Design of Interior Permanent Magnet Machines for Hybrid Electric Vehicles

Size: px

Start display at page:

Download "Design of Interior Permanent Magnet Machines for Hybrid Electric Vehicles"

Augusta Madison Lyons
5 years ago
Views:

1 Design of Interior Permanent Magnet Machines for Hybrid Electric Vehicles S.Stanton Technical Director Z.Tang Application Engineer, Detroit Ansoft Corporation Acknowledgements: William Cai Remy International

2 Stator Lamination and Winding 2004 Toyota Prius Power = 50 kw Bus Voltage = 200 V Locked Rotor: Torque = 350 N/m. Current = 250 Amps. Speed ~ 3600 rpm (up to 6000 rpm).

3 Rotor Lamination 2004 Toyota Prius Q Axis D Axis Groove for Permanent Magnet

4 2D Structure 3D Structures Actual Prius Motor More Complex Design: Stacked PM Rotors with Offset

5 Objectives Analytical vs. FEA Approach Proposed Method Geometry creation Reduced FEA modeling techniques Distributed computing Equivalent Circuit Model Complete system analysis Basis: 2004 Prius IPM motor

6 Analytical Method Based off of X d and X q Non-Linear permeability is needed X d R1 Id U cos E0 R1 X q Iq U sin How to accurately take into account for local saturation?

7 Transient FEA Method of choice Dynamic effects Voltage source FEA is the only method to accurately account for local saturation

8 Transient FEA Time consuming when analyzing the full operating spectrum of the load angle During design phase, many topologies are considered Speed = 3000 rpm = 20 degree

9 Trade Offs for IPM Analysis Analytical Fast Calculation of Local Saturation is Challenging Transient FEA Accurate Time consuming

10 Proposed Solution Equivalent Circuit Model Static FEA Solutions System Simulator Reduced modeling techniques Distributed computing Ideal during the design phase of 2D & 3D structures

11 Geometry Creation User Defined Primitives (UDP)

12 Geometry Creation User Defined Primitives (UDP)

13 UDP for Arbitrary Winding Configurations Concentric winding Lap winding with coil pitch=1 Double-layer lap winding DC winding

14 ECE Method 1 Create a State Space type model from a magnetostatic FEA parametric solution IA Input Variables IB IC Output Parameters - La, Lb, Lc - a, b, c - Torque

15 Back EMF System Representation Current Ia, Ib, Ic Inputs: Ia, Ib, Ic, Theta Torque Outputs: EMF: Va, Vb, Vc Torque Where: Theta

16 ECE Method 1 All possible combinations of input variables Sweep Three Phase currents IA, IB, IC: Sweep +/- Peak: 5-9 steps Rotor Angle: 0-90 deg: steps Total number of solution: 5 * 5 * 5 * 31 = minimum 9 * 9 * 9 * 61 = higher fidelity

17 ECE Method 1 All possible combinations of input variables Sweep Three Phase currents IA, IB, IC: Sweep +/- Peak: 5-9 steps Rotor Angle: 0-90 deg: steps Total number of solution: 5 * 5 * 5 * 31 = minimum 9 * 9 * 9 * 61 = higher fidelity 2D ~1 minute per solution: days 3D ~30 minutes per solution: days

18 Proposed Method Assume Wye connected winding IA = Im * cos( * PolePairs - ) IB = Im * cos( * PolePairs - 120) IC = Im * cos( * PolePairs - 240) Let phi = * PolePairs - Independent Dependent Iy IB IA Ix IC

19 Proposed Method, Full Sweep Im from 0 to Imax: 4-9 steps Beta from 0 to 360 deg: 12 steps Rotor Angle: 0-90 deg: steps Total number of solutions: 4 * 12 * 31 = Compared to * 12 * 61 = Compared to Can further be improved for rapid design

20 Periodic Flux Linkage 15 Waveform can be reconstructed from 15 deg

21 Periodic Flux Linkage 15 Waveform can be reconstructed from 15 deg

22 Periodic Torque Ripple Waveform can be reconstructed from 15 deg

23 The Power of Scripting 15

The Power of Scripting Torque Flux A Flux B Flux C -3.73E-03-4.81E-05 2.75E-02-2.74E-02 1.67E+00-3.31E-03 2.91E-02-2.58E-02-4.81E+00-6.09E-03 3.03E-02-2.42E-02-1.99E+01-8.74E-03 3.09E-02-2.20E-02-2.

24 The Power of Scripting Torque Flux A Flux B Flux C -3.73E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E+01 * * * -2.53E-02 * * * E-02 * * * -3.21E-03 * * * VB Script Java Script Excel

25 User Defined Simplorer Models

26 Optimized Input Variations Im from 0 to Imax : 4-9 steps Beta from 0 to 330 deg: 12 steps Rotor angle: 0 to12/13.5 deg: 5-10 steps Total number of solutions: 4 * 12 * 5 = Compared to * 12 * 10 = Compared to 6588 Good for rapid design using DSO

27 Maxwell 3D DSO Setup Each Design has 72 variations for a total 720 total variations Just click on Analyze All!

28 DSO Distributed Solve Option Host Nodes Computation Characteristics CPU time 622 hours 10 min Real time 66 hours 40 min Speed-up factor 9.3 Number of Nodes: 10 CPU type: Intel Dual Processors 2.8 GHz, 2 Gig RAM OS: Windows XP Average mesh size 79,000 elements

29 V_DC*cos(PP*(Rotor.PHI-pi/12)-phi) Simplorer ECE Circuit RA + VA V RB A B C Im_IN Torque_OUT beta_in TTheta_IN T T w + V_DC*cos(PP*(Rotor.PHI-pi/12)-2*PI/3-phi) VB RC Im FL_PMA_OUT FL_PMB_OUT FL_PMC_OUT ECE - LINK V_DC*cos(PP*(Rotor.PHI-pi/12)-4*PI/3-phi) VC ICA: ICA: phi := delta * pi/180 period := 1/freq Turns := 72 phi := delta * pi/180 period := 1/freq PP := 4 V_DC := := 200*0 speed_rpm speed_rpm := 6000 := 3000 delta := 90 0 L_End := 1u RPhase := := 170k 17m freq freq := speed_rpm := * PP*2 * / PP*2 120 / 120 L_End := 1u PhA_IProbe TR Probe V_Line_Line TR Probe Torque_Probe TR Probe Torque Phase Current RA.I [A] RB.I [A] RC.I [A] IPM_SM.Torque_OUT 75.00m 80.00m 85.00m m 80.00m 85.00m

30 Simplorer ECE Circuit IA APos ANeg L_End_A A EMF_A + V L_AA E1 10 L_AA 10 KAB M_AB 10 KAC M_AC L4 I4 10 Lambda_PM_A 10 Turns macro model A B C BPos BNeg L_End_B IB A + V EMF_B L_BB E2 10 L_BB M_BC 10 KBC L5 I5 10 RampA 10 L_End 10 Lambda_PM_B IC A Im CPos L_End_C EMF_C + V L_CC E3 10 L_CC L6 I6 10 Lambda_PM_C CNeg EQU Ixa := IA.I Ixb := IB.I * cos(2*pi/3) 10 Im 10 beta Iya := 0 Iyb := IB.I * sin(2*pi/3) 10 Theta gain1 GAIN 10 Theta_Deg Ixc := IC.I * cos(4*pi/3) Iyc := IC.I * sin(4*pi/3) Ix := Ixa + Ixb + Ixc Iy := Iya + Iyb + Iyc Im := sqrt(ix^2 + Iy^2)*Turns / 1.5 beta := (atan2 (Ix, Iy)) * 180/PI

31 Line to Line Back EMF Simulated Tran. FEA Equiv. Ckt. Measured: 250 V/div

32 BEMF (V) Back EMF vs. Speed Simulated vs. Oakridge Testing - BEMF, 2004 Prius IPM FE-Peak Test-Peak ECE Peak SPEED (RPM) 2% < Error < 8% Computation Time Transient 2D FEA: ~4 Hours Equivalent Circuit Method 2D: 10 min

33 Sweep Load Angle; Speed Held Constant at 3000 rpm 140 Torque (N-m) Torque_Probe.MEAN Phase Current (RMS) PhA_IProbe.RMS Torque (N-m) 0 Phase Current (RMS) Load Angle (degrees) 0 Solution Time: 1 Hour

34 Future Work: Complete System Analysis

35 Complete System

36 Conclusions New Methodology Reduced Modeling Techniques Ia, Ib, Ic, Im,, Distributed Processing FEA during Design Phase: Rapid Design Very Good Accuracy Same Methodology for other Motor Types

37 References Report on Toyota/Prius Motor Torque Capability, Torque Property, No-Load Back EMF, and Mechanical Losses, J. S. Hsu, Ph.D., C. W. Ayers, C. L. Coomer, R. H. Wiles Oak Ridge National Laboratory Report on Toyota/Prius Motor Design and manufacturing Assessment J. S. Hsu, C. W. Ayers, C. L. Coomer Oak Ridge National Laboratory Evaluation of 2004 Toyota Prius Hybrid Electric Drive System Interim Report C. W. Ayers, J. S. Hsu, L. D. Marlino, C. W. Miller,G. W. Ott, Jr.,C. B. Oland Oak Ridge National Laboratory

Maxwell v Study of a Permanent Magnet Motor with MAXWELL 3D: Example of the 2004 Prius IPM Motor. Motor Application Note 11.

Maxwell v Study of a Permanent Magnet Motor with MAXWELL 3D: Example of the 2004 Prius IPM Motor. Motor Application Note 11. Study of a Permanent Magnet Motor with MAXWELL 3D: Example of the 2004 Prius IPM Motor Study of a Motor The Electro Mechanical software package provided by Ansoft enables extensive motor simulation. This