Electro-mechanical Interactions In the Design of Integrated EV drivetrains Dr. Melanie Michon October 2016
Agenda Requirements for an electro-mechanical design process o Electro-mechanical interactions add complexity to the process Setting out a defined design process and establishing design parallels o For gears and electrical machine Electro-mechanical system interactions o o What interactions can and should be captured at each stage in the design process What methods do we use to capture these A suggested toolchain Slide 2
Questions.. How do you make sure the product is FIT FOR PURPOSE How do you make sure the product meets its PEFORMANCE AND DESIGN TARGETS? How do you make sure the product AVOIDS FAILURE How do you make sure you meet your PRODUCT DESIGN CYCLE TARGET Project Set-up PRODUCT DESIGN SPECIFICATION Requirements for your DESIGN & SIMULATION PROCESS System optimisation for ELECTRO-MECHANICAL DESIGN Slide 3
Design process is driven by performance targets Target driven design trade-off: holistic assessment of all targets for electrical, mechanical AND system interactions Efficiency, energy economy Driveline NVH, driveability Durability, duty cycle Packaging, lay-out, gear stages, physical interfaces Thermal performance Mass, cost Limited design data is available Simulation speed is important to assess many concept designs Simulation accuracy should be sufficient to make concept down-selection Large design loops should be avoided by using a defined hierarchy in the design process The analysis (CAE) process determines how these targets can all best be met and what simulation speed and accuracy to use at different stages of the design process Slide 4
Current simulation toolchain Simulation of Gearbox e.g. Concept + RomaxDesigner Simulation of Electrical Machine e.g. RMxprt + Maxwell Implements multi-fidelity modelling Assessment of all performance targets mechanical engineering Implies multi-fidelity modelling Assessment of all performance targets electrical engineering However, what about gearbox-motor interactions?? Slide 5
Requirements for the EV Design Process Simulate and eliminate (identify and avoid) failure modes as early as possible Use multi-fidelity analyses for maximum insight with optimum simulation speed at each stage of the process and to manage data flow between different departments Consider all performance targets and trade-off performance using a holistic design approach Consider electro-mechanical interactions within the drivetrain from the start Slide 6
DESIGN PROCESS For Gears and Electrical Machine Slide 7
Design flow within an EV Design Process Down-selection, from many to one. Then optimise System Level Definition System Architecture Number of gear stages Power/Energy ratings Component Level Definition Topology selection Ratio split, macrogeometry Sizing of electrical machine parts Part Level Definition Gear microgeometry, Bearing confirmed Stator, Rotor geometry, winding layout Detailed housing x1000 Concept Design Options x10 Concept Design Options x Design Options Final Design Simulation toolchain Capture any electro-mechanical interactions Sound engineering decision making tools Requires the use of multi-fidelity models Slide 8
Main Design Stages in a Concept Design Process x1000 Concept Design Options Vehicle Concept Layout Design x10 Concept Design Options Concept Design x Design Options Detailed Design Tolerances & Sign off Final Design Slide 9
Design process for Gearbox Vehicle Concept Lay-out Concept Design Detailed Design Tolerances & Sign-off Ratios Topology Sizing - 1 Sizing - 2 Bearings Concept Housing Detailed Housing CAD Slide 10
Design Process for Gearboxes Analysis methods are well-defined Target Efficiency Packaging Durability Vehicle Concept Gearbox Layout Gearbox Concept Design Topology Sizing 1 Sizing 2 Concept Bearings Concept Housing Detailed Design Detailed Housing Tolerances & Sign-off Driveability Thermal Noise Increasing model fidelity Increasing input data requirement Decreasing simulation speed Weight Cost Slide 11
Design process for Electrical Machine Vehicle Concept Lay-out Concept Design Detailed Design Tolerances & Sign-off Rating Topology Sizing - 1 Sizing - 2 Materials Winding 2D/3D EM FEA & Thermal Final specification Slide 12
Design Process for Electrical Machines Analysis methods are well-defined Target Vehicle Concept Electrical Machine Layout Electrical Machine Concept Design Detailed Design Tolerances & Sign-off Topology Sizing 1 Sizing 2 Concept Material Efficiency Packaging Durability Driveability Thermal Noise Increasing model fidelity Increasing input data requirement Decreasing simulation speed Weight Cost Slide 13
Discussion The design methods are well established for gearboxes and motors separately. But what are the electromechanical influences that need to be considered? Slide 14
VEHICLE CONCEPT Slide 15
Gearbox-Motor Interactions through the Design Process Step 1: Vehicle Concept Target Efficiency Packaging Durability Driveability Thermal Noise Weight Cost Vehicle Concept Vehicle drive cycle simulation Acceleration and driveability simulation Cost for gearbox versus cost for electrical machine Interaction Trade-off between overall gear design/ratio and machine torque/power rating, including driving behaviour, hence vehicle acceleration, driveability and overall efficiency Slide 16
Efficiency-driven architecture selection Concept down-selection trade-offs to be made Complex, multi-speed gearbox - electrical machine operating range is limited and can easily be designed to operate at efficiency hot spot Simple, single-speed gearbox electrical machine efficiency requires optimisation over wide operating range High overall gear ratio small, high speed machine Low overall gear ratio large, low speed machine Driving behaviour and drive cycle selection affects fuel/energy economy, and design trade-offs Slide 17
Concept down selection case study Toyota Prius (HEV) case study Sensitivity studies to assist concept down selection Assess many drive cycles Urban Extra-urban Highway Close to real world driving Slide 18
Planetary gear ratio Final drive + planetary ratio The values of the final drive and planetary gear ratios were varied for different drive cycles New European Driving Cycle o Fuel consumption close to optimum Artemis Urban: low speed, starting and stopping o Higher ratios are better Highway Fuel Economy Test: high speeds, low accelerations o Lower ratios are better Combination of eight drive cycles: o Existing ratios are a good compromise 4 3 2 Toyota Prius 1 2 3 4 5 Final drive ratio Slide 19
LAY-OUT Topology Sizing Basic Concept Slide 20
Gearbox-Motor Interactions through the Design Process Step 2: edriveline Lay-out (Topology) Target Efficiency Packaging Durability Driveability Thermal Noise Weight Cost Interaction edriveline Layout (Topology) Qualitative efficiency match between electrical machine and gear topology (single speed versus multi-speed) Space for gearbox versus space for electrical machine Does gearbox space claim affect machine thermal capacity Mass for gearbox versus mass for electrical machine (qualitative) Cost for gearbox versus cost for electrical (qualitative) Space claim for gearbox versus space claim for electrical machine Slide 21
Qualitative assessment of topologies Topology 1 Topology 2 Topology 3 Topology 4 Gear Efficiency ++ ++ + 0 Electrical machine Efficiency ++ - + 0 Packaging ++ + + 0 Electrical machine thermal performance ++ + + 0 edriveline Weight ++ ++ + + edriveline Cost - + ++ + Slide 22
Gearbox-Motor Interactions through the Design Process Step 3: edriveline Layout (Sizing-1) - PACKAGING Target edriveline Layout (Sizing-1) Efficiency Packaging Airgap stress method: quickly assess if electrical machine efficiency dictates available space for gear Initial space claim using basic sizing for gear and electrical machine 75 mm Durability Is gear durability affected by machine size requirements Driveability Thermal Qualitative: can thermal performance gearbox affect machine cooling requirement? Noise Weight Cost Mass for gearbox versus mass for motor Cost for gearbox versus cost for motor D s Interaction Gearbox/machine sizing within space envelope, affecting thermal and durability requirements L Slide 23
Space Claim and lay-out of edriveline All within target packaging constraints D s L Initial design assessment of Gear Centre Distance required to achieve durability targets (contact) Combined with electrical machine sizing based on initial thermal assessment (shear stress) Slide 24
Gearbox-Motor Interactions through the Design Process Step 3: edriveline Layout (Sizing-1) - DYNAMICS Target edriveline Layout (Sizing-1) Efficiency Packaging Durability 75 mm Driveability 1 st driveline torsional mode based on machine and gear inertia, reference mount stiffness, response to e.g. shock load Thermal Noise Lowest powertrain bending mode provides initial mount characteristics Weight Cost Interaction Initial dynamic interaction between motor and gearbox D L s Slide 25
Driveability and 1 st Torsional Mode Torsional model (e.g. Matlab) Torsional mode shape from 6 dof model in RomaxDesigner Actual profile of average torque Simulation set up and target parameters (oscillation and duration of oscillation) Slide 26
Low Frequency Powertrain bending modes and mounts These modes fundamentally about mount stiffness, not stiffness of housing/shafts Romax has verified that the same results come from 3D Romax model and 1 st principles, single mass at C of G on springs Mode f n Motor excitation Unbalance Corresponding rotor speed 1 13.0 Hz 195 rpm 2 20.0 Hz 1200 rpm 3 31.6 Hz 474 rpm Slide 27
Gearbox-Motor Interactions through the Design Process Step 3: edriveline Layout (Sizing-2) GEAR RATIO SELECTION/EFFICIENCY Target Efficiency Packaging Durability Driveability Thermal Noise Weight edriveline Layout (Sizing-2) Analytical system efficiency calculation: Selection of gear ratios affects both gear and electrical machine efficiency Space envelope dictates limits in gear ratio/machine torque Is gear durability affected by tooth number selection, Qualitative assessment: machine manufacturability may restrict slot number selection Lumped parameter model gives initial thermal estimate Cost Interaction Gear ratio selection affects system efficiency and durability. Space envelope puts further constraint on optimisation Slide 28
Second stage ratio Gear ratio optimisation for all design targets Remaining design space NVH motor + gearbox System efficiency Packaging Ratio limits Overall ratio Smaller, lighter electrical machine Higher gear ratio Slide 29 System durability assessment
Gearbox-Motor Interactions through the Design Process Step 3: edriveline Layout (Sizing-2) GEAR RATIO SELECTION/NOISE Target Efficiency Packaging Durability Driveability Thermal Noise Weight edriveline Layout (Sizing-2) Analytical system efficiency: quick assessment if gear/machine efficiency is affected by selection of tooth/slot/pole numbers Is gear durability affected by tooth number selection, Qualitative assessment: machine manufacturability may restrict slot number selection Order plot of electrical and mechanical excitations Cost Interaction Order analysis of system excitations drives selection of gear tooth numbers and machine pole/slot combination Slide 30
Order plot analysis Order analysis including both gearbox AND motor excitation orders Analytical calculation of excitation frequencies with minimum required input data Concept design recommendations can be made (change tooth numbers, number of poles/slots, topology) Significant overlap No significant overlap Slide 31
E-DRIVELINE CONCEPT Full concept design Slide 32
Gearbox-Motor Interactions through the Design Process Step 4: edriveline Concept HIGH SPEED BEARING DESIGN Target Efficiency Packaging Durability Driveability Thermal Noise Weight Cost Interaction edriveline Concept RxD incl. e-machine mechanical model: High speed shaft bearing pre-load affects efficiency RxD incl. e-machine mechanical model: High speed shaft bearing pre-load affects durability RxD: Gear forces -> bearing stiffness -> lateral modes of rotor shaft Lateral modes of the high speed shaft are affected by gear forces, coupling and the rotor design. Bearing pre-load can avoid unwanted resonances Slide 33
Gear loads, bearing pre-loads, shaft modes and efficiency Critical speeds of rotor shaft are dependent on bearing stiffness Low bearing stiffness can lead to resonant frequency being within the operating range irrespective of the shaft stiffness Pre-load increases bearing stiffness but also increases drag; What is the trade-off? Complex interaction with the gear load, depending on the axial constraints of the bearings etc. How to select bearing constraints, pre-load, helix angle etc., given their impact on shaft critical speed and efficiency Slide 34
Effect of Bearing Stiffness on Motor Resonance Romax Simulation Critical Speed map Test Data Resonance of 1 st Mode (RPM, no pre-load) Slide 35 0.1% 10% 30% 100% 12 227 35 531 42 531 50 656
Gearbox-Motor Interactions through the Design Process Step 4: edriveline Concept - NOISE Target Efficiency Packaging Durability Driveability Thermal Noise Weight Cost Interaction edriveline Concept Analytical assessment of initial system response to electrical and mechanical excitations by the sound power through the bearings Initial system response to electrical and mechanical excitations Slide 36
How to assess system dynamic response at the concept stage (no housing!) Initial measure for NVH performance of concept design Identify potential NVH issues early Right First Time design Gearbox (SOURCE) Bearing 2 Bearing j Bearing 1 Acoustic power transmitted through bearings Powertrain definition, no housing definition Representative model of generic housing is used Bearing n Sound pressure at a given location Slide 37
Gearbox-Motor Interactions through the Design Process Step 4: edriveline Concept TORQUE RIPPLE Target Efficiency Packaging Durability Driveability Thermal Noise Weight Cost Interaction edriveline Concept Dynamic Fusion/Simulink/Orchestra: Driver can feel torque ripple driving uphill (high T, low speed) RxD: Torque ripple can excite 1 st torsional vehicle mode Electrical machine torque ripple may affect vehicle driveability Slide 38
Investigate driveability through dynamic model of electro-mechanical system Intelligent creation of multi-body dynamics model: Appropriate degrees of freedom for each dynamic problem Optimal computation time Integrates with Adams, Modelica or Simulink This enables driveability investigations and control system design, e.g. anti-jerk control response to high torque ripple at low speed/high torque Slide 39
E-DRIVELINE DETAILED DESIGN Housing design and detailed electromagnetic analysis Slide 40
Gearbox-Motor Interactions through the Design Process Step 4: edriveline Detail Unbalanced Magnetic Pull Target Efficiency Packaging Durability Driveability Thermal Noise Weight Cost Interaction edriveline Detail Analytical/FEA & RxD to assess if airgap length is affected by static UMP and may affect efficiency Bearing selection and position may be affected by UMP Analytical/FEA & RxD to assess if high speed shaft bearing selection is affected by static UMP in electrical machine Static Unbalanced Magnetic Pull (UMP) in electrical machine interacts with gear deflections and may affect bearing selection and/or airgap length Slide 41
Static Unbalanced Magnetic Pull interacts with mechanical system and may affect bearing selection Static system analysis: Deflection of the shaft/bearing/ housing system due to mechanical loads (e.g. gear loads) Rotor eccentricity leads to Unbalanced Magnetic Pull Bearing selection may be affected by UMP Airgap length may be affected by UMP Motor stator Air gap Gearbox housing Motor rotor Slide 42
System and multi-physics interactions for the rotor shaft assembly Durability Efficiency Noise/Dynamics Gearbox related Gear loads affect bearing life Gear loads deflects the rotor Bearings pre-load affects bearing drag and mechanical efficiency UMP also affects gear misalignment UMP also affects gear misalignment and hence micro-geometry & and hence micro-geometry & stress efficiency Bearings pre-load affects bearing durability UMP also affects the bearing loads and hence durability All causes of rotor displacement give rise to need for air gap which affects electrical efficiency Gear loads also affects bearing stiffness and motor lateral dynamics UMP also affects gear misalignment and hence micro-geometry & TE Bearings require pre-load to avoid no-load dynamics problems UMP acts as a negative stiffness and changes eigenvectors Motor related UMP causes the deflection to increase UMP also affects the bearing loads and hence mechanical drag UMP also affects the bearing loads and hence stiffness and eigenvalues Slide 43
Gearbox-Motor Interactions through the Design Process Step 5: edriveline Detail NOISE Target Efficiency Packaging Durability Driveability Thermal Noise Weight Cost Interaction edriveline Detail RxD and electro-magnetic FEA to calculate the system response to electrical and mechanical excitations System response (including housing) to mechanical AND electrical excitations Slide 44
Harmonic analysis to determine electrical machine excitation force shapes Calculation of the excitation from gears (transmission error) and motor (torque ripple, imbalance and radial force shapes) Complex stator radial force shape 8 th harmonic 16 th harmonic 24 th harmonic Harmonic analysis Slide 45
System response to gear AND e-machine excitations H=48, 936 rpm H=48, 8472 rpm 2 nd stage, 6516rpm 1 st stage, 8208rpm Slide 46
Initial NVH analysis is already possible with a concept housing design to guide the design Concept housing design, e.g. space claim functionality Unit excitations for TE, analytical calculation of dominant force shapes Determine where peaks occur in response Torque ripple Radial force shapes Transmission Error Slide 47
Identify potential design actions from the simulation: 1 st Gear Mesh Transmission Error Gearbox casing wall excited by vibration through left hand bearing Potential Design Actions: o o Remove left hand bearing Introduce ribs on offending panel to reduce vibration Slide 48
Gearbox-Motor Interactions through the Design Process Step 5: edriveline Detail DYNAMIC SYSTEM RESPONSE INCLUDING CONTROL Target Efficiency Packaging Durability Driveability Thermal Noise Weight Cost Interaction edriveline Detail Advanced dynamic simulation to assess system dynamics, e.g. shock load response Advanced dynamic simulation to assess effect of control strategies on noise Dynamic system response due to electro-mechanical interactions and control system influences Slide 49
Electro-mechanical system analysis and control design Mechanical system Controlled Electro-mechanical system Electrical & Control system Dynamic Fusion: Discretised model with correct number of DOF Combined system description Including electrical and mechanical representations DQ transformation of machine quantities Include control of electrical machine Slide 50
Control of electrical machine affects electro-mechanical interactions SPM including field weakening Effect of control parameter design on torsional response of combined electro-mechanical system Te* Torquecurrent conversion id* iq* id iq PI controller vd vq D-Q transformation is used for system analysis and control design PI controller is implemented for torque control Electro-mechanical interactions introduce additional LF torsional modes Control parameter selection affects torsional modes Slide 51
Gearbox-Motor Interactions through the Design Process Step 5: edriveline Detail ELECTRO-MAGNETIC/MECHANICAL MACHINE DESIGN Target Efficiency Packaging Durability Driveability Thermal Noise Weight Cost Interaction edriveline Detail Any effects on airgap can affect efficiency, stresses on lamination pack can affect iron losses Shock load response can affect peak torque and may lead to demagnetisation Any longer transients can affect thermal performance Dynamic UMP can cause additional excitations in electrical machine, also axially Interactions of electro-magnetic and mechanical design in electrical machine, e.g. dynamic UMP Slide 52
Areas of ongoing research Unbalanced Magnetic Pull in electrical machines: M. Michon, R. Holehouse, K. Atallah, G. Johnstone, Effects of rotor eccentricity in large synchronous machines, IEEE Transactions on Magnetics 2014 M. Michon, R. Holehouse, K. Atallah, J. Wang, Unbalanced magnetic pull in permanent magnet machines, IET International Conference on Power Electronics, Machines and Drives 2014 M. Michon, K. Atallah, G. Johnstone, Effects of unbalanced magnetic pull in large permanent magnet machines, IEEE Energy Conversion Congress and Exhibition 2014 Smart co-simulation of electro-magnetic and mechanical interactions derive multi-fidelity methods to incorporate into concept design process Knowledge Transfer Partnership with the University of Sheffield System dynamics including electro-mechanical and control interactions: B. Wang, M. Michon, R. Holehouse, K. Atallah, Dynamic behaviour of a multi-mw wind turbine, IEEE Energy Conversion Congress and Exhibition 2015 Slide 53
Gearbox-Motor Interactions through the Design Process Step 5: edriveline Detail MECHANICAL DESIGN FOR TORQUE Target Efficiency Packaging edriveline Detail RxD: Torque -> bolting pattern/mounts -> deflections Durability RxD/Analytical: additional loads on bolted joints between e- machine and gearbox need to be accounted for Driveability Thermal Noise Weight Cost Interaction Electro-magnetic torque affects bolted joints between e- machine and gearbox Slide 54
Torque -> bolting pattern/mounts -> deflections Need to take care of how the structure responds to the torque action/reaction with regard to the gearbox and motor o o o o o Integrity of bolted joint Gear misalignment Air gap (with UMP etc.) Spline rating/misalignment Bearing loads along motor axis 2, 3 or 4 bearings Torque from wheels Reactions from mounts Split plane, shear of bolted joint Slide 55
Gearbox-Motor Interactions through the Design Process Step 5: edriveline Detail SYSTEM THERMAL PERFORMANCE Target Efficiency Packaging Durability Driveability Thermal Noise Weight Cost Interaction edriveline Detail Advanced efficiency calculations to assess how temperature affects gearbox losses FEA for electrical machine efficiency Assessment of temperature hotspots (e.g. windings) Lumped parameter model of complete system, gear meshes and machine losses as source Assessment of system thermal performance including mechanical and electrical parts Slide 56
Thermal modelling of edriveline Heat flow System-level thermal models used to evaluate heat flow through the drivetrain Evaluation of drivetrain thermal performance and efficiency o o Total energy loss over each drive-cycle Fluid and component temperature limits EM losses Splitter EM energy to/from coolant EM heat exchanger EM thermal model GB loss heat exchanger GB losses Conduction between EM and GB GB bulk heat exchanger GB energy to/from coolant GB bulk thermal model Sump Cooling Cooling Environment Slide 57
Thermal/Efficiency/Tribology modelling of edriveline Understanding inter-relationship between the different physics Drive Cycle Heat generated Temperature dependent efficiency models LTCA Oil properties Slide 58
Gearbox-Motor Interactions through the Design Process Summary changes in the methods applied Target Vehicle Concept edriveline Layout Topology Sizing 1 Sizing 2 edriveline Concept Design edriveline Detailed Design edriveline Tolerances & Sign-off Efficiency Packaging Durability Driveability Thermal Noise Increasing model fidelity Increasing input data requirement Decreasing simulation speed Weight Cost Slide 59
Summary A multi-physics, multi-fidelity approach to simulation is necessary for robust design of an edrives system o o Many different functional targets with different physics Multi-fidelity to provide different needs within the design process Products for simulating gearboxes and motors have been developed along this approach, but nothing exists for motorgearbox interactions Slide 60