Software Solutions for the Design and Simulation of Electric Machines Dr. Markus Anders, CD-adapco
Agenda Software for Electric Machine Design and Simulation: About SPEED SPEED& JMAG SPEED& STAR-CCM+, 2D EMAG SPEED& HEEDS: An optimization case SPEED & STAR-CCM+, Thermal: Workflow STAR-CCM+, Thermal: Link with other software suppliers STAR-CCM+ & JMAG: Example
About SPEED SPEED is the leading initial design software for electric machines Development started early 1980 s on PC s under MS-DOS platform using Pascal as the programming language at the SPEEDLab, University of Glasgow by Prof. TJE Miller. Detailed analytical analysis with finite-element links or finite-embedded solver for motors, generators and alternators including inverters and other electronic controls. The analytic approach provides nearly instantaneously results (in seconds/minutes). The embedded FE solver (using PC-FEA) is a unique capability, enables our customers to do things SPEED could not otherwise do and improves the accuracy of SPEED by a substantial margin. GoFER using PC-FEA
About SPEED 6 main machine programs for Synchronous machines: Induction machines: Switched reluctance machines: Brushed PM-DC machines: Wound-field commutator machines: Axial flux machines: PC-BDC, PC-IMD, PC-SRD, PC-DCM, PC-WFC and PC-AXM Fully scriptable (ActiveX) Over 1000 corporate accounts worldwide Can link to other FE programs, such as JMAG Flux Opera
SPEED & JMAG SPEED models can be easily imported into JMAG by using the GoFERRun Option Other FEA links. The types that are currently supported are synchronous machines (through PC-BDC) and induction machines (through PC-IMD). http://www.jmag-international.com/
SPEED & STAR-CCM+ STAR-CCM+ has been added as well to the list of Run options in the BDC GoFER dialog: Single-load-point Bgap distribution (OC) Btooth waveform (OC) Cogging torque i-psi calculation i-psi Polygon PCFEA
SPEED & STAR-CCM+ SPEED s Bgap GoFER with STAR-CCM+ The GoFER s using STAR-CCM+ makes arbitrary machine geometry manipulation very easy to fit the actual shape by using the incorporated CAD-modeller in STARCCM+ resulting in improved accuracy of the model.
SPEED & HEEDS Multi-disciplinary Design Optimization The main features: Multi-disciplinary, multi-objective parametric design optimization using SHERPA Multiple search methods simultaneously, A combination of global and local search methods, No tunable parameters; all parameters are automatically adjusted in an adaptive manner, Adaptive Automated Design of Experiments Sensitivity studies SPEED Robustness and reliability assessments Design Sweep (post processing) Create a variety of plots and tables Best illustrate relationships among variables and design goals Automated by scripting Parallel Plot showing design trends among designs evaluated during an optimization.
SPEED & HEEDS Very simple optimization example for minimum cogging torque and magnet volume Baseline Design Concept A Optimized Design: 94% reduction in cogging torque 52% reduction in magnet volume Concept B Optimized Design: 93% reduction in cogging torque 59% reduction in magnet volume Appr. 2000 calc., 3 hours
SPEED & HEEDS Design Exploration Using the US06 Drive Cycle Motor Design Parameters Geometry Motor Efficiency Results SHERPA Note: Only FEASIBLE Design Concepts are Displayed Feasible Infeasible
STAR-CCM+: General 3-D Multi-physics & Multi-purpose software STAR-CCM+ is a powerful, all-in-one tool that combines: Ease of use, All-in-one software package, Automatic meshing, Extensive modelling capabilities, Powerful post-processing. CAD STAR-CCM+ Report Developed since 2004: Uses the latest numeric and software technologies. Designed from the outset to handle very large models (100M+ cells). Full process integration: CAD to CAE in one package. Rapid development cycle: new releases every for months Started originally as a CFD software package
STAR-CCM+: General 3-D Multiphysics & Multipurpose software Integrated engineering solution for solving multidisciplinary problems CAE Integration Multidisciplinary Analyses Surface Preparation Meshing Geometry For electrical machines: CHT (Conjugate Heat Transfer including Conduction, Convection and Radiation) Windage losses (e.g. surface PM, no rotor sleeve or Switched Reluctance machines) EMAG 2D/3D Stress analysis (FE based solver) & vibro acoustic
Electric Machine Simulations A true multi-physics problem Electromagnetics Thermal / CFD Losses / Heat Loads
CD-adapco Tools For Electric Machines SPEED to STAR-CCM+ Workflow (for thermal design) Electric machine design solution Template based geometry, analytic tool + models for 3D effects, 2D FEA solver. 3. FE-analysis and fitting 1. Create SPEED model of the analytical model based on geometry, parameters, & winding scheme 4. Preparation of the geometry 2. Desing check with in STAR-CCM+ by reading the xgdf file static and dynamic analytical analysis 7. Solving and post processing in STAR-CCM+ 6. Mapping process for rotor and stator heat losses is carried out separately and automatically with transfer of the values from neighbor grid node in SPEED to STAR-CCM+ Multi-physics, general purpose simulation solution General geometry, 3D finite volume CFD solvers 5. Transfer of the heat loss distribution from the FEanalysis to STAR-CCM+ via the sbd-file
Thermal Considerations Heat flows through coils fast along the direction of the copper, slow perpendicularly to it. The Material is then anisotropic Wire Bundles: copper conducts heat better than insulation, varnish, potting material or air. Lamination Stack: Steel conducts faster than insulation coating Both physically modeled by setting 2 values for thermal conductivity: and How to determine the direction field? Set Direction field from coil geometry Bulk Coil model can use analytic expression for the direction field. (streamlines of the direction field are shown and look like winding pattern)
Thermal Modeling SPEED/Motor-CAD/STAR-CCM+ 3. Run thermal calculations in Motor-CAD to check the model 1. Creation of the Motor-CAD model based on geometry parameters and winding scheme or import from SPEED 4. Preparation of the geometry in STAR-CCM+ by running a Java script 2. FE-analysis and fitting of the analytical model FE-grid SPEED FV-grid STAR-CCM+ Data transfer 7. Solving and post processing in STAR-CCM+ 6. Mapping process for rotor and stator heat losses is carried out separately and automatically with transfer of the values from neighbor grid node in SPEED to STAR-CCM+ 5. Transfer of the heat loss distribution from the FEanalysis to STAR-CCM+ via the sbd-file
Thermal Modeling Links with other software supplier: JMAG, FLUX, Motor-CAD, FE EMAG Loss Calculation From SPEED From JMAG (JSOL, Japan) From Flux (Cedrat/Magsoft, France/US) From Motor-CAD (MDL, UK) STAR-CCM+ cooling analysis Conjugate heat transfer using liquid and/or gaseous coolants Import of thermal loading from EMAG tool 2D or loss 3Ddistribution loss distribution data isdata is mapped onto STAR-CCM+ grid
JSOL/CD-adapco Announcement July, 2014
JMAG/STAR-CCM+ Co-sim possibilities Legacy Methods Automatic Methods Firstdevelopment deliverable Future developments JMAG > Nastran files > STAR-CCM+ STAR-CCM+ > Nastran > JMAG JAVA based coupling Injector example Volume to Volume API Electric Machine (3DEMAG > 3D Thermal > 3DEMAG > 3D Thermal > ) Possible2D EMAG > 3D Thermal > 2D EMAG > 3D Thermal >
Example: Loss import from JMAG The model, losses and load cases Low speed: 600 rpm High speed: 8,000 rpm Loss density JMAG model Copperloss density distribution JMAG Ironloss density distribution JMAG Magnet loss density distribution JMAG
JMAG Example Losses vs. speed Low speed (600 rpm) Copper losses are dominating. Medium speed (4,000 rpm) Iron losses are slightly higher than copper losses High speed (8,000 rpm) Iron losses are dominating. Magnet losses are negligible Low speed Speed Current Torque Output Copper loss Iron loss Magnet loss rpm A Nm kw W W W 600 84.8 22.8 14.3 334.4 17.2 0.20 4000 60.0 18.5 7.7 167.4 268.1 2.14 Medium speed High speed 8000 30.4 9.1 7.6 43.0 345.3 0.74
Example 2: Model Set-up in STAR-CCM+ Simulation goal: Steady state temperatures Water cooled housing, coolant induced at 40 C, 0.25 m/s Winding region with bulk end windings Venting holes in the rotor at representative shaft radius of JMAG model Moving reference frame model to allow for the rotation of the rotor Runtime Mesh: 2,7Mio. polyhedral cells Converges within 200 iterations Computation time: runs on 5 cores in 36 minutes #19.IGS: 3D cad model for thermal analysis (iges format)
Example 2: Loss import from JMAG Mapped losses and temperature distribution Low speed: 600 rpm Mapped imported heat loss distribution in STAR-CCM+ Temperature distribution in STAR-CCM+ High speed: 8,000 rpm
Example 2: Transfer of Temperatures back as.nas files STAR-CCM+ solution can be mapped back onto NASTRAN grid Export of solution data only to reduce file size Temperature is written as vertex data Unit is specified by user 600 RPM 4000 RPM 8000 RPM
SPEED & STAR-CCM+ Questions?