© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

Motor Design OptimizationIncluding electromagnetic performance

and mechanical stress

Christian Kremers – Dassault Systèmes

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

2

Electric Drive EngineeringElectric Machine

• High energy density

• High speed (20k+ rpm)

• High efficiency

• Acoustic comfort

• Thermal management

Gearbox

• Lightweight structure

• Compact design

• Well lubricated

• Efficient

• Durable

Structural housing

• Protection

• Lightweight structure

• Heat dissipation

• Noise mitigation

• Chassis integration

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

3

Electric Drive EngineeringElectric Machine

• High energy density

• High speed (20k+ rpm)

• High efficiency

• Acoustic comfort

• Thermal management

Gearbox

• Lightweight structure

• Compact design

• Well lubricated

• Efficient

• Durable

Structural housing

• Protection

• Lightweight structure

• Heat dissipation

• Noise mitigation

• Chassis integration

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

4

Electric Drive Engineering

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

5

Long-term Commitment to Simulation

Design

Simulation

2005 20102000

CATIA

Analysis

2017

Abaqus

fe-safe

Simpoe

SimpackCST

Wave6XFlow

Geensoft

Dymola

Isight

Tosca

SFE

EXAOpera

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

6

Long-term Commitment to Simulation

Design

Simulation

2005 20102000

CATIA

Analysis

2017

Abaqus

fe-safe

Simpoe

SimpackCST

Wave6XFlow

Geensoft

Dymola

Isight

Tosca

SFE

EXAOpera

Structural Analysis

System Level Simulation

Process AutomationElectromagnetic Analysis

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

7

Outline Optimization phases

Electromagnetic performance analysis

Mechanical stress analysis

Optimization / design space exploration

Outlook & conclusion

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

8

Electrical Machine Optimization Optimization Goals

Maximize Power-to-weight ratio

Maximize Efficiency

Minimize costs

Design torque-speed characteristic

Constraints

Mechanical stiffness

Thermal limits

Noise and vibration limits

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

9

Electrical Machine Optimization Optimization Goals

Maximize Power-to-weight ratio

Maximize Efficiency

Minimize costs

Design torque-speed characteristic

Constraints

Mechanical stiffness

Thermal limits

Noise and vibration limits

IPMSM, 48 slots, 8 poles,

distributed winding with 8 turns

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

10

Multiphase Optimization

Current waveforms

Phase 1:

• Electromagnetic performance

• Simplified loss calculation

• Mechanical stress

Start

Phase 2.1:

• System level simulation

including controller, inverter

Phase 2.2:

• Detailed loss/force calculation

• Thermal/CFD calculation

• Noise and vibration analysis

Dymola

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

11

Optimization in Phase 1 Goal: Obtaining “good” design candidate(s)

Desired:

Short simulation time per design point (huge design space)

Parametrized CATIA model

on 3DEXPERIENCE

CST

Abaqus

DOE / Optimization

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

12

Electromagnetic Performance Goals:

Maximize torque and power

Minimize torque ripple

Maximize efficiency at selected operating points

Constraint:

Demagnetization under fault condition

IPMSM, 48 slots, 8 poles,

distributed winding with 8 turns

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

13

Electromagnetic Performance Simplified geometry, 2D mesh, Subvolume

1. Uniform current distribution,

loss less materials, linear magnets

2. Individual conductors,

conductive linear magnets

3. Nonlinear magnets

incr

easi

ng c

urre

nt

1+3

2

Efficiency at

Operating Points

Demagnetization

DQ-Model

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

14

Electromagnetic Performance: DQ ModelMain motor parameters are calculated for dq-current range

incr

easi

ng c

urre

nt

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

15

Electromagnetic Performance: DQ ModelMain motor parameters are calculated for dq-current range

incr

easi

ng c

urre

nt

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

16

Main motor parameters are calculated for dq-current range

Electromagnetic Performance: DQ Model

Based on the DQ-Model the torque-speed characteristic is obtained

Pea

k cu

rren

t / A

Ld Lq 𝜓𝑑

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

17

Electromagnetic Performance: DQ ModelDynamic characteristic: Torque vs. speed & Operating Points

Vol

tage

/ V

P

eak

Cur

rent

/ A

OP 1OP 2

OP 1

OP 2

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

18

Electromagnetic Performance: DQ ModelDynamic characteristic: Torque vs. speed & Operating Points

Vol

tage

/ V

P

eak

Cur

rent

/ A

OP 1OP 2

OP 1

OP 2

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

19

Efficiency @ Operating Points Individual conductors Eddy current losses

Eddy currents in magnets

Iron losses based on fitted loss density curves

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

20

Demagnetization Fault condition: 1.5x short circuit current calculated based on DQ-model (rated

torque, no-load flux) applied to negative d-axis Maximal demagnetization

Magnets at 120°C

Voltage reduction

Fault condition

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

21

Mechanical Stress Analysis Stress due to

Press fit mounting of the shaft in the rotor core initial stress

Centrifugal force at 20000 rpm

End effects are ignored

2D simulation

Maximum von Mises stress of 400 MPa is accepted

Hexahedral elements mesh

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

22

Optimization in Phase 1

Goals:

Maximize torque and power

Minimize torque ripple

Maximize efficiency

Minimize costs, i.e. minimize magnet mass

Constraints:

Stress < 400 MPa

Demagnetization voltage reduction ratio > -5%

Parametrized CATIA model on

3DEXPERIENCE

EM

Stress

DOE / Optimization

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

23

Optimization in Phase 1 Latin Hypercube DOE with 2000 iterations has been performed (16 input

parameters)

Single thread run time ≈ 20 min per design point

Workload can be distributed (cores & machines)

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

24

Design Space Exploration: Scatter Plots

0

100

200

300

400

500

600

0 2 4 6 8

Max

imum

Tor

que

/ Nm

Torque Ripple (6th) / %

Design Limit

(Pareto Front)

0

100

200

300

400

500

600

0 1 2 3 4

Max

imum

Tor

que

/ Nm

Magnet Mass / kg

Design Limit

(Pareto Front)

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

25

Design Space Exploration: Self-Organizing Map

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

26

Design Space Exploration: Compare Compare “Good” candidates

Efficiency

Power & Torque

Torque ripple

Cost ≡ Magnet mass

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

27

Design Space Exploration: Compare Compare “Good” candidates

566

67

CoggingTorque

TorqueMax

TorqueRipple 6th

TorqueRipple 12th

PowerMax

MagnetMass

Eff. OP2

Eff. OP1

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

28

Multiphase Optimization

Current waveforms

Phase 1:

• Electromagnetic performance

• Simplified loss calculation

• Mechanical stress

Start

Phase 2.1:

• System level simulation

including controller, inverter

Phase 2.2:

• Detailed loss/force calculation

• Thermal/CFD calculation

• Noise and vibration analysis

Dymola

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

29

Multiphase Optimization

Current waveforms

Phase 1:

• Electromagnetic performance

• Simplified loss calculation

• Mechanical stress

Start

Phase 2.1:

• System level simulation

including controller, inverter

Phase 2.2:

• Detailed loss/force calculation

• Thermal/CFD calculation

• Noise and vibration analysis

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

30

Multiphase Optimization

Current waveforms

Phase 1:

• Electromagnetic performance

• Simplified loss calculation

• Mechanical stress

Start

Phase 2.1:

• System level simulation

including controller, inverter

Phase 2.2:

• Detailed loss/force calculation

• Thermal/CFD calculation

• Noise and vibration analysis

MQS –FD (linear)

MQS –FD (linear)

MQS –TD (non-linear, 2d)

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

31

Summary Fast initial optimization using 2D FEM

KPI’s from mechanical and electromagnetic domain are considered

3DEXPERIENCE platform manages geometry, simulation data and results

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

32

3DEXPERIENCE Conference for Design, Modeling & Simulation 2020

WHEN November 10 - 12, 2020

WHERE Darmstadtium, Darmstadt, Germany

WEB 3ds.com/events/

WE ARE LOOKING FORWARD TO SEEING YOU AGAIN!

© D

assa

ult S

ystè

mes

| C

onfid

entia

l Inf

orm

atio

n |

12/1

2/20

19| r

ef.:

3DS

_Doc

umen

t_20

19

33