Titelmasterformat Klicken bearbeiten ANSYS Solutions durch for Electric Drives Daniel Bachinski-Pinhal, Johannes Raitmair, Wolfgang Freiberger, Christof Gebhardt, Jochen Haesemeyer
© CADFEM 2014
ANSYS Conference & 2nd CADFEM Ireland Users’ Meeting 24th & 25th September 2015 - Engineers Ireland, Dublin
Agenda
Component & Assembly Design
System Simulations
Customized Solutions
Electromagnetic
Power train
Stretching the limits
Circuits
Automatization
Signal integrity
Behavioral models
Thermal Mechanical Fluids
© CADFEM 2014
ANSYS Workbench
© CADFEM 2014
Workflow & Process Management • Flexible • Scalable
© CADFEM 2014
• Parametrised • Reproducible
Agenda
Component & Assembly Design
System Simulations
Customized Solutions
Electromagnetic
Power train
Stretching the limits
Circuits
Automatization
Signal integrity
Behavioral models
Thermal Mechanical Fluids
© CADFEM 2014
Electric Drives – Design Process
Analytical
Electro-magnetic
Thermal
Stress & vibration
Fatigue & lifetime © CADFEM 2014
Analytical Design • Easy and fast to configure • Datasheet/specification data as input • Winding editor: Definition of winding schemes
Source: ANSYS Inc.
© CADFEM 2014
Electromagnetic Field - Full Range • High-performance software package • Uses finite element analysis (FEA)
• Static solution • Magnetostatic • Electrostatic
• Time-Dependent • Time domain (Transient) • Frequency domain (AC) • Electric, Magnetic, Electromagnetic
Source: CADFEM
• Numerical modeling • 2D(Planar, Axisymmetric) • 3D
• Motion • Linear motion • Rotational motion • (cylindrical,non-cylindrical)
© CADFEM 2014
-8-
Electric Drives – Design Process Analytical
Electro-magnetic
Thermal
Stress & vibration
Fatigue & lifetime
© CADFEM 2014
Thermal - CHT Simulation • Losses from EM simulation used as realistic loads • Fluid flow & heat transfer in single coupled simulation • Detailed modelling of physical effects
© CADFEM 2014
- 10 -
CFD – Realistic Loads • Coupling of ANSYS Maxwell with ANSYS CFD • Electromagnetic analysis computes electromagnetic power loss • Coupling to CFD for considering heat transfer by surrounding fluid
Maxwell: Power Loss
CFD: Thermodynamics
© CADFEM 2014
Thermally Induced Motor Failure • Winding overheat à most common cause of motor failure • Hot-Spots usually not accessible for measurement • Accurate simulation requires detailed winding geometry • ANSYS equivalent material data model
Acessible for sensors
Hot-Spot “inside“ structure
• Detailed simulation of piece of winding • Extraction of equivalent material data (anisotropic model)
• Objective: Reduce simulation time maintaining result accuracy
Source: CADFEM GmbH
© CADFEM 2014
Thermal – Realistic Loads • Losses from EM simulation as realistic loads • Transient and steady-state heat transfer analysis with estimated or CFD-computed heat transfer coefficients • 2D-3D interpolation
Source: CADFEM GmbH
© CADFEM 2014
Thermal – Efficient Fluid Models • Components often cooled by fluid flow • Fluid flow by detailed CFD requires computing power
Oil flow channels
• ANSYS heat pipe model with semianalytical approach
• Objective: Reduce simulation time maintaining global result accuracy • Restriction: Exact knowledge of fluid flow not of interest
Axial air flow Axial Tgradient
Source: CADFEM GmbH
© CADFEM 2014
Example: Predict Electric Motor Performance • Objective: Predict Electric Motor Performance
Maxwell
Fluent
Mechanical
• real-life operating conditions
• ANSYS Solution 200.00
16% drop in predicted performance
Torque (N)
180.00
Y1 [NewtonMeter]
• Electromagnetic loss modeling with ANSYS Maxwell and thermal effects with ANSYS Fluent • Look at further thermal and stress analysis in ANSYS Mechanical
160.00
140.00
123.75 2.00
• Value of Simulation • Magnet temperature was more than 30 K higher than assumed • Combined simulation provides more understanding of product operation © CADFEM 2014
3.00
4.00
5.00
6.00 Time [ms]
7.00
8.00
Time (ms)
Single physics simulation, assuming a magnet temperature of 22 °C 3-way Multiphysics simulation shows that the actual magnet temperature: 53 °C
Temp Depeneded Demagnetization • Temperature determined by simulation • Arbitrary value
© CADFEM 2014
16
Dynamic Temperature Dependent Coupled Demagnetization • Maxwell Transient
1st thermal iteration
• 3A current pulse
• First thermal iteration on original curve
2nd thermal iteration
• Second iteration on lower level
© CADFEM 2014
17
Electric Drives – Design Process Analytical
Electro-magnetic
Thermal
Stress & vibration
Fatigue & lifetime
© CADFEM 2014
Structural - Upfront Improvement • Designer level analysis • • • •
Dimensioning Robust Design Sensitivity Single physics analysis
• Standardized assessments • Workbench FKM • ANSYS nCode DesignLife workflows
• Persistent data structure • Safety through knowledge transfer • Economic usage through customization • Re-use of data
© CADFEM 2014
Source: CADFEM GmbH
Stress & Fatigue Life • Centrifugal and magnetic forces • Nonlinear contact between magnet and steel • Definition of rotor shape according to • Deformation • Stresses • Fatigue life
© CADFEM 2014
Pressfit • Geometric modeling of overlap • Numerical modeling of overlap • Easy to define • Simple variation without touching the geometry
• Nonlinear contact combined with • No additional load • Rotation speed • Thermal deformation
© CADFEM 2014
Time to Frequency Domain Force coupling
ANSYS Workbench Magnetic Field Structural Dynamics Acoustic Field
Forces
Displacements
Source: CADFEM
© CADFEM 2014
22
Vibration of Rotating Electrical Machines • Automated load transfer from magnetic field analysis • Transfer of transient magnetic loads to frequency domain by DFT
• Structural dynamics simulation • Harmonic response based on magnetic forces
• Evaluation of structure borne sound
Sources: CADFEM © CADFEM 2014
- 23 -
Structure Bourne Sound to Air Bourne sound • Result of Structure borne Sound Analysis: Surface Velocity • Integral of Surface Velocity = ERP = Equivalent Radiated Power
ERP = r × c0 × A × v 2 • Airborne Sound is radiated, when the Surface Waves excite Air Waves • Radiation efficiency σ is a Function of Frequency and <1 for low Frequencies [Dr. Neher, MAN Turbo &Diesel]
SPWL = r × c0 × A × v 2 × s à Airborne Sound Analysis is required to get the True Sound Power
© CADFEM 2014
24
CADFEM Test & Simulation correlation solution Simulation: 1295Hz
Simulation Tests: 755Hz
Testing
Calibration
CADFEM can help Nr. Test Sim. Diff.
Opt.
Diff.
§1 Reverse Engineering Solution: 92 112 21% 92 0% 2 3 4 5
© CADFEM 2014
Ø To close the gap between tests and simulation 403 758 88% 399.5 0,8 % Ø To understand and find the unknown parameters 755 1295 71% 753.4 0,2 % Ø To replace Try-and-Error with process automation 1341 51%the key 874.8 1,5of % an unsuccessful Ø888To understand issues 1341 1755 31% 1305 improve 2,7 % design and subsequently the design 25
Rotordynamics • Dynamic response of an elastic rotor key task: avoid instable behavior in operation mode • Instable Behavior – Reasons most common: • internal rotor friction & internal (= rotating) damping mechanisms (!) • Bearing characteristics • Contact between rotor and stator • …
© CADFEM 2014
Rotordynamics in ANSYS: Campbell Diagram Frequency
© CADFEM 2014
27
ANSYS Workbench: Siemens Rotordynamic Simulation • Generator test setup with superconductors 2.5D model of the rotor
• Rot
Modal analysis with 3D model
• Identification of critical speeds, resonance behavior and stability of the rotating machine • Consideration of gyroscopic effects • 2.5 axiharmonic elements combines computation speed and convenient modeling • Transfer of simulation know-how from CADFEM to Siemens by pilot project
© CADFEM 2014
By courtesy of Siemens AG
Campbell diagram: natural frequencies as function of the rotational speed
28
Production History of Injection Molded Parts §
MoldSim container (Project Schematic) - Attach MoldSim container(s) to Setup cell(s) of e.g. Static Structural: Fiber orientation and initial stress data transfer to ANSYS Workbench Mechanical
© CADFEM 2014
- 29 -
Adhesion Films • Modeled by contact/interface elements • Debonding occurs by adhesion film fracture • Debonding start and progression • Limitations: • One-time separation only • No further closing • No fatigue cycling
• Modeled by material • Detailed modeling: viscoelasticity, creep, fatigue… • Stress distribution inside adhesive layer availible • Stress strongly depends on accurate material model © CADFEM 2014
Quelle: Henkel
- 30 -
Structural - Validation & Optimization • Expert level analysis • Virtual prototyping • • • •
Advanced material modelling Nonlinear statics & dynamics Robust Design Coupled physics
• Deployment of simulation methods • Reusable on designer level • Quality assurance
© CADFEM 2014
Structural - Connected Fatigue Life • Established solution from HBM • Certified by Germanischer Lloyd
• Multiaxial assessment • Material libraries • SN & EN libraries
• Load mapping • • • • •
Multi-channel time series Transient Vibration Temperature dependent Duty cycles
• Integration into Workbench
© CADFEM 2014
WB/FKM: Fatigue Assessment of non-welded solid parts • Fatigue Analysis using the local stress approach of FKM Guideline • Standard fatigue evaluation in the German machine building industry
• Fully integrated into ANSYS Workbench • Associative, parametric Workflow for sensitivity and optimization studies • Low effort by automatic determination of local stress gradients • Specific material data base included
*) 6th, revised edition, 2012, VDMA
© CADFEM 2014
WB/FKM-Weld: Fatigue Assessment of welded shell structures • Fatigue Analysis using the nominal stress approach of FKM Guideline • Standard fatigue evaluation in the German machine building industry
• Fully integrated into ANSYS Workbench • Reduced effort by automatic detection of weld connections • Specific material data base included
*) 6th, revised edition, 2012, VDMA
© CADFEM 2014
Agenda
Component & Assembly Design
System Simulations
Customized Solutions
Electromagnetic
Power train
Stretching the limits
Circuits
Automatization
Signal integrity
Behavioral models
Thermal Mechanical Fluids
© CADFEM 2014
System Simulation • Component interaction • Control logic & software • Electronics • Electro-mechanical components
System Component 1
…
Component N
• Multi-Level • Field 1 System • 0-D 1 2-D/3-D
Mechanical Thermal
Mechanical Interaction
Thermal
• Multi-Domain • Electro-thermal • Electro-mechanical • Electro-magnetic compatibility
© CADFEM 2014
Electric
Electric
Controls
Controls
Electric Powertrain System Interactions Controls Battery Inverter
Actuator
Energy Transfer Information © CADFEM 2014
Mechanics
Electric Powertrain - Thermal Simulation Heat-Sink + Fan à Model Extraction from CFD
Power Transistor à Electrothermal Characterisation © CADFEM 2014
Motor à Magnetic FEM Co-simulation
Heat-Sink + Fan - Model Extraction from CFD
1. Geometry & Boundary Conditions
2. 3-D CFD Model 3. Equation Model for “0-D” Simulation
© CADFEM 2014
Motor Magnetic FEM Co-Simulation Solved each time step
System simulation
Field simulation
• Source voltage • Control logic / software • Mechanics, Fluids, Thermal
• Induced voltage • Magnetic force / torque • Losses: Iron, copper, magnet
Solved each time step EQUL_I ERS
EQUL_D ERS
Inertia
LoadTorqueDamping
J
A B C
ROT1
N
ROT2 RMX
STF
LoadTorqueRamp
© CADFEM 2014
System Simulation - Control System - Simulink Co-Simulation Simulink: Control system
© CADFEM 2014
Simplorer Power electronics, actuator, Digital
System Simulation - Control System Embedded Systems Software Development C/C++ code
Compiler
Binary
Hardware Target Real-Time Simulator (HIL, SIL) © CADFEM 2014
System Simulation C interface Compiler
Code Generation
System model • Power Electronics • Actuator • Mechanical/Thermal • Parasitics/EMC • Digital Control (VHDL) • ...
EMC Issues for Power Electronics • Simulation of EMC for Power Electronics is a new discipline • Motivation • Closer proximity of power electronics to humans and sensitive electronics • Increased switching frequencies for loss optimization => More radiated energy • Tighter EMC regulations
© CADFEM 2014
Electro-Magnetic Compatibility • Electric-electric interaction of neighbouring devices • Coupling types: • Radiated: EM wave • Conductive: Capacitive, inductive, resistive
© CADFEM 2014
Electronic Design Issues
Display Panel
EMI/EMC
Power Integrity
Common mode radiation - Differential mode radiation -
Switching Noise (SSO/SSN) - Impedance Profile - Power/GND Plane Resonance - Return Path discontinuity -
Pwr Noise
Control IC
Driver IC
Memory
Signal
PCB
Signal Integrity Power Supplies
© CADFEM 2014
Impedance - Crosstalk - Reflection - Termination - Dielectric Losses -
Agenda
Component & Assembly Design
System Simulations
Customized Solutions
Electromagnetic
Power train
Stretching the limits
Circuits
Automatization
Signal integrity
Behavioral models
Thermal Mechanical Fluids
© CADFEM 2014
Electric Drives – Design Process
Stretching the Limits
Automatization
Behavioral Models
© CADFEM 2014
Example 1: Fast Battery Pack Simulation - BMWI Project
• „Zentrales Innovationsprogramm Mittelstand” (ZIM) • 2-year project à 350,000 € • Project partners • CADFEM • Zentrum für Sonnenenergie- und Wasserstoff-Forschung (ZSW) Baden-Württemberg • LionSmart
© CADFEM 2014
Example 1: Fast Battery Pack Simulation • Car battery pack with 3 layers of 33 cells à 99 cells pack • Each cell is a source of heat • Temperature of the center of each cell was chosen as output
Simulation Information § §
© CADFEM 2014
4000s, adaptive time step 33 min computing time
Example 2: Elastohydrodynamics (EHD) • Bartel, D., IMK, Uni Magdeburg:
Reibungsreduzierung von mischreibungsbeanspruchten Tribosystemen durch Simulation. 8. CADFEM CAE Forum, Stuttgart (2011). • Approximation of Navier Stokes Equation: Reynold’s Approach for Thin Fluid Films in Gaps à Tribo-X
roller on glass plate
fluid EHD result: pressure distribution
structural EHD result: shaft orbit © CADFEM 2014
50
Electric Drives – Design Process
Stretching the Limits
Automatization
Behavioral Models
© CADFEM 2014
Automatizaton – Advanced Vibration Simulation of Electric Drives • Applicaton Examples • • • •
Claw pole machines Skewed teeth Axial flux machines … Source: CADFEM
• Individual force calculation mechanism
EMVIB
1_1_Teilmodell_Test1
6
• More complex load distribution • Periodic re-use of data for improved performance
5 4
Y1
3 2 1 0 -1 4.00
© CADFEM 2014
5.00
6.00
7.00 Time [ms]
8.00
9.00
10.00
- 52 -
• Automated script collection • Has an extension character • Open source • Users encouraged to change and adapt
FluxLinage(d-axis) [Wb]
UDOs – User Defined Outputs
0.2 0 -0.2 -0.4 300
300 200 100
100 0 0 -100 -100 -200 -200 Iq [A] Id [A] -300 -300
• UDOs are scripts to define typical outputs from standard Maxwell data
400
Torque [N.m]
• Dq-Transform • Currents, fluxes, inductances • Matrix multiplication in background
200
200 0 -200 -400 300 200
200
100 0
0 -100 -200
Iq [A] © CADFEM 2014
-200 -300
Id [A] 53
Machine Design Toolkit • Toolkit • Has an extension character • Open source • Users encouraged to change and adapt
• Simulation for motor design • Automatic generation of models with different operation point • Torque Speed Curve Computation • Efficiency Map Computation
Existing Design
Create Reports
User Input Data
Simulate Txn Curve (parametric sweep)
Create designs
Solution Scheme
Current/Gamma Sweep
Retrieve Data from Sweep
• Supports all types of synchronous machines (also with Magnets)
© CADFEM 2014
54
Electric Drives – Design Process
Stretching the Limits
Automatization
Behavioral Models
© CADFEM 2014
Function Models - Enabled by Automatization • Re-Use knowledge of 3-D Field Simulation for efficient analysis of full systems • Extract behavioral models from 3D field simulation • RLC Models for cables and connectors • Modal Reduction State Space Model for linear structural components by SPMWRITE • Linear Time Invariant (LTI) for cooling applictions with CHT • Krylov Model Order Reduction (MOR inside ANSYS) for cooling, thermoelectric & piezo applications • Equivalent Circuit Elements (ECE) for electromagnetic actuators • Universal nonlinear behavioral models by DoE • Efficent DoE scheme, automated model selection and verification © CADFEM 2014
Behavioral Models – Strategy for Industrie 4.0 • The system integration generates value • Customers require numerical description of product properties • 3D geometry is a standard • Physical properties are likely to follow
• Behavioral models of your components • Will give additional customer satisfaction • Increase your competition advantage • Seal and secure your IP
© CADFEM 2014
Agenda
Component & Assembly Design
System Simulations
Customized Solutions
Electromagnetic
Power train
Stretching the limits
Circuits
Automatization
Signal integrity
Behavioral models
Thermal Mechanical Fluids
© CADFEM 2014
ANSYS Workbench – Multiphysics Simulation Platform
© CADFEM 2014
59
Parametric Simulation • What is the benefit of a single simulation? Material ± 5-10%
• Variation gives most understanding • • • • • • •
Geometric shape Material Current Windings Circuits Controller …
Boundary Conditions ± 1-20%
Geometry ± 0.1-10%
Result ± ??%
Manufacturing ± 5- 30%
Loads, Signals ± 0.1-50%
• DoE, Optimization, Robust Design
© CADFEM 2014
60
Example • Synchronous machine with internal permanent magnets (IPM) • Defined Power • High efficiency • Low ripple • Cost minimization due to reduced material usage • Magnet • Steel
© CADFEM 2014
- 61 -
Is a manual variation efficient? • Take 1 Parameter: Thickness of Magnet • The evaluation of the results is quite simple. • Just use two graphs in Excel.
mechanical Power (W)
Power 9350 9300 9250 9200 9150 9100 9050 9000 8950 8900 8850 4.9
5 5.1 5.2 5.3 Thickness of magnet (mm)
5.4
Mass Area of materia (mm²l
8.60E-05 8.50E-05 8.40E-05 8.30E-05 8.20E-05 8.10E-05 8.00E-05 7.90E-05 4.9 © CADFEM 2014
5 5.1 5.2 5.3 Thickness of magnet (mm)
5.4
Manual variations • Manual variation: normally 3 designs (low-mid-high) • • • •
Do you want to set this up manually? Can you ensure that all designs can be regenerated? Is this efficient? Ist the gathered information useful?
© CADFEM 2014
If 5 p.: 35 = 243 designs! If 7 p.: 37 = 2187 designs!
Benefit of stochastic sampling • Manual variation: normally 3 designs (low-mid-high) • Failed design: loss of large amount of information • Stochastic sampling: • No loss of information, best representation of variation space!
© CADFEM 2014
Technical Know How By Systematic Variation • Sensitivity = Understanding • What are the most important design variables? • Is there a correlation between parameters? • What is the range of results?
• Optimization • How to find the best result • Handle conflicting goals • Correlation of simulation and test
• Robust Design • Consider scatter in input
© CADFEM 2014
Sensitivity Study
B-Field
S1-Stress
© CADFEM 2014
66
Sensitivity Study
© CADFEM 2014
67
Sensitivity Study • Correlation Matrix • What is important? Positive correlation
• Where do we have correlations? • Design varialbe à result • Result à result No correlation
• Filter designs
Negative correlation © CADFEM 2014
68
Optimization • Define multiple targets • Use advanced algorithms without beeing a mathematician • Find „the“ best design • Use the results of the sensitivity study • Fast – only seconds instead of hours • Accurate - verified prognosis quality
© CADFEM 2014
69
Optimization • Define multiple targets • Use advanced algorithms without beeing a mathematician • Find „the“ best design • Use the results of the sensitivity study • Fast – only seconds instead of hours • Accurate - verified prognosis quality
• Re-do the optimization with different goals
© CADFEM 2014
70
Robust Design • Consider scatter • Manufacturing tolerances • Material properties • Operation conditions
• Check the robustness • How safe is my design? • How many samples will fail?
© CADFEM 2014
Save
Failure
71
CADFEM – Simulation is more than Software PRODUCTS Software and IT Solutions SERVICES Advice, Support, Engineering KNOW-HOW Transfer of knowledge CADFEM in D, A, CH • 1985 founded • 2,300 customers • 12 locations • 185 employees (worldwide > 250) © CADFEM 2014
72