SPARK Introduction | February, 2016 | 1 Design Optimization of Traction Electric Machines for EV and HEV – Perspectives from Compact to Racing Cars Dan M. Ionel, Ph.D., IEEE Fellow [email protected] CWIEME Chicago, October 4, 2017
SPARK Introduction | February, 2016 | 1
Design Optimization of Traction Electric Machines for EV and HEV –
Perspectives from Compact to Racing Cars
Dan M. Ionel, Ph.D., IEEE [email protected]
CWIEME Chicago, October 4, 2017
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 2SPARKLaboratory
Outline
• Introduction• PEIK and SPARK at University of Kentucky• Design optimization with CAI – Differential Evolution (DE)
• Hybrid Electric Vehicle (HEV) optimization examples• Rated performance• Influence of cooling• Design for driving cycle
• Without (ferromagnetic) core…• In-wheel axial flux coreless machines for EV solar PV racing cars• Cored vs. coreless• Flux vs. current weakening
• Without magnets…• Lucid Motors: new luxury EV, induction motor, multi-physics analysis• NASA and OSU: 10MW ring electric motor for aircraft engines
• Conclusion.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 3SPARKLaboratory
Dan M. Ionel, Ph.D., FIEEEDan M. Ionel is Professor and L. Stanley Pigman Chair in Power atUniversity of Kentucky (UK). At UK he also serves as the Director of thePower and Energy Institute of Kentucky (PEIK) and of the SPARKLaboratory.
Previously, he worked in industry for more than 25 years, most recentlyas Chief Engineer for Regal Beloit Corp., and before that as the ChiefScientist for Vestas Wind Turbines. He contributed to technologydevelopments with long lasting industrial impact, designed machinesand drives with ratings between 0.001 and 10,000hp, published morethan 150 technical conference and journal papers, including 5 winnersof IEEE Paper Awards, and holds more than 30 patents, including amedal winner at the Geneva Invention Fair.
Dr. Ionel is an IEEE Fellow, was the Chair of the IEEE Power and EnergySociety Electric Motor Subcommittee, the General Chair of IEEE IEMDC2017 Conference, and is currently the Chair of the IEEE WG for 1812Test Guide revision.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 4SPARKLaboratory
SPARK and PEIK at University of Kentucky (UK)
• UK enjoys a longstanding tradition in electric machines and drives• Early developments on linear and PM motors, and vector control• Many learned machines using the Nasar and Boldea text books
• PEIK, Power and Energy Institute of Kentucky, launched with DOE grant in 2010• Core faculty in electric power engineering and many others in related fields• Endowment established and inaugural L. Stanley Pigman Chair started in 2015• SPARK and other laboratories; faculty: 10+; graduate research students: 60+• Strategic collaborators include ANSYS Inc., NREL, Center for Applied Energy Research
(CAER) at UK and others.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 5SPARKLaboratory
Systematic Optimization and Comparative Studies
• Multi-objective optimization problems, for example: • minimum cost, and• maximum efficiency (min. losses), and• minimum torque ripple• Etc.• …“many-many” objectives
• Evolutionary process• hundreds of generations• thousands of candidate designs
• Aims• establish Pareto-sets and fronts, i.e. “best
compromise” designs• systematic comparison of different
solutions, e.g. design topologies.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 6SPARKLaboratory
Differential Evolution Optimization
)( 210 rrr xxFx −+
))1,0(rand(if rC≤
Randomly select three designs
Apply crossover and mutation
Crossover:
Mutation:
Selection:• Single objective• Multi-objective (Pareto-dominance)
Repeat until convergence
Objective function evaluation (model execution)
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 7SPARKLaboratory
Ultra-fast Time Stepping Electromagnetic FEA
Courtesy of ANSYS, Inc.
Bigger Faster
Higher Fidelity
Traditional approach – solve all time steps sequentially:
New approach (TDM)– solve all time steps simultaneously:
t1 t2 t3 t4 tnt0𝟎𝟎
t0 t1 t2 t3 t4 tn…
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 8SPARKLaboratory
Past Example – Optimization for Formula E Racing Cars
Published paper: A. Fatemi, D. M. Ionel, M. Popescu, N. A. O. Demerdash, “Design Optimization of Spoke-Type PM Motors for Formula E Racing Cars”, IEEE ECCE 2016.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 9SPARKLaboratory
Single Point Performance Optimization for Reference Design• Rating: 400Nm 1,500rpm (typical peak for HEV automobiles)• Topology: 48-slot 8-pole IPM• Set values: 22.5 Arms/mm2, 0.475 slot fill factor• Independent Variables: ten (10) total for stator and rotor geometry• Objectives:
• Minimum losses and• Minimum cost of active materials
• Constraints:• Torque ripple < 15%• Magnet B_min > 0.3 Br
• Optimization method:• CMODE – Differential Evolution (DE) based• Generations: 60; members per generation 80.
xD Description
ksi Rsi/Rso
g Air-gap
Kwt wT/Slot-pitch
kwtt tip/Slot-opening
kdPM dPM/dPM,max
kwPM wPM/wPM,max
kwq wq/wq,max
hPM PM height
αPM Magnet pole arc
dy Yoke depth
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 10SPARKLaboratory
Possible Improvements through AI – DE Optimization
Design G59M12
Active material cost [pu] 93.4
Stator losses [W] 5680
Input voltage THD [%] 11
Torque ripple [%] 14
BPM,min [T] 0.40
Torque angle at MTPA [deg.] 134o
Power factor 0.80
Reference
88.7
7841
19
25
0.41
138o
0.67
ReferenceG59M12
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 11SPARKLaboratory
NC (natural cooled) FC (forced cooled) and LC (liquid cooled).
Optimal Designs with Different Cooling Systems
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 12SPARKLaboratory
Advanced Vehicle Simulator (ADVISOR); developed by NREL
System Level Design
Driving cycle
Vehicle model
Torque& Speed
Speed profile
Torque profile Energy
𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 =𝐹𝐹𝑎𝑎 + 𝐹𝐹𝑚𝑚 + 𝐹𝐹𝑐𝑐 + 𝐹𝐹𝐷𝐷 . 𝑟𝑟𝑤𝑤
𝑛𝑛𝑑𝑑
Rolling Resistance𝐹𝐹𝑚𝑚 = 𝑘𝑘𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚(𝜃𝜃)
Aerodynamic Resistance𝐹𝐹𝐷𝐷 = 0.5𝜌𝜌(𝜐𝜐 + 𝜐𝜐0)2𝐶𝐶𝑑𝑑𝐴𝐴𝑓𝑓
Climbing force𝐹𝐹𝑐𝑐 = 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑛𝑛 𝜃𝜃
Acceleration𝐹𝐹𝑎𝑎 = 𝑚𝑚𝑚𝑚
Energy distribution functionCombined Dynamometer Driving Schedule
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 13SPARKLaboratory
K-means Clustering Algorithms for Representative PointsCyclic representative points
Urban Dynamometer Driving ScheduleCyclic representative points
Highway Fuel Economy Driving Schedule
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 14SPARKLaboratory
Multi-objective Optimization Considering Driving Cycle
• Objective 1: Minimization of loss over the representative load operating points
The ratio of loss over net output power multiplied by the energy weights obtained through selection of representative points is summed up to minimize the dissipated energy. ∑𝑳𝑳𝑳𝑳𝑳𝑳𝑳𝑳𝒊𝒊
𝑻𝑻𝒊𝒊𝝎𝝎𝒊𝒊𝒘𝒘𝒊𝒊
• Objective 2: Minimization of material cost
Machine stack length is adjusted to deliver the rated torque corresponding to the maximum current density. Therefore only the rated operating point directly contributes to this objective.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 15SPARKLaboratory
Two Designs from the Pareto Front vs. Reference (P)
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 16SPARKLaboratory
UK Solar Car – Gato Del Sol
Major on-going projects
• Design and build our first 4-wheel challenger class solar car
• Re-engineer Gato Del Sol V to set the world speed record for solar-capable electric vehicles
• Design a solar cruiser class solar car.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 17SPARKLaboratory
Solar Car Construction and Operation
Sources: UK Solar Car; Ali Emadi, “Advanced Electric Drive Vehicles”, CRC Press, Boca Raton, FL, 2015 (lower left); and UK SPARK Lab.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 18SPARKLaboratory
Conventional In-wheel AFPM Machine
• Original NGM stator-rotor kit• Redesigned rotor prototyped; the
sponsorship of Arnold Magnetics is gratefully acknowledged
• High speed constant power operation possible only by mechanically increasing the airgap (flux weakening).
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 19SPARKLaboratory
Coreless In-wheel AFPM Machines
• MARAND Precision with CSIRO design (top)• Commercially available• One stator two rotors• Distributed winding (3ph)• Litz wire
• Our UK designs (right)• Multiple rotor stator disks• Concentrated coils• Steel only on end rotors
• Not to exact scale.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 20SPARKLaboratory
FEA of Coreless Machine Designs
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 21SPARKLaboratory
MARAND Machine with Halbach PM Rotor
• Each rotor pole comprises multiple magnets with different directions of magnetization
• Total magnets per rotor 4x40=160
• No back iron – even in the rotor
• Highest achievable airgap flux density – sinewave with a peak close to PM remanence
• Low(est) mass• High(est) efficiency
• High torque• Low speed. Source: “High Efficiency Permanent Magnet Motor,” http://www.ata.org.au/ wp-
content/uploadsmarand_high_efficiency_motor.pdf
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 22SPARKLaboratory
Pros and Cons of Coreless Machines
Pros•No core losses, i.e. no
“fixed” losses•Lowest rotor losses•High(est) efficiency
throughout a wide speed range
•Virtually no cogging and ripple torque
•Low(est) noise.
Cons•High(est) cost because of
large PM quantity•High AC winding losses
(special wire and windings)•Heating of inner stators•Ultra-low inductance (special
electronics)•Unsuitable for high-speed
constant power operation.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 23SPARKLaboratory
Optimal Design Studies
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 24SPARKLaboratory
Performance Comparison – One and Two Active Wheels
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 25SPARKLaboratory
Power Electronics and PM Machine Control
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 26SPARKLaboratory
Field (Flux) or Current Weakening?
• Traditional field (flux) weakening for PM machines• Design the machine (IPM) with “suitable” inductance, hence
core• Advance the torque angle at rated current to reduce main flux
and achieve constant power at high speed
• Current weakening• Design the inverter with voltage overload for given current
and power rating• Reduce the current at 90 deg constant torque angle to
diminish torque and achieve constant power at high speed• May not work for machines with core and/or for very cost
competitive applications.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 27SPARKLaboratory
Traction Characteristics
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 28SPARKLaboratory
LUCID – Car and Induction Motor
Courtesy of Lucid Motors and ANSYS, Inc.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 29SPARKLaboratory
The Induction Motor at the Heart of the Powertrain
Two cooling mechanisms• Water jacket in the motor case• Oil splashing onto the end-windings and the rotor.
50% Glycol, 50% Water
Transmission Oil
Courtesy of Lucid Motors and ANSYS, Inc.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 30SPARKLaboratory
Multi-physics Electromagnetic – Thermal Coupled Analysis
All contours use Local Values, Auto Range
EM Volume Losses Used in Computational
Fluid Dynamics
Rotor Volumetric Losses
Stator Volumetric Losses
Rotor Bar Losses
191.6 Nm @ 3000 rpm, 59.7 kW Equivalent car speed @ 25mph,
20% grade slope Slip = 0.0156258 Current Source = 492 A Core loss = 477 W Winding Loss = 2749 W Rotor Bar Loss = 1175 W Total EM Loss = 4353 W
Courtesy of Lucid Motors and ANSYS, Inc.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 31SPARKLaboratory
Oil-Transient Conjugate Heat Transfer
End Windings and Insulation Lateral Cut
Rotor Bars
Axial Cut
282F
139F
Courtesy of Lucid Motors and ANSYS, Inc.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 32SPARKLaboratory
Thermal Analysis – Oil Transient and CHT Simulations
Oil Transient Simulation Results (VOF 0.1 Iso-Surface, 0 - 0.5s)
Courtesy of Lucid Motors and ANSYS, Inc.
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 33SPARKLaboratory
Electric Machine for NASA Sponsored ProjectBoeing 737- 800 Aircraft with CFM56-7B Engines
Credits: Codrin-Gruie (CG) Cantemir and Adrian Munteanu, “10 MW Ring Motor; OSU NNX14AL87A”, EnergyTech16 Conf., Cleveland, OH, Nov 2016; NASA Sponsored Project.
Grey - unchanged partsDirect drive10 MW cont. @ 5,000 rpmFirst implementation770kg; approx. 8hp/lb
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 34SPARKLaboratory
Electric Machine for NASA Sponsored Project
• Outer rotor induction machine
• Air-gap diameter approx. 1.2m
• Modules of MV winding and power electronic switches
• Cooling is implemented inthe booster’sblades.
Credits: Codrin-Gruie (CG) Cantemir and Adrian Munteanu, “10 MW Ring Motor; OSU NNX14AL87A”, EnergyTech16 Conf., Cleveland, OH, Nov 2016; NASA Sponsored Project.
Boeing 737- 800 Aircraft with CFM56-7B Engines
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 35SPARKLaboratory
Electric Machine for NASA Sponsored Project
Credits: Codrin-Gruie (CG) Cantemir and Adrian Munteanu, “10 MW Ring Motor; OSU NNX14AL87A”, EnergyTech16 Conf., Cleveland, OH, Nov 2016; NASA Sponsored Project.
Build of the First Demonstrator
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 36SPARKLaboratory
Conclusion
•Electric transportation applications are… different •We exemplified machines with
•Core and without core•Magnets and without magnets
•The ideal electric machine will have, as always: •Virtually no cost ;)•Virtually no losses ;)
• Innovation remains key•How to best combine the achievements from different areas?
Dan M. Ionel, HEV EV Optimization | CWIEME, Chicago, October 4, 2017 | 37SPARKLaboratory
Selected Recent Papers• Rallabandi, Vandana, Taran, Narges, and Ionel, D. M., “Multilayer Concentrated Windings for Axial Flux PM Machines”, IEEE
Transactions on Magnetics, Vol. 53, No. 6, 10.1109/TMAG.2017.2661312, 4p (2017).• Rallabandi, Vandana, Taran, Narges, Ionel, D. M., and Eastham, J. F., “Coreless Multidisc Axial Flux PM Machine with Carbon Nanotube
Windings”, IEEE Transactions on Magnetics, Vol. 53, No. 6, 10.1109/TMAG.2017.2660526, 4p (2017).• Zhang, Peng, Ionel, D. M., and Demerdash, N. A. O., “Saliency Ratio and Power Factor of IPM Motors with Distributed Windings
Optimally Designed for High Efficiency and Low-Cost Applications”, IEEE Transactions on Industry Applications, Vol. 52, No. 6, pp.4730-4739 (2016).
• Wang, Yi, Ionel, D. M., Rallabandi, Vandana, Jiang, M., and Stretz, S., “Large Scale Optimization of Synchronous Reluctance MachinesUsing CE-FEA and Differential Evolution”, IEEE Transactions on Industry Applications, Vol. 52, No. 6, pp. 4699-4709 (2016).
• Fatemi, A., Ionel, D. M., Demerdash, N. A. O., and Nehl, T., “Large-scale Design Optimization of PM Machines over a Target OperatingCycle”, IEEE Transactions on Industry Applications, Vol. 52, No. 5, pp. 3772-3782 (2016).
• Fatemi, A., Ionel, D. M., Demerdash, N. A. O., and Nehl, T. W., “Optimal Design of IPM Motors with Different Cooling Systems andWinding Configurations”, IEEE Transactions on Industry Applications, Vol. 52, No.4, pp. 3041-3049 (2016).
• Fatemi, A., Ionel, D. M., Demerdash, N. A. O., and Nehl, T.W., “Fast Multi-Objective CMODE-Type Optimization of PM Machines UsingMulticore Desktop Computers”, IEEE Transactions on Industry Applications, Vol. 52, No. 4, pp. 2941-2950 (2016).
• Taran, Narges, Rallabandi, Vandana, Ionel, D. M., and Heins, G., “A Comparative Study of Coreless and Conventional AFPM Machinesfor Low and High Speed Operation”, IEEE ECCE 2017 Congress, Cincinnati, OH, 6p (Oct. 2017).
• Liu, X., Cramer, A. M., Rallabandi, Vandana, and Ionel, D. M., “Switching Frequency Selection for Ultra-Low-Inductance Machines”,IEEE IEMDC 2017 Conference, Miami, FL, 6p (May 2017).
• Taran, Narges, Rallabandi, Vandana, Heins, G., and Ionel, D. M., “A Comparative Study of Conventional and Coreless Axial FluxPermanent Magnet Synchronous Motors for Solar Cars”, IEEE IEMDC 2017 Conference, Miami, FL, 7p (May 2017).
• Rallabandi, Vandana, Taran, Narges, Ionel, D. M., and Eastham, J. F., “On the Feasibility of Carbon Nanotube Windings for ElectricalMachines: Case Study for a Coreless Axial Flux Motor”, IEEE ECCE 2016 Congress, Milwaukee, WI, 7p (Sep. 2016).
• Fatemi, A., Ionel, D. M., Demerdash, N. A. O., and Popescu, M., “Design Optimization of Spoke-Type PM Motors for Formula E RacingCars”, IEEE ECCE 2016 Congress, Milwaukee, WI, 8p (Sep. 2016).
• Fatemi, A., Ionel, D. M., Demerdash, N. A. O., Stretz, S., and Jahns, T., “RSM-DE-ANN Sensitivity Analysis of Material Cost in PM Motorswith Distributed and Concentrated Windings”, IEEE ECCE 2016 Congress, Milwaukee, WI, 7p (Sep. 2016).