Considering Power Electronics. Mark Solveson Application Engineer

Transcription

1 Electric Machines Considering Power Electronics Mark Solveson Application Engineer 1

2 Outline Machine Design Methodology Introduction RMxprt Maxwell Advance Capabilities Core Loss Demagnetization / Magnetization Field Circuit Co Simulation Maxwell Circuit Editor Simplorer Capabilities, Switches, IGBT Characterization Simplorer Examples Multi Physics Force Coupling Thermal Coupling 2

3 Introduction: Machine Design Methodology 3

4 Maxwell Design Flow Field Coupling ANSYS CFD Fluent RMxprt Motor Design HFSS Maxwell 2 D/3 D Electromagnetic Components ANSYS Mechanical Thermal/Stress PExprt Magnetics Field Solution Model Generation 4

5 D2D GAIN J Simplorer Design Flow System Coupling ANSYS CFD Icepack/Fluent Simplorer System Design RMxprt Motor Design IA PMSYNC Torque A IB ICA: A A IC PP := 6 A HFSS, Q3D, SIwave PExprt Magnetics ANSYS Mechanical Thermal/Stress Maxwell 2 D/3 D Electromagnetic Components Model order Reduction Co-simulation Push-Back Excitation 5

6 Analytical solution RMxprt Initial Motor Design 16 different Motor/Generator types Input data geometry, winding layout saturation, core losses comprehensive results machine parameters performance curves 6

7 RMxprt Motor Design Parametric Sweep: Stack_Length Skew/no Skew Stator_ID AirGap Monitor: Torque Power Efficiency Determine the Best Design Create FEA Model Export Circuit Model 7

8 Integrated EMDM Foundations Auto Setup Maxwell lldesign from RMxprt 8

9 Maxwell/RMxprt V15 Axial Flux Machine AC or PM Rotor Single or Double Side Stator Sample Inputs Sample Outputs 9

10 Maxwell/RMxprt V15 Axial Flux Machine Maxwell 3D auto setup (Geometry, Motion, Master Slave, Excitations, etc. ) 10

11 Design Exploration Maxwell Project P1 cond Workbench Schematic P2 parallel 11

12 Design Exploration 12

13 Design Exploration Six Sigma 13

14 Integrated Motor Solution More Than 30 UDP Machine Components for 2D and 3D 14

15 RMxprt Dynamic Link to Simplorer 15

16 Maxwell TRW / Ansoft 1.40 Position & Current Hysteresis Control Close/Open1 Curve Info Position Coil Current 3.00 Diode Current Position [mm m] Coil Current [me eter] Time [ms]

17 Automatic Adaptive Meshing 17

18 Advanced Capabilities Coreloss Computation 18

19 Lamination Core Loss in Time Domain Instantaneous hysteresis loss 1 db db ph( t) khbm cos Hirr dt Instantaneous classic eddy current loss dt 1 db ( t) k 2 c 2 dt Instantaneous t excess loss p c 2 where p e ( t ) 1 C k c db dt 2 C e cos 0 e 1.5 / cos d 2 21

20 Observation As a post processing step Including eddy, hysteresis with minor loop and excess losses Applicable to soft ferromagnetic and power ferrite materials Practical as it is based on available manufacturerdddata sheet provided Computed core loss with time 22

21 Core Loss Effects on Field Solutions Basic concept: the feedback of the core loss is tk taken into account by introducing i an additional component of magnetic field Hin core loss regions. This additional component is derived based on the instantaneous core loss in the time domain 23

22 Model Validation by Numerical Experiment The effectiveness of the model can be validated by the power balance experiment from two test cases: considering core loss feedback and without considering core loss feedback. The increase of input electric power and/or input mechanical power between the two cases should match the computed core loss Lo oss (W) Loss (W) Three phase h transformer Three phase motor Time (ms) Core loss 2 Input power increase Input power increase Core loss Time (ms) 25

23 Advanced Capabilities Demagnetization Modeling 26

24 Modeling Mechanism The worst demagnetization point for each element is dynamically determined from a full transient process The demagnetization point is source, position, speed and temperature dependent Each element uses its own recoil curve derived at the worst demagnetization point in subsequent transient simulation B K p Recoil lines Worst demagnetizing point Hc 0 Br Br' H 27

25 Irreversible Demagnetization If a demagnetizing point P goes below the knee point K, even after the load is reduced or totally removed, the subsequent working points will no longer along the original BH curve, but along the recoil line. B Br Br' K p Recoil line Hc 0 H 28 The animation shows how the demagnetization permanently occurs with varying load current

26 Temperature Dependent Model Work on intrinsic B i H, instead of B H curve B = B i + μo H Two temperature t dependent d parameters: Remanent flux density B r and Intrinsic coercivity H ci 2 1 T T 0 2 T T 0 Br ( T ) P ( ) 2 )1 T T T T H ( T ) Q( Br ( T ) Br ( T ) 1 0 H ci 0 T ) ( T) Hci( T ci 0 0 T where T 0 is the reference temperature, and α 1, α 2, β 1 and β 2 are coefficients which are provided by vendors 30

27 Temperature Dependent Model Once a model at T 0 is constructed, any B i H curves at other T can be recovered in terms of P(T), Q(T) B H curve in the 2nd and 3rd quadrants can be further recovered by B = B i + μo H Copied from vendor datasheet Derived from our implemented temperature dependent model 31

28 Benchmark Example 8-pole, 48-slot, 50 KW, 245 V, 3000 rpm Toyota Prius IPM motor with imbedded NdFeB magnet Two steps in 3D transient FEA: 1. Determine the worst operating point element by element during the entire transient process 2. Simulate an actual problem based on the element based linearized model derived from the step 1 To further consider the impact of temperature, elementbased average loss density over one electrical cycle is used as the thermal load in subsequent thermal analysis The computed temperature distribution from thermal solver is further feedback to magnetic transient solver to consider temperature impact on the irreversible demagnetization 32

29 H c ' change in one element during a transient process: The 1 st cycle (0 to 5ms) doesn t consider temperature impact. The 2nd cycle (5 to 10ms) has considered d the feedback kfrom thermal solution based on the average loss over the 1st cycle Observation: H c' has dropped from 992,755 A/m to 875,459 A/m, which is derived from the worst operating condition 33

30 Contours of loss density distribution Static temperature distribution (K) 34

31 Torque profiles showing demagnetization and temperature dependence: 35 Torque profiles derived from without considering demagnetization, considering demagnetization but no temperature impact and considering demagnetization as well as temperatures dependence

32 Magnetization Compute magnetization based on the original non remanent B H curve Find operating point p from nonlinear solutions Construct line b at the operating point p, which is parallel to the B Slope of line a at saturation point line a at saturation point Br is the intersection of line b with B axis Element by element Br p Line b 36 0 H

33 What is the Difference between Using Magnetostatic and Transient solver? Magnetostatic case: the operating point used for computing magnetization (Br) is from single source point; B Br p Transient case: the operating point used for computing magnetization (Br) is the maximum operating point with the largest (B,H) during the entire transient simulation. 0 B Br p H 0 H 37

34 Anisotropic or Isotropic Magnetization Anisotropic magnetization: magnetization direction is determined by the orientation of the magnet material and the direction is specified by a user; Isotropic magnetization: magnetization direction is determined by the orientation of the magnetizing field and is determined during the field computation. P(T) input Q(T) input For isotropic magnetization, all three components have to be set to zero 38

35 Field Circuit Co simulation 39

36 Co simulation Mechanism Thevenin equivalent (impedance matrix, source voltages) Convert node to loop FE Simulator Lumped field parameters (inductances, induced Internal voltages) ETA ETB ETC J UA.VAL TH11 TH12 TH13 UB.VAL UC.VAL TH14 TH15 TH16 StfTachoShaft R w TH24 TH25 TH26 TH21 TH22 TH23 M StfMotorShaft ww E w J Norton equivalent (conductance matrix, source currents) STF DCMP STF MasTacho J := 0.15m DcmpMotor J := 2.1m Circuit Simulator MasCouplingLeft J := 0.9m 40

37 Maxwell Circuit Editor Example Commutator bar: model position WidC WidB (a) (b) (c) (d) Commutating model: model parameters G LagAngle Period Gmax WidC-WidB b c a d 0 WidC+WidB Position 41

38 Case Example for Commutating Circuit PMDC Motor Winding currents Torque Brush commutation circuit 42

39 Simplorer: Power Electronics 43

40 Simplorer Technology Highlights 44

41 State of the Art Drive System: A Multidomain Challenge ANSYS provides a comprehensive toolset for multidomain work: Drive systems Simplorer conservative structures (electrical circuits, mechanics, magnetics, hydraulics, thermal,...) Simplorer non conservative systems (blocks, states, digital, n th order differential equations. Drive components Maxwell with motion and circuits RMxprt and PExprt (incl. thermal) Maxwell with ANSYS Thermal. HFSS, Q3D, SIwave with circuits (Designer/Nexxim), ANSYS Mechanical, ICEPACK, etc.... A11B11C11 A12 M A2 B12 3~ B2 J ROT C12 C2 ROT1 ROT2 ASMS ROT1 STF ROT2 ROT1 ROT2 ROT1 M SV = RS STF ROT2 J ROT ROT1 ROT2 I GAIN GAIN I GAIN GAIN LIMIT LIMIT 45

42 Multi Domain System Simulator Electrical circuits Analog Simulator Magnetics Mechanics Hydraulics, Thermal,... J A11B11 C11 A12 B12 C12 3~ M A2 B2 C2 MMF M(t) STF GND L + ROT1 ROT2 JA - ASMS F(t) m STF GND H Simplorer Simulation Data Bus / Simulator Coupling Technology State space Models Block Diagram Simulator State Machine Simulator Digital/VHDL Simulator INV 46 x Ax Bu y Cx state EIN SET: TSV1:=1 SET: TSV2:=0 SET: TSV3:=0 SET: TSV4:=1 (R_LAST.I <= I_UGR) (R_LAST.I >= I_OGR) AUS transition SET: TSV1:=0 SET: TSV2:=1 SET: TSV3:=1 SET: TSV4:=0 PROCESS (CLK,PST,CLR) BEGIN IF (PST = '0') THEN state <= '1'; ELSIF (CLR = '0') THEN state <= '0'; ENDIF; INV PST J Q K Flip flop QB CLR

43 Electromechanical Design Environment Matlab RTW UDC MathCAD Matlab Maxwell Simulink Co Simulation C/C++ Programming Interface (FORTRAN, C, C++ etc.) Simulation Data Bus/Simulator Coupling Technology Maxwell Circuits Block Diagram State Machine VHDL AMS Model Database Electrical, Blocks, State Machines, Automotive, Hydraulic, Mechanics, Power, Semiconductors 47

44 Analog Circuit Simulator Multi domain simulation example Digital Electrical Electrical supply Digital Control bjt1 bjt2 ctrl1 CTRL1 BS=>Q Digital control ctrl2 75 CTRL2 BS=>Q TRIG Mechanical / fluid - + TRIG Battery PLUNGER behavioural models I DETECT plunger_control 75 A Solenoid p1 p p2 Solenoid m F em_force spring F spacer gravity v alue := *9.8 plunger s0 := m := limit accumulator orifice Mechanical sul := sll_ := 0.0 Hydraulic 48

45 Multi Physics Co Simulation Transient Electromagnetic FEM Co simulation Maxwell 2D/3D Digital Digital Control CTRL1 BS=>Q ctrl1 Electrical bjt1 bjt2 CTRL2 BS=>Q ctrl2 75 TRIG TRIG PLUNGER DETECT I - + Battery 75 A plunger_control Solenoid p1 p p2 Solenoid m F em_force plunger s0 := m := orifice Future: Multidomain model extraction and co simulation spring F spacer gravity v alue := *9.8 limit accumulator Mechanical sul := sll_ := 0.0 Hydraulic 49

46 Semicondutor Modeling In Simplorer IGBT Device model Semiconductor device model on Simplorer IGBT Device model : Average / Dynamic Capability of IGBTmodel Thermal management for Inverter Thermal model in Simplorer s semiconductor model. Extract thermal network from ANSYS Icepak Heat / Power loss coupling with device model Inverter surge and conduction noise Extract parasitic LCR from Q3D Extractor Inverter surge and conduction noise in Simplorer 50

47 Semiconductor device model in Simplorer Ideal switch model ON:short, OFF:open Semiconductor system level Modeled as nonlinear resistance in consideration of a static characteristic. Semiconductor devicelevel Dynamic characteristics, therma and physical characteristics are modeled. BJT / MOSFET /JFET / IGBT / Diode / Thysistors SPICE compatible spice 3f5 compatible MOSFET (spice3 Lv.1 6, BSIM1 4, EKV,JFET) 51

48 IGBT model 1) 2) 1. System model Nonlinear resistance verification of operation 3) 4) 2. Average model Static char. & average loss. Heating & temp. rise 3. Basic Dynamic model Dynamic char.& Energy Switching loss & heating. 4. Advanced Dynamic model Detailed dynamic char. Inverter surge & noise 52

49 IGBT Characterization Average model is developed for system simulation and is integrated t dinto the extraction ti tool Common thermal model is used among the IGBT family members 53

50 Average IGBT model A switching waveform (current and voltage) is systematic. Calculate a switching loss for every cycle. DC loss and turn ON/OFF loss pulse is an input to a thermal network. Losses compute as an averaged rectangle pulse. A thermal network is calculable in the independent sampling time. PON/POFF switching loss EON/EOFF switching energy loss PDC conduction loss TON/TOFF turn on, turn off time Vce,sat collector emitter saturation voltage. 54

51 Dynamic IGBT model Static characteristic modeled the same as Average model. Switching energy is derived by the integration of a current cross voltage waveform. The Dynamic model can obtain an exact switching waveform. It can applies also to EMI/EMC and a noise simulation Eon Eoff (VCE=600V IC=300A VGE=15V T=25 ) 55

52 IGBT device circuit model Internal thermal network Internal equivalent circuit i IGBT junction THERMO_T Zth_IGBT RT1 RT2 RT3 RT4 CT1 CT2 CT3 CT4 ST IGBT chip bottom typ_therm>2 Current, Voltage, Temp., VgeSlope dependency modeled for each capacitance. Independent tail current source. RC snubber are implemented. PT Diode junction PD switches to typ_therm +10 RD1 CD1 RD2 CD2 RD3 CD3 Zth_diode RD4 CD4 THERMO_D SD Diode chip bottom impedance to ambient RDT CDT 56

53 IGBT Characterization 57

54 IGBT inverter design Circuit design (loss) + thermal model Ambient temperature = 20 cel Package temperature 1T 1D 1T, 1D junction temperature Examination of temperature cycle Line current 1T, 1D SW loss + DC loss 58

55 Simplorer + Icepak = Detailed modeling of thermal system Q3D Extractor CAD Import ANSYS Icepak Parasitism LCR extraction Design of the cooling technique and arrangement Device property and loss consultation Simplorer The simulation in consideration of change of detailed temperature environment Design of substrate radiating route

56 Induction Motor FEA Coupled with Simplorer Frequency controlled speed G_R1 := SA.VAL G_R2 := -SA.VAL G_S1 := SB.VAL G_S2 := -SB.VAL G_T1 := SC.VAL G_T2 := -SC.VAL 1400 rpm 3PHAS A * sin (2 * pi * f * t + PHI + phi_u) ~ ~ PHI = 0 PHI = -120 B6U D1 D3 D5 + V 2L3_GTOS g_r1 g_s1 g_t1 PhaseA1 PhaseA2 PhaseB1 Rotor1 Rotor2 + ~ PHI = -240 D2 D4 D6 g_r2 g_s2 g_t2 PhaseB2 PhaseC1 FREQ := 800 Hz AMPLITUDE := 800 V FREQUENCY := 60 Hz Fed by ac-dc-ac inverter AMPL := 800 PHASE := 0 deg FREQ := 50 Hz AMPL := 500 PHASE := -315 deg PHASE := -75 deg SA SB ICA: LL:=237.56u RA:= m LDUM:=100m CDC:=10m LDC:=10m RDC:=10 VZENER:=650 PhaseC2 Name Value SIMPARAM1.RunTime [s] k SIMPARAM1.TotalIterations 40.51k FEA PHASE := -195 deg SC SIMPARAM1.TotalSteps 10.00k FEA1.FEA_STEPS Current LA.I [A] LB.I [A] LC.I [A] Torque 1.50k 1.00k * LD.I [A] VDC.V [V] 0 0 Speed m m m m m m 60

57 BLDC motor FEA Coupled with Simplorer Inverter fed three phase BLDC motor drive Chopped current control ICA: LL:=922u RA:=2.991 PWM_T:=60 PWM_PER:=180PER I_TARG:=9 I_HYST:= Output torque m 30.00m 1500 rpm Q1 Q3 Q5 RA Ohm LL H sourcea1 Magnet01 + sourcea2 Magnet02 sourceb1 400 V sourceb2 sourcec1 Q4 Q6 Q2 sourcec2 FEA THRES := PWM_T QS1 VAL[0] := mod( INPUT[0],INPUT[1] ) INPUT[1] := PWM_PER GAIN + ANGRAD LA.I QS2 EQUBL 57.3 CONST -30+PWM_PER -LC.I QS3 EQUBL CONST -60+PWM_PER LB.I EQUBL QS4 -LA.I EQUBL Chopped currents QS5 CONST CONST -90+PWM_PER -120+PWM_PER LC.I QS6 EQUBL INPUT := -LB.I THRES1 := I_TARG - I_HYST EQUBL 8.50 CONST -150+PWM_PER m 30.00m m 30.00m m 30.00m

58 SRM FEA Coupled with Simplorer ICA: LL:= u RA:=203m A1 AirRotor rpm + A2 AirRotor2 140 V 100u F B1 B2 C1 C2 FEA Name Value FEA1.FEA_STEPS 1.00k SIMPARAM1.RunTime [s] 6.90k SIMPARAM1.TotalIterations 4.05k SIMPARAM1.TotalSteps 1.00k current L1.I [A] L2.I [A] L3.I [A] E1.I [A] control variable * QA.VAL * QB.VAL * QC.VAL ROTA.VAL[0] ROTB.VAL[0] ROTC.VAL[0] QA QB QC VAL[0] := mod( INPUT[0],90 ) EQUBL EQUBL EQUBL m m 10.00u * FEA1.OMEGA V_ROTB1.TORQUE [Nm] GAIN 57.3 CONST CONST mechanical + ANGRAD m u 1.00m u 1.00m m u 1.00m 62

59 Electric Machine Design: Maxwell Simplorer Co Simulation 3 ph Windings Stator & Rotor Co simulation Permanent Magnets 63 3ph Line Currents Flux Linkages

60 Multi physics 64

61 Multiphysics Coupling through WB Maxwell 3D provide volume/surface forces to ANSYS Structural Solver improvements Surface forces are supported Thermal Stress with Electromagnetic Force load The electromagnetic force density from Maxwell is used as load in Structural 65 Deformation of the stator Deformation of coils

62 Force Coupling Maxwell to Mechanical Tangential Force on Tooth Tips 02_DC-6step_IPM ANSOFT Force (Newtons) Curve Info ExprCache(ToothTipTangent_Full1) ExprCache(ToothTipTangent_2) ExprCache(ToothTipTangent_3) ExprCache(ToothTipTangent_4) ExprCache(ToothTipTangent_5) ExprCache(ToothTipTangent_6) Time [ms] Radial Force on Tooth Tips 02_DC-6step_IPM ANSOFT Force (Newtons) Curve Info ExprCache(ToothTipRadial_Full1) ExprCache(ToothTipRadial_2) ExprCache(ToothTipRadial_3) ExprCache(ToothTipRadial_4) ExprCache(ToothTipRadial_5) ExprCache(ToothTipRadial_6) Time [ms] 66

63 Force Coupling Maxwell to Mechanical Max Deformation vs time Case 1 0% Eccentricity 67 Case 2 50 % Eccentricity

64 Maxwell Couplings 2D/3D Losses Mapped Losses Temperature 68 Forced water cooling Forced air cooling Natural air cooling

65 Two Way CFD Thermal Analysis, R14 CFD Model Temperature Geometry 69 Losses Maxwell Model Mapped Losses

66 Power Loss Mapped into FLUENT Power Loss in windings are not displayed. 70

67 Results Temperature Distribution 71

68 Thank you 72