Practical Issues In The Manufacture Of High Efficiency Electric Motors

Practicallssuesin the Manufactureof High EfficiencyElectricMotors PeterJohn Tavnerr,PaoloCaselotti2 I TechnicalDirector,FKI EnergyTechnology 2 TechnicalDirector, Marelli Motori S.p.A.

Abstract European motor manufacturersseeking to deliver motor ranges which comply with the current voluntary efficiency agreement have to choose between developinga new range,improving an existingrangeor obtainingtheir rangefiom anothersource. Some manufacturershave introduced new ranges and in general these motors incorporatean increasein active materials,electricalsteeland copper,to increase their effìciency.Theseincreasesin materialsnaturallyincreasethe basecost of the motor. Other manufacturers,including the authors, have taken an existing range and modified aspectsof the designto optimiseperformanceand achievea higher efficiency. Further manufacturershave sought to buy higher efficiency motors from a lower labour cost source,where they can offset the increasein material cost of the higher efficiency motor with a substantialreduction in labour cost for its manufacture.[n the latter case an important componentin achievingthe higher efficiencyis relatedto maintainingthe quality of the deliveredproduct. This paper will describethe issuesassociatedwith optimisingthe efficiency of an existing range of induction motors, utilising changesin design,new materials and optimisedmanufacturingprocesses.tt will pay particularattentionto practical problems which occltr, their elimination and the economic impact of those improvements.

lntroduction The TEFC motor has evolved, over 80 years, with designsto International Standarddimensionsand outputsin a tough, blow-over cooling enclosure,which gives a rugged, interchangeablemachine that can be used in a wide variety of applications.

203 Manufacturers have evolved their products to achieve requirements as economically as possible and a previous paper, Ref 1, has shown that this has achievedmachineswith progressivelyimproving weight-powerratios,as shown in Fig. l, taken from Glew, Ref 2 thereforebecomingintrinsicallycheaper. However, Fig. 1 also shows that the weight-powerof TEFC motors has not reduced at the same rate as larger machines. In recent years there has been a tendency for the weight-powerof TEFC motors to increase.This is due to the constraintsof International Standardsand to the poor heat transfer in the TEFC machine, compared to modern large machines, which have air-blown internal active parts. So it has been progressivelyharder for manufacturersto make TEFC motors more competitively, by lowering the weight-power ratio, without deteriorating the noise and efficiency of the machine. In other words the limit of present knowledge of thermal performance is being reached and economic improvementscan only be achievedby: 1. Lowering performance,or 2.By using more materials,increasingcost to improve performance but then finding a cheaperplace to make the motor, or 3. By increasingpresentknowledge,using modern methodsfor optimal designto achieve better performancewith less material or changingto better but cheaper materials. The challengeof the ef iciency initiatives in North America and the European Union is that option i) is not possible.Manufacturersare deploying a variety of methodsin ii) above,the effect on costshasbeenanalysedby Parasitiliet al in Ref 3 but this was done on the basis that manufacturersincreasedweight/power. We are not aware of any activity in iii). But have we reachedthe true limit of weightpower?This paperbelievesnot. The situation is analogousto that in which Japanesecar manufacturersfound 2lttre, compact themselvesin USA in the 1980s,wheretheir conventionally-sized, The easy demanded. market American the power than less produced engines car solution was to build bigger engines, in the American fashion. However, the Japaneseanswerwas to use the latest design techniquesto get more power out of thà existing 2 litre package. This is now widely acceptedas the appropriate solution and those same engines have shown the way to achieve more stringent environmentalrequirements.This paper follows from Ref I to describewhat can be done practically to address the fundamental problems in design and manufacture fbr an existing range of TEFC motors from an established manufacturer.

InteractionBetween Losses, Efficiency& Weight-Power The lossesin a motor are classifiedas follows:

rt*

-l

204 l. Load DependentLossesincluding: . Joule lossesin the statorwinding; o Joulelossesin the rotor winding; o Straylosses. 2. Load Independentor ConstantLosses,including: o lron loss, o Mechanical lossesincluding fan, bearing and shaft seal. ln small and medium sized motors at normal load the Joule lossesrepresentthe significant proportion of total losses, typically above 50%. Therefore their reductionshould be one of the principal objectivesin raising motor efficiencies. Stray lossesalso representa significant proportion of losses,occurring in both rotor and stator,but they are difficult to predict and thereforeto reduce. Haatajain Ref 4 implied the reduction of lossesis the only way to develop high efficiency motors and this must increasethe volume of active parts and therefore increasethe weight-power. But this is only so if currents, fluxes and materials remain constant. No self-respecting manufacturer improves his machines by keeping everythingthe same.

Heat Transfer & Losses in TEFC Machines The problem with the TEFC geometryis that all lossesfrom the active parts, the core, rotor and stator windings, must be extractedfrom the interior of the machine, passedto the ribbed barrel and removedby convectionby the blow-over air. There are four weaknessesin this arrangement: l. Internal air does not remove much heat from the active parts, so there is little assistanceto heat transferfrom internal fluid circulation. 2. The heat from the rotor to the ribbed barrel crosses 3 thermal resistance interfaces,the air gap, the stator insulation and the core/barrelinterface. 3. The lossesfor smaller motors are dominatedby Joule lossesdevelopedin the windings which is mostly extractedacross2 of theseinterfaces. 4. A large proportion of the second major source of loss, the strays, are concentratedin the rotor and must be extractedacross3 of theseinterfaces. 5. Production factors can worsen the thermal resistance across each of the interfaces. 6. The convective heat transfer acrossthe final interface, the ribbed barrel, isnot optimal. In consequencethe bulk heat transfer coefficient for a TEFC motor, relating total lossesto stator winding temperaturerise and motor surfacearea, is typically 15 W/m4K, much lower than that for larger electrical machines,typically 300 W/m4K.

205 This paper is arguing that one can lower the weight-power, by raising the bulk heat transfercoefficient in the machinefrom 15 W/m4K. This cóuld increaseloss density, with the threat of reducing efficiency. But lowering temperature will decreaseloss. If the materials& designare also improved then efficiency can be increased.The following sectionsdescribesthe changesthat can be made.

What to Do ? To reacha higherperformancecompetitivelyone must optimise: 1. Performanceparameterssuch as: o Losses& efficiency. . Noise. o Starting current & torque. 2. Productionfactors r Improvedprocessesto reduceloss. r Improve the choice of core frame diameters. o Lower the cost of productionby removingunnecessary processes. To improve the variability of design & manufacturingprò."rr", and materials the following areasfor optimisationshouldbe addressed: L Designcalculations; o Identificationand predictionof stray losses; o Thermal& ventilationmodellingto improveperformance. 2. Materials; o Use improvedsteels; o Reducematerial variance; 3. Processes; o Better impregnationto improve heat transfer; o Limit tool wear to reduceloss& wear variance; r Improved statorwinding to reduce loss; o Improved rotor cage construction; o Improved core/barrelfit.

Reducing Losses It has been shown that Joule lossesare an important part of machine loss. These can be reducedby using high-permeability,rather than low-loss, steels,punched

206 steel, with sharptools. Additional benefit can be obtainedby using semi-processed annealedafter punching,to eliminatethe punchinglosses. It is also clear that one of the crucial problemsfor reducingstray losseslies in the rotor from which heat has to be removedacross3 thermal interfaces. A significantcontributorto theselossesis high frequencycurrentcirculationin the smooth surfaceof the rotor, due to stator slot harmonics.Usually rotors are machined after diecasting the cage in the closed slots. Such machining is frequentlydone at high speedin one cut, leadingto distortion and burring of the rotor lamination edges,enhancingthe lossesdue to these currents.Experiments have shown that by reducingthe speedand cut size it is possibleto reducethese surfacelosses. There are also papersthat addressthe problem of stray lossesdue to rotor slot skew.For exampleWilliamson et al, Ref 5, has shownthe substantialimprovement in efficiency that can be achieved by eliminating interbar currents, Fig 2. Experimentswith motors without skew, where the rotor and stator slots numbers have beenchosento avoid noise and cogging,show a significantreductionin stray losses,which reducesmotor temperatureand raisesefficiency.

Optimising Heat Transfer There are some referencesthat addressthe heat transfer weaknessesof TEFC machines,for example Pickering et al Ref 6. This dealt with the major area for improvement, the heat transfer from the ribbed body, which depends upon the dimensionsof the rib in relation to the frame and the airflow over the ribs. This airflow should be high enoughfor scrubbingaction but not so high as to reduce heat transfer.The fan dimensionsneedto be optimisedin relationto the fan-cowl, which must have a good exit shapeover the ribs. The work describedby Hayward in Ref 7, on the motor rangeconsideredhere, showedhow the fan, fan-cowl and rib dimensionscan be optimisedby modelling and test. Fig. 4 shows the experimental Air Flow Test Rig based on the current range of motors on which ComputationalFluid Dynamics predictionswere also made. Fig. 5 showsgood comparisonbetweenthe full size Air Flow Test Rig and the CFD predictions, confirming the modelling. The CFD was then used to optimise rib dimensions and air flow so that the heat transfer coefficient on that surfacecould be improved.

Productionlssues An importantissuein productionprocessis to limit the variability of efficiency values attributable in a population of motors to periodic variations in tools,

207 this to be a vital issue in improving processesand materials. we consider ttti#;ffie

production issuesof reducing section has referred to the important iool' and if possible eliminating such core lossesby maintaining sharp punching punching' lossesby annlaling the laminationsafter stray losses by improving rotor reduce can We have also seen that we directions rather than one severecut' A machining,using two light cuts in opposite the machining at the stator outer similar effect can be achieved by'minimising there, away,from the air gap' is small' diameter but the stray loss intensity high the core back is important to ensurea However the resultantsurfacefinish on machining the is as to the cast stator frame' heat transfer coeffrcient from the core Ra 1'5 are recommended' of figures of the frame itself.Roughness is the quality and repeatability of transfer heat An important issue for i m p r e g n a t i o n o f t h e s t a t o r w i n d i n g . T h e u s u a l . c r i t e r i o nhigh forim p r erepeatable gnationisthe and impregnation with a dielectric withstand but of .or.rrr". coeffrcientof heat transferis essential' effect of manufacturing processeson There are some papers that addressthe thesematters,for exampleby Walters'

Conclusions doesnot necessarilymean increasing Improvingthe efficiency of TEFC motors th e v o l u me o fma te ri a l a n d thereforecost,butdoesnecessi tateopti mi si ngdesi gn are: and production factors' The areasto attack 1 ' R e d u c e c o r e & J o u l e l o s s e s b y o p t i m i s i n g m a g n e t i c s t e epunching' lmaterialtogivehigh by annealingafter permeabilityand eliminate put.ttittg losses the the rotor by reviewing machining and 2. Reduce stray losses particufutf' oí are selections slot stator and rotor that need to skew the slots, making sure problems' optimisedto avoid noiseand cogging give good heat transfer particularly in io pro".rré, manufacturing 3. lmprove resPectof: o o p ti m i s i n g ro to rma c h i ni ngtomi ni mi serotorsurfacel ossesandguaranteea rePeatableand accurateatrgaP' to ensure good stator to o lmprove stator core and J,u,ot frame machining framefit. rl mp ro v e s ta to rw i n d i n gi mpregnati ontoensuregoodheattransfer. o Reducingvariability in manufacturingprocess' fan, fan-cowl and rib design to 4. Optimise siator frame cooling UV uppÀpriate stator' . 1r ,raisethe convectiveheat transferat the . with a low weight-power ratto' can motors of range existing The result is that an 6' now alsoachievea higher effrciency'Fig'

208 These motors have also been incorporatedinto integratedproducts, which can et al, obtain high driven equipmentefficienciesfrom invertersupply,seeShakweh Ref 8. Acknowledgements This paper is publishedwith the permissionof the Directors of MarelliMotori SpA and FKI EnergYTechnologY.

References P J, High efficiency motors - performance'economy t1] Caselotti P, Conchetto A, Tavner 20-22 September & reliability, by optimisation,2ndEEMODS Conference,London, Verlag' 1999,Springer PP95-101. of energy efficient and environmentally friendly l2l Glew N, Design unJ -unrfucture largemachines.IEE Colloquium 1999,ppl ll-ll5' the designoptions and cost impact of improving l3l ParasitiliF, villani M, Evaluationof Conference,London, 20-22 September EEMODS 2"d efficiency, motor induction 1999,SpringerVerlag,PP5l4-528' three phase induction motor efficiency in Europe' t4] HaatajaJ, Py,rhonenJ, Irnproving EMD Conf, cambridge, 1997, Publn 444, ppl90IEE the challengefor manufacturers, 194. of interbar currentson harmonic lossesand skew t5] Williamson S & Smith AC, Influence pp 369-374' in cage induction motors, IEE PEMD Conf, &ath,2002, Publn 487, the stator end from transfer Heat T F, Roylance t6] Pickering S, Lampard D, Hay N, EMD Conf, IEE motor, induction wound concentric voltage windings of a low ' 471-481 412, Publn 1995, Durham, PP of an Electric Motor, Nottingham University, Ul HayrvardJ, The Ventilation Performance MSc ExternalProject,F34SP6,2000' Miller H, Plug and play integratedmotor drives' IEE t8l ShakwehY, Owen G H, Hall D J, 655-661' 487,pp Publn Bath,2A02' Conf, PEMD

209 Wt/Power, kg/kW

50 40

;tui.

30 20

1

,*GSNERAL I

l0 0 1900 1920 1940 19ó0 1980 2000 Year Fig. l. Weight per unit output as a function of year, from Glew

o

----------

Doubleskew Nomlskry Halfskw Noskw

Fig. 2. Effect on motor efficiency of reducing rotor skew, taken from Williamson

Fig. 3. Example of Finite Element Analysis of the Flux Distribution in an existing Design of 4 pole Motor, Showing the Opportunity to Optimise the Flux Pattem

2t0

{

Fig. 4. Photographof an Air Flow Test Rig usedto Optimise Ribbed FrameHeatTrans\er 6.5{s+00 5.89e+00 5.23e+00 4 58e+00 3 93s+00 3.270+00 2.62e+00

. :

1 96e+00 1.31400 54€-01

I x é

r VeloolyVeclorsColoredByVeloqly[4agnitude lm,s)

'5 May 200, ]

5113d lLyElT Ts!'glll! !"|

Fig. 5. Top-SmokeTest in the Original Geometry;Bottom-CorrespondingCFD data from FLUENT, showing closeagreementwith Air Flow Test Rig.

Fig. 6. Exampleof a new high efficiencymotor optimisedfrom anexistingrange