowcost.
Finallytheattentionwillbefocusedonpermanentmagnet(PM)tubularlinearmotors.
1.2 Structureoflinearactuators
A. Geometries
Linear electric actuators can drive payloads in a straight line providing thrust force
directly, without any intermediate mechanical transmission such as gears, screws or crank
shaft[2],[3].
Therangeoflinearmachinetypesavailabletodaycorrespondsalmostexactlywiththe
rangeofrotatingcounterpart[4],[5],[6],infactItisknownthatcuttingarotatingmachine
witharadialemiplaneandunrollingit,asdepictedinFigure1,aflatsinglesidedtopologyis
obtained.IftherotatingmotoriscutbyaradialplaneandsquashedasshowninFigure2,
a flat doublesided topology occur. Finally, the tubular linear topology is derived from the
linearflatsinglesidedconfigurationrollingitupalongthedirectionofmotion,asdepictedin
Figure3.Thesetransformationsturnthestatoroftherotatingmachineintotheprimaryof
thelinearmachine,whichisfedbyavariableorfixedfrequencysupply,andtherotorinto
thesecondary,whichisequippedwithpermanentmagnet,withamultiphasewindingoritis
11|P a g e
BasicsofTubularLinearMachines
Figure1Imaginaryprocessofsplittingandunrollingarotary
machinetoproduceasinglesidelinearmotor.
characterizedbyavariablereluctanceferromagneticstructuredependingonthekindofthe
machine.
It is evident that, due to the finite length of the machine, linear or reciprocating
motionsareachievableonlybyanextensionofthelinearstatororofthelinearrotor.The
result, shown in Figure 4, is the so called long stator or short stator machine. In practical
applicationsbothofthemareuseddependingontheappliancerequirements,forexample,
short stator solutions are often used in packaging and manufacturing sectors, while long
statortypesareadoptedfortransportationsystems.Themovingpartofthelinearmachine
is often called forcer, mover or slider and can coincide both with the primary or with the
secondaryofthemachine.
B. Forcesandfundamentalprincipleofoperations
In classical rotary machines there are two forces acting on the rotor surface [7]: a
tangentialforcewhichproducesthetorqueandaradialforce,actingperpendicularlytothe
rotorsurface.Thislatterforcesumstozeroaroundthecircumferenceduetothecylindrical
Figure 2Imaginary process of cutting and squashing a rotary machine to

produceadoublesidelinearmotor.
12|P a g e
Figure 3Conceptual transformations to obtain a tubular motor starting to a

rotarymachine,passingthroughaflatlinearactuator.
symmetryofthemachine.Whenthemachineistransformedtocreatealinearmachine,
thetangentialforcessumeachothertoconstitutethedrivingforce.Thenormalforcesinflat
configurationsarenotnowbalancedandalargenetforceactsbetweentheunrolledrotor
and stator. The ratio between the latter and the thrust force is worthy of considerations,
typical value vary in the range 810. The propulsion force on the moving part of a linear
actuatorisgeneratedindifferentways,accordingtothemachinesoperationprinciple:
linearinductionmotorsproducethrustforcebytheelectromagneticinteraction
between the primary translating magnetic field, generated by an alternate
currentmultiphasewinding,andtheinducedcurrentonthemultiphasewinding
oronthebarofthesecondary;
linearsteppingorswitchedreluctancemotorscreatethrustforcebytheaction
ofamagneticfield,producedbyelectronicallyswitcheddirectcurrentwindings,
andanarrayofferromagneticpolesoravariablereluctanceferromagneticrail;
linearsynchronousmotorsproducethrustbytheactionoftravellingmagnetic
field, produced by an alternating current multiphase winding, and an array of
magneticpolesoravariablereluctanceferromagneticrail;
The aforementioned list suggests structural differences among the range of linear
machine types; in this way the presence/absence of windings in the primary and in the
secondary,theuseofpermanentmagnets,theutilizationofinnovativematerialsdetermine
thespecificlinearmotorcharacteristics.
C. Applications
Theuseoflinearactuatorshasreceivedagrowinginterestinseveralapplications[8],
[9],sincealargenumberoflinearorreciprocatingmotionsisrequired.
13|P a g e
(a)
(b)
Figure4Linearmotorwiththesocalledshortstator(a)orlongstator(b).
Linear machines show performances comparable with the ones of the rotating
counterpart; developments in materials, design methodologies and in manufacture
technologieshaveincreasedforcedensity,dynamicbehaviour,robustnessandreliability.In
someapplicationsthedesignconstraintlimitsorreduceperformancesoflinearmotors,viz.
in transport sector where large airgap are obliged, but these restrictions are more than
compensated by many characteristics of linear motors that cannot be achieved by rotary
equivalents.
Historically the first linear motor topology that took place in industrial and transport
applications was the induction one, but in the last 25 years the improvements in power
electronics and permanent magnet materials has driven the linear synchronous machine
towardincreasingnumberofutilizations.
Nowadaysflatandtubularlinearmotorsareappliedforcontrolledmotionsandservo
actuation in highspeed packaging and pickandplace machines, in automated structure
such as industrial robots [10], fast manipulators and automatic machine tool feeder [11].
Other applications are reciprocating compressors and refrigerators [12]. The
aforementioned applications share relatively low power requirement and they constitute
the main commercial demand of linear drives, furthermore they are often customized
solutions.
Highmediumpowerapplicationsarelessfrequent,howeverresearcharebeingcarried
outinthisparticularfield.Lineardriveshavebeenproposedinenergyproductionsystems,
freepiston energy converter was studied in [13] while tidal and waveenergy conversion
systemswereanalyzedin[14],[15].
An important application field concerns with transportation systems for public and
factories [16]; linear peoplemovers are currently used in operations at several
international airports, while magnetically levitated or wheels supported trains are used in
Japan[1]andEurope.
Linear motor drives are strong candidates for both lowspeed and highspeed mass
transitapplications.Systemsusinglinearinductionandsynchronousmachinesareprevalent
inindustry.
14|P a g e
The future prospects of application of linear drives are expected to involve not only
traditionalsectors,suchastransport,materialprocessingandautomationequipments,but
alsonewone,likeinaerospaceandautomotivesectors.
1.3 Briefcomparisonamongdifferentlinearactuators
Researchesandstudieshavebeencarriedoutforalltypeoflinearmachines,butonly
someofthemhavebeenfocusedforrealapplications.Thefeasibility,thecosteffectiveness,
the robustness and the reliability are key aspects that must be owned by an industrial
product.Forsomekindofmotorsmostofthesepropertiesareinherentlysatisfiedbutfor
other motor types these requirements have been satisfied with the developments of
materialsandofdigitalcontrollers.
Thelatterscoveredanimportantrolefortheindustrialdiffusionoflinearactuators.In
fact, control strategies for direct drive systems and servo actuations have become more
complexandprecise,capabletogeneratesmoothmotioneveninharddynamicoperations.
Theoperationalprinciplestronglyaffectsthemotorstructure,thusitdeterminestheir
attitudetobeusedforspecificapplications.
A. Linearinductionactuators
Linearinductionactuators(LIAs)havebeenbuiltwithana.c.multiphasewindinginthe
primary, whilst the secondary houses a cage or a shortcircuited three phase winding.
However, conventional secondary configurations have a conducting plate on a solid iron
plateormayhaveonlytheformer[4],[17].Theplatesolutiongivesaneasymanufactured
machine,witharobustandrelativelycheapsecondary,suitedforhostileenvironmentorfor
lowmaintenanceapplications.ThecoreofprimaryandsecondaryofflatLIAs,exceptforthe
plate secondary solution, is laminated in order to reduce core losses and magnetizing
current,furthermoreopenslotorsemiclosedslotgeometryfacilitatetheirconstruction.For
tubular structure, open slot geometry has been used since recent developments of soft
magneticmaterials,viz.softmagneticcomposites.Infact,laminationsoftubularcorewith
standardtechnologiesmanifestproblemsforfluxpathsorforconstructionfeasibility.
TheLIAsconfigurationdependonthetravellength:forshorttravelapplications,viz.0.5
m,thetubularstructureisfavouritewhilstfortravellengthupto23mthemovingprimary
flatconfigurationispreferred,andaflexiblecablesuppliesthewindings.Forgreatertravel
lengthlongprimaryflatconfigurationsareused;theprimarysectionsareplacedalongthe
trackandthesecondaryactsasthemover,theformerareturnedononlywhenthelatteris
intheirproximity.
B. Linearpermanentmagnetsynchronousactuators
Linear permanent magnet synchronous actuators (LPMSAs) has penetrated into the
market and established the practical application, since developments in hard and soft
15|P a g e
magnetic material occurred [16]. Progress of highenergy rareearth magnets, whose

remanencecanbeupto1.2Tat20Candwhoseoperatingtemperaturecanbeupto150C,
haspermittedtoreachproperperformancesforseveraldirectdriveapplications.Thereisno
doubt that permanent magnet machines are more expensive than induction machines.
However,theyofferimprovedperformanceespeciallyintermsofforcedensityanddynamic
response.Thisisparticularlytrueinsmallsizesactuators,wherethemagnetisingcurrentfor
induction machines becomes important to the disadvantage of the thrust force.
Unfortunately,thepresenceofmagnetscanbeaprobleminpollutedenvironmentdueto
straymagnetfields,particularlyinpresenceofironparticles,sincetheycanbeattractedand
causeamachinefault.
PrimarycorestructureforflatandtubularLPMSAsissimilartotheprimaryofLIAs.The
LPMSAs secondary is made up of permanentmagnets, used for excitation flux production,
and a laminated core. Magnets can be buried in the core structure or can be surface
mounted.
MovingprimaryflatLPMSAsareusedfortraveluptoabout3m,whilstforlongertravel
moving magnets flat configurations are used. For travel length of less than 1 m tubular
configurations may be preferred, since tubular structures make better use of materials,
resultingincompactactuators.
C. Linearreluctancesynchronousactuators
Linearreluctancesynchronousactuators(LRSAs)havethesameprimarycharacteristic
of LIAs and LPMSAs, but the secondary does not house windings or PMs. The thrust force
production is entrusted to the interaction between the secondary magnetic circuit
anisotropy and the primary travelling magnetic field. These actuators are specifically
designed to emphasize the magnetic anisotropy between qd axes, in fact the force
productionisproportionaltothesecondarysaliency.Recentresearchinrotatingreluctance
machine has led to development also in the linear counterpart, resulting in a relatively
inexpensivemotorwithgoodperformanceindices.
Good energy conversion is achieved by thin airgap, but this feature implies large
normal forces, viz. large attractive force between the stator and the mover. Thus the
mechanicalpartofflatconfigurationmeritsspecialtreatment,inordertoobtainasufficient
reliabilityandrobustness.
LSRAs can constitute an alternativeto LPMSAs and LIAs. Compared with the formers,
they have lower costs and can operate at very high speed because of the easier field
weakeningcapabilityandtheruggedrotor.Itdoesn'thavesecondarywindingthattravelsat
synchronousspeed,sothecontrollerissimplerthanotherofACmachines.Ifcomparedto
LIAs,ithastheoreticallynosecondarylossesandacomparableforcedensitydependingon
thedesignofthemotors.
16|P a g e
In conclusion, LRSA are a rarity in industrial drives, although they are not completely
unknown[18],[19].
D. Linearswitchedreluctanceandstepperactuators
Linear switched reluctance motor (LSRM) is an electric machine which posses simple
structurewithawindingonlyonthestator.Thecostsarelowerthanothermotorsduetoits
simpleconstruction[20].
Further,thewindingsareconcentratedratherthandistributed,makingthemidealfor
lowcostmanufacturingandmaintenance.Concentratedwindingsenableanaturallyfailsafe
system that can operate even with a phase shorted or open [21]. The availability of
numerouslowcostconvertertopologiestodrivethemmakestheLSRAadvantageousover
linearinductionandpermanentmagnetsynchronousmotors,whichincurahighercostand
havethermaldrawbacks[22],[23].AlthoughLSRAshaveplentyofadvantages,thefactthat
the force ripple deteriorates the performance restricts its application to precision devices.
Muchresearchhasbeenreportedtoreduceforceripple,andimprovedperformanceshave
been enhanced [24], [25]. Tubular construction of this kind of motor permits to overcome
theattractionforcebetweenexcitationandmover,sincetheaxialsymmetryofthemachine
allowsainherentneutralizationofthisforce[26].
In linear stepper actuators (LSAs) the motion is caused by the tangential force which
tendstoalignpolesofthemovingpartwiththeexcitationpolesofthestaticpart[27],[28].
Incontrasttorotarycounterpart,wherethenormalforceofattractionbetweenstatorand
rotorpolefacesisneutralizedbythesameforceactingbetweenthediametricallyopposite
pairofstatorandrotorpoles,thenormalforceinthelinearcounterpartishighandcanbea
seriousmechanicalproblem.Theproblemmay,tosomeextent,beovercomebytheuseofa
symmetricaldoublysidedversionoftherectilinearcounterpart.
1.4 Technicalcharacteristicsoftubularlineardrives
Tubular linear drive offers several advantages in comparison with both the flat linear
andtherotarycounterpart.Somebenefitsareinherentlyintroducedbythetubularstructure
ofthemotor,hencetheybelongonlytothiscategoryofmotor,whereasotheradvantages
arecommonlypresentintheotherlinearmachines.
A. Lineartubularactuator
Tubular linear actuator is obtained through the transformation depicted in Figure 3,
whichawardsspecialfeaturetothemotorifcomparedtotherotatingcounterpart.Thefirst
one concerns the motor windings, indeed they do not have the endwindings, thus all the
conductorsandthephasecurrentactivelyparticipatetothemagneticfieldproduction[6].
The endwinding Joule losses disappear, hence an increased coil efficiency and a better
powertoweight ratio compared to both the flat and the rotating topology results. The
17|P a g e
windingshapelookslikearingofconductors,soitiseasytoimplementanditissuitedtobe
industriallyproduced.Furthermore,thetubulargeometryariseswithinherentfeatureofself
neutralization of normal forces acting between the stator and the mover, since they are
symmetricallydistributedoverthemovercircumference.Thepistonlikeappearanceofthe
tubular linear machine gives the designer a firm foothold to connect the mechanical load,
andtheclosedframemakesitcapabletooperateinpollutedandhostileenvironments[29].
Ontheotherhand,thetubulartopologyimposeslimitsinthemaximummoverstroke,since
excessivemoverblendingmaycausevibrationsandmechanicaldamageofthemotor.
Among the various linear actuators, tubular topologies with permanent magnet
excitationareparticularlyattractive,sincetheydonotshowthetypicalassemblyproblems
oflinearopenmachines,havingindeedacompactstructureandahighforcedensity[30]
[31], [32]. Unfortunately, the permanent magnets interact with stator core determining a
disturbance force, which can cause negative effects such as vibrations, noises, positioning
errorsandnonuniformmovementatlowspeed.Suchaforceinlinearactuatorsisthesum
oftwocomponents:thecoggingforceandtheextremityforce[33].Theformerisgenerated
bythetendencyofthemagnetstoalignwiththestatoratpositionswherethepermeanceof
the magnetic circuit is maximized and has a spatial wavelength depending on the pole
number,theslotnumberandthepolepitch,whereasthelatterdependsontheinteractions
between the finite length of the armature core and the moving magnet, and has a spatial
wavelengths of one polepitch. The effects of disturbance force can be minimized at the
motordesignstage[34],[35],employingmethodssuchasskewingandoptimallydisposing
themagnets,andoptimizingthelengthofthearmaturecore[36].Furtherimprovementsto
getsmoothmotioncanbeobtainedbyemployingacurrentshapingcontrolstrategy[37].
The production of thrust force makes all linear drives suited to simplifying the
mechanicsinautomatedmachinethroughdirectdriveactuators,jointlytotheadoptionof
advancedcontrolmethodologies.
Toachieveprecisemotioncontrol,mostofthesemachinesuserotaryelectricalmotors
as their prime motion actuators, and couple their output shafts to mechanical motion
transmissions as reduction gear, belt, ball screw, pinionrack, etc. Though this is the most
widely used method, mechanical transmissions suffer of several drawbacks, which are
emphasizedthroughthehighdynamicperformancesofmodernPMdrivesandcontrols.
The use of mechanicaltransmissions increasethe whole dimensions of machinesand
their costs. Figure 5 shows some applications where the linear drives can replace the
rotatingcounterparts.
Themostevidentdisadvantagesofindirectdrivesare:
deadpointsinmechanicalchainthatcancauseproblemforpositioncontrol.
backlashes, which decrease the machine precision particularly during the
inversionofthemotion.
18|P a g e
stiffness
torsional elasticity and bending of the shafts, imputable of resonance which
represents in many cases an insuperable limit both for the accuracy of the
positioningcontrolandforthemaximumspeed.
additionalfrictionswhichcausepowerlosses,hencereducedefficiency.
additionalmechanicalpartsreducethesystemreliability.
All these aspects engrave negatively on the performance of the system in terms of
accuracy, resolution and repeatability. Applications with short stroke enlarge these limits,
since the reduction of the positioning times increase the accelerations, viz. the dynamic
aspects.
The directdrive philosophy, which provide thrust force directly to a payload, offer
numerous advantages over their rotarytolinear counterparts. All the previous drawbacks
areovercame,sincethemechanicaltransmissionisnoweliminated.Linearmotorsbecome
an integral part of the automatic machines, which are capable of higher dynamic
Figure5Exampleofsomeapplicationswherethelineardrivescanreplacetherotating
counterpart.
19|P a g e
performance, improved reliability and higher control bandwidth than the traditional
machines.
B. Lineartubularactuatorcompetitors
Pneumatic and hydraulic pistons have the same appearance of tubular electric
actuators,thustheyconstituteanalternativechoicetotheelectriccounterpart.
Industrial applications often require controlled motion and, in some cases, also
controlled force. Traditional hydraulic and pneumatic thrusters often use openloop force
control.Theforceiscreatedbyapplyingpressuretoaworkinggasorfluidinahydraulicor
pneumaticcylinder.Apistonthenconvertsthepressuretoaforceappliedtotheload.
Thissolutionshowsseveraldisadvantagesifcomparedwithtubularlinearelectricone,
firstofall,anelectrictubulardrivehaslargeraccuracy,readinessandcontrolreliability,itis
freemaintenanceandithaseasierenergysupplyavailability,andsecondlyanelectricdrive
doesnotpresentanyfluiddripping.
1.5 IroncorePMtubularlinearactuator
Among the various linear actuators, tubular topologies with permanent magnet
excitationareparticularlyattractive,sincetheydonotshowthetypicalassemblyproblems
oflinearopenmachines,havingindeedacompactstructureandahighforcedensity.Figure
6depictsasimplifiedthreedimensionalsightofaPMtubularmotorwithaxiallymagnetized
magnets.
Uptofewyearsagothetubularmotorwasnotusedinindustrybecauseofconstructive
andperformancesreasons.Ironlessactuatorsexhibitedlowspecificforcesincehighenergy
rareheartmagnetshadbeendeveloped,whileironcoremotorshadtherealizationproblem
ofamagneticcircuitsuitedforthefluxlinepath.
Usually,inthetraditionalrotatingmotorstheplanarlaminationofthemagneticcore
Figure6ThreedimensionalsightofaPMtubularmotorwithaxiallymagnetizedmagnets.
20|P a g e
interrupts the circulation of eddy current induced by timevariable magnetic fields. In a

tubular motor eddy currents flow around the actuator circumference, hence their
interruption should be realized with pieslice lamination [38], as shown in Figure 7. The
latter is practically unfeasible, thus other solution were adopted. Core lamination able to
shortenthepathsoftheinducedcurrentcanbelongitudinalortransversewithrespectof
the direction of motion [5], [6]. Both of them show drawbacks: the former exhibits the
practicaldifficultytopreciselyalignthelaminationsofeverycoreallaroundthemotionaxle,
as shown in Figure 8, while the latter determines an increase of the magnetic circuit
reluctanceduetothesubstantialtotalthicknessoftheinsulationsamongthelaminationson
the motor backiron which has to be crossed by the flux lines. These problems have been
overcamethankstothescientificandindustrialdevelopment,whichhasproducedtheSoft
MagneticComposites(SMC).Thisclassofmaterialsisbasicallythatofironpowderparticles
coatedwithanelectricallyinsulatedlayerasshownschematicallyinFigure9.Thesepowders
canbecompactedincomplexformand,afterathermaltreatment,arereadytouseshowing
isotropicmagneticproperties.
Finally, windings of this kind of motor have not frontal connections, resulting in a
compactstructureandagreatdensityofforce.
TheprincipalpartsthatformaPMtubularlinearactuatorwillbeshortlydescribedin
thefollowingsections.
A. Permanentmagnets
Permanentmagnettubularactuatorscanbeconstructedwithsurfacemountedradially
magnetized magnets, depicted in Figure 10, or with axially magnetized magnets, shown in
Figure 11. An interesting comparison between the two solutions was carried out in [39],
where it is concluded that the axially magnetized machine has a higher force density, but
morepermanentmagnetmaterialisrequired.Ifthesamevolumeofpermanentmagnetis
used, the two topologies lead to the same force density. However, axially magnetized
machineshouldbepreferredbecauseaxiallyanisotropicrareearthmagnetsareusuallyless
Figure7conceptualpieslicelaminationformagneticcoreoftubularactuator.
21|P a g e
Figure8Practicalarrangementofstatorcoreusinglongitudinallaminates
withrespectofthedirectionofmotion.
expensiveandwidelyavailable.
Magnet typology for this actuators have to be selected considering the motor
performancetargetfixedindesignstage.However,themostdiffusedmagnetsareinFerrite
or in NdFeB. The first ones represent the most economic choice, having low remanence,
lowenergyproductandastrongtemperaturederating.
The NdFeB magnets have at present the most elevated specific energy, similar
consideration concerns the remanence value and the intrinsic coercive field, even with
magnettemperatureof100150C.TypicalvalueofmainparameterforFerriteandNdFeB
magnetaresummarizedinTableI.
Alloftheseelementshavetobeconsideredduringthemotordesigntomakesurethat
the magnet doesn't suffer permanent damages that jeopardize the reliability and the
performancesofthemotor.
Figure9SchematicpictureofSMCmaterial.
22|P a g e
TableIComparisonamongmainparameterofferriteandNdFeBmagnets.
Parameter
Ferrite
NdFeB
Br[T]
0.20.5
1.11.3
Hc[kA/m]
150295
7001000
Relativerecoil
permeability
1.1
1.08
Temperature
coefficientof
Br[%C]
0.11/0.12
0.54/0.60
Temperature
coefficientof
Hc[%C]
0.2
+0.3
Figure10Axialsectionofatubularmovercomposed Figure11Axialsectionofatubularmovercomposed
ofradiallymagnetizedmagnets.
ofaxiallymagnetizedmagnets.
B. Magneticcircuit
TheuseofSMCmaterialallowstoovercometheproblemofapropermagneticcircuit
for tubular actuators, since flux paths have both radial and axial direction. Magnetic
23|P a g e
propertiesofsuchamaterialhavetobeanalyzedinordertofullyunderstanditspotentiality
[40].
Ingeneral,SMCshavelowerinitialpermeability,lowersaturationinductionandhigher
iron losses at 5060 Hz than the laminated steel. These limitations can be cancelled if the
isotropic 3D behaviour of SMC will be exploited [41], [42]. Further, the aforementioned
characteristics are strongly dependant on the kind of powder and on the transformation
process used to obtain the finished component. In particular, the electromagnetic and
mechanical properties will depend on the added lubricant/binder and on the process, for
example,coldorwarmcompaction.
A low amount of additive results in a powder mix that can be compacted to higher
densities and heat treated with higher temperature, which is beneficial for the magnetic
properties, as shown in Figure 12. During compaction stress is introduced in the particles,
whichdeterioratesthesoftmagneticproperties.Thehighertheheattreatmenttemperature
isset,thehigherthedegreeofstressrelief.However,itiscrucialthatnosinteringtakeplace
Figure12HeattreatmenteffectinhysteresisloopcharacteristicofSomaloy500.
Figure13EffectofmaterialdensityonBHcurvesofSomaloy500.
24|P a g e
betweentheparticles,asthedynamiclosseswouldincreaserapidlywithfrequency.
AspresentedinFigure12,themagneticinductiondependsheavilyonthecompaction
pressure of the material, because it determines the material density. A higher magnetic
inductionforacertainappliedfieldmeansimprovedperformanceofthemotor.Compared
tolaminations,SMChasnormallyaslightlylowerinductionlevel,thoughtthedifferenceis
small at higher fields. Hysteresis losses are worse if compared to that one of laminations,
insteaddynamicbehaviourisbetterduetoreducededdycurrentlosses.Thismeansthatat
acertaincriticalfrequency,typicallyaroundacoupleof100Hz,thelosseswillbelowerin
the SMCmaterial than in laminations. The maximum relative permeability is significantly
lowerinSMCscomparedtotheinplanepropertyofsteellaminations.
OneofthemainadvantagewithSMCisits3Dproperties;thematerialisisotropic,thus
magneticfluxcanflowequallywellinalldirections.Anotheradvantageistheopportunityto
obtain3Dshapesbyacompactionprocess.Thecombinationofthesetwofeaturesapplied
to particular machines, such as the PM tubular one, can compensate the lower magnetic
properties than laminations. Tubular PM machines have high natural reluctance due to
presenceofmagnets,thusthepermeabilityisoflessimportanceandgoodperformancecan
bestillachieved.
TheinfluenceofthechoiceofSMCsontheperformanceoftubularpermanentmagnet
machines was carried out in [43], where it is shown that SMC machines have a slightly
inferiorperformanceifcomparedwithlaminatedone,buttheyareeasiertomanufacture,
and,therefore,potentiallybelowercost.
C. Windings
Theprimarywindingoftubularmotorshassimplediskshape,likethatonedepictedin
Figure14.Thetubularmotor,foritsownnature,hasaverysimplestatoricstructure,thusits
constructionconsistsonlyinanalternativelystackofSMCdisksandwindings.
The winding can be preassembled and precompacted in order to obtain good slot
fillingcoefficient.
Thebenefitsarosebythisoperationaremanifold:theelectricexploitationoftheslotis
increased, and finally the thermal exchange of winding is improved, bringing to a more
uniformdistributionofthewindingtemperature.
Figure14Diskshapedstatorwindingsoftubularlinearactuator.
25|P a g e
1.6 Conclusion
An overall comparison among different topologies of linear machines has been well
developed, showing merits and demerits of each type of motor. Hence, it results that,
among the various linear actuators, tubular topologies with permanent magnets are
particularlyattractive.Infacttheydonotshowthetypicalassemblyproblemsoflinearopen
machines, then they have a compact structure, moreover soft and hard magneticmaterial
development bring them to higher force density. These machines have excellent
characteristicsasservomotor,whichmakethemsuitedtotheuseinautomatedindustrial
machines.
26|P a g e
CHAPTER
2
BasicsofMultiphaseDrives
2.1 Introduction
The conventional structure for variablespeed drives consists of a threephase motor
supplied by a threephase voltage source inverter (VSI). The converter can be viewed as a
decoupling interface between the threephase mains and the motor. Thus, the need for a
specific number of phases, such as three, disappears. Furthermore, the development of
power electronics makes it possible to consider the number of phases as an additional
designvariable.
Nowadays, there is an increasing attention towards multiphase drives. Indeed they
offer a greater number of degrees of freedom compared with threephase motor drives,
which can be utilized to improve the drive performance [44], [45]. In this chapter the
fundamentalaspectsofmultiphasedriveswillbepresented,focusingtheattentiontotheso
calledmultimotormultiphasedrive.
2.2 Multiplespacevectorrepresentation
As known, a multiphase motor drive cannot be analyzed using the space vector
representationinasingledqplane.Itisknownthat,tocompletelydescribeamultiphase
electromagnetic system, the space vector representation in multiple dq planes (multiple
space vectors) must be adopted [46]. Some modern approach to the modulation of
multiphase inverters have overcome the inherent difficulty of synthesizing more than one
independentspacevectorsimultaneouslyindifferentdqplanes,inordertofullyexploitthe
potential multiphase motor drives [47]. They have shown that the multiple space vector
notationiswellsuitedtorepresentandtoanalyzemultiphasesystems.
In a nphase system, for a given set of n real variables x1, x2, x3,...xn, a new set of
variablescanbeobtainedbymeansofthefollowingsymmetricallineartransformations[48]:
1 n
x h = x k h(k 1) ,(h=0,1,2,...,n1)
(2.1)
n k =1
where = exp( j 2 / n ) .
27|P a g e
Withtheexceptionfor h = 0 (and h = n / 2 forevennumberofphases),thequantity

x h is a complex number and it is called space vector component of sequence h, or space
vectoronthe(dq)hplane.
Itcanbeshownthatthespacevectorcomponentofsequence h isrelatedtothespace
vector component of sequence n h through a complex conjugate relationship,
furthermorethespacevectorcomponentofsequence n h coincidewiththespacevector
component of sequence h . These properties can be synthesized by the following
equation:
x h = x *h
(2.2)
Therealvariable x 0 = x 0 isobtainedfor h = 0 andiscalledzerosequencecomponent,

whereastheadditionalrealspacevectorcomponentofsequence n / 2 appearsforaneven
numberofphasesonly.
Using the aforementioned classifications and applying (2.2), the space vector
transformationscanberewrittenasfollows:
2 n
x h = x k h(k 1) (h=1,2,...,r)
(2.3)
n k =1
x0 =
1 n
xk
n k =1
(2.4)
(2.5)
xn / 2 =
1 n
x k (1) k 1
n k =1
where r = n / 2 1 for an even phase number and r = (n 1) / 2 for an odd phase

number.Thecorrespondinginversetransformationsare:
x k = x 0 + x n / 2 (1) k 1 + x h
h ( k 1)
(k=1,2,...,n)
(2.6)
h =1
wherethesymbolrepresentsthedotproduct.
Intheparticularcaseofbalancedandsinusoidaloperatingconditions(sequence1),the
spacevector x1 assumesaspecialrelevancebeingtheonlyspacevectordifferentfromzero.
However,itisopportunetoemphasizethat,inthegeneralcase,alltherspacevectors
aswellasthezerosequencecomponentarenecessarytocompletelydescribethenphase
system.
28|P a g e
2.3 MathematicalmodelofmultiphasePMsynchronous
machines
ThebehaviouroftherotatingnphasesynchronousmachinewithsurfacemountedPM
canbedescribedintermsofmultiplespacevectorsbythefollowingequationswritteninthe
statorreferenceframe:
d
v sh = Rs ish + sh
(2.7)
dt
sh = Lsh ish + M se,h ie e jh
(2.8)
(2.9)
n r
M se,h e[ish ]
2 h =1
(2.10)
ve e j = Re ie e j +
d
e e j
dt
e e j = Le ie e j +
Fromthemagneticcoenergy,thetorqueequationcanbeobtainedasfollow:
n r
T = p h M se,h i sh jie e jh
(2.11)
2 h =1
wherepisthepolepairsnumber,histhemultiplespacevectorindex,andistheelectrical
positionoftherotorreferenceframe.
Multiphase drives enable the independent control of a series of multiphase motors
supplied by one multiphase inverter, provided that the machines are designed to produce
sinusoidalfield distribution in their airgap. The latter feature simplifies the torque
expression(2.11),infactinthiscaseonlythefirstcurrentspacevectorinfluencesthetorque
production. Moreover, considering a PM brushless tubular linear actuator, the equations
(2.7)(2.10) remain unchanged, whereas equation (2.11) becomes the expression of the
thrustforce(2.12)undertheassumptionofnegligibleendeffects.
n
F=
M set ,1 is1 jie e j
(2.12)
2p
where p is the pole pitch, M set ,1ie is the voltage constant expressed in V /(m / s ) and the
relationbetweenelectricposition andlinearposition x is:
29|P a g e
x
p
(2.13)
2.4 Mainfeatureofmultiphasedrives
Multiphasevariablespeeddrivehasreceivedgrowinginterestsincethesecondhalfof
the1990s.Theareaofelectricshippropulsionandtheirdevelopmentsactedinthatperiod
as propeller for the research [49], [50]. Nevertheless, oldest records concerning this topic
datesbacktothesecondhalfofthe1960s.Historicaltechnicalreasonsthaturgedtoadopt
themultiphasedrivesolutioninsteadofthreephaseonearelistedbelow[51]:
1) Multiphase variable speed drive reduces the stator current per phase, for a
givenmotoroutputpower.
2) Theuseofmorethanthreephasesofferanimprovedreliabilityduetothefault
tolerancefeaturesofmultiphasedrives
3) Multiphase machines present reduced pulsating toques produced by time
harmoniccomponentsintheexcitationwaveform.
As stated at point n1, the use of multiphase drive instead of the threephase one
helped to overcome the problems related to highpower applications with current limited
devices.Foragivenmotorpower,anincreaseinphasenumberdeterminesareductionin
powerperphase,enablingtheuseofsmallerpowerelectronicdevicesineachinverterleg,
withoutincreasingthevoltageperphase[52].Thisisstillasolutionadoptedforhighpower
applications,suchastransportandshippropulsiondrives.Infact,largemultiphasemachines
forshippropulsionhavealreadybeenprototypedindustrially,andarecurrentlyundergoing
commercialevaluation.
Theimprovedreliabilityfeaturesofmultiphasedrives,listedatpointn2,enabletheir
use also in faulty conditions, in fact if one phase of a multiphase machine become open
circuit,themachineisabletoselfstartandtorunwithonlyaderating,whichdependson
postfaultcontrolstrategyandinthenumberofthephases.Inthethreephasecase,theloss
of one phase determine an important derating of the machine while it is running,
furthermore the machine is not selfstarting and, for this purpose, it requires an external
mean.Furtherdetailswillbegiveninparagraph2.5and2.6.
Finally,theadvantagesderivedfromstatementatpointn3wasveryimportantinthe
60s,whenthreephaseinverterfedacdrivesoperatewithsixstepmode.Timeharmonicof
voltages and currents introduced by this operationmode produced low frequency torque
ripple, leading to difficulties on speed control and noise production. Since in a nphase
machine torque pulsations are caused by supply timeharmonics of the order 2n1, which
resultintorquerippleharmonic2ntimeshigherthanthesupplyfrequency,anincreasein
30|P a g e
the number of phases seem the best solution to the problem. This aspect of multiphase
drives has lost importance since the discover of PWM of VSI, which allows the control of
inverteroutputvoltageharmoniccontent.
Another feature of multiphase machines concerns their efficiency [53], indeed stator
windings create a magnetic field in the machine airgap with a lower spaceharmonic
content. It results, for two identical machines with the same airgap field and the same
fundamental component of stator current, a reduction of stator Joule losses, hence an
increased efficiency. The stator Joule losses saving obtained by increasing the number of
phasesfromthreeisrelativelylow.Unlikely,itapproachesanasymptote,thustheuseofa
fifteenphase machine instead of twelvephase one produces a stator Joule losses save of
only0.2%[1].
Someoftheaforementionedtechnicalcharacteristicsarenowadaysstillasrelevantas
theywereintheearlydays,suchasthatoneatpointn1andatpointn2oftheprevious
list.Overtheyears,multiphasedrivecharacteristicshavebeenexploredmoreindetailsand
otherbeneficialfeatureshavebeenrecognized,suchasthepossibilitytoenhancethedrive
torque capability [54], [55] and the opportunity to realize a multiphase multimotor drive
systemwithsingleinvertersupply[56].
The two latter aspects, together with the fault tolerant feature, appear as natural
solutions for some industrial problems, thus they will be described more in details in the
followingparagraphs.
Allthefeaturesabovepresentedaresuitedbothformultiphaserotarymachineandfor
multiphaselinearones.
2.5 Multiphasedrivesfortorquecapabilityenhancement
A possible way to use the large number of degrees of freedom of multiphase drives
consistsintheutilizationofthespatialthirdharmoniccomponentofthefluxdensityinthe
air gap for increasing the torque production capability. This feature could be particularly
interestinginapplicationswherevolumetricorweightconstraintsareimportant,indeedfor
a given torque value, the volume and the weigh of the multiphase machine controlled as
statedabovearesmallerthantheonesofthethreephasemotor.
The relation between time and spaceharmonic fields produced by a multiphase
windingpresentedin[53]explainsclearlytheincreasingtorquecapabilityfeature.Underthe
hypothesis of stator current with fundamental component of frequency f together with
timeharmonic components at integer multiples of f, the excited 2 P pole nphase stator
winding can be modelled by blocks of current and then resolved into rotating harmonic
surfacecurrentdistributions,thusitresultsoftheform:
J S ( , t ) = Re 2 J Sk e j ( k t P )
k
(2.14)
31|P a g e
where:
= 2 f
(2.15)
In(2.14),kisapositiveintegerwhichrangesoveralltimeharmonicsproducedbythe
inverter, and is the azimuthal coordinate. The rms phasor of stator surface current
densityisrepresentedby J Sk ,ithas 2 P polesandrotatesataspeedof k = P radians
persecondwithrespecttothestator.
Theexpressionfor J Sk is:
J Sk =
n Ik Z
K d K p
D
(2.16)
where Z is the number of seriesconnected conductors per phase, I k is the kth time
harmonic component of phase current, and D is the mean airgap diameter. Pitch and
distributionfactorforthe thharmonicarerespectively K d and K p .
Itisshownin[53]that J Sk isnonzeroonlyforvaluesof thatarerelatedtokandto
thephasenumbernbytheexpression:
i = 0, 1, 2, 3,...
= k 2ni
(2.17)
Equation (2.17) contains the basic information about the time and space harmonic
fieldsproducedbyannphaseexcitation.Inafivephasemachine(n=5),thefundamental
frequency excitation ( k = 1 ) produces spaceharmonic fields of order = 1, 9, +11, ...
whereasthethirdtimeharmonic( k = 3 )producesspaceharmonicfieldsoforder = 3,7,
+13, ecc. It is shown that the fundamental timeharmonic in excitation controls the
fundamental spaceharmonic in the airgap of the machine, moreover the third time
harmonicinexcitationcontrolsthethirdspaceharmonic.
The independent control of the first and third spatial harmonic component of the
magneticfieldintheairgapofamachinewithconcentratedwindings(oneslotperpoleper
phase,i.e.q=1)canbeusedtosynchronizethethirdspatialharmonicwiththefirstonein
order to obtain an airgap flux density nearly trapezoidal, leading to backEMF voltages
havingnonsinusoidalwaveforms,sothatthetorqueperamperecanbeincreasedby10%
[53],[55],[57].Ahighperformanceinductionmotordrivewithhightorquedensitycanbe
obtained using an Extended Field Oriented Control [58], [59] or an Extended Direct
TorqueControl[60].
32|P a g e
Furthermore, multiphase PM synchronous motors with surface mounted magnets in

therotorhaveanearlyrectangularairgapfluxdensitydistribution,whereasthearmatureis
realized with fractional slot windings in order to minimize the cogging torque effect.
Nevertheless, the use of concentrated winding with a fullpitch is able to maximize the
torque per ampere. These motors have the advantages of both BLDC (rectangularfed
brushless dc motors) motors and PMSMs (sinusoidal fed PM synchronous motors). This
means that it has the controllability of a PMSM while having the high torque density of a
BLDCmotor[61].
2.6 Faulttolerantdrives
The increased number of phases in multiphase drives offers considerable benefits
becauseofthecapabilitytocontinueoperationwhenasingleormultiplephaselossoccurs
[62].
Threephasedriveissensitivetodifferentkindsoffaults,bothinmotorphaseandin
inverterleg.Whenoneofthesefaultsdoesoccurinonephase,thedriveoperationhastobe
stopped for a nonprogrammed maintenance schedule. The motor in faulty conditions is
abletorunbutitisnotstillselfstarting.Thecostofthisschedulecanbehigh,thusjustifying
thedevelopmentoffaulttolerantmotordrivesystems.
On the contrary, multiphase machines in postfault condition can continue to be
operated with an asymmetrical winding structure and unbalanced excitation, producing a
higher fraction of their rated torques with little pulsations when compared to the three
phasemachines[63],[64],[65].
Such a feature is very useful in safety critical applications, where the whole system
reliability has to be high, even when a fault occurs. Applications which use the by wire
technology, naval propulsion systems, moreelectric aircraft and electric and hybrid
vehiclesaretypicalexamples.
Fault tolerant feature can be easily obtained in multiphase machine designed as n
phasemachinecomposedofbgroupsorwindingsofaphasesystems[66],[67],[68][69].
The whole motor has n=ab phases, but exists b isolated neutral points and each aphase
system is fed by b aphase drives. The postfault strategy for this kind of motor is very
simple, the aphase winding in which the fault has occurred is disconnected, whereas the
remaining aphase windings can continue to operate without any control algorithm
modification.Thetorqueandpowerderatingcoefficientforeveryaphasesystemdetached
is(b1)/b.
A multiphase machine with single neutral point allows to fully exploit the degrees of
freedom of multiphase drives, thus it offers better characteristics in postfault operations
thanthemultineutralpointsolution[70].Theexclusionofonlythefaultyphasewithinthe
possibilitytocontinuetooperatethehealthyphases,leavingoutproblemsrelatedtofault
isolation,permitstominimizethetorqueandpowerderating.Indeed,foragivennphase
33|P a g e
motor, the single neutral point solution permits the exclusion of the only faulty phase,
whereas in multineutral point solution is necessary to take out of service the aphase
systemwhichcontainsthefaultyphase,thusaphasesaredetached.
The control strategy for postfault operation in a single neutral point machine has to
determine the relations between the currents in the remaining healthy phases in order to
pursue a specific aim. The effect of the strategies used, for a given multiphase drive,
depends on the load operating point and on the load torque characteristic. Keeping
magnitudeand phase of the currents in the remaining unfaulted phases at their prefault
values, it results in a stator joule losses reduction by a factor (n1)/n and a consequential
reduction in developed torque, but the slip increase along with the rotor loss. This easy
control strategy, does not exploit the whole number of degrees of freedom of the
multiphasedrive.
Analternativesolutioncanallowanincreaseinthemagnitudeofthecurrentineach
unfaultedphase.Hence,thewholestatorjoulelossremainsatitsprefaultvalueandlow
torquepulsationsresult.Thissolutiongivesbothasmallerdropinspeedandasmallerrotor
joulelossesthanthepreviousstrategy.
A third strategy could pursue the aim to maintain the torque and the power at their
prefaultvalue.Anincreaseofthemagnitudeofthecurrentineachunfaultedphase,higher
than the previous strategy, is necessary; consequently it determines also an increase in
statorjoulelosses.Thusthissolutionisnotsuitedforoperationssustainedforalongperiod
oftime,sinceoverheatingproblemsreveal.
The practical implementation of any of these strategies might prove other problems,
suchastotheratingofthepowerelectronicswitchesandthelimitationimposedbytheDC
linkvoltage.
2.7 MultiMotordrives
Anotherpossibilityofferedbymultiphasedrivesisrelatedtothesocalledmultimotor
drives. A welldefined number of multiphase machines, having series connected stator
windings with an opportune permutation of the phases, as shown in Figure 1, can be
independentlycontrolledwithasinglemultiphaseinverter[71][72].
Avectorcontrolalgorithmcanbeappliedtoeachmachineseparatelytogeneratethe
inverterlegcontrolsignals,andtosupplythestatorwindingsofthemultimachinesystem
fromasinglecurrentcontrolledbyavoltagesourceinverter(VSI).Invertercurrentcontrolis
performedinthestationaryreferenceframe,usinginverterphasecurrents.Thenumberof
multiphasemotors connected in series, or rather the number of driving axles, depends on
the number of the phases of the multiphase inverter. The maximum number of motor
connectableinseries,forasystemwithnphases,is M = (n 2) / 2 ,incaseofnis aeven
number,whileitis M = (n 1) / 2 ifnisanoddnumber.
34|P a g e
Figure1Conceptualwindingconnectionsof ( n 1) / 2 machinesinmultimotordrivessuppliedbyanphase
inverter.
The necessity to control several electrical machines with a coordinated motion is an

usual requirement in automatic industrial applications. Some examples are the XY placer,
highspeed packaging machine, pickandplace machine used for electronic boards and
siliconmonocrystalproductionequipments.
Thetraditionalsolutionsusethreephasedrives,orratheronethreephasemotorfed
by one inverter for each driven axle shown in Figure 2. The control architecture of such a
system can be centralized or decentralized. In the first case the control of each motor is
elaborated by a central control unit or by one control unit belonging to a more complex
hierarchicalcontrolsystem,andeachinverterbehavesinmanycasesasapoweramplifierof
thecontrolsignalselaboratedbythecentralizedcontrol.Inthesecondcasethecontrolof
Figure2Multimotorsconfigurationwiththreephasedrives.
35|P a g e
each motor is directly implemented on its own inverter, and the motion synchronization
amongthewholedrivingaxlesismanagedbyonemasterdrivethroughanelectroniccam.
Taking into account the power connection point of view, each inverter can be fed by
the threephase grid or can be connected to a DC link as shown schematically in Figure 3,
enablingacompleteenergyrecoverysystem.
Multiphasemultimotorsdrivesprovidedifferentsolutionsratherthanthethreephase
one, both for power feature and for control structure. The former results in a hardware
saving, indeed taking as example a twomotor drives, it results that the minimum inverter
leg number required for multiphase solution is five, whereas the classical threephase
solutionsusessixinverterlegs.Thehardwaresavingconcerns,besidesdiscretecomponents
oftheinverterleg,alltheauxiliarycircuitandcomponentsusedtocontrolthecomponents
of the leg, such as drivers, protection circuits, electronic power supplies etc. The saving
existsforaphasenumbergreaterthanfive,andingeneral,higheristhenumberofphases,
greateristhenumberofsavedlegsis.
Another excellent feature of the multiphase multimotor drives consists in a control
components saving, in fact the implementation of the vector control algorithm for all the
motorsconnectedinseriescanbemanagedwithinasingleDSP.Thecontrolofeachmotor
can be executed in parallel to give the proper control references that, after appropriate
summation, results in the phase current control references. If the threephase solution is
considered,oneDSPwillbeusedforeachdrivebecauseeachinverteriscontrolledwitha
DSP.
Theonlydrawbackofmultiphasemultimotordrivesisanincreaseofthewholestator
windinglossesduetotheflowofthephasecurrentsthroughthestatorwindingsofallseries
connectedmotors.Consideringthesituationwhereonlyonemotorisrunning,itresultsin
Figure3Multimotorsconfigurationwiththreephasedrivesconnectedwith
commonDCbus.
36|P a g e
statorJuolelossesinthemotoritselfandintheothermotorsconnectedinthemultimotor
drive. The efficiency ofthe multimotor drive decreases when compared to theequivalent
threephase case, but the advantages above discussed seem to overcome the efficiency
drawback.
A. MultiMotordriveconcept
Vectorcontrolofmultiphasemachinerequiresatminimumonlytwocurrent,i.e.only
one space vector current, if only the fundamental field is utilized, as shown by equation
(2.12). The additional degrees of freedom offered by multiphase drives can be utilized to
controlindependentlyothermachineswithinamultiphasemultimotordrivesystem.There
arenotanyconstraintsconcerningwiththemotortypologiesconnectableinseries,thenthe
multimotor drive can be composed of induction and/or permanent magnet synchronous
and/orsynchronousreluctancemotors.
If a machine is designed to produce sinusoidalfield distribution [51], then standard
modellingassumptionappliesandonlythefirstharmonicofinductancetermsexistsinthe
phasevariable model. The applications of Clarkes transformation, given in the expression
(2.18) for an odd phase number, to the phase voltage equations produces a set of n
equations in the stationary frame reference. The first pair of equations is identical to the
correspondingoneforathreephasemachine,andtheyarethetwocomponentsofthefirst
spacevector.Thelastrowin(2.18),orthelasttworowsforevenphasenumbers,represents
the zerosequence equation. The other couple of rows in expression (2.18), which are
(n3)/2foroddphasenumberand(n4)/2forevenphasenumber,correspondtothehth
spacevectors,wherethespacevectorindexhhasthesamemeaningexpressedin(2.3).
Asdiscussedabove,thefirstspacevectorofthecurrentistheonlyoneabletoproduce
torque/forceinthemotor.Theothercurrentspacevectorsdonotaffectthetorque/force
andtheirvaluesareinfluencedbytheirrelativevoltagespacevectorsandthestatorwinding
leakageimpedances.Thesecurrentspacevectorsrepresentdegreesoffreedomthatcanbe
usedtocontrolothermachinesconnectedinserieswiththefirstmachine.Ifthecontrolof
themachineshastobedecoupledonefromtheother,torque/forceproducingcurrentspace
vector of one machine must not produce torque/force in all the other machines in the
group.Thiscanbeachievedwithaproperseriesconnectionofthemotors,abletotransform
each current space vector into the first current space vector for its relative motor of the
group.Inotherwords,thestatorwindingconnectionsofmultiphasemachinesmustbesuch
thatwhatonemachineseesasthefirstcurrentspacevector,theothermachinesseeashth
currentspacevector,andviceversa.
In order to do so, it is necessary to connect stator windings of all the multiphase
machines in series with an appropriate phase transposition, and finally close all phases of
thelastmotorinstarconnection,asshowninFigure1.Thenecessaryphasetranspositionis
takendirectlyfromtheClarke'stransformationmatrix(2.18),readingitpercolumnsinorder
37|P a g e
to understand the spatial angle among the rows, which represents the phase step. Thus,
phases1ofallthemachineswillbeconnectedinserieswithouttransposition,indeedthe
phasestepamongtherowsincolumn1iszero.Phase2ofthefirstmachine,viz.column2,
hasaphasestepequalto;thismeansthatitwillbeconnectedtophase3ofthesecond
machine,whichwillbefurtherconnectedtophase4ofthethirdmachineandsoon.Inthe
same way, phase 3 of the first machine is connected to the phase 5 of the second
machine,whichfurthergetsconnectedtophase7ofthethirdmachine,andsoon,because
of the phase step is 2. This explanation provides all the information to construct the
connectiontableorconnectivitymatrixproposedin[56],[75],andreportedhereinTableI
inthegeneralcaseofanphasesystem.Thefirstcolumnin
TableIshowsthemachinenumberconnectedinseries,thefirstrowshowsthephase
number.
Then it becomes possible to completely independently control the machines while
supplyingthedrivesystemfromasinglecurrentcontrolledvoltagesourceinverter.
Afterthedescriptionofthemaincharacteristicsofmultiphasedrives,itisworthnoting
that the multiple space vectors assume different meaning case by case. As an example, in
fivephasemachineswithconcentratedwindings,thefirstspacevectorofthestatorcurrents
isresponsibleofthefirstspatialharmoniccomponentofthemagneticfieldintheairgap,
whereasthethirdspacevectorisrelatedtothethirdharmoniccomponent[57],[58],[74].
On the other hand, in multiphase multimotor drives, the first space vector of the output
currents is responsible for the first spatial harmonic component of the first machine,
whereas the third space vector generates the first spatial harmonic component of the
secondmachine[71].
C=
1
0
cos( )
sin ( )
cos(2 )
sin (2 )
cos(3 )
sin (3 )
...
...
1
0
cos(2 )
sin (2 )
cos(4 )
sin (4 )
cos(6 )
sin (6 )
...
...
1
0
cos(3 )
sin (3 )
cos(6 )
sin (6 )
cos(9 )
sin (9 )
...
...
...
1
0
1
2
...
...
...
...
(n 1)
(n 1)
(n 1)
cos
2 cos
3 ...
cos
2
(n 1)
(n 1)
(n 1)
sin
2 sin
3 ...
sin
2
1
1
1
...
2
2
2
cos(2 )
sin (2 )
cos( )
sin ( )
cos(4 )
cos(2 )
sin (4 )
sin (2 )
cos(6 )
cos(3 )
sin (6 )
sin (3 )
...
...
(n 1)
(n 1)
cos
2 cos
(n 1)
(n 1)
2 sin

sin
2
1
1
2
2
38|P a g e
(2.18)
TableIConnectivityMatrixforgeneralnphasecase.
M1
...
M2
b+1
c+2
d+3
e+4
f+5
g+6
h+7
i+8
l+9
...
M3
b+2
c+4
d+6
e+8
f+10
g+12
h+14
i+16
l+18
...
M4
b+3
c+6
d+9
e+12
f+15
g+18
h+21
i+24
l+27
...
...
...
...
...
...
...
...
...
...
...
...
...
B. Machineconnectivity
The phase number n influences the number of connectable machines and their
individualphasenumbers.Thisissueiswelldescribedin[75],wherethemultimotordriveis
dividedinthreecategoriesdetailedasfollows.
1) Thenumberofphasesnisaprimenumber,thenthenumberofmachinesconnected
inserieswithphasetranspositionis:
(n 1)
M=
(2.19)
It corresponds to the number of independent space vectors obtained with

transformation(2.3).Everymachineofthegrouphasanumberofphasesequalton.
Exampleofmultiphasesystemsthatbelongtothiscategoryaren=3,5,7,11,13,17,
19,etc.
2) The number of phases does not belong to the previous category, but it can be
calculatedas:
(2.20)
n = l m (m=2,3,4,...)
The number of machines in series is still calculated with (2.19), but not all M
machines have n phases. For example, with a ninephase system, the multimotor
driveiscomposedofM=4machines:oneofthemisathreephasemachine,whereas
39|P a g e
the remaining three machines have nine phases. The connection diagram for this
case is shown in Figure 4. For the general case of m>1, the machines of the series
haveanumberofphaseswhichbelongtothefollowingsuccession(2.21):
n n
n
n, , 2 , .... , m 1
(2.21)
l l
l
Multiphasesystemsbelongingtothiscategoryhaveanumberofphasesn=9,25,27,
49,81,etc.
3) The number of phasesdoes not belong to the category listed at pointn1 and n2.
However, n is divisible by two or more prime numbers, denoted as n1, n2, n3, etc.
Thenn=n1n2n3nj.Thus,themaximumnumberofmachinesconnectableinseries
iscalculatedby(2.22):
(n 1)
M<
(2.22)
Orderingrulesrequirethatallthenphasemachinesareatfirstconnectedinseriesto
the source, with phase transposition. Then all the other machines follow, ordered
withadecreasingphasenumber.Thus,ifthesequenceofthelargestprimenumber
whichdividesnisn1,n2,n3,...,thenthesequenceofthemotorsconnectedinseriesis
ordered as the prime number sequence. The observation of this rule permits the
seriesconnectionofthewholegroupofmotors,otherwiseitshouldbeimpossibleto
Figure4Connectiondiagramwithphasetranspositionfortheninephasecase.
40|P a g e
connectamotorwithhighernumberofphasesafteralowerphasenumbermachine.
Thus the latest nphase machine has to be connected with the first n1phase
machine, the latest n1phase machine has to be connected with the first n2phase
machine,andsoon.Theconnectionofmachineswithdifferentprimenumbers,such
asn1andn2,seemstobeimpossible,sincetheration1/n2isnotaintegernumber.An
attempt to connect these kind of machines leads to the shortcircuiting of some
terminals. The group of machineswill becomposed of motors with different phase
number,ingeneralthemachinephasesareequaltonandoneoftheprimenumber
n1,n2,n3,etc.
Thatis,thenumberofphasesofmachinesaren,n1orn,n2orn,n3orn,...orn,nj.
This class of odd phase numbers encloses the situation where, besides the same
primenumber,thereisatleastoneotherprimenumberinthesequence,suchasn=
n1n2n3njlm.Thephasenumbersbelongingtothiscategoryaren=15,21,33,35,
39,45,51,55,57,63,65,69,75,77,etc.
Categoriesatpointn2andn3ofthelistcanbeconsideredasubcategoriesofamore
general case, where n is not a prime number. The common feature among these two
categories is that the series connection is composed with motors with two or more phase
numbers.
C. Control
The torque/force produced by a nphase PM synchronous machine is directly
controllablethroughthecurrentspacevectors,asshownin(2.11).Amachinedesignedfor
multimotordrivehastoproducesinusoidalfielddistributionintheairgapinordertomake
the machine controllable though the only first current space vector, whereas the other
current space vectors must not influence torque/force, viz. the motion control. Current
controlVSIisthusrequiredtosupplythemultimotorsystemwithdecoupledcontrolofeach
motor. Any of the available vector control algorithm is applied in conjunction with each
machineinordertogenerateindividualmachinephasecurrentreferences,whicharefurther
summed in an appropriate way to obtain the inverter phase current references. Vector
control of multimotor drives can be achieved by using either current control in the
stationaryreferenceframeorintherotatingreferenceframe[73].Thelattershowsin[76]
[77]anincreasedparametersensitivity.Thisheppenssincethedecouplingvoltagesbecome
functionsofparametersofbothmachines,duetotheneedforcompensatingvoltagedrops
in one machine caused by the flow of the torque/force producing currents of the other
machinethroughoutitswindings.Thus,controlinthestationaryreferenceframeisbetter
suitedseriesconnectedmultiphasemultimotordrivesystems.
41|P a g e
2.8 Conclusion
Multiphasedriveshavelargenumberofdegreesoffreedomifcomparedtotheclassic
threephase system. Though the multiphase drive concept is not new, it has attracted the
attention of the research community since the 90s. Thus, new features have been
developed, such as the drive power capability enhancement, faulttolerant characteristics
and, last but not least, the multimotor drive configuration. The latter represents an
interestingandpromisingsolutionforautomaticmachinewherecoordinatedmotioncontrol
amongseveralmotorsisrequired.Thenitwillbeinvestigatedinthefollowingchapters.
42|P a g e
CHAPTER
3
PMTubularMotormodellingand
optimizeddesignalgorithm
3.1 Introduction
AnoptimaldesignofPMtubularlinearactuatorspermitstofullyexploitthelimitsof
softandhardmagneticmaterial,resultinginhighperformancemotors.Thisobjectivecanbe
reachedpayingattentionnotonlytotheelectromagneticissues,butalsotothethermaland
mechanicalconstraints.
Anintegratedelectromagnetic,thermalandmechanicaldesignisdevelopedinorderto
maximize the thrust force or the dynamic performance of a series of PM tubular linear
actuators. Furthermore, the dynamic behaviour of the motor is very important due to the
inherentreciprocatingmotionofthemover,thusanoptimizednumberofwiresperslotis
determined through numerical simulations which take into account also the inverter
constraints.
3.2 Reviewofdesignoptimizationmethodologies
A variety of techniques has been employed to facilitate the design optimization,
prediction of magnetic field distributions through analytical field computation was carried
out in [78], [79], [4], whereas the most common approach employs a lumped equivalent
circuit[80],[81],[82],[83],finallyFiniteElementMethod(FEM)representsanotheranalysis
tool. The first one gives good calculation results of field distribution and thrust force,
considering also cogging force effect, but it does not allow to manage the relationship
betweencriticaldesignparametersandmachineperformance,ashappenswiththesecond
methodology. Finally, FEM analysis provide an accurate mean to determine the field
distribution,thattakesintoaccountthesaturation,butitremaintimeconsuming.
Itdoesnotexistabettersolution,becauseeachtechniquehavemeritsanddemerits,
dependingalsoonthemotordesignconstraintsandinthemotortopology.
43|P a g e
PMTubularMotormodellingandoptimizeddesignalgorithm
3.3 Electromagneticmodel
ThestructureofthePMtubularactuatorisshowninFigure1.Thesliderisconstituted
by a tube containing axially magnetized NdFeB magnets alternated with ferromagnetic
disks. The stator is composed of a magnetic yoke, and several inner disks. The stator
windings,placedbetweenthedisks,havetheformofcylindricalcoils.
A. Preliminaryconsiderations
In order to define an optimized algorithm, the main technical issues of tubular linear
actuatorshavetobeconsidered.
Magnetic circuit lamination cannot be adopted because of the 3D flux paths, but the
use of Soft Magnetic Composites (SMCs) can solve the problem to mould the yoke, the
statorinnerringsandthesliderpoles.AlthoughSMCshavelowerstrength,higherlossesat
lowfrequenciesandlowermaximumpermeabilitythanlaminations,insulatedironpowders
present isotropic magnetic properties combined with the possibility to mould complex 3D
magneticcircuits.
A preliminary machine design can be carried out on the basis of a simplified one
dimensionalfieldanalysisinordertodeterminethemainactuatordimensions,oncesome
keyparametersareprefixed.Thisallows,forgivenvolumespecifications,thedetermination
of the main actuator dimensions and performance, considering thermal and mechanical
issues.
B. Thrustforcecontributions
IntheactuatortopologyshowninFigure1,theconstantcomponentofthethrustforce
is produced by the permanent magnet and winding current interaction. There is also an
additionalconstantcomponentofforce,thereluctanceforce,duetothevaryingreluctance
ofthemagneticcircuit.Inthistypeofactuators,therearefoursourcesoftheforceripple:
1) thecoggingeffect,i.e.thetendencyofpermanentmagnetstoalignwiththestator
teethatpositionswherethereluctanceofthemagneticcircuitisminimized,
2) theendeffects,duetotheopenmagneticcircuitnature,
3) thepresenceofharmonicsintheairgapfluxdensitydistributionofthepermanent
magnets,
4) thepresenceofhighorderharmonicsinthereluctanceofthemagneticcircuits.
44|P a g e
Figure1Basicschemeofatubularpermanentmagnetmotorwithaxialmagnets.
Thefirstcontributionproducesthecoggingforce,thesecondcontributionthesocalled
endeffect force, and the third contribution is responsible of the field harmonic
electromagneticforce.ThecausesofforcepulsationscomingfromthePWMsupplyarenot
consideredinthispaper.
Unfortunately,thetraditionalmethodsusedinrotaryPMbrushlessmachinestoreduce
the cogging torque, i.e. use of skewed or asymmetrical distributions of the permanent
magnets[34],arenoteasilyapplicableintubularmachines.Asaconsequence,inthesetypes
ofmachines,thecoggingforcecanbereducedactingonlyonthegeometriesofstatorslots,
magnetsandsliderpoles.
Theendeffectforce,instead,canbecompensatedaddingferromagneticdisksinone
or in both extremity of the stator assembly: this aspect will be discussed paragraphs 3.6,
sectionD.
C. Fundamentalhypothesis
Some hypothesis has used to simplify the definition of the model equation for the
consideredactuatortopology.Inparticular,ithasbeenassumed:
onedimensionalfieldanalysis;
infiniteironpermeability;
cylindricalsymmetry;
negligibleendeffects;
negligibleironsaturation.
Despitesomeoftheaforelistedassumptionsseemtobeveryrestrictive,theyresults
fromapproximationssuitableforthekindofmotoranalyzed.Thestatementinthefirstpoint
simplifying the problem modelling, whereas the simplification stated in the second point
consider that PM tubular machines have low natural permeance due to presence of
magnets, thus the iron reluctance is less important and can be neglected. The cylindrical
symmetrylistedinthethirdpointisainherentpropertiesofthemotor.Theendeffectcan
be neglected as stated in the fourth point if proper strategies are used. Finally, iron
saturationdoesnotoccurwhenthemachineisdesignedsuchthatthemagneticloadisin
thelinearpartofthefirstmagnetizationcurveofiron.
45|P a g e
D. Designofthemagnets
The main geometric parameters used to define the electromagnetic model of the
motoraredepictedinFigure2.
The magnetization curve of the NdFeB magnet can be expressed by the following
relationship
Bm = r H m + Br
(3.1)
where Bm is the magnet flux density in axial direction, Hm the magnetizing force, Br the
magnetremanenceandrthemagnetpermeability.
Themagnetdiametercanberelatedtothepolepitchbyequatingtheslidercorefluxto
theairgapfluxasfollows
2 ( Bm
2
(d si2 d sib
)) = B g d g t p
(3.2)
wheretpisthepolepitch,Bgistheairgapfluxdensitycalculatedinthemeandiameterdg,
thelatterdefinedas
d g = d si + 2ht + g
(3.3)
The magnet width can be expressed in terms of the airgap flux density and magnet
magnetizingforceasfollows:
tp
wp
wm
wep
hb
dsib
dse
hep
ht
hs
wt
dex
ws
Figure2Drawingofthetubularactuatorshowingthegeometricalparameters.
46|P a g e
wm =
2 Bg
0 KC
H m 0
where0isthemagneticairgapgivenby
0 = ht + g
(3.4)
(3.5)
and K C istheCarterscoefficientthattakesthestatorandslidersloteffectintoaccount.It
canbeexpressedby
tS
tP
.
KC =
2
2
(
(
(t s wep ) / 0 ) 0
wM / 0 ) 0
t
tS
(5 + ((ts wep ) / 0 )) P (5 + (wM / 0 ))
(3.6)
(3.7)
E. Sliderandstatordesign
TheslidercorefluxdensityBpcanberelatedtothemagnetfluxdensityby
2
B p d si w P = 2 ( B m 4 (d si2 d sib
)) .
Thetoothfluxisequaltothepolefluxdividedbythenumberofteethinapolepitch,
andcanbeexpressedas
t = B g d g t p /N sp
(3.8)
where
N sp = N S / N P .
(3.9)
Theradialcrosssectionalareaofapolarshoefacingtheairgapandthatofthetooth
canbecalculatedrespectivelyby
Aep = (d g + g ) wep
(3.10)
and
47|P a g e
At = (d g + g + 2 hep ) wt .
Thetoothfluxdensityis
Bg d g t p
Bt = t =
At wt (d g + g + 2 hep ) N sp
whereasthestatorpoleshoefluxdensityresults
Bg d g t p
Bep = t =
.
Aep wep (d g + g ) N sp
Theslotwidthcanbecalculatedasfollows:
w s = wact / N S wt
(3.11)
(3.12)
(3.13)
(3.14)
where wact isthetotalactuatorlength.

Takingintoaccountthepolefluxtobackironfluxratio,assumingthetoothfluxdensity
equaltotheslidercorefluxdensity,thecrosssectionalareaofthebackironisgivenby
Ab = At N sp / 2
(3.15)
andthebackironheightbecomesasfollows:
Ab
hb =
.
(d se + 2g + 2h s + hb )
Finally,theslotheightcanbeexpressedby
hs = 12 (d ex d se 2g 2hb )
(3.16)
(3.17)
(3.18)
F. Thrustforcecalculation
Thethrustforcecanbecalculatedasfollows:
1 3 d
F = k ik
v s k =1 dt
48|P a g e
where kisthemagnetfluxlinkedwiththekthphase, ikisthecorrespondinglinecurrent

andvsthesliderspeed.
Assumingsinusoidalwindingcurrents,athrustforceisconstantwhenalsotheinduced
EMFs are sinusoidal. Applying Steinmetzs transformation to (3.18) leads to the following
expressionofthethrustforce:
1 3
F = j k I k
(3.19)
v s k =1
Thesliderspeedin(3.20)isrelatedtotheangularfrequencyofthesupplyvoltageas
follows:
tp
(3.20)
v s = .
Themaximumvalueofthecoilfluxlinkagecanbecalculatedas
Bg d g t p
c =
2 N sp
andthemaximumvalueofthephasefluxlinkageis
M = Ng nQ Kd c
(3.21)
(3.22)
where N g is the number of seriesconnected groups of coils, Q is the number of coils per
group, n isthenumberofwirespercoil,and K d isthewindingfactor.
ThecurrentdensitycanbeexpressedintermofJoulelossesandgeometricparameters
by
J=
Pcu
N s (d si + 2 + hs ) w s h s K s
(3.23)
where Pcuisthewindinglosses, K s istheslotfillfactorand isthecopperresistivity.Itis

worthnotingthatthevalueofPcuin(3.23)shouldbechosentakingintoaccountthecooling
capabilityofthetubularactuator.
Thepeakvalueofthetotalslotcurrentvalueis
I c = 2 w s hs K s J .
(3.24)
49|P a g e
Finally, assuming that the machine operates in field oriented control conditions, the
electromagneticthrustforcecanbeexpressedas
I
3
M c
F=
(3.25)
2 tp
n
3.4 Thermalmodel
The thermal analysis allows to verify the compatibility between the temperatures
insidetheactuatorandtheselectedinsulationclass.Thethermalanalysiscanbeperformed
by means of lumped equivalent thermal circuit or FEM analysis of the temperature
distributions.
The first method is particularly suitable in the preliminary design to determine the
averagetemperatureofdifferentinnerpartsoftheactuator,buttheprecisionoftheresults
isinfluencedbytheaccuracyofthelumpedcircuitschematization.
Forsimplicity,thetemperatureoftheframe,mouldedinaluminiumorcastiron,canbe
considered uniform because of the high value of thermal conductivity coefficient.
Considering the actuator fixed on an adiabatic surface, with the simplified frame shape
depicted in Figure 3, and with the slider in horizontal position, the frametoenvironment
temperaturerise f e canbecalculatedbymeansofthefollowingequation:
Qtot =
f e
Rconv
+ Q rad
(3.26)
where Qtotisthetotalpowerlosses, Rconvthethermalframetoenvironmentresistancedue

to natural convection, and Qrad the radiated thermal power. Rconv can be expressed as
follows:
1
R conv =
(3.27)
vc S V + hc S H
SH
hc
SV
vc
Figure3Simplifiedframeshapeusedforthermalmodelling.
50|P a g e
where SV and SH are the values of the vertical and horizontal frame surfaces, whereas vc
and hc are the natural convection coefficients for vertical and horizontal surfaces. The
valuesofvcandhccanbecalculatedasfollows:
avc = 5.6 4
f e
h Tenv
hc = 0.5 vc
(3.28)
(3.29)
(3.30)
wherehistheframeheightandTenvtheambientabsolutetemperature.
Finally,theradiatedpowercanbecalculatedasfollows:
Q rad = rad S rad (Tenv + f e ) (Tenv )

4
whereradistheframeradiationcoefficientandSradistheradiationsurfacevalue,calculated
asthesumofSVandSH.
Undertheassumptionofwindinglossesuniformlydistributedinthecoils,theaverage
valueofthewindingtoframetemperaturerisecanbeexpressedas
P s 2 ws 2s
w f = cu +
(3.31)
N s S ' 3 4
wheresisthethicknessoftheslotinsulation,Sisthelateralsurfaceofthewinding, isthe
thermalconductivityoftheslotinsulation,and istheequivalentthermalconductivityof
thewinding,dependingontheenamelling,theimpregnationandthefillfactoroftheslot.
The main slot parameters are highlighted in Figure 4, where a axial sight of the motor is
shown(a)anda3Dviewofonewindingisdepicted(b).
Finally,thewindingtoenvironmenttemperatureriseis
w e = w f + f e
(3.32)
51|P a g e
ws
s
hs
(a)
(b)
Figure 4Drawing of the tubular actuator showing the geometrical

parametersofonehighlightedcoil(a),and3Dsightofthecoil(b).
3.5 Mechanicalmodeloftheslider
The slider of tubular PM motors usually presents a slight eccentricity with respect to
thesymmetryaxisofthestator,duetotheassemblingmisalignmentandclearanceofthe
bearings[84][85].Asaconsequence,theresultantradialforceactingontheslider,dueto
themagnets,isnotzeroandthestaticequilibriumconfigurationoftheslideraxistendsto
bend.Thecalculationofthemaximumdisplacementoftheslideraxisisextremelyimportant
in order to verify that there is no interference between slider and stator. In addition, the
reaction forces on the bearings have to be determined, because they increase the friction
reducingthelifeoftheactuator.
The analysis of the static equilibrium configuration of the actuator is reported under
theassumptionofelasticmaterialsandsmalldisplacements.
A. Staticequilibriumanalysis
Toanalyzethestaticequilibriumoftheslider,itisconvenienttousetheschemeofa
simplysupportedbeamloadedwithadistributedforceq,showninFigure5.Asknown,the
staticequilibriumequationofthedeflectedbeamisasfollows:
EI
d 4v ( z )
+ q( z ) = 0
dz 4
52|P a g e
(3.33)
q(z)
v(z)
L
z
Figure 5Scheme of the simplysupported beam for the

studyofthesliderstaticequilibriumconfiguration.
ThebendingmomentM(z)isrelatedtothedisplacementvasfollows:
d 2v ( z )
(3.34)
M ( z ) = EI
dz 2
Equations(3.33)and(3.34)canbesolvedimposingtheboundaryconditionsthat v(z)
andM(z)vanishforz=0andz=Loncethedistributedloadqisknown.
Inordertodetermine q,referenceismadetoFigure6,showingatransversesectionof
thetubularactuator,wherethesliderislocatedinaneccentricpositionwithrespecttothe
statorsymmetryaxis.Undertheassumptionofsmalldisplacements,themagneticgapcan
beexpressedasfollows:
(3.35)
=0+ycos.
df
0
R
si
Figure6Transversesectionofthetubularmotor
showingtheeccentricityoftheslider.
53|P a g e
TheelementaryforceactingontheelementarysurfacedSoftheslideris:
B( ) 2
df ( ) =
dS .
2 0
(3.36)
Thedistributedforceq,resultantofallelementarycontribution,isdirecteddownwards
andisgivenby
q = d si
0
B( ) 2
cos( )d .
4 0
(3.37)
Toexplicittheexpression(3.37),thefunction B ( ) hastobedetermined.Considering
themotorsectiondepictedinFigure7andcalculatingtheMMFdropalongthedashedlineit
follows:
1 W H
B( ) = 0 M M .
(3.38)
2
( )
Substituting (3.35) in (3.38), the resulting equation can be expressed in Mc Laurens

series,asfollows:
y
cos + ...
B ( ) = B0 1
(3.39)
whereB0(z)isthevalueofthefluxdensityintheoriginalconfiguration.Substituting(3.39)in
(3.37)leadstothefollowingexpressionofq(z):
S-N
N-S
S-N
wM
Figure7Longitudinalmotorsection.
54|P a g e
q=
R si B02
y .
2 0 0
(3.40)
Inordertotakethenonuniformityofthemagneticfieldintheactuatorintoaccount,it
can be demonstrated that in (3.40) B0 can be approximated by BRMS. In addition, the
displacementycanbeexpressedasthesumoftwoterms,namelytheeccentricityv0andthe
displacementvduetobending.Inthisway,thedistributedloadqbecomes:
2
R si B RMS
(v + v 0 ) .
q=
2 0 0
(3.41)
Substituting(3.41)in(3.33)andsolvingthedifferentialequationitispossibletoobtain
theanalyticalexpressionofv(z).Addingthemaximumvalueofv(z)totheinitialeccentricity
v0leadstothemaximumdisplacementymaxoftheslideraxis.
B. Maximumsliderdisplacement
Assumingthat v0istheeccentricityoftheslideraxiswithrespecttothesymmetryaxis
ofthestator,themaximumdisplacementoftheslidercanbeexpressedas
(3.42)
ymax=v0
whereisacoefficientgreaterthan1definedas
1
1
1
KL
KL
= cosh
.
+ cos
2
2
2
In(3.43)ListhedistancebetweenthebearingsandKisdefinedasfollows
2
Rsi BRMS
K=
2 0 0 E I
4
(3.43)
(3.44)
whereBRMSistheRMSvalueinapolepitchofthespatialdistributionofthefluxdensity,Eis
the Youngs modulus and I is themoment of inertia of the slider tube with respect to the
bendingaxis.
55|P a g e
Thebehaviourof isshowninFigure8asafunctionoftheproduct KL.Accordingto

(3.42), the maximum displacement is proportional to the slider eccentricity. Therefore
attentionhastobepaidtothechoiceofbearingstypeandtheassemblingaccuracy.
Finally,itisworthnotingthatif KLisamultipleof ,then(3.42)hasaninfinitevalue,
i.e.astaticequilibriumconfigurationisnotpossible.ThecaseKL=1givesthecriticaldistance
betweenthebearings,thatis
Lcr =
=4
0 0 3E I
2
R si B RMS
(3.45)
Obviously,thestatorlengthshouldbefarenoughfromthecriticallengthLcrinorderto
avoidinterferencebetweenstatorandslider.
C. Reactionforcesonthebearings
Theresultantforce Rduetotheradialmagneticforcescanbecalculatedbymeansof
thefollowingequations:
R=
2
d se Brms
v0 L
0 0
(3.46)
whereisaconstantgreaterthan1,depictedinFigure9,whoseexpressionisasfollows:
1
KL
KL
=
+ tan
tanh
(3.47)
KL
2
2
10
0.5
1.5
2.5
Figure8Behavioroftheparameter.
56|P a g e
KL
10
0.5
1.5
2.5
KL
Figure9Behavioroftheparameter.
Thereactionforcesonthebearingsareequalto R/2.Theyproducefrictionforcesthat
are opposed to the motion. It is necessary to verify that these friction forces are not too
relevantwithrespecttotheratedthrustforce.
3.6 Optimizeddesignalgorithm
AtwostepprocedureisappliedtoobtaintheoptimaldesignofaseriesofPMtubular
linearactuator.Inthefirststeptheintegratedmodelofthemotorisutilizedfordetermining
therelationshipsbetweencriticaldesignparametersandmachineperformance,facilitating
the rapid assessment of a large number of potential designs, satisfying magnetic, thermal
andmechanicalconstraints.Inthesecondstep,atwodimensionalFEManalysisisadopted
torefinethepreliminarydesign,allowingtheforcerippletobeminimized.
Thedesignoptimizationcanpursuitseveraltarget,suchasthemaximumforcedensity,
themaximummoveracceleration,etc..Assumingthatthemotorwillbeusedinautomated
industrial machine, then the dynamic performance represents the most meaningful
parameter. Thus, the design objective is to select the machine parameters in order to
maximize the force density/acceleration and minimize the total force ripple. In the same
time,thermalandmechanicalconstraintshavetobesatisfied.Thethermallimitdependson
theinsulatingmaterialadoptedforthestatorwindings,whereasthemechanicalconstraints
arerelatedtothemaximumdisplacementduetotheelectromagneticforcesinpresenceof
assemblingmisalignment.
A. Inputparameteranddesignconstraints
The integrated model of the motor can be used to define different design algorithm,
eachonediffersfromtheothersontheparameterschosenasinputoftheproblem.
Inparticular,inputparameterswillbeusedinthedesignalgorithmindifferentmanner
onthebasisoftheirmeanings.Someinputsdefinetheproblemconstraints,thustheydefine
onespecifiedvaluethatwillbeusedinthealgorithm,whileotherinputsdefinearangeof
57|P a g e
valueswhereonedesignparametermayvary.Finally,otherinputparameterssetthelimit
value of certain design parameter, so they represent a condition which determines the
discardingoftheproblemsolutionnotcompatiblewiththeaforementionedcondition.
Thefirsttypologyofparameterscanbenamedfixconstraints,thesecondonemay
be named rangeconstraints while the third one can be called limitconstraints. The
choice of the input parameters had been done considering the technical issues which
concerntherealizationofthetubularmotorwithSMCstechnology,viz.consideringissuesof
theSMCcompactionprocess.Further,somemechanicalconstraintshadtobefixedinorder
tousecommercialindustrialcomponents,thusobtainingamotorwithpotentiallowcost.
Amongthefixconstraints,itispossibletolist:
Numberofpoles Np;
NumberofslotsNs;
Statorexternaldiameter dex;
Actuatorlength wact;
Toothshoepolewidth wep;
Shoepolehigh hep;
Slidertubewidth ht;
Mechanicalairgaplength g;
Innersliderborediameter dsib;
Toothfluxdensity Bt;
Initialslidereccentricity v0;
Maximumwindingsovertemperature w e .
These constraints define the motor volume, the main electromagnetic specifications,
the maximum windings overtemperature and the initial slider eccentricity due to the
assemblingmisalignmentandclearanceofthebearings.
The input parameters variable in a reasonable range, defined rangeconstraints,
considerthefeasibilityissuesrelatedtoSMCsandmagnets,andare:
Maximumandminimummagnetwidth wm,maxandwm,min;
Maximumandminimuminnersliderdiameter dsi,maxanddsi,min;
Finally,theparameterdefinedaslimitconstraintsare:
Maximumslidercorefluxdensity Bp,max;
Maximumsliderdisplacement ymax.
58|P a g e
Keepingthetoothfluxdensityandtheslidercorefluxdensitylimitsunderthekneeof
the BH curve of the magnetic material, iron saturation is avoided. The maximum slider
displacementhastobefixedasasmallpercentageofthemechanicalairgap,letssay10%.
B. Preliminarydesign:solutionprocedureoftheintegratedmodel
equations
Thefixconstraintsandtherangeconstraintsareaprobleminput,andtheycanbe
used to develop the problem solution. For a given set of fixconstraints and limit
constraints,thefollowingsolutionprocedurecanberepeatedforeachvalueof wmand dsi
includedintherangeconstraints:
1) Solvingthesystemofequationformedwith(3.1),(3.2),(3.4)and(3.6)then Bm,
HmandBgcanbecalculated.
2) Thevalueofwtisobtainedusingthe(3.8),the(3.12)andBg.
3) Substitutingwtin(3.14),thenresultsthevalueofws.
4) Using(3.11),(3.15),(3.16),(3.17)and wt,itispossibletodefinethevaluesof hb
andhs.
5) Through(3.13) and (3.7), Bep (3.11)and Bp are calculated; Bp is then compared
withBep,maxinordertodiscard/donotdiscardthecurrentsolution.
6) Thepreviousresultpermittocalculateanumericalvaluefor(3.21)and(3.22).
7) Thesolutionofthethermalmodel,asillustratedinsubsequentsection,permits
tocalculatePcu.
8) Using Pcu in (3.23) and substituting the result in (3.24), the slot current is
determined.Thelattercanbeusedtocalculatethethrustforceby(3.25).
9) Finally,alldataareavailabletocalculate(3.44),then(3.43)andfinallytheslider
displacementby(3.42),whichiscomparedwith ymaxinordertodiscard/donot
discardthecurrentsolution.
C. Thermalmodelsolution
The substitution of (3.28) and (3.29) in (3.27) define the relation between f e and
Rconv .Usingthelattertermandsubstituting(3.30)in(3.26)yield(3.48).
5/ 4
f e
Tenv + f e
+ B
100
+ C = Qtot
(3.48)
59|P a g e
where
5.6 S V
5.6 S H
A =
+
4 hT
0.5 4 h Tenv
env
B = rad ( S V + S H )
(3.49)
Tenv
( SV + S H )
C = rad
100
D = 1 s + 2 ( ws 2 s )
N s S ' 3
4
Equation (3.48) can be solved using NewtonRaphson iterative method to calculate

f e .Theiterativeformulawhichpermitstocalculatetheframeovertemperatureis:
fn+1e = fne
f ( fne )
f ' ( fne )
(3.50)
(3.51)
where
Tenv + f e f e
5/ 4
+
f ( f e ) = A f e + B
+ C we Qtot
D
D
100
T
+
5
1
1
env
f e
1/ 4
f ' (C ) = 4 A f e + 4 B 100 100 + D
D. FiniteElementAnalysis
Thepreliminarydesignobtainedwiththeintegratedmodelofthemotorcanbefurther
improved using an axissymmetric finiteelement analysis to calculate the flux density
distributionintheactuatorandthelocalizedsaturationintheironcore.Theeffectofload
currents can be evaluated in order to refine the main actuator dimensions, such as tooth
width, tooth polar shoe width, etc.. This analysis allows the evaluation of the cogging and
endeffectforces,furthermoreitpermitsthecorrectsizingandthecorrectpositioningofthe
compensationdisks,asshowninFigure10.
The finiteelement analysis can be applied also to the thermal design, in order to
determine the temperature distributions in the different part of the actuator, as a
consequenceofthecorelossesandwindinglosses.
60|P a g e
Figure10Magneticfieldnearacompensatordisk.
E. Numericaloptimizationofthenumberofwireperslot
Thedynamicbehaviourofthetubularmotorstronglyinfluenceitsservocharacteristics,
whichareveryimportantforautomatedmachineinindustrialapplications.Thus,thedrive
performancearelimitedbyboththeinverterlimitsandthemotorlimit.Theformerconsists
in Dc bus voltage, nominal current and overload current, while the latter is the maximum
slot current, which define the maximum current before that magnet demagnetization
occurs. Thus the number of wires per slot, viz. the number of turns in series per phase,
influence the dynamic behaviour of the motor, since the phase parameters and the main
motorconstantsvarywithit.Itiswellknownthatphaseresistanceandinductancedepend
onthesquareofthenumberofwiresperslot,whilethebackEMFconstantandtheforce
constants depend on the number of wires per slot. Finally, the phase current is inversely
proportional to the number of wires per slot for a fixed nominal slot current, which
guarantiestheinvariabilityoftheproducedthrustforce.
In classic rotary machine the number of wires per slot is sized for a specified voltage
and frequency, with constant rotor speed. Tubular linear actuators never operate at
constantspeedbecauseofitsreciprocatingmotion,thusthenumberofwiresperslotcanbe
consideredasadesigndegreeoffreedomwhichcanbeoptimizedpursuingaspecifictarget
foragivensetofinverterandmotorlimits.
The following constraints are fixed: the DC bus voltage, the nominal current of the
inverter,itsoverloadcurrent,thenominalslotcurrentandfinallythemaximumslotcurrent.
The proposed optimization is based on a numerical simulation of the tubular drive with
variable number of wires per slot, assuming that the inverter feeds the motor with its full
voltageandfixingthemoverstroke.Thesubsequentinertialstartupisregisteredintermsof
position, velocity, phase current and slot current. The position transient, together with
phaseandslotcurrent,isusedtoselectthebetternumberofwiresperslot.Theselection
criterion considers at first the position curve: in first instance, the motor with number of
wiresperslotwhichreachesforfirstthehalfstrokeisconsideredthebetter.Thentheslot
61|P a g e
currenttransientisanalyzedtoensurethatitslimitisnotreached,oratleastitlastsfora
shortperiod.
The drive model used for this simulation is depicted in Figure 11, where the motor
parameters with * change their value in concordance with the number of wires per slot
andmisthemovermass.Theblockcurrentlimitimplementsthemaximumslotcurrent
and the overload drive current, thus it limits the phase current in agreement with the
aforementioned current constraints. The inverter block considers the DC bus voltage and
calculates the vq voltage considering the amount of vd voltage necessary to keep the id
currentequaltozero.Thecalculationrequiredforthelatterpurposeuses iqand ,which
enterintheinverterblock.
Anexampleofoptimizationisreported:foragivenDCbusvoltage,inverteroverload
currentandmaximumslotcurrentthetransientbehaviourofmainelectricandmechanical
characteristicshavebeenregistered.Thenumberofwiresperslotismeasuredinperunit
(p.u.),andthebasevalueisthenumberofwiresperslotcalculatedinthesamemannerofa
rotarymachine.
Figure12showsatypicalpositiontransient,fortheconsideredexample,areductionof
wires per slot results in a increased dynamic performance, since the halfstroke is quicker
reachedbymotorswithslownumberofwiresperslot.Figure13showthephasecurrentin
p.u., where the base value is the inverter overload current: the overload constraint never
occurs, in fact the maximum phase current value remains under the unity. The current
limitationshighlightedwithadashedlineinFigure13isintroducedbytheslotcurrentlimit.
Figure 14 depict slot current transient referred to the maximum slot current: the figure
highlightswithadashedellipsetheeffectoftheslotcurrentlimit,whichitlastsforaperiod
growingwiththereductionofthenumberofwiresperslot.
(Mseie)*
vq +
DCbus
Voltage
iq
1
Rs*+s Ls*
current
limit
Kf*
ms
tp
1
s
Ls*
inverter
Ls*
iq
vd + +
id
1
Rs*+s Ls*
Figure11Blockdiagramofthedriveusedtooptimizethenumberofwiresperslot,payingattentionatdrive
andmotorconstraints.
62|P a g e
0.08
Stroke [m]
0.05
0.8
Phase current [p.u.]
0.06
n=3.5e-001
n=4.6e-001
n=5.7e-001
n=6.7e-001
n=7.8e-001
n=8.9e-001
n=1.0e+000
n=3.5e-001
n=4.6e-001
n=5.7e-001
n=6.7e-001
n=7.8e-001
n=8.9e-001
n=1.0e+000
0.07
0.04
0.03
0.6
0.4
0.02
0.2
0.01
0
0.005
0.01
0.015
0.02
0.025
time [s]
0.03
0.035
0.04
0.045
0.005
0.01
0.015
0.02
0.025
time [s]
0.03
0.035
0.04
0.045
Figure 12Mover position transient, for a variable Figure 13Phase current in p.u. (base value: inverter
numberofwiresperslot.
overloadcurrent),foravariablenumberofwiresper
slot.
Slot current [p.u.]
0.8
0.6
n=3.5e-001
n=4.6e-001
n=5.7e-001
n=6.7e-001
n=7.8e-001
n=8.9e-001
n=1.0e+000
0.4
0.2
0.005
0.01
0.015
0.02
0.025
time [s]
0.03
0.035
0.04
0.045
Figure 14Slot current in p.u. (base value:

maximumslotcurrent),foravariablenumberof
wiresperslot.
3.7 Conclusion
Thischapterpresentsthemathematicalmodelandanoptimaldesignalgorithmusedto
design of a prototype of permanent magnet actuator, paying attention not only to the
electromagnetic issues, but also to the thermal and mechanical constraints. The model
allows to manage the relationship between critical design parameters and the machine
performance. Further, it permits to develop a large number of optimized motor design
quickly, with a precision smaller than 7%. Detail regarding the motor design and their
relativeexperimentalresultswillbepresentedinnextchapter.
63|P a g e
CHAPTER
4
OptimizeddesignofaPMtubular
linearactuatorand
experimentalresults
4.1 Introduction
The design methodology proposed in the previous chapter, which consider
electromagnetic, thermal and mechanical issues, will be used to size a threephase PM
tubularlinearactuator.
Initially,apreliminaryoptimizeddesignwillbeappliedinordertoanalyzetheinfluence
of critical design parameters on the machine performance. Hence, a FEM analysis will be
usedtorefinethedesignandtotakeintoaccounttheendeffectproblem.Itwillfollowthe
optimizationofthenumberofwireperslot.
Finally,experimentalresultsofaprototypeoftubularlinearactuatorwillbepresented
andtheywillbeusedtovalidatethedesignprocedure.
4.2 Definitionofthedesignconstraints
The optimized design algorithm can be applied after the definition of the constraints
describedandlistedinParagraph3.7AinChapter3.
The first two parameters, Np and Ns, are crucial for the motor design, indeed they
strongly affect the amplitude and the space periodicity of the cogging force. Brushless
motors use fractional pitch windings in order to minimize the cogging torque, thus this
solution will be adopted to define the number Np and Ns, since other technical solutions
cannotbeequallyeffectiveforthetubulartopology.ThechoiceofNp=8andNs=9results
an optimum compromise for the cogging force issues, since the space periodicity of the
coggingforceresults1/9ofthepolepitch.Inadditionfurthertheequivalentnumberofslots
perphaseandperpoleisquitehigh,inparticularitis3.
Otherdesignconstraints,suchasdexandwact,definethevolumeofstator.Thesetwo
parametershadbeenchoseninordertoobtainacompactmotor,sincetheratiowact/ dexis
1.6.
65|P a g e
OptimizeddesignofaPMtubularlinearactuatorandexperimentalresults
Thenthestainlesssteeltubewidthhtthatcontainsthemagnetsandthesliderwiththe
SMC poles had been chosen considering the commercial dimensions of such an industrial
product. The mechanical airgap length g represents a crucial parameter, since it strongly
influencestheperformanceandthedesignofthemachine.Indeed,thesmallertheairgap
is, the highest the thrust force is, but at the same time the radial force and the slider
bending become critical. Thus it had been repeated the preliminary optimized design for
three different values of g, which are respectively 0.5, 0.75 and 1 mm. The inner bore
diameterofthesliderdsibhasbeensetinordertoguaranteeaeasysliderfeasibilityandto
improvethemotordynamicfeature.Thewidthofthetoothshoepolewephadtoleaveaslot
openinglargerthanthesumbetweenthestainlesssteeltubewidthhtandthemechanical
airgapg,inordertominimizeleakagefluxexinslotopenings.Thethicknessoftoothshoe
polehasbeenfixedconsideringtechnicalissuesonthecompactionprocessoftheSMCs.The
initial slider eccentricity v0 has been determined considering the typical value of bearing
clearance and assuming the presence of mounting misalignment of the slider, fixed at 0.1
mm.Themaximumwindingsovertemperaturedependsonclearlybytheinsulationclassof
thewindings,whichisclassB.
ThetoothfluxdensityBtandthemaximumslidercorefluxdensityBp,maxhasbeenfixed
at 1.1 T in order to avoid the saturation of the iron core. Figure 1 shows the BH curve of
Somaloy500.Itcanbeseenthatthehighlightedoperationalmagneticpointisunderthe
kneeofthecurve.
Then,magnetwidthlimits,viz.wm,maxandwm,min,havebeenfixedwiththeaimtoobtain
theratiowm/tpwithinintherange0.20.8,whilethelimitsoftheinnersliderdiameterdsi,max
and dsi,min have been selected in order to obtain the ratio dsi/dex whithin in the range
0.160.65.Thelatterlimitsarerelatedtotheminimummoverdiameterandtotheminimum
slotdepth.
Finally,themaximumsliderdisplacementymaxofthesliderwassettomaintaintheratio
2.1
2
1.9
1.8
1.7
1.6
1.5
1.4
1.3
B [T]
1.2
1.1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
10'000
20'000
30'000
40'000
50'000
60'000
H [Asp/m]
70'000
80'000
90'000
100'000
110'000
Figure1BHcurveofSomaloy500,themagneticoperationalpoint
ishighlightedinthecurve.
66|P a g e
ymax/v0underthevalue0.1.
4.3 Analysisoftheinfluenceoftheairgaplengthonmotor
performance
Thepreliminarydesignpresentedinthepreviouschaptercanbeusedtocarryoutan
analysisoftheinfluenceoftheairgaplengthonthemotorperformance.
Asintherotatingmachine,theairgaplengthrepresentsafundamentalparameterfor
the designer, and its value influences both the electromagnetic and the mechanical
behaviour.Fromthemagneticpointofview,thedesignerhastodesignthemachineswith
thiny airgap, thus higher force density can be achieved. At the same time, from a
mechanical point of view, thin airgap may emphasize the slider bending because of the
bending force depends from the initial bearing clearance, on the slider mechanical
characteristics and from the square of the airgap flux density. The bending effect of the
sliderinatubularlinearmotormustnotbeneglected,sincethesliderbehaveslikeathin
beamloadedwithdistributedforce.Thus,athinairgapiscritical,infactitcanbefurther
reducedbythesliderbendingand,insomecase,creepingbetweenthesliderandthestator
occurs.
Figure2depictstheoptimizedforcedensityasafunctionoftheratiodsi/dex,calculated
astheratiobetweenthethrustforcecomputedwiththepreliminarydesignequationsand
the stator volume (wact d2ex/4). It can be observed that the maximum force/volume is
achievedforinnersliderdiametertoexternalstatordiameterincludedintherange0.40.5.
Further,themechanicalairgaplengthinfluencesnoticeablythethrustforcedensity.
thrustforcadensity[N/m3 ]
3.50E+05
3.00E+05
2.50E+05
2.00E+05
g=0.5mm
1.50E+05
g=0.75mm
1.00E+05
g=1mm
5.00E+04
0.00E+00
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
dsi/dex
Figure 2Forcetostator volume ratio as function of the ratio dsi/dex, computed for three different
valueofthemechanicalairgap.
67|P a g e
Figure 3 and Figure 4 depict respectively the radial force density, which produce the
sliderbending,andtheratioymax/v0,bothasafunctionoftheratiodsi/dex.
Figure3showsarapidgrowthoftheradialforcedensity,indeeditsvaluedependson
thesquareoftheairgapfluxdensityand,forafixedairgaplengthandwithgrowingdsi/dex
ratio,itsvaluequicklyincreases.Consideringtheequation(3.1),(3.2) and(3.4)inorderto
obtaintherelationbetweentheairgapfluxdensityBgandthemagnetremanenceBr,and
bendingforcedensity[N/m3 ]
1.00E+07
1.00E+06
g=0.5mm
g=0.75mm
1.00E+05
g=1mm
1.00E+04
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
dsi/dex
Figure3Radialforcetostatorvolumeratioasfunctionoftheratiodsi/dex,computedforthreedifferentvalues
ofthemechanicalairgap.
1.11
1.10
1.09
ymax/v0
1.08
1.07
g=0.5mm
1.06
1.05
g=1mm
1.04
g=1mm
1.03
1.02
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
dsi/dex
Figure 4Maximumtoinitial slide displacement ratio as function of the ratio dsi/dex, computed for three
differentvaluesofthemechanicalairgap.
68|P a g e
simplifyingthefinalequation,itresults(4.1).
1
B g = Br
2tp
0 KC
(4.1)
d +2 w
m
si
Equation(4.1)showsthatforagivenairgaplengthg,viz.foragiven0,agrowthinthe
innersliderdiameterdsiresultsinanairgapfluxdensityincrease.
Figure4showsthemaximumtoinitialsliderbendingasafunctionoftheratiodsi/dex,
althoughtheincreaseofthebendingforcewiththeratiodsi/dexisimportant,themaximum
sliderdisplacementdecreaseswithdsi/dexsincethemomentofinertiaoftheslidertubewith
respecttothebendingaxisincreaseswiththefourthpowerofdsi.
ThefinalanalysisofFigure3andFigure4allowstostatethatmotorswithdsi/dexratio
smaller than 0.380.40 exhibit larger slider displacement and lower bending force density
thanmotorswithhigherdsi/dexratio.Theformerhavetobepreferredthanthelatter,since
large radial force can damage the mechanical sliding system and can strongly reduce the
motorreliability.
4.4 Optimizeddesignofatubularlinearactuator
Theanalysiscarriedoutinthepreviousparagraphgivessomeimportantguidelinesfor
thechoiceofthemotorsizeamongthevariousproposedbythepreliminarydesign.Thus,it
was decided to develop a motor with an airgap length of 0.5 mm, since the other two
solutions do not produce significant advantages in terms of maximum slider displacement
andradialforcedensity,whilethethrustforcedensityresultshigher.
A. Preliminarydesign
The target of the design of the motor is to obtain a tubular actuator suited for
automatic industrial application, thus the thrust force do not represent the only choice
parameter, but also the dynamic performance must be considered. Figure 5 depicts
respectivelythenormalizedthrustforceofthemotor(thebasevalueis35N)asafunctionof
the dsi/dexratio,whileFigure6showstheslideraccelerationasafunctionofthe dsi/dexratio,
assuminganinertialpayloadvariableintherange00.5kg.ThelatterFigureshowsthatan
increasing in the additional inertial mass shifts the point of maximum acceleration toward
growingvaluesofthedsi/dexratio.
The dsi/dex ratio range to be selected for the maximum thrust force is 0.40.5, but in
this range the dynamic behaviour, measured by the slider acceleration in Figure 6, is very
low. The dsi/dex ratio range which maximize the dynamic performance is 0.20.32. Then, a
compromise value of the dsi/dex ratio equal to 0.28 was selected. That value is highlighted
bothinFigure5andinFigure6.
69|P a g e
normalizedthrust forca[p.u.]
2
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
dsi/dex
Figure5Normalizedforceasfunctionoftheratiodsi/dex,computedforthemechanicalairgapof0.5mm.
90
acceleration[m/s2 ]
80
70
60
M=0.0kg
50
M=0.1kg
40
M=0.2kg
30
M=0.3kg
20
M=0.4kg
10
M=0.5kg
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
d si /dex
Figure6Maximumslideraccelerationasfunctionoftheratiodsi/dex,computedforthemechanicalairgapof
0.5mmandwithanadditionalinertialmassvariableintherange00.5kg.
B. FEMrefining
The main motor dimensions calculated with the preliminary motor design have been
usedtodefineaFEMmodelofthemotor,showninFigure7.Thatmodelwasusedtoverify
thepreliminarydesigncalculation.Furthersimulationspermittedtoanalyzetheendeffect
force disturbance, which has been minimized through the use of a compensator disk. The
70|P a g e
windings
stator
Slider pole
magnet
Figure7Femmodelofthemotorsizedwiththepreliminarydesign.
latter aspect results very important in PM tubular linear actuator, since end effect force
rippledeterioratetheslidermotion,producesnoise,vibrations,etc.
Figure8depictstheendeffectforcecomputedwithFEManalysis,imposinganullvalue
ofthecurrentinthewindings,forthemotorchoseninthepreviousparagraphasafunction
ofthesliderposition;theforceisinp.u.andthebasevalueisthemotornominalthrustforce
of35N,whilethesliderpositionisreferredtothepolepitch.
Figure9showsthep.u.reluctanceforceofonestatordisk,computedwithFEMmodel
showninFigure10,asafunctionofsliderpositioninp.u..ThecomparisonbetweenFigure8
and Figure 9 shows that the force produced by one disk has the same waveform and the
sameamplitudeoftheendeffectforceofthewholemotor.Thus,onediskcanbeplacedin
a particular position referred to the stator in order to produce the same force waveform
shiftedofhalfspaceperiodicity,thusthenetforcewithnullcurrentinthewindingsshould
be minimized. The position of the compensator disk was investigated through a series of
0.5
0.4
endeffectforce[p.u.]
0.3
0.2
0.1
0
0.1 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.2
0.3
0.4
0.5
(sliderposition)/tp
Figure8Endeffectforce,normalizedtothethrustforce,computedbyFEManalysisasfunctionoftheslider
position,expressedinp.u.ofthepolepitchtp.
71|P a g e
0.5
0.4
diskforce[p.u.]
0.3
0.2
0.1
0
0.10.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0.2
0.3
0.4
0.5
(sliderposition)/tp
Figure 9Reluctance force produced by one stator disk, normalized to the thrust force, computed by FEM
analysisasfunctionofthesliderposition,expressedinp.u.ofthepolepitchtp.
Stator disk
Slider pole
Magnet
Figure10MagneticfluxpathoftheFEMmodelofonestatordisk.
FEM simulations, where the compensator disk has been shifted in order to determine the
optimalposition.ThefinalresultsofthisactivitiesareshowninFigure11,wheretheoptimal
compensationoftheendeffectforceresultsinaverylowforceripple.
Finally,ithasbeencalculatedthethrustforcewithaFEMmodelcompletedwiththe
compensation system. The sinusoidal stator currents have been imposed to change in
synchronismwiththesliderposition,inordertosimulaterealoperativeconditions.Figure
12showsthethrustforceproducedbythemotor:theforcerippleistheresidualendeffect
force,showninFigure11,whilethecirclerepresentsthemeanvalueofthecomputedthrust
force.
C. Optimizationofthenumberofwiresperslot
Themodelproposedinparagraph3.7sectionEhasbeenusedtooptimizethenumber
ofwiresperslotinordertofullyexploitthetubularlineardrivecapability.Themainresults
72|P a g e
compensatedenfeffectforce[p.u.]
0.5
0.4
0.3
0.2
0.1
0
0.1 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.2
0.3
0.4
0.5
(sliderposition)/tp
Figure11Compensatedendeffectforceinp.u.asafunctionofthesliderpositionreferredtothepolepitch.
thrustforce[p.u.]
1.1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
(sliderposition)/tp
Figure12ThrustforcevaluesobtainedbyFEManalysiswithsinusoidalstatorcurrentsvaryingsynchronously
withthepositionoftheslider.
ofthisoptimizationaredepictedinFigure13,Figure14,Figure15,whichshowrespectively
themoverpositiontransient,thephasecurrenttransientandtheslotcurrenttransient.
The base value for the slot and for the phase current are the maximum slot current,
whichavoidsmagnetdemagnetization,andthemaximumphasecurrentoftheinverter.The
base value for the number of wires per slot is that one calculated in same manner of a
rotatingmachine.
Thenumberofwireswhichgivesbetterdynamicperformanceisthesmallerone,but
this solution results very hazardous, since slot current limit occurs for a long period, as
showninFigure15.Amongthevarioussolutionproposed,thatonewhichappearstobea
goodcompromiseispointedoutinFigure13,Figure14andFigure15.Thus,thenumberof
wires per slot has to be reduced by 33% in comparison to the that one calculated in the
traditionalway.
73|P a g e
0.08
Stroke [m]
0.05
0.8
Phase current [p.u.]
0.06
n=3.5e-001
n=4.6e-001
n=5.7e-001
n=6.7e-001
n=7.8e-001
n=8.9e-001
n=1.0e+000
n=3.5e-001
n=4.6e-001
n=5.7e-001
n=6.7e-001
n=7.8e-001
n=8.9e-001
n=1.0e+000
0.07
0.04
0.03
0.6
0.4
0.02
0.2
0.01
0
0.005
0.01
0.015
0.02
0.025
time [s]
0.03
0.035
0.04
0.045
0.005
0.01
0.015
0.02
0.025
time [s]
0.03
0.035
0.04
0.045
Figure 13Mover position transient, for a variable Figure 14Phase current in p.u. (base value: inverter
numberofwiresperslot.
overloadcurrent),foravariablenumberofwiresper
slot.
Slot current [p.u.]
0.8
0.6
n=3.5e-001
n=4.6e-001
n=5.7e-001
n=6.7e-001
n=7.8e-001
n=8.9e-001
n=1.0e+000
0.4
0.2
0.005
0.01
0.015
0.02
0.025
time [s]
0.03
0.035
0.04
0.045
Figure 15Slot current in p.u. (base value:

maximumslotcurrent),foravariablenumberof
wiresperslot.
4.5 Experimentalresults
Figure 16 shows the prototype of a PM tubular motor realized in laboratory in
compliancewiththedesignalgorithmbasedontherelationshipsdescribedintheprevious
chapter.ThestatorandthemagneticcylindersofthesliderarebuiltinSomaloyTM500with
0.5%KenolubeTM,compactedtothedensityof7.16g/cm3.TheNdFeBpermanentmagnets
(BR=1.1Tat20C)areassembledinsideanonmagneticstainlesssteeltube.
A. Thrustforce
Figure17showsthebehaviouroftheforce(normalizedwithrespecttotheratedforce)
producedbythemotorwithnullcurrentsandcalculatedbymeansofFEManalysiswhenthe
motoriswithoutcompensationdisks.
74|P a g e
Figure16Prototypeoftubularmotor.
Normalized end-effect force (p.u.)
0.6
0.4
0.2
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
-0.2
-0.4
-0.6
Slider position [p.u.]
Figure17Endeffectforce(normalizedwithrespecttheratedthrustforce)producedbythemotorwithnull
currents,asafunctionofthesliderposition(normalizedwithrespectthepolepitch).Behaviorofthemotor
withoutcompensationdisks,calculatedwithFEManalysis
Figure 18 shows the force measured on the prototype. The average force value in
Figure18isduetothefrictionofthesliderbearings,whereasthesuperimposedalternative
forcegivesvaluesclosetothoseshowninFigure17.
The measured values of the force in presence of compensation disks are shown in
Figure19.Ascanbeseen,thecompensationdiskshavepracticallycancelledtheforceripple.
Figure 12 shows the behaviour of the thrust force calculated by means of finite
elementanalysis,undertheassumptionthatthestatorcurrentsvarysynchronouslywiththe
sliderpositionxaccordingtothefollowingrelationships,takingtothefieldorientedcontrol
principlesintoaccount.
x 2
I k = I rated sin
(k 1) k=1,2,3
t
3
p
(4.2)
75|P a g e
TheaveragevalueoftheforceinFigure12correspondstotheratedvalue.Toconfirm
the analytical results, the following constant currents have been injected in the stator
windings:
I2=I3= 12 I 1
(4.3)
whereI1hasthemaximumpositivevalue.
Figure20showstheforceproducedbytheactuatorfordifferentsliderpositions,apart
from the contribution of the friction force shown in Figure 19. It's worth noting that the
maximum value of the force measured with constant currents (highlighted by the circle)
correspondstothevalueoftheforcecalculatedwithcurrentssynchronizedwiththeslider
position,showninFigure12.Theagreementbetweenexperimentalandnumericalresultsis
satisfactory.
Normalized end-effect force (p.u.)
1
0.8
0.6
0.4
0.2
0
0.0
0.1
0 .2
0.3
0.4
0.5
0 .6
0.7
0.8
0.9
1.0
-0.2
Slide r position [p.u.]
Figure18Endeffectforce(normalizedwithrespecttheratedthrustforce)producedbythemotorwithnull
currents,asafunctionofthesliderposition(normalizedwithrespectthepolepitch).Behaviourofthemotor
withoutcompensationdisks,measured.
Normalized thrust force (p.u.)
1
0.8
0.6
0.4
0.2
0
0
0.1
0.2
0.3
0.4
0.5
0 .6
Sl ider positi on [p.u]
0.7
0 .8
0.9
Figure 19Compensated endeffect force (normalized with respect the rated thrust force) produced by the
motor with null currents, as a function of the slider position (normalized with respect the pole pitch),
measured.
76|P a g e
Normalized thrust force [p.u]
1.2
1
0.8
0.6
0.4
0.2
0
-0.2
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
slid er position [p.u.]
Figure20Measuredforceproducedbytheactuatorwithconstantvaluesofthestatorcurrents,apartfromthe
contributionofthefrictionforce.
B. Thermalbehaviour
A test of thermal characterization in which the power losses are equal to the rated
losseshasbeenperformed.Thecalculationoftheframeovertemperaturebymeansofthe
thermalmodelproposedinthepreviouschapterleadstovaluessimilartothosemeasured
bythermocouplesfixedontheexternalsurfaceofthemotor,asshowninTableII.
Finally, the windingstoenvironment overtemperature rise was calculated (assuming
=0,2Wm1K1).TableIIIcomparesthenumericalresultswiththemeasuredtemperatures.
Ascanbenoted,thereisagoodagreementamongallresults.
TableIIFrametoenvironment overtemperature.
Valuecalculatedwiththermal
model
36.2C
Averagevalueofthemeasures
34.1C
TableIIIAveragewindingtoenvironmentovertemperature.
Valuecalculatedwiththermal
model
46.9C
Measuredvalue
43.7C
4.6 Conclusion
The mathematical model and an optimal design algorithm presented in the previous
chapterwasusedtodesignaprototypeofpermanentmagnettubularlinearactuator,paying
77|P a g e
attention not only to the electromagnetic issues, but also to the thermal and mechanical
constraints. The measures carried out on the prototype show a good agreement with the
resultsoftheproposedmodel.
78|P a g e
CHAPTER
5
Fivephasedualmotorseries
connecteddrive:simulations
andexperimentalresults
5.1 Introduction
Two fivephase tubular linear actuators, having seriesconnected stator windings of
distributedtype(numberofslotperphaseandperpolegreaterthan1),withanopportune
permutation of the phases, can be independently controlled with only one fivephase
inverter.Thisdrivestructurewillbedescribedandanalyzed,payingattentionontheinverter
voltage limit. Then a position control, which is necessary due to the finite length of the
slider,willbepresented.
Finally,experimentalresultsoftheproposedpositioncontrolofthedualmotorseries
connecteddrivewillbepresented.
5.2 Descriptionofthefivephasedualmotorseries
connecteddrive
Thefivephasetubularlinearmotorusedforthemultimotordrivehastobedesigned
with special characteristics. Sinusoidal flux density distribution produced by the stator
windingsisrequired,thentheinvertervoltagelimithastobeinvestigatedinordertofully
exploitthespeedrangeofthetwomachines.
A. Motorconnections
The independent control of the two fivephase motor is achieved using two
independentcurrentspacevectorsofthefivephasesystem.
Applyingthetransformations(2.3)and(2.4)tothefivecurrentsofthesystemi1,i2,,i5,
thenthreeindependentspacevectorsresult: i0 , i1 and i 2 .Relation(2.2)yieldsthefollowing
equation: i1 = i4* and i2 = i3* . The zerosequence of the current is null due to the neutral
pointconnectionofthephases,whilethesequencen1andn3canbeusedtocontrolthe
twofivephasemotorsconnectedinseries.
79|P a g e
Fivephasedualmotorseriesconnecteddrive:simulationsandexperimentalresults
The connection scheme with the correct phase transposition can be obtained by the
connectivitymatrixofTableIinChapter2,thusitresultstheschemedepictedinFigure1.
M1andM2arethetwotubularlinearmotors,whilea,b,c,d,erepresentthefivewindings
ofeachmachine.
The proposed phase connection is not the only one which allows the independent
controlofthefivephasedualmotordrives,sincetheindependentcontrolofthetwomotors
is achieved using each couple of independent current space vectors. All the possible
connectionschemescanbederivedbythecombinationsoftheindependentspacevectors,
whichisshowninTableI.
5phaseinverter
M1
M2
a
Figure 1Connection scheme with the phase transposition for the twomotor series
connecteddrive.
TableICombinationsofindependentspacevectorsforthemultimotordrive.
80|P a g e
M1
i1
M2
i1
i3
i2
i1
i2
i4
i3
i1
i3
i4
i4
i2
i4
i3
i2
B. Invertervoltagelimit
The inverter voltage limit can be determined using DutyCycle Space Vector (DCSV)
approach,infactitpermitstodefineamodulationstrategyofafivephaseinvertersuited
fortheproblemsolution.
Following the wellknown carrierbased PWM approach, the DCSV approach directly
determines,ineachswitchingperiod,theswitchingsequenceofeachinverterleg.Thelatter
canbeobtainedcomparingatriangularcarriersignalwithfivemodulatingsignalsassumed
constantineachswitchingperiod.Themodulatingsignalsaredefinedbythequantitiesmk(k
=1,2,...,5),whichrepresentalsothedutycyclesofthefiveinverterlegs.
Themodulatingsignalsmustsatisfythefollowingconstraints:
mk [0,1] (k=1,2,...,5).
(5.1)
Inordertoanalyzethemodulationstrategy,anopportunemodelisintroduced,which
is valid considering ideal switches and a switching frequency much higher than the
fundamentalfrequency.Undertheseassumptions,thehigherfrequencycomponentsofthe
variablescanbeneglected,andtheinput/outputquantitiesarerepresentedbytheiraverage
valuesoveraswitchingperiodTc.
Thegenericpolevoltagevk0canbewrittenas
v k 0 = Vc mk ,(k=1,2,...,5),
(5.2)
beingVcthevoltageoftheDCsource.
The relationships between the pole voltages vk0 and the linetoneutral load voltages
vkNare
v kN = v k 0 v N 0 ,(k=1,2,...,5).
(5.3)
Note that, owing to the symmetry of the starconnected load, the zerosequence
componentofthelinetoneutralloadvoltagesiszero.
Taking(2.3),(5.2)and(5.3)intoaccount,leadsto
v h = Vc mh ,(h=1,3),
(5.4)
(5.5)
vN 0 =
Vc m 0
,
2
81|P a g e
where the DCSVs mh and the zerosequence component m0 of the fivephase inverter are
givenby
2 5
m h = mk h(k 1) ,(h=1,3),
5 k =1
(5.6)
(5.7)
m0 =
2 5
mk .
5 k =1
Ascanbeseenfrom(5.4)and(5.5),thetwolinetoneutralloadvoltagesexpressedin
terms of space vectors are directly proportional to the corresponding DCSVs, whereas the
zerosequence component of the dutycycle m 0 does not affect the load voltages, but
determinesthevoltagevN0.
A general solution to the modulation problem of the fivephase inverter can be
obtainedbyusing(2.6)and(5.4)leadingto
1
1
m k = m0 +
(5.8)
v h,ref h(k 1) ,(k=1,2,...,5).

Vc h =1,3
2
Assuming the voltage space vectors v h , ref known as reference quantities, then (5.8)
allows the calculation of the modulating signals of all the inverter legs in each switching
period.Thevalueofm0in(5.8)isadegreeoffreedomwhichcanbeutilizedtofullyexploit
the dc voltage, and to optimise some characteristics of the modulation law, such as the
numberofswitchcommutationsinaswitchingperiodandtheoutputvoltagespectrum.
Thedutycyclesconstraints,emphasizedin(5.1),introducecomplicatedlimitationson
thepossiblevaluesof v1, ref and v 3,ref ineachswitchingperiod.
In threephase inverters this problem involves only the space vector v1,ref , and it has
been already completely solved. In fivephase inverters the analytical solution of this
problemrepresentsaveryhardtask.
Themodulationconstraintsexpressedin(5.1)become:
mk mi 1 ,(k=1,2,...,5),(i=1,2,...,5).
(5.9)
Substituting(5.8)in(5.9)andtakingtherelationships(5.4)intoaccount,itispossibleto
transformthevoltagelimitproblemintheDCSVlimitproblem,whichisindependentofVc,
leadingto
82|P a g e
[mh ( h (k 1) h (i 1) )]
1 ,(k=1,2,...,5),(i=1,2,...,5).
h =1,3
(5.10)
AmoreusefulexpressioncanbeobtainedbyrepresentingtheDCSVsinpolarformas
follows:
mh = mh e j h
,(h=1,3).
(5.11)
(5.12)
Taking(5.11)intoaccount,(5.10)canberewrittenas
3
3 1
m1 sin1 (k + i 2) sin(k i ) + m3 sin3 (k + i 2) sin(k i ) .

5
5
5
5 2
Theanalysisof(5.12),writtenforeachcouple(k,i)(k=1,2,...,5andi=1,2,...,5),
allowsthedeterminationofthevoltagelimitineachswitchingperiod.
In the case of multimotor drives, the DCSVs must be completely independent, with
arbitrary behaviour, being related to the first spatial harmonics in the two different
machines.Ingeneral,theycanbedescribedasfollows:
m1 = m1 e j1 ,
(5.13)
(5.14)
m3 = m3 e j3 .
It is very interesting to derive the validity domain of m1 and m3 regardless of the

valuesofthephaseangles1and3.
Takingintoaccounttheworstconditionfor1and3allows(5.12)toberewrittenas
3 1
m1 sin (k i ) + m3 sin (k i ) ,(k=1,2,...,5),(i=1,2,...,5).

5
5 2
(5.15)
The five inequalities in (5.15) can be summarized by the two following simultaneous
constraints
2 1
m1 sin + m3 sin
,
5
5 2
(5.16)
83|P a g e
2
1
m1 sin
+ m3 sin .
5
5 2
(5.17)
Thecorrespondingvaliditydomainof m1 and m3 isrepresentedbytheshadedareain

Figure 2. This figure can be used by the designer to set the voltage amplitude to size the
windingsofthefivephasemotor.Forexample,ifthedualmotorapplicationrequirethatthe
twomachinesworkatdifferentspeedatthesametime,suchasthenominalspeedforone
machine and quasi null speed for the second machine, then mi can be selected equal to
0.5;finallythevoltageamplitudecanbecalculatedby(5.4).
The windings of the fivephase tubular linear motor have been designed for
m1 = m3 =0.326, which corresponds to point D in Figure 2, in order to fully exploit their
speedrangeatthesametime.
C. Magneticdesignofthefivephasetubularmotor
The optimal motor design described in chapter 4 has carried out the optimal
dimensions to realize a prototype of PM tubular motor. The stator and the magnetic
cylinders of the slider are built in Somaloy500 with 0.5% KenolubeTM, compacted to the
density of 7.16 g/cm3, while axial NdFeB magnets are assembled inside a nonmagnetic
stainless steel tube, which constitutes the slider. Due to the high cost of the prototype,
mainlydeterminedbythemouldcostsofthestatorSMCdisks,ithasbeendecidedtouse
oneofthemainfeatureofthetubularactuator:themodularityofstatorandoftheslider.A
constrained deign of the motor has been carried out, using the same stator disks and the
samemagnetsofthemotordesignedinchapter4.
Thedesignaimistoproducesinusoidalfielddistributioninthemachineairgap,thusa
propernumberofslotperphaseandperpolehavetobechosen.Atthesametime,aproper
ratiobetweenthepolenumberandtheslotnumberhavetobesetinordertominimisethe
0.7
0.6
0.5
0.4
m3
0.3
0.2
0.1
0
0
0.1
0.2
0.3
0.4
m1
0.5
0.6
0.7
Figure2Validitydomainof m1 and m3 forfivephasedualmotorseriesconnecteddrive.
84|P a g e
coggingforce.Further,thisratioshouldbesuchthatthepolepitchofthefivephasemotor
resultssimilartothatoneofthethreephaseprototypeofchapter4.Theaccomplishmentof
all of these constraints produces a motor similar to three phase one, considering the
magneticpointofview.
Ifthemaingeometricparametersofthefivephasemotoraredistinguishedthrough*
fromthatonesofthethreephasemotor,thenthefollowingrelationsresult:
N S t S = N P t P ,
(5.18)
N S* t S = N P* t P* .
(5.19)
(5.20)
CalculatingtSfrom(5.18)andsubstitutingitin(5.19)yields(5.20).
t P* =
N P N S*
N S N P*
tP
The choice of the NP=14 and Ns=15, t P* / t P results in =0.9523, that is a 4.78% of
reduction in the pole pitch. Hence, the slider magnetic cylinder width of the fivephase
motordecreaseby12%ifcomparedtotheslidermagneticcylinderwidthofthethreephase
motor. Further, the proposed numbers of poles and slots result in a winding having an
equivalent number of slots per phase and per pole equal to 3, which provides a quasi
sinusoidalfluxdensitydistributionintheairgapofthemachine.
FEManalysishasbeen performedinordertocompensatetheendeffectforce,using
the same methodologies used for the threephase motor, and also for analysing the
Compensatedendeffectforce
[p.u.]
1.2
1
0.8
0.6
0.4
0.2
0
0.2
0.4
0.6
0.8
1
1.2
0
0.2
0.4
0.6
0.8
1.2
1.4
1.6
1.8
sliderposition[p.u.]
Figure3Compensatedendeffectforce,normalizedwiththefivephasemotorthrustforce,asafunctionofthe
sliderpositioninp.u..
85|P a g e
Compensatedendeffectforce
[p.u.]
0.1
0.08
0.06
0.04
0.02
0
0.02
0.04
0.06
0.08
0.1
0
0.2
0.4
0.6
0.8
1.2
1.4
1.6
1.8
Figure4Zoomofthecompensatedendeffectforce,normalizedwiththefivephasemotorthrustforce,asa
functionofthesliderpositioninp.u..
independent control of the motor. Figure 3 shows the compensated endeffect force as a
function of the slider position, normalized with the thrust force of the motor. The
effectivenessoftheproposedcompensationmethodisfurtherconfirmedbyFigure4,which
isazoomofFigure3.
Figure5depictstheFEMcalculationofthecompensatedthrustforceproducedbythe
injection of a constant first current space vector, which corresponds to a set of constant
currents i1, i2,,i5 of sequence 1, as a function of the slider position. The thrust force is
normalizedtothemaximumvalueofitsvalue.Thespatialperiodicityoftheforceis2tp,asin
thethreephasemotor.
Finally, Figure 6 and Figure 7 depict the FEM calculation of the compensated thrust
forceproducedbytheinjectionofaconstantthirdcurrentspacevector,whichcorresponds
Thrustforce[p.u.]
1.2
1
0.8
0.6
0.4
0.2
0
0.2
0.4
0.6
0.8
1
1.2
0
0.2
0.4
0.6
0.8
1.2
1.4
1.6
1.8
Figure5Compensatedthrustforceproducedbythefirstcurrentspacevector,normalizedwiththefivephase
motorthrustforce,asafunctionofthesliderpositioninp.u..
86|P a g e
toasetofconstantcurrents i1,i2,,i5ofsequence3,asafunctionofthesliderposition.The
spatial periodicity of this force contributions is one third in comparison to that one of the
force produced by the fundamental space vector, further its amplitude is similar to the
compensated endeffect force. Thus the effect of the third space vector on the force
production can be neglected if compared to the force produced by the first current space
vector,andindependentdualmotorseriesconnecteddrivecanberealized.
Thrustforce[p.u.]
1.2
1
0.8
0.6
0.4
0.2
0
0.2
0.4
0.6
0.8
1
1.2
0
0.2
0.4
0.6
0.8
1.2
1.4
1.6
1.8
Figure6Compensatedthrustforceproducedbythethirdcurrentspacevector,normalizedwiththefivephase
motorthrustforce,asafunctionofthesliderpositioninp.u..
0.1
Thrsutforce[p.u.]
0.08
0.06
0.04
0.02
0
0.02
0.04
0.06
0.08
0.1
0
0.2
0.4
0.6
0.8
1.2
1.4
1.6
1.8
Figure7Zoomofthecompensatedthrustforceproducedbythethirdcurrentspacevector,normalizedwith
thefivephasemotorthrustforce,asafunctionofthesliderpositioninp.u..
87|P a g e
D. Controlscheme
Thefundamentalcontrolschemeofthefivephasedualmotorseriesconnecteddriveis
depicted in Figure 8. The motor n1 (M1) is controlled by means of the first space vector,
while the motor n2 (M2) is controlled through the third space vector. The scheme
implementsapositioncontrolforeachmotor,whichwillbepresentedmoreindetailslater.
The position reference is compared to the slider actual position, thus each motor control
generatethecorrespondingvoltagespacevectorinrotatingframe,thentheyarecalculated
inthestationaryframebymeansofeachslidermeasuredposition.Finally,thesereferences
enterintheblockoftheinverseClarketransformation,whichgivesthemodulatingsignals
fortheinverterlegs.StatorcurrentsaremeasuredusingLEMsensors,andtheyareusedto
calculatethefirstandthethirdcurrentspacevectors.
EachmotoriscontrolledbytheblockcorrespondingMMmotorcontrolinFigure8,in
particulareachblockcontainsthecontrolschemeshowninFigure9.
Theoutercontrolloopconcernsthepositioncontrol:thesliderposition,afterachange
of measure units, is compared to the actual position and the error passes through a P
regulator.Furthermore,afeedforwardregulator,withkgain,actsonthesubsequentnode,
improving the dynamic behaviour of the drive. Then, the speedregulator elaborates the
error signal resulting from the speedreference, the actual speed and the feedforward
action.Then,theinnercontrolloopconcernsthe qcomponentofthecurrent,viz.therated
force,whilethe dcomponenthasanullreference,sincethemotoroperateswithinitsbase
speed range. Finally, the block for the backemf compensation provides the voltage space
vectorinrotatingframe.
rectifier
M1position
reference
Xref,1
m,1
vq2
vd2
v1
d-q
ma
v1
d-q
m,2
m,2
M2motor
control
vq1
vd1
v2
v2
C-1
mb
mc
md
me
5-phase
inverter
Xref,2
M1motor
control
m,1
M2position
reference
ia, i b, ic, id, ie
M1
M1 measured
slider position
M2
M2 measured
slider position
Figure8Blockdiagramofthecontrolscheme.
88|P a g e
E. Tuningoftheregulator
Thetuningoftheregulatorsofthepositioncontrolscheme,namelytheregulators(a),
(b)and(c)inFigure9,willbeexaminedwithreferencetothemotorparametersof
TableII.
Thetuningoftheregulators(a)and(b)isquitesimpleifthebackemfaresupposedto
be compensated, as shown in Figure 9. The tuning for both regulators is usually obtained
withzeropolecancellations.
Figure10showstheBodediagramoftheopenlooptransferfunctionfromtheoutput
of the regulator (a) to the current iqm, the Bode diagram of the regulator (a) and of their
product. The zero of the PI regulator (a) is selected so that it cancels the system pole at
lower frequency, due to the stator resistance and the inductance, whereas the gain is
selectedsothatthephasemarginisabout85degrees.Thepoleathighfrequencyisdueto
thepresenceoftheinverterandtoothernonidealbehaviours.
For the tuning of the regulator (b) the aforementioned methodologies can not be
applied,sincethemechanicalpoleisnotknownbecauseofthelackofinformationaboutthe
viscous friction coefficient. Thus an experimental tuning has carried out, fixing the zero at
the pulsation of 1/15 rad/s and increasing the gain till a satisfactory speed response is
reached. The same methodology has been used to set the gain of the position regulator,
keeping inactive the feedforward action. The gain of the position regulator is 50. Finally,
actinginparameterkofthefeedforwardaction,thepositionresponseisoptimizedinorder
tohavenegligiblepositionerror.
ref,M +
-
P(c)
ref,M +
m,M
PI(b)
Iq-ref,M +
PI(a)
Iq-m,M
m,M
Xref,M
Id-ref,M +
-
PI(a)
vq,M
vd,M
Id-m,M
Compensationof
thebackemf
ks
Lowpass
filter
(d-q)1
ab
c de
Ia, Ib, Ic, Id, Ie
m,M
MM measured
slider position
Figure9PositioncontrolschemeimplementedinthetheblockMMmotorcontrolof.Figure8
TableIIMotorparameters.
RS[]
LS[mH]
Mseie[Vs/m]
Slidermass[kg]
6.5
17.7
14.2
1
89|P a g e
Bode Diagram
100
Magnitude (dB)
50
-50
-100
-150
0
Phase (deg)
-45
-90
-135
-180
1
10
10
10
10
10
10
Figure10Bodediagramplotofthetransferfunctionfromtheoutputoftheregulator(a)intheblockdiagram
ofFigure9tothecurrentiq-m(solid),regulator(reddashed)andcompensatedsystem(bluedashed).
Frequency (rad/sec)
5.3 Experimentalresults
In order to verify the effectiveness of the proposed fivephase dualmotor series
connecteddrive,depictedinFigure11,someexperimentaltestshavebeenperformed.
The experimental setup consists of a customdesigned fivephase voltage source
inverter,showninFigure12,feedingafivephasedualmotorseriesconnecteddrive.
Figure11Fivephasedualmotorseriesconnecteddrive.
90|P a g e
Figure12Fivephaseinverter.
ThecontrolalgorithmisimplementedinaDigitalSignalProcessor(DSP)TMS320F2812.
Theswitchingperiodis100s,correspondingtoaswitchingfrequencyof10kHz.
The position references used to test the multimotor drive have sinusoidal waveform
settableinfrequencyandamplitude.
A. Startuptest
(a)Experimentaltest.Behaviourofthemultimotor
drive with growing position references. Yellow: M1
actual speed (5(m/s)/div). Fuxia: M2 actual speed
velocity (5(m/s)/div)). Green: M1 actual position
(20cm/div).Ciano:M2actualposition(20cm/div).
Figure 13Start up test: the position references are

dynamicbehaviourofthecontrolscheme.
(b) Experimental test. Behaviour of the multimotor

drivewithdecreasingpositionreferences.Yellow:M1
actual speed (5(m/s)/div). Fuxia: M2 actual speed
(5(m/s)/div)). Green: M1 actual position (20cm/div).
Ciano:M2actualposition(20cm/div).
increased and then decreased in order to test the
91|P a g e
B. Positionreferencevariation
Figure14Experimentaltest.Behaviourofthemultimotor
drive with reduction of the M1 position reference, while
M2 position reference is null. Yellow: M1 actual speed
(5(m/s)/div). Fuxia: M2 actual speed (5(m/s)/div)). Green:
M1 actual position (20cm/div). Ciano: M2 actual position
(20cm/div).
C. Velocityvariation
(a)Experimentaltest.Behaviourofthemultimotor
drive during an acceleration of both motors, for a
fixed position amplitude of the sinusoidal position
references. Yellow: M1 actual speed (5(m/s)/div).
Fuxia: M2 actual speed (5(m/s)/div)). Green: M1
actualposition(20cm/div).Ciano:M2actualposition
(20cm/div).
(b) Experimental test. Behaviour of the multimotor

drive during an deceleration of both motors, for a
fixed position amplitude of the sinusoidal position
references. Yellow: M1 actual speed (5(m/s)/div).
Fuxia:M2actualspeed(5(m/s)/div)).Green:M1actual
position (20cm/div). Ciano: M2 actual position
(20cm/div).
Figure 15Velocity variation test: the position references have a fixed amplitude, while the velocity
(freqeuncy)isincreasedandthendecreased.
92|P a g e
D. Independentcontroltest
Figure16Experimentaltest.Behaviourofthemultimotor
drive with sinusoidal position reference for M1, while M2
position increase its frequency (velocity). Yellow: M1
position references (20cm/div). Fuxia: M1 actual position
(20cm/div). Green: M2 position references (20cm/div).
Ciano:M2actualposition(20cm/div).
5.4 Conclusion
The effectiveness of the fivephase dualmotor series connected drive with tubular
actuatorhasbeenproved,andtheconsistencyofthecontrolschemehasbeenvalidatedby
someexperimentaltests.
The position control implemented permits to have an independent control of each
motor, characterized by a low position error. Further improvement can be achieved using
motorwithanumberofslotperpoleandperphasegreaterthan3.
93|P a g e
Conclusions
Linearelectricmachineareappliedinagrowingnumberofindustrialapplications,since
theycanbeeasilyintegratedinthestructureoftheapplication,furtherenablingthedirect
drivecontrol.Amongthevariouslinearmotortopologies,thetubularstructureseemstobe
particularlyattractiveforindustrialpurposesduetobothitsclosedformandtheinherently
absence of attractive force between the stator and the mover. Furthermore, technologic
developments in soft and hard magnetic materials have solved the feasibility problems of
theironcoreandhaveenhancedtheforcedensityofsuchkindofactuator.
To fully exploit the aforementioned technologic developments and to obtain high
performancemotors,accuratemotordesignhavetobedeveloped.Thus,thedesignerhasto
payattentionnotonlytotheelectromagneticissues,butalsotothethermalandmechanical
ones.
Nowadays, multiphase drives exhibit excellent feature, especially for the industrial
sector.Thelargernumberofdegreesoffreedomcomparedwiththreephasemotordrives
enable the implementation of new and promising drive structure, such as the so called
multimotormultiphasedrive.Multiphasemachineswithspecialdesigncanbeconnectedin
series and then independently controlled with a single multiphase inverter, resulting in a
more compact structure, with less inverter legs than the threephase counterpart.
Furthermore,thecentralizedcontrolofsuchastructurereflectsinahighperformancedrive.
The features of the tubular linear actuator and that one of the multiphase drives can be
combinedinordertoobtainautomatedmachinewithbettercharacteristics.
In this Thesis work, a mathematical model of the electromagnetic, thermal and of
mechanicalissueshasbeendevelopedforPMtubularlinearactuators.Ithasbeenusedasa
startingpointforanoptimaldesignalgorithmdefinition.Theproposeddesignequationscan
be used to investigate the behaviour of a the motor as a function of some crucial
parameters, such as the airgap length, the total volume, etc.. The design methodologies
describedinthisThesisrevealtobeeffectiveandprecise.Validationwasobtainedbymeans
ofexperimentalresultscarriedoutonaprototypeoftubularlinearmotor.
Thecontrolstrategiesformultimotordriveshavebeendefined,andtheeffectiveness
of a position control of a fivephase dualmotor series connected drive with tubular
actuatorshasbeenverified.Theindependentcontrolofasuchamotorshasbeenverified
throughdynamictests,whichalsodemonstratetheconsistencyofthecontrolscheme.
95|P a g e
BIBLIOGRAPHY
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
E.Masada,LineardrivesforindustrialapplicationinJapanhistory,existingstateand
futureprospective,ProceedingsofLIDIA95,Japan,pp.912.
AmitavaBasak,PermanentmagnetDClinearmotors,OxfordUniversityPress,1996,
ISBN0198593929.
HamidA.Toliyat,G.B.Kliman,HandbookofElectricMotors,CRCPress,2004,ISBN
0824741056.
I.Boldea,S.A.Nasar,Linearelectricactuatorsandgenerators,CambridgeUniversity
Press,1997.
G.W.McLean,Reviewofrecentprogressinlinearmotors,IEEProc.B,1988,135,(6),
pp.380416.E.R.
Laithwaite, Linear electric machines A personal view, Proceedings of the IEEE
Volume63,Issue2,Feb.1975Page(s):250290.
J.F.Eastham,Novelsynchronousmachines:linearanddisc,IEEProc.B,1990,31,pp.
4958.
I. Boldea, Linear electromagnetic actuators and their control: a review, EPEPEMC
2002,Dubrovnik&Cavtat,Croatia,911Sept.2002,P.1P.13.
J.F.Eastham,F.Profumo,A.Tenconi,G.Gianolio,Lineardriveinindustrialapplication:
stateoftheartandopenproblems,ICEM2002,Bruge,Belgium,2528August2002.
MarignettiF.,ScaranoM.ALinearTubularActuatorForRobotArm.INES2001IEEE
International834,ConferenceonIntelligentEngineeringSystems.(vol.1,pp.267272)
A.K.Buding,Precisionengineeringdemandsfornearlineardrivesinthemicrometer
andsubmicrometerrange,Proc.I.Mech.Eng.,March1985,pp.203213.
K.Park,E.P.Hong,andK.H.Lee,Developmentofalinearmotorforcompressorsof
householdrefrigerators,inProc.LDIA2001,Nagano,Japan,2001,pp.283286.
W.R.Cawthorneetal.,Developmentofalinearalternatorengineforhybridelectric
vehicle applications, IEEE Trans. Veh. Technol., vol. 48, no. 6, pp. 17971802, Nov.
1999.
P. Cancelliere, F. Marignetti , V. Delli Colli, R Di Stefano, M. Scarano, A Tubular
Generator For Marine Energy Direct Drive Applications, IEEE International Electric
MachinesandDrivesConference,May1518,2005,SanAntonio,TX.
DansonM.JosephandWillieA.Cronje,DesignandAnalysisofaDoubleSidedTubular
Linear Synchronous Generator with Particular Application to WaveEnergy
Conversion, IEEE PES PowerAfrica 2007 Conference and Exhibition Johannesburg,
SouthAfrica,1620July2007
D.Howe,Z.Q.ZhuandR.E.Clark,StatusoflineardrivesinEurope,InvitedPaperat
LinearDrivesforIndustryApplications(LDIA2001),Nagano,2001,pp.468473.
97|P a g e
Bibliograpy
[17] V. Di Dio, M. Montana, State of art of tubular linear induction motor,

ElectrotechnicalConference,1996.MELECON'96.,8thMediterranean,Volume1,13
16May1996,Page(s):285288vol.1.
[18] G. Stumberger, B. Stumberger, D. Dolinar, Position and current dependent flux
linkages, thrust and friction force of linear synchronous reluctance motor, Industry
ApplicationsConference,2001.ThirtySixthIASAnnualMeeting.ConferenceRecordof
the2001IEEE,Volume4,30Sept.4Oct.2001,Page(s):23102317vol.4.
[19] I.BoldeaandS.A.Nasar,Thrustandnormalforcesinasegmentedsecondarylinear
reluctancemotor,Proc.IEE,vol.122,no.9,pp.922924,1975.
[20] Jang SeokMyeong, Park JiHoon, You DaeJoon, Cho HanWook, Sung HoKyung,
Designofhighspeedlinearswitchedreluctancemotor,InternationalConferenceon
ElectricalMachinesandSystems,2007.ICEMS.811Oct.2007,Page(s):16681671.
[21] Lee ByeongSeok, Bae HanKyung, P. Vijayraghavan, R. Krishnan, Design of a linear
switchedreluctancemachine,IEEETransactionsonIndustryApplications,Volume36,
Issue6,Nov.Dec.2000,Page(s):15711580.
[22] N.S. Lobo, Hong Sun Lim, R. Krishnan, Comparison of Linear Switched Reluctance
Machines for Vertical Propulsion Application: Analysis, Design, and Experimental
Correlation,IEEETransactionsonIndustryApplications,Volume44,Issue4,JulyAug.
2008,Page(s):11341142.
[23] T.Ishii,Elevatorsforskyscrapers,IEEESpectr.,vol.31,no.9,pp.4246,Sep.1994.
[24] Sang Heon Lee, Yoon Su Baek, Contactfree switched reluctance linear actuator for
precision stage, IEEE Transactions on Magnetics, Volume 40, Issue 4, Part 2, July
2004,Page(s):30753077.
[25] Husain,MinimizationoftorquerippleinSRMdrives,IEEETrans.Ind.Electron.,vol.
49,pp.2839,Feb.2002.
[26] N. Bianchi, S. Bolognani, J. Corda, Tubular linear motors: a comparison of brushless
PM and SR motors, International Conference on Power Electronics, Machines and
Drives,2002.(Conf.Publ.No.487)47June2002,Page(s):626631.
[27] J. Corda, Cylindrical linear stepper motor, IEE Colloquium on Stepper Motors and
TheirControl,25Jan1994,Page(s):8/18/4.
[28] D. Casadei, C. Rossi, G. Serra, A. Tani, Optimal design of a tubular hybrid stepping
motorundervolumetricconstraints,Proc.ofSPEEDAM2004,Capri(Italy),June1618,
2004,pp.857863.
[29] D.J. de Groot, C.J. Heuvelman, Tubular linear induction motor for use as a servo
actuator, Electric Power Applications, IEE Proceedings B, Volume 137, Issue 4, July
1990,Page(s):273280.
[30] J.F. Eastham, R. Akmese, H. C. Lai, Optimum design of brushless tubular linear
machines,IEEETrans.onMagnetics,Vol.26,No.5,September1990,pp.25472549.
[31] J. Wang, G.W. Jewell, d. Howe, Design optimisation and comparison of tubular
permanentmachinetopologies,IEEProc.Electr.PowerAppl.,Vol.148,No.5,Sept.
2001,pp.456464.
98|P a g e
[32] L.ElAmraoui,F.Gillon,A.Castelain,P.Brochet,M.Benrejeb,Experimentalvalidation
ofalineartubularactuatordesign,ICEM2002,Bruge,Belgium,2528August2002.
[33] P.J. Hor, Z.Q. Zhu, D. Howe, J. ReesJones, Minimization of cogging force in a linear
permanentmagnetmotor,IEEETransactionsonMagnetics,Volume34,Issue5,Part
1,Sept.1998,Page(s):35443547.
[34] C.A. Borghi, D. Casadei, A. Cristofolini, M. Fabbri, G. Serra, Application of a multi
objectiveminimizationtechniqueforreducingthetorquerippleinpermanentmagnet
motors,IEEETrans.onMagnetics,Vol.35,No.5,September1999,pp.42384246.
[35] T. Yoshimura, H.J.Kim, M. Wanata, M. Torii, D. Ebihara, Study on the reduction of
detentforceofpermanentmagnetlinearsynchronousmotor,IEEETrans.onMagn.,
vol.31,no.6,pp.37283730,Nov.1995.
[36] Z.Q.Zhu,X.P.Xia,D.Howe,P.H.Mellor,Reductionofcoggingforceinslotlesslinear
permanentmagnetmotors,IEEProcElecfr.PowerAppl.,vol.144,no.4,pp.277282,
July1997.
[37] J.Y.Hung,Z.Ding,Designofcurrentstoreducetorquerippleinbrushlesspermanent
magnetmotors,ElectricPowerApplications,IEEProceedingsB,Volume140,Issue4,
July1993,Page(s):260266.
[38] A.Canova,G.Gruosso,M.Otella,M.Repetto,N.Schofield,Softmagneticcomposites
fortubularlinearactuators,ICEM2002,Bruge,Belgium,2528August2002.
[39] J. Wang, G.W. Jewell, d. Howe, Design optimisation and comparison of tubular
permanentmachinetopologies,IEEProc.Electr.PowerAppl.,Vol.148,No.5,Sept.
2001,pp.456464.
[40] Lars Hultman, Zhou Ye, Soft Magnetic Composites Properties and Applications,
PMTEC2002inOrlando,USA.
[41] LarsHultman,MatsPerssonandPerEngdahl,SoftMagneticCompositesforAdvanced
MachineDesign,PMAsia2005,inShanghai,China.
[42] LarsO.Hultman(HgansAB),AlanG.Jack(UniversityofNewcastleuponTyne,United
Kingdom), Soft Magnetic Composites Motor Design Issues and Applications,
PMTEC2004,inChicago,USA.
[43] Jiabin Wang, D. Howe, Influence of soft magnetic materials on the design and
performance of tubular permanent magnet machines, IEEE Transactions on
Magnetics,Volume41,Issue10,Oct.2005,Page(s):40574059.
[44] E.Levi,R.Bojoi,F.Profumo,H.A.Toliyat,S.Williamson,"Multiphaseinductionmotor
driver a technology status review", in IEE Electr. Power Appl., 2007, Vol. 4, No. 1,
pp.489516.
[45] E. Levi, Multiphase Electric Machines for VariableSpeed Applications, IEEE
Transactions on Industrial Electronics, Volume 55, Issue 5, May 2008, Page(s):1893
1909.
[46] G.Grandi,G.Serra,A.Tani,Generalanalysisofmultiphasesystemsbasedonspace
vectorapproach,inProc.of12th PowerElectronicsandMotionControlConference
(EPEPEMC),2006,CDROMpaperT2409.
99|P a g e
Bibliograpy
[47] D.Casadei,F.Milanesi,G.Serra,A.Tani,L.Zarri,Spacevectormodulationbasedona
multidimensionalapproachformultiphaseinverterswithanoddnumberofphases,
Power Electronics Specialists Conference, PESC 2008. IEEE, 1519 June 2008,
Page(s):13511357.
[48] Grandi, G.; Serra, G.; Tani, A., General Analysis of MultiPhase Systems Based on
SpaceVectorApproach,PowerElectronicsandMotionControlConference,EPEPEMC
2006.12thInternational,Volume3,Aug.302006Sept.12006,Page(s):834840.
[49] M.Benatmane,T.McCoy,T.Dalton,andT.L.Cooper,Electricpowergenerationand
propulsion motor development for US Navy surface ships, in Proc. All Electric Ship:
DevelopingBenefitsforMaritimeApplicationsConf.,London,U.K.,Sep.1998,pp.89
94.
[50] P.T.Norton,P.E.Thompson,Thenavalelectricshipoftodayandtomorrow,inProc.
3rdAllElectricShipSymp.,Paris,France,Oct.2000,pp.8086.
[51] E. Levi, Multiphase Electric Machines for VariableSpeed Applications, IEEE
TransactionsonIndustrialElectronics,Volume55,Issue5,May2008,Page(s):1893
1909.
[52] L.Parsa,Onadvantagesofmultiphasemachines,inProc.ofAnnualConferenceof
theIEEEIndustrialElectronicsSociety(IECON),2005,pp.15741579.
[53] S. Williamson and S. Smith, Pulsating torques and losses in multiphase induction
machines,inProc.IEEEInd.Appl.Soc.Annu.MeetingIAS01,Chicago,IL,2001.
[54] H.A. Toliyat, T.A. Lipo, J.C. White, Analysis of a concentrated winding induction
machineforadjustablespeeddriveapplications.I.Motoranalysis,IEEETransactions
onEnergyConversion,Vol.6,Issue4,Dec.1991,pp.679683.
[55] H.A. Toliyat, T.A. Lipo, J.C. White, Analysis of a concentrated winding induction
machine for adjustable speed drive applications. II. Motor design and performance,
IEEETransactionsonEnergyConversion,Vol.6,Issue4,Dec.1991,pp.684692.
[56] E.Levi,M.Jones,S.N.Vukosavic,andH.A.Toliyat,Anovelconceptofamultiphase,
multimotor vector controlled drive system supplied from a single voltage source
inverter,IEEETrans.PowerElectron.,vol.19,no.2,pp.320335,Mar.2004.
[57] L. Parsa, H.A. Toliyat, Fivephase permanentmagnet motor drives, IEEE Trans. on
IndustryApplicat.,vol.41,no.1,2005,pp.3037.
[58] H. Xu, H.A. Toliyat, L.J.Petersen, Fivephase induction motor drives with DSPbased
controlsystem,IEEETrans.onPowerElectron,vol.17,no.4,2002,pp.524533.
[59] Y.Zhao,T.A.Lipo,SpaceVectorPWMControlofDualThreePhaseInductionMachine
Using Vector Space Decomposition, IEEE Transactions on Industry Applications, Vol.
31,Issue5,Sept./Oct.1995,pp.11001009.
[60] S. Lu, K. Corzine, Direct Torque Control of FivePhase Induction Motor Using Space
Vector Modulation with Harmonics Elimination and Optimal Switching Sequence,
AppliedPowerElectronicsConferenceandExposition,2006,March19,2006,pp.195
201.
100|P a g e
[61] L.Parsa,N.Kim,H.A.Toliyat,FieldWeakeningOperationofHighTorqueDensityFive
Phase Permanent Magnet Motor Drives, IEEE International Conference on Electric
MachinesandDrives,May15,2005,pp.15071512.
[62] C.B. Jacobina, I.S. Freitas, T.M. Oliveira, E.R.C. da Silva, A.M.N. Lima, Fault tolerant
controloffivephaseACmotordrive,PowerElectronicsSpecialistsConference,2004,
Vol.5,2025June2004,pp.34863492.
[63] H.A. Toliyat, Analysis and simulation of fivephase variablespeed induction motor
drivesunderasymmetricalconnections,IEEETransactionsonPowerElectronics,Vol.
13,Issue4,July1998,pp.748756.
[64] Y. Zhao, T.A. Lipo, Modelling and Control of a MultiPhase Induction Machine with
Structural Unbalance, Part. I, Machine Modelling and MultiDimensional Current
Regulation,IEEETransactionsonEnergyConversion,Vol.11,Issue3,Sept.1996,pp.
570577.
[65] Y. Zhao, T.A. Lipo, Modelling and Control of a MultiPhase Induction Machine with
Structural Unbalance, Part. II, FieldOriented Control and Experimental Verification,
IEEETransactionsonEnergyConversion,Vol.11,Issue3,Sept.1996,pp.578584.
[66] M.BenatmaneandT.McCoy,Developmentofa19MWPWMconverterforU.S.Navy
surface ships, Proc. Int. Conf. on Electric Ship ELECSHIP, Istanbul, Turkey, 1998, pp.
109113.
[67] C.L.FerreiraandR.W.G.Bucknall,Modellingandrealtimesimulationofanadvanced
marine fullelectrical propulsion system, Proc. IEE Power Electronics, Machines and
DrivesConf.PEMD,Edinburgh,Scotland,2004,pp.574579.
[68] S.Mantero,E.DePaola,andG.Marina,Anoptimisedcontrolstrategyfordoublestar
motors configuration in redundancy operation mode, Proc. European Power
Electronics and Applications Conf. EPE. Lausanne, Switzerland, 1999, CDROM paper
no.013.
[69] S. Mantero, A. Monti, and S. Spreafico, DCbus voltage control for double star
asynchronous fed drive under fault conditions, Proc. IEEE Power Electronics
SpecialistsConf.PESC,Galway,Ireland,2000,pp.533538.
[70] A.N. Golubev, and V. Ignatenko, Anomalous operation of multiphase asynchronous
electricdrive,Russ.Electr.Eng.,2001,72,(10),pp.2228.
[71] A.Iqbal,S.N.Vukosavic,E.Levi,M.Jones,H.A.Toliyat,Dynamicsofaseriesconnected
twomotorfivephasedrivesystemwithasingleinvertersupply,inProc.ofIEEEInd.
Appl.Soc.AnnualMeeting(IAS),2005,pp.10811088.
[72] A. Iqbal, E. Levi, M. Jones, S.N. Vukosavic, A PWM scheme for a fivephase VSI
supplying a fivephase twomotor drive, in Proc. of Annual Conference of the IEEE
IndustrialElectronicsSociety(IECON),2006,pp.25752580.
[73] E.Levi,M.Jones,S.N.Vukosavic,ASeriesConnectedTwoMotorSixPhaseDriveWith
InductionandPermanentMagnetMachines,IEEETransactionsonEnergyConversion,
Vol.21,Issue1,March2006,pp.121129.
101|P a g e
Bibliograpy
[74] D.Casadei,G.Serra,A.Tani,L.Zarri,Generalinvertermodulationstrategyformulti
phase motor drives, Proc. of ISIE, Vigo (Spain), CDROM, pp.11311137, June 47,
2007.
[75] E. Levi, M. Jones, S.N. Vukosavic, H.A. Toliyat, Operating principles of a novel
multiphase multimotor vectorcontrolled drive, IEEE Transaction on Energy
Conversion,Volume19,Issue3,Sept.2004,Page(s):508517.
[76] M. Jones, S. N. Vukosavic, and E. Levi, Independent vector control of a sixphase
seriesconnected twomotor drive, in Proc. 2nd IEE Int. Conf. Power Electronics,
MachinesandDrivesPEMD,Edinburgh,U.K.,2004,pp.879884.
[77] M. Jones, E. Levi, and A. Iqbal, A fivephase seriesconnected twomotor drive with
currentcontrolintherotatingreferenceframe,inProc.IEEEPowerElec.Spec.Conf.
(PESC),Aachen,Germany,2004,pp.32783284.
[78] 8 N. Bianchi, Analytical field computation of a tubular permanentmagnet linear
motor, IEEE Transactions on Magnetics, Volume 36, Issue 5, Part 2, Sept 2000,
Page(s):37983801.
[79] 11 Jiabin Wang, D. Howe, G.W. Jewell, Analysis and design optimization of an
improved axially magnetized tubular permanentmagnet machine, IEEE Transaction
onEnergyConversion.,Volume19,Issue2,June2004,Page(s):289295Artn8.
[80] 9 N. Bianchi, S. Bolognani, F. Tonel, Design criteria of a tubular linear IPM motor,
ElectricMachinesandDrivesConference,2001.IEMDC2001.IEEEInternational2001,
Page(s):17.
[81] 10 N. Bianchi, S. Bolognani, D. Corte, F. Tonel, Tubular linear permanent magnet
motors: an overall comparison, IEEE Transactions on Industry Applications, Volume
39,Issue2,MarchApril2003,Page(s):466475.
[82] 5R.E.Clark,D.S.Smith,P.H.Mellor,andD.Howe,Designoptimizationofmoving
magnet actuators for reciprocating electromechanical systems, IEEE Trans. Magn.,
vol.31,pp.37463748,Nov.1995.Artn8.
[83] 6T.MizunoandH.Yamada,Magneticcircuitanalysisofalinearsynchronousmotor
withpermanentmagnets,IEEETrans.Magn.,vol.28,pp.30273029,Sept.1992.Art
n8.
[84] F. Marignetti, P. Cancelliere, M. Scarano, Analytical formulation of the noload
magnetic field in IPM tubular machines with translator eccentricity, IEEE Industrial
Electronics,IECON200632ndAnnualConferenceon,610Nov.2006,Page(s):829
835.
[85] F. Marignetti, P. Cancelliere, V. Delli Colli, R. Di Stefano, M. Scarano, Analytical
FormulationofRadialMagneticForcesinPMLinearTubularMachineswithTranslator
Eccentricity, Industry Applications Conference, 2007. 42nd IAS Annual Meeting.
ConferenceRecordofthe2007,IEEE2327Sept.2007,Page(s):530536.
102|P a g e

Filippo Milanesi Tesi

Diunggah oleh

Informasi Dokumen

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Filippo Milanesi Tesi

Diunggah oleh

Hak Cipta:

Format Tersedia

University of Bologna

DEPARTMENT OF ELECTRICAL ENGINEERING

Design optimization and

Final Dissertation in 2009

Figure 2Imaginary process of cutting and squashing a rotary machine to

Figure 3Conceptual transformations to obtain a tubular motor starting to a

magnetic material occurred [16]. Progress of highenergy rareearth magnets, whose

interrupts the circulation of eddy current induced by timevariable magnetic fields. In a

Withtheexceptionfor h = 0 (and h = n / 2 forevennumberofphases),thequantity

Therealvariable x 0 = x 0 isobtainedfor h = 0 andiscalledzerosequencecomponent,

where r = n / 2 1 for an even phase number and r = (n 1) / 2 for an odd phase

sh = Lsh ish + M se,h ie e jh

Furthermore, multiphase PM synchronous motors with surface mounted magnets in

The necessity to control several electrical machines with a coordinated motion is an

It corresponds to the number of independent space vectors obtained with

where wact isthetotalactuatorlength.

where kisthemagnetfluxlinkedwiththekthphase, ikisthecorrespondinglinecurrent

where Pcuisthewindinglosses, K s istheslotfillfactorand isthecopperresistivity.Itis

where Qtotisthetotalpowerlosses, Rconvthethermalframetoenvironmentresistancedue

Q rad = rad S rad (Tenv + f e ) (Tenv )

Figure 4Drawing of the tubular actuator showing the geometrical

Figure 5Scheme of the simplysupported beam for the

Substituting (3.35) in (3.38), the resulting equation can be expressed in Mc Laurens

Thebehaviourof isshowninFigure8asafunctionoftheproduct KL.Accordingto

Equation (3.48) can be solved using NewtonRaphson iterative method to calculate

Slot current [p.u.]

Figure 14Slot current in p.u. (base value:

Slot current [p.u.]

Figure 15Slot current in p.u. (base value:

Normalized end-effect force (p.u.)

Slider position [p.u.]

Slide r position [p.u.]

Normalized thrust force (p.u.)

Normalized thrust force [p.u]

slid er position [p.u.]

v h,ref h(k 1) ,(k=1,2,...,5).

m1 sin1 (k + i 2) sin(k i ) + m3 sin3 (k + i 2) sin(k i ) .

It is very interesting to derive the validity domain of m1 and m3 regardless of the

m1 sin (k i ) + m3 sin (k i ) ,(k=1,2,...,5),(i=1,2,...,5).

Thecorrespondingvaliditydomainof m1 and m3 isrepresentedbytheshadedareain

Figure2Validitydomainof m1 and m3 forfivephasedualmotorseriesconnecteddrive.

ia, i b, ic, id, ie

Ia, Ib, Ic, Id, Ie

Figure 13Start up test: the position references are

(b) Experimental test. Behaviour of the multimotor

increased and then decreased in order to test the

(b) Experimental test. Behaviour of the multimotor

[17] V. Di Dio, M. Montana, State of art of tubular linear induction motor,

Anda mungkin juga menyukai