Considering Thermo-mechanical Modeling and Design of ...publish.illinois.edu/grainger-ceme/files/2014/06/... · Considering Thermo-mechanical Modeling dD i fEl ti lM hi Prof. J. Rhett

CEME Tele-SeminarMonday, December 7, 2009

Considering Thermo-mechanical Modeling d D i f El t i l M hi

Prof. J. Rhett Mayor

and Design of Electrical Machines

yGeorgia Institute of Technology

Atlanta, GA

-0 04

-0.03

-0.02

-0.01

0

0.01

0.02

0.03

0.04

0.05

Temperature Distribution of Stator [C]

245

246

247

248

249

250

251

252

253

-2

0

2

4

6

8

10

12

14

x 10-3

Temperature Distribution of Half of the Tooth [C]

245

246

247

248

249

250

251

252

253

-0.06 -0.04 -0.02 0 0.02 0.04 0.06

-0.05

-0.04

244

245

0.035 0.04 0.045 0.05 0.055

-4

2

244

245

Thermal Management of MPG’s

Key thermal management issues are related to internal gap control and magnetic circuit thermal managementI d i Increased temperatures on magnetic material and stator windings have deep impact on overall system performanceMagnet and Stator temperatures must be Magnet and Stator temperatures must be maintained at <150 to avoid large efficiency losses

1.60

R/Rn

0.80

1.00

1.20

1.40

n , N

/N0

η/η0

0.20

0.40

0.60

R/R

n

0.000 20 40 60 80 100 120 140 160

Temperature ('C)

Case 1: Benchmark study

Benchmarked thermal response of existing MPG design through thermal steady state effectiveness of coolingNatural convection cooling from all surfaces

Max Temps (˚C)

Component 1 2 3

Core 541

Swing Arm 695

Rotor/Stator 351

Case 2: Finned MICSE core

Average engine temperature drops by almost 200˚

Max Temps (˚C)

Component 1 2 3

Core 541 354

Swing Arm 695 516

Rotor/Stator 351 226

Transient Thermal Analysis Correlation Study

Transient coupled thermal-stress FEM models implemented in ALGOR FEA package are utilized to study the thermal response of the systemFLIR A20 Thermal imaging system has been used to determine IR signature of MPG-1 system during start-up transientsMPG-1 system during start-up transientsCorrelation studies between the thermographic data and ALGOR transient thermal analysis have validated the accuracy of the MPG-1 thermal models

IR temperature profile corresponds within 85% of ALGOR simulationT t l t t ithi ± 10˚CTemperatures correlate to within ± 10 C

Outline

1. Introduction2 Review of Machine Design2. Review of Machine Design

1. SM-PMAC Generator Case Study

3. Generic FD Thermal Modeling Approachg pp4. G-FD/FEA Benchmarking Studies5. Experimental Validationp6. Integrated Thermo-electromagnetic MDO 7. Summary & Conclusionsy

6

Conventional Machine Design Methodology

1

Machine Design(SM-PMAC Generator)Electromagnetic

S l t J EM Design T < Trial

No2

Select J EM Design T < Tlimit ? Design

YesThermo-mechanical

7

450W PMAC Generator Design Optimization

Generator scaling study included beneficial scaling effects to achieve optimal power density in 450W PMAC generator designMTD 1 t d i ti i d i hi h fid lit 5

6

7

8

Wi/W

)

LinearPower scaling

( ) ω⋅⋅∝ LDP 2 2

MTD-1 generator design was optimized using high-fidelity Maxwell™/PSPICE™ FEA to simulate expected no-load an loaded performanceExtensive materials selection study considered trade-offs 0

1

2

3

4

5

0 1 2 3 4 5

Nor

mal

ized

Pow

er (W

between high-frequency performance and high magnetic saturation limits for silicon, cobalt and amorphous ironsImproved thermal management of the rotor shaft enable implementation of NdFeB magnets, higher field strength

Gearing (x:1)

Initial scaling studies based on 30W design using fundamental scaling laws

p g , g gover SmCo rotors => increased power density

Material Saturation Flux Density

Core Loss at 400Hz (@ 1T)

Maxwell™ FEA confirmed optimization of stator design for minimum mass

y ( )

M19 – 26 Gauge (0.47 mm) 1.7 T (17 kG) 24.48 W/kg

Cogent NO 005 (0.12 mm) 1.8 T (18 kG) 11.8 W/kg

Metglas™ 2605C0 (0.023 mm) 1.8 T (14 kG) 6.0 W/kg

Hi ® 50 2 2 T (22 kG) 17 64 W/k

8

of stator design for minimum mass without saturation (1.7T in teeth)

Hiperco® 50 2.2 T (22 kG) 17.64 W/kg

MTD-1Gx Swing-optimized PMAC Prototype

Thermo-mechanical design optimization studies resulted in integrated cooling fins and to allow >6A/mm2 current densitiesStator windings were potted with thermally conductive epoxyStator windings were potted with thermally conductive epoxy improve winding thermal management Two 450W PMAC swing-optimized generators were fabricated with different winding configurations for maximum copper fill factor Stator ring and

spider laminates

MTD-1G1 MTD-1G2

Stator outer diameter (mm) 62

spider laminates

( )Rotor outer diameter (mm) 32.5Axial length (mm) 40Rotor Inertia (kg.m2) 2.615 x 10-5

Air Gap (mm) 0.25Number of stator slots 30Number of poles 10Winding AWG 22 bifilarNo. turns per phase 130,130,130 140,130,140Phase Resistance at 100oC (Ω) 0.681 0.61,0.58Max. RMS current (A) 5

9

Mass (kg) 1.4

MTD-1Gx Oscillatory Performance

Oscillatory testing of the MTD-1Gx prototypes utilized a 4-bar linkage to approximate swing-engine motion

MTD 1G2: Maxwell 2D Back EMF vs Time

10

20

30Phase A Maxwell Phase B MaxwellPhase C Maxwell

MTD-1Gx Model Validation Oscillatory Power Testing

Actual power measured at frequencies up to 16Hz Estimated power at 55Hz is >700W based on FEA simulation

-30

-20

-10

0

Volta

ge (V

)

Test Freq. Vrms Power

1 2.15 2.5 3.8

2 4 5.0 15.1

simulation

0 0.02 0.04 0.06 0.08 0.1 0.12

Time (sec)

MTD-1G1 No-load 8.4Hz

20

30Phase APhase BPhase C

Simulated Back EMF at 8.66Hz3 8.66 10.3 63.2

4 16 18.9 213.5

55 ~700

-20

-10

0

10

Volta

ge (V

)

10

-300.000 0.020 0.040 0.060 0.080 0.100 0.120

Time (s)

Measured Back EMF at 8.4Hz (MTD-1G1)

MTD-1Gx Rotational Performance

M TD-1G1 Rotational Test

30

40

50Phase 1 Phase 2 Phase 3

MTD-1G2 Power vs Speed (5ohm)y = 5E-05x2 + 0.0249x

R2 = 0.9994

1000

1200

-30

-20

-10

0

10

20

Volta

ge (V

)

400

600

800

Pow

er (W

)

-50

-40

-0.010 -0.005 0.000 0.005 0.010Time (s)

MTD-1G2 Thermal response40.00 140.00

MTD-1G2 Cooling effect120.0

0

200

0 500 1000 1500 2000 2500 3000 3500 4000

Speed (rpm)

20.00

25.00

30.00

35.00

age

(V) 80.00

100.00

120.00

Tem

p (°

C)

Phase 1 rms voltageWinding Temp

60.0

80.0

100.0

mpe

ratu

re ('

C)

300W - Ambient300W - Cooling

300W400W

500W600W

0 00

5.00

10.00

15.00

Volta

0 00

20.00

40.00

60.00

Win

ding

0.0

20.0

40.0

Win

ding

Tem

11

0.000 1000 2000 3000 4000 5000 6000 7000 8000 9000

Time (sec)

0.000 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

Time (s)

Generator Summary

MTD-1Gx Specifications

Rated Power (W) 600

Rated Current rms (A) 6.25

Max Power (W) 750( )

Max Current rms (A) 7.5

Power Factor 0.84

Efficiency 85%y

Max Winding Temp (ºC) 135

Dimensions (LxWxH) 5” x 3” x 3”

Mass (kg) 1.4

Max Specific Power (W/kg) 535.7

12

1

Machine DesignMachine Design(SM-PMAC Generator)Electromagnetic

2

Select J EM Design T < Tlimit ?

Trial Design

No2

Sequential Design Process is fundamentally not suited to

coupled multi-physics problem

Advantage Disadvantage

limit

YesThermo-mechanical

coupled multi physics problem

13

Lower J Lower copper loss, cooler Thicker wire, need larger slot, larger weight

Higher J Thinner wire, less slot, less weight Larger copper loss, hotter,

Outline




14

Rationale

• Thermal management in electric machines is a critical design issue

• Computational techniques to evaluate high fidelity temperature distributions in temperature sensitive electrical machines are required in thesensitive electrical machines are required in the design stage

• The proposed model uses a finite difference p papproach to accurately and quickly simulate steady state and transient heat transfer in electrical machineselectrical machines

• This study will consider PM machines where the rotor does not contribute to thermal effects

15

Transient Thermal Modeling Approaches

Quasi-Transient: Using an applied constant

t thi t i t

ture

Quasi-Transient

current this transient simulation type shows how the temperature changes with time

Tem

pera

t

Fully Transient

with time

Fully Transient:Using time varying current

Time

yUsing time varying current (defined by either IEC standards or user input) this simulation shows how the temperature changes with time

16

Existing Thermal Modeling Techniques

• Classical thermal electric machine design uses various simplifications to approximate the motor as a cylinder to carry out thermal analysisy y

• More recent advances in thermal modeling of electrical machines include Thermal Circuits and FEA

Thermal Circuits FEA

Ad tAdvantages

Disadvantages

17


18

[Boglietti, A.; Cavagnino, A.; Staton, D., "Determination of Critical Parameters in Electrical Machine Thermal Models," Industry Applications, IEEE Transactions on , vol.44, no.4, pp.1150-1159, July-Aug. 2008]

Case 3: Ducted Air Flow

Air flow ducted axially along body with flow at 2 m/s92mm cooling fan adds g1.18W power draw and 80 g to system massThermal management objectives are achieved with minimal power and mass penalties

Max Temps (˚C)

Component 1 2 3

Core 541 354 251

Swing Arm 695 516 412

Rotor/Stator 351 226 146


• Classical thermal electric machine design uses various simplifications to approximate the motor as a cylinder to carry out thermal analysisy y

• More recent advances in thermal modeling of electrical machines include Thermal Circuits and FEA

Thermal Circuits FEA

Ad t Mi i l C t ti l ti • High AccuracyAdvantages • Minimal Computational time High Accuracy• Generic

• Cumbersome setupDisadvantages • Requires experimental data fit

• Low Accuracy (within ± 5 °C)

• Cumbersome setup• Large computational

time

20

Generic Model Design Parameters

• Complete description of the geometry is achieved using parametric model that takes advantage ofusing parametric model that takes advantage of symmetry

• 8 parameters for modeling a PM machine8 parameters for modeling a PM machine

•ΘOuter•ΘToothΘTooth•Θfoot•RStator Outer •RSt t IRStator Inner •Rwinding bottom •Rfoot outer •R

half the tooth pitch

21

•RFoot Inner tooth pitch

Automated Mesh Generation

RStator Outer

RStator Inner

Rwinding

Rf

RFoot Inner

ΘOuter

Rfoot outer

• Start by modeling half the tooth pitch

ΘOuter

ΘTooth

Θfoot

22

Transformation of Conventional Stator Slot Design

Actual Slot Modeled Slot

Actual and Modeled Slot Overlay

[“Maxwell v11 user’s guide”, Ansoft Corporation, June. 2006]

23

Automated Mesh Generation

• 2D polar coordinate mesh is generated using a

Tinf outer , hinf outer RStator Outer

dθis generated using a center node distribution

• Mesh is segmented such th t th b d i f

RStator Inner

Statordθ

that the boundaries of nodes correspond to the parameters of the model

Rwinding

Rfoot outer

Windings

Air

dr

• Boundary conditions applied to mesh RFoot Inner

ΘOuter

Air

Tinf inner , hinf inner

ΘTooth

Θfoot

24

Finite Difference Approach

1st Law Energy Conservation

Steady State Polar Finite Difference Equation

Transient Polar Finite Difference Equation

Heat Generation in Windings Equation

25

Development of Generic Model

T1 CK1,1 K1,n

X =

1 C1

Tn CnKn nKn 1

26

n,nKn,1

Exterior Boundary Conditions

Smooth Natural Horizontal

Smooth Natural Vertical

Smooth ForcedT∞

R

tframe00

00 00

VRequivalent

tframe

0

tframe

0

Finned Natural Finned Natural Finned ForcedHorizontal VerticalFinned Forced

V

00 00

tframe

27

tframetframe


Smooth Natural Horizontal

Smooth Natural Vertical

Smooth ForcedT∞

R

tframe00

00 00

VRequivalent

tframe

0

tframe

0

28


T∞

RRequivalent

Finned Natural Finned Natural Finned ForcedHorizontal VerticalFinned Forced

V

00 00

tframe

29

tframetframe


Smooth Equivalent Resistance

TTb

Requivalent

T∞

R T∞TbRequivalent

Finned Equivalent Resistance

T

Requivalent

T∞Tb

30

Frame Thickness

L00

FrameFace

Mount

LMotor

TMotor

Foot TMotorootMount

TMotor

LMotorLMotor /2Motor

TMotorTMotor /2

T /2 T /2 T /2

31

TMotor /2 TMotor /2 TMotor /2

Outline




32

Numerical Validation of FD Technique

• Mesh convergence tests were performed on a wedge consisting of a single material and compared with results from FEA.

• Comparative performance was evaluated based on accuracy• Comparative performance was evaluated based on accuracy, mesh size, and computational time

0.05

0.06

0.07

190

190.5

0 02

0.03

0.04

189

189.5

0.02 0.04 0.06 0.08 0.10

0.01

0.02

188.5

33

Numerical Validation of FD Technique

Number of Nodes

Computational Time [sec]

Min. Temp [C]

Max. Temp [C]

FEA

331.00 30.00 101.78 105.51562.00 180.00 101.82 105.55

1145.00 780.00 101.84 105.58

Proposed Model

315.00 0.40 101.47 105.20676.00 0.50 101.79 105.521173.00 1.10 101.63 105.37

Analysis

Number of Nodes ~320 ~600 ~1150320 600 1150

Precent error [%] 0.30 0.03 0.20

Time Reduction [%] 98 67 99 72 99 86

34

98.67 99.72 99.86

PM Simulation Studies: Case 1

• Steady State model run with a heat flux boundary condition imposed on the = 10 kW/m3condition imposed on the boundaries that contact the windings

• Maximum temperature

T∞ = 25 Ch = 100 W/m2-K

x 10-3


252

253


252

253

• Maximum temperature within 0.2° C

• Minimum temperature within 0 1° C

q” = 10 kW/m2

247

248

249

250

251

252

247

248

249

250

251within 0.1 C

• Computational Times:• FEA: 60 Sec

T∞ = 225 Ch = 150 W/m2-K

244

245

246

244

245

246

Time red ction 95%

• Proposed Algorithm: 3 Sec

35

Time reduction: 95%


• Transient model run with a50

55Winding Boundary Temperature vs Time

heat flux boundary condition imposed on the boundaries that contact the windings 45

50

e (C

)35

40

Tem

pera

ture

6 A/mm2 4 A/mm22 A/mm2• Transient temperatures

within 0.2° C Computational Times

T∞ = 25 Ch = 10 W/m2-K

= 10 kW/m3

25

30 Algor v19Proposed Algorithm

• Computational Times• FEA: 210 s• Proposed Algorithm:

FEA

0 50 100 150 200 250 30025

Time (min)

Time reduction: 98 5%

3 s

36

Time reduction: 98.5%


• Steady model run with a uniform heat generation imposed on the windings and contact resistance between the windings and stator

T t f d t b ithi 0 1° C

Time reduction: 95.7%

• Temperature found to be within 0.1° C• Computational Times

• FEA: 35 s

T∞ = 25 Ch = 80 W/m2-K

= 10 kW/m3

• Proposed Algorithm: 1.5 s

= 64.8 kW/m3

R = 10-3 m2-K/WFEA - Typical Slot GeometryRt,c = 10 m -K/W

37

FEA - Modeled Slot Geometry

Summary of Results

Computational Time (sec)

FEAProposed

ETime

FEAp

AlgorithmError

Reduction

Case 1 60 3 ± 0 2° 95 00 %Case 1 60 3 ± 0.2 95.00 %Case 2 210 3 ± 0.2° 98.57 %Case 3 35 1.5 ± 0.2° 95.71 %

Average 96.43

38

Outline




39

Class Item Unit Value

General

Pole Number [‐] 8Rated / Max Power [kW] 10 / 20Rated / Max Torque [Nm] 40 /80Rated / Max Speed [rpm] 2450 / 8000p p

Slot Number [EA] 12Core Type [‐] Separated

Core Thickness [mm] 0.35Core Material [‐] RM 8

Stator Core Length [mm] 80Winding Type [‐] Concentric

Stator

Winding Type [ ] ConcentricCoils per phase Winding [EA] 4

Turns per Coil [Turns] 24Resistance [mohm] 4.85

Leakage Inductance [uH] 33

Magnetizing Inductance [uH]L1: 198 L2: 73 3L2: ‐73.3

Rated / Max Current [Apeak] 120 / 250Winding Insulation [‐] H

Rotor

Magnet Material [‐] Nd‐Fe‐BrPM Flux [Wb] 0.0534

Core Thickness [mm] 0.35C M i l [ ] RM 8Core Material [‐] RM 8

Cooling ‐ ‐Natural

ConvectionFrame Material [‐] 6061 Al

DimensionsLength of Frame [mm] 167

Inner/Outer Radius of Frame [mm] 110 / 93.5

40

Inner/Outer Radius of Frame [mm] 110 / 93.5

[9]Youngkook Lee and T. G. Habetler, “Current-Based Condition Monitoring and Fault Tolerant Operation for Electric Machines in Automotive Applications,” International Conference on Electrical Machines and Systems, pp. 2011-2016, October 2007.

Thermal Circuit Approach

Psw – Stator Winding Loss

Psc – Stator Core Loss

41[9]Youngkook Lee and T. G. Habetler, “Current-Based Condition Monitoring and Fault Tolerant Operation for Electric Machines in Automotive Applications,” International Conference on Electrical Machines and Systems, pp. 2011-2016, October 2007.

Calculation of G-FD Model Parameters

Item Unit ValueAngle of Foot [rad] 1.09E‐02Angle of Tooth [rad] 1.20E‐01Outside Angle [rad] 0 2618Outside Angle [rad] 0.2618

Inner Radius of Foot [mm] 57.9Outer Radius of Foot [mm] 59.883Radius of Windings [mm] 60

Inner Radius of Stator [mm] 87.359

42

[ ]Outer Radius of Stator [mm] 100

G-FD Steady-State Temperature Results

Item Unit Case 1 Case 2

Rotating Speed [rpm] 500 1000

Load [Nm] 9 0

q‐axis current [A] 25.4 32.6 95

100


37.06

37.08

d‐axis current [A] ‐1.05 1.8

Copper Loss [W] 5.034 0.072

Core Loss [W] 18 12 37 575

80

85

90

37.02

37.04

Core Loss [W] 18.12 37.5Θ Stator ‐Experimental

[K] 13 20.1

Θ Stator – FD l

[K] 16.6 26.160

65

70

36.98

37

Simulation[K] 16.6 26.1

Θ Stator – FD Simulation (h=10)

[K] 14.2 22.2

55-20-10010203040

36.96

43

G-FD Transient Solution

Case 1: 500rpm, zero load Transient Temp of Stator [C]

5

10

Del

Tem

pera

ture

Sta

tor

0 0.5 1 1.5 2

x 104

0

Time Sec

20

25Transient Temp of Stator [C]

Case 2: 1000rpm, ~20% Load

10

15

Del

Tem

pera

ture

Sta

tor

440 0.5 1 1.5 2

x 104

0

5

Time Sec

Outline




45

Coupled Multi-physics Machine Design

Advantages:

Eliminates costly design iteration steps

Avoids heuristic parameter selection

Multi-physics solution accounts for power supply, ambient, thermal condition, material, and load

( ) 10210110 34 ⋅−⋅+⋅−++⋅=Ζ ANmmVol BHη

46

Temperature Constraint: T<90°C

Results from Preliminary Integrated Design

Machine design specification:

15 kW 1800 rpm 60 HzConv. PSO

15 kW, 1800 rpm, 60 Hz

Temp. Limit: 90 OC

Amb. Temp.: 30 OC

Diameter mm 80 104.7

Length mm 75 87.7

Magnet Length mm 6.5 6.83Forced cooling: 5 m/s

Orientation: Horizontal

Magnet Length mm 6.5 6.83

Volume cm3 440 410

Mass kg 40.4 32.7

C t D it A/ 2 6 5 6 83Current Density A/mm2 6.5 6.83

Efficiency % 94.7 94.8

Temperature C 79.4 89.4V

Torque/Ampere Nm/A 3.06 3.84

Power Density W/kg 371 459

tframe

00

47

frame

Outline




48

Summary

DiaSGapLength

InductanceThichMag• Results have shown that the proposed algorithm is both

AirGap Flux Density

Number of turns per phase

Tooth WidthStator and Rotor Yoke Thickness

accurate and computationally efficient as compared with FEA

• Algorithm is also easy to use Back EMF

Output Power

with minimal setup time• Algorithm was designed for

easy integration with an Current

Current Desnity

Slot Fill Factor

electromagnetic optimization algorithm

• Heuristic current density Design

Parametersselection can be replaced with a fast computational thermal simulation

49

Weigth VolumeLoss

Conclusions

1. A Generic model for thermal analysis of electric machines has been developed

2. A technique for transforming typical stator geometries to a simplified geometry in polar coordinates was developed

3. When base lined against an FEA package, the proposed algorithm has shown an average time reduction of 96% and equivalent accuracy.

4. The direct integration of the thermo-mechanical and electromagnetic physics in the design process has been demonstrated

5. Results from PM design case study show a 20% increase in torque-density over existing techniques

50

References

[1] V. Subrahmanyam, Electric Drives: The McGraw-Hill Companies, INC., 1996.[2] A. Boglietti and A. Cavagnino, "TEFC Induction Motors Thermal Models: A Parameter Sensitivity

Analysis," IEEE Transactions on Industry Applications, vol. 41, pp. 756-763, May/June 2005.[3] M. Baggu and H. Hess, "Evaluation of an Existing Thermal Model of an Induction Motor and its [ ] gg , g

Further Application to an Advanced Cooling Topology," Proceedings of IEEE International Electric Machines and Drives Conference, vol. 2, pp. 1079-1083, May 2007.

[4] Y. K. Chin and D. A. Staton, "Transient Thermal Analysis using both Lumped-Circuit Approach and Finite Element Method of a permanent magnet traction motor," IEEE Africon, vol. 2, pp. 1027-1036 S t b 20041036, September 2004.

[5] S. K. Chowdhury, S. P. Chowdhury, and S. K. Pal, "An Interactive Software for the Analysis of Thermal Characteristics of Capacitor-Run Single-Phase Induction Motors," Electric Power Components and Systems, vol. 29, pp. 997-1011, October 2001.

[6] C Liao C L Chen and T Katcher "Thermal management of AC induction motors using[6] C. Liao, C.-L. Chen, and T. Katcher, Thermal management of AC induction motors using computational fluid dynamics modeling," Electric Machines and Drives, vol. 99, pp. 189 -191, May 1999.

[7] F. P. Incropera, D. P. Dewitt, T. L. Bergman, and A. S. Lavine, "Fundamentals of Heat and Mass Transfer," vol. 6, 2007., ,

[8] Harley, Y. Duan, " Method for Multi-objective Optimized Designs of Surface Mount Permanent Magnet Motors with Concentrated or Distributed Stator Windings." unpublished.

[9] Youngkook Lee and T. G. Habetler, “Current-Based Condition Monitoring and Fault Tolerant Operation for Electric Machines in Automotive Applications,” International Conference on Electrical Machines and Systems, pp. 2011-2016, October 2007.

51

Th k Thank you.

QUESTIONSQUESTIONS?


250

251

252

253

244

245

246

247

248

249

250

52

244

Documents

Considering Thermo-mechanical Modeling and Design of ...publish.illinois.edu/grainger-ceme/files/2014/06/... · Considering Thermo-mechanical Modeling dD i fEl ti lM hi Prof. J. Rhett