Thermal Analysis & Design Improvement of an Internal Air-Cooled - PowerPoint PPT Presentation

Thermal Analysis & Design Improvement of an Internal Air-Cooled Electric Machine Dr. James R. Dorris Application Specialist, CD-adapco

Thermal Analysis of Electric Machines Motivation – Thermal challenges in electric machines Workflow – SPEED & STAR-CCM+ applied to electric machines Project Description – Overview of internal fan-cooled induction machine Electromagnetic Modeling – SPEED model and computation of losses Thermal modeling – STAR-CCM+ CHT model and results of simulations Motors / Cooling systems of particular interest – Electric machine applications of particular interest Conclusions

Motivation: – Electric Machine Industry Focus The past 10 years shows industry focus on: – high torque / power density / high efficiency – and reduction in size, weight, cost This combination leads to more performance from a smaller package – and a thermal challenge. Source graphics: NREL Increased operating temperatures would result in: 20 ºC – Requirement for better insulating materials 110 ºC 140 ºC 200 ºC – Reduced lifetime due to higher risk of 230 ºC thermal damages to insulations, bearings, etc – Higher risk of demagnetizing permanent magnets Demagnetization Improved cooling systems via CFD thermal simulation provide engineering value. Source graphics: Integrated Magnetics

Electromagnetic Analysis: – The SPEED suite of programs The following machine types are available: PC-BDC: Brushless permanent magnet and wound-field AC synchronous PC-IMD: Induction PC-SRD: Switched Reluctance PC-DCM: Direct Current (PM) PC-WFC: Wound-field and commutator PM 4

CD-adapco Tools For Electric Machines – SPEED and STAR-CCM+ Workflow 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, & 4. Preparation of the geometry winding scheme in STAR-CCM+ by reading 2. Stability check with the xGDF file static and dynamic analytical analysis 7. Solving and post processing FE-grid SPEED in STAR-CCM+ FV-grid STAR-CCM+ Data transfer 6. Mapping process for rotor and stator heat losses is carried out separately and automatically with transfer of the 5. If stable results, transfer values from neighbor grid node in of the heat loss distribution SPEED to STAR-CCM+ from the FE-analysis to STAR-CCM+ via the sbd-file Multi-physics, general purpose simulation solution General geometry, 3D finite volume solvers

Project Description – Induction Motor Thermal Performance Challenge : – A North American Motor Company’s (NAMC) internal fan cooled split-phase induction machine is not sufficiently cooled, expensive to test many design configurations. Solution: – Develop a steady-state thermal analysis of a machine using the CD- adapco SPEED  STAR-CCM+ workflow. Procedure: – Compute losses using SPEED – Import SPEED geometry, losses, NAMC CAD for non-active components into STAR-CCM+ – Define appropriate physics, boundary conditions – Solve Conjugate Heat Transfer (CHT) simulation at the specified load point – New load point only requires losses to be re-computed and re-imported into STAR-CCM+ - can be ready in ~15min. – Geometry changes can be performed by swapping out parts and re-meshing – can be ready is less than 1 day.

Electromagnetic Analysis – Computing losses using SPEED 2-D Geometry, winding definition Materials – Lamination Steel, rotor cage Controller definition, simulation settings Analytic Calculations (< 1s) 2D FEA electromagnetic solution (<1min) Fast computation, can connect with Losses computed from FEA B-field HEEDS optimization engine distribution via Steinmetz Model Heat Loads

Physics Modeling Windings modeled as bulk material SPEED Geometry: – I 2 R loss slot dependent – Active Components – Anisotropic Thermal Conductivity Rotor Cage I 2 R Losses – Uniform distribution over rotor bars I 2 R loss Temperature Dependent – SPEED model losses computed at average temperature – STAR-CCM+ model adjusts for local temperature dependent resistivity 𝜍 𝑊 = 𝑅 𝑈 ∗ (1 + 0.00393 ∗ 𝑈 − 𝑈 NAMC Geometry: 𝑠𝑓𝑔 ) – Non-Active Components 𝑊 Where 𝝇 𝑾 is the volumetric heat load, 𝑅 𝑈 is the total heat load, 𝑼 is the local region temperature, 𝑼 𝒔𝒇𝒈 is the average temperature (80 C) and 𝑾 is the region volume. End-winding surface roughness Core losses spatially dependent

Simulation Steady State Temperatures 2 1 2 1 • Rotor Bar Avg=116.7 C, End Ring 1 Avg=114.1 C, End Ring 2 Avg=115.9 C • Shaft Min Temp=44.8 C, Shaft Max Temp=116.5 C • SPEED model with rotor temp @ 116 C requires 52.5 % of copper conductivity for consistent losses and performance at this load point.

Comparison with Measurements • NAMC Measurements on aux and main winding at 2 circumferential locations, both for the fan (cold side) and exhaust (hot side) of the end winding. • Do not resolve aux vs. main winding due to NAMC’s end -winding geometry. • Compare with mean and standard deviation of temperature in outer 5mm of end- winding End Winding 2 (cold side) End Winding 1 (hot side) Measurement Simulation % Error Measurement Simulation % Error Mean 70.6 C 71.6 C 1.40 % Mean 88.0 C 88.5 C 0.57 % STD 3.38 C 1.94 C STD 3.16 C 1.08 C

Heat Flow: Rotor and Stator – Easy Reports in STAR-CCM+ Stator: Air 84 W Copper Loss Aux + Main 171 W Winding 87 W 49 W Air Stator Lams Iron Loss 120 W Housing 82 W Rotor: Air 36 W Copper Loss Rotor 102 W Cage 66 W 42 W Air Rotor Lams Iron Loss 41 W Shaft 17 W

CD-adapco Tools For Electric Machines – SPEED and STAR-CCM+ Workflow 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 Modified Design Original Design of the analytical model based on geometry, parameters, & 4. Preparation of the geometry winding scheme in STAR-CCM+ by reading 2. Stability check with the xGDF file static and dynamic analytical analysis 7. Solving and post processing FE-grid SPEED in STAR-CCM+ FV-grid STAR-CCM+ Data transfer 6. Mapping process for rotor and stator heat losses is carried out separately and automatically with transfer of the 5. If stable results, transfer values from neighbor grid node in of the heat loss distribution SPEED to STAR-CCM+ from the FE-analysis to STAR-CCM+ via the sbd-file Multi-physics, general purpose simulation solution General geometry, 3D finite volume solvers

Phase 2: Vented Stator – Design Improvement • Geometry of Stator swapped • Approximately 1 day work for vented stator design: required by intermediate user to swap geometry • New geometry part created in STAR-CCM+ • Conformal interfaces rebuilt • Entire model re-meshed • Physics / boundary conditions reset • Expecting Temperatures to drop across entire model: • Lower Copper Temp / Heat load • Lower Al Cage Temp / Heat load

Simulation Steady State Temperatures – Original Stator Design (reference) – Vented Stator Design • 2 1 Rotor Avg=96.7 C • End Ring 1 Avg=93.7 C 2 1 • End Ring 2 Avg=96.6 C • Shaft Min Temp=36.0 C • Shaft Max Temp=96.7 C • End Winding Temperatures decreased by nearly 20 C. End Winding 2 (cold side) Orig Design Vented Stator % Mean 70.6 C 59.2 C 16.1 % STD 3.38 C 2.13 C End Winding 1 (hot side) Orig Design Vented Stator % Mean 88.0 C 67.9 C 22.8 % STD 3.16 C 0.97 C

Heat Flow: Rotor and Vented Stator Stator: Air 62 W Copper Loss Aux + Main 37 W 156 W Winding 15 W 94 W 86 W Air 20 º C 20 Stator Lams Iron Loss 90 W Housing 82 W Rotor: Air 33 W Copper Loss Rotor 93 W Cage 60 W 9 W 42 W Air 20 20 º C Rotor Lams Iron Loss 35 W Shaft 17 W

Cooling Systems of particular interest for CFD analysis Loss Mechanisms – Copper losses, spatially distributed iron losses, friction and windage losses Heat transfer – Conduction, Radiation – Convection (natural or forced) Simulation of fluids moving in and around objects – Liquids and/or gases

Conclusions Computation of losses in SPEED – copper (I 2 R) losses on Stator windings and rotor cages – Spatial distribution of iron losses (eddy + hysteresis loss in laminations) Efficient workflow form SPEED to STAR-CCM+ – Geometry, losses are easily imported – New load points or geometric changes are easily studied. Detailed physics easily defined – Temperature dependent copper losses – Anisotropic thermal conductivity of windings CFD can provide heat-flow analysis that measurements cannot – quicker and less expensive – leads to better insights for design improvements Workflows also available from other Emag codes – Flux 2D/3D, JMAG, or STAR-CCM+ 2D Emag solver