CONTENTS

Acknowledgments

Preface

Readers' Advantage

Chapter Description

Framing the Multiphysics Design Methodology

CHAPTER 1 Basics of Electrical Machines Design and Manufacturing Tolerances

1.1 Introduction
1.2 Generic Design Flow
1.3 Basic Design and How to Start
1.4 Efficiency Map
1.5 Thermal Constraints
1.6 Robust Design and Manufacturing Tolerances
References

CHAPTER 2 Fem-Based Analysis Techniques for Electrical Machine Design

2.1 T–Ω Formulation
2.2 Field-Circuit Coupling
2.3 Fast Ac Steady-State Algorithm
2.4 High Performance Computing—Time Domain Decomposition
2.5 Reduced Order Modeling
References

CHAPTER 3 Magnetic Material Modeling

3.1 Shape Preserving Interpolation of B–H Curves
3.2 Nonlinear Anisotropic Model
3.3 Dynamic Core Loss Analysis
3.4 Vector Hysteresis Model
3.5 Demagnetization of Permanent Magnets
References

CHAPTER 4 Thermal Problems in Electrical Machines

4.1 Introduction
4.2 Heat Extraction Through Conduction
4.3 Heat Extraction Through Convection
4.4 Heat Extraction Through Radiation
4.5 Cooling Systems Summary
4.6 Thermal Network Based on Lumped Parameters
4.7 Analytical Thermal Network Analysis
4.8 Thermal Analysis using Finite Element Method
4.9 Thermal Analysis Using Computational Fluid Dynamics
4.10 Thermal Parameters Determination
4.11 Losses in brushless permanent magnet machines
4.12 Cooling Systems
4.13 Cooling Examples
References

CHAPTER 5 Automated Optimization for Electric Machines

5.1 Introduction
5.2 Formulating an Optimization Problem
5.3 Optimization Methods
5.4 Design of Experiments and Response Surface Methods
5.5 Differential Evolution
5.6 First Example: Optimization of an Ultra High Torque Density PM Motor for Formula E Racing Cars: Selection of Best Compromise Designs
5.7 Second Example: Single Objective Optimization of a Range of Permanent Magnet Synchronous Machine (PMSMS) Rated Between 1 kW and 1 MW Derivation of Design Proportions and Recommendations
5.8 Third Example: Two- and Three-Objective Function Optimization of a Synchronous Reluctance (SYNREL) and PM Assisted Synchronous Reluctance Motor
5.9 Fourth Example: Multi-Objective Optimization of PM Machines Combining DOE and de Methods
References

CHAPTER 6 Power Electronics and Drive Systems

6.1 Introduction
6.2 Power Electronic Devices
6.3 Circuit-Level Simulation of Drive Systems
6.4 Multiphysics Design Challenges
References

Index

IEEE Press Series on Power Engineering

EULA

List of Tables

Chapter 1

Table 1.1
Table 1.2
Table 1.3
Table 1.4
Table 1.5
Table 1.6
Table 1.7
Table 1.8
Table 1.9
Table 1.10
Table 1.11
Table 1.12
Table 1.13
Table 1.14
Table 1.15
Table 1.16
Table 1.17
Table 1.18

Chapter 2

Table 2.1
Table 2.2
Table 2.3
Table 2.4
Table 2.5
Table 2.6

Chapter 3

Table 3.1
Table 3.2
Table 3.3
Table 3.4
Table 3.5
Table 3.6

Chapter 4

Table 4.1
Table 4.2
Table 4.3
Table 4.4
Table 4.5
Table 4.6
Table 4.7
Table 4.8

Chapter 5

Table 5.1
Table 5.2
Table 5.3
Table 5.4

Chapter 6

Table 6.1
Table 6.2

List of Illustrations

Preface

Figure 1 Multiphysics design methodology.

Chapter 1

Figure 1.1 Typical steps for the manufacturing of an electric machine, in this case a line-fed permanent magnet (PM) synchronous motor, which includes in the rotor a die-cast aluminum cage.
Figure 1.2 Example of stator modular construction with segmented core and concentrated coils forming a multiphase winding [2,3]. Versions with a single tooth and coil modules are also possible.
Figure 1.3 Exploded view of a general-purpose National Electrical Manufacturers Association (NEMA) frame squirrel-cage induction motor. Courtesy of Regal Beloit Corp.
Figure 1.4 PM synchronous motors with interior permanent magnet (IPM) rotors and concentrated (left) and distributed windings, respectively [4]. Courtesy of Regal Beloit Corp.
Figure 1.5 Record-breaking ultra-high density torque 100 hp spoke-type IPM motor for Formula E racing cars showing, from the left, magnetic field in the motor cross section, axial cut view and photo [4,5]. Courtesy of Equipmake, Ltd.
Figure 1.6 Multiphysics and multidomain electric machine design flow.
Figure 1.7 Eight-pole rotor topology.
Figure 1.8 48 slots/8-pole IPM machine configuration.
Figure 1.9 (a) Winding pattern; (b) winding slot detail.
Figure 1.10 Electric drive topology of three-phase inverter, star-connected electric motor with controller.
Figure 1.11 Operation modes in IPM machines.
Figure 1.12 D-Q solution for interior permanent magnet machine design.
Figure 1.13 Maximum torque per ampere unit.
Figure 1.14 Field distribution on rotor and interior permanent magnets at the time instant of maximum input phase current:
(a) flux lines with magnetic field and induced current density distribution on PM,
(b) flux lines with power loss and induced current density distribution on PM.
Figure 1.15 Electromagnetic torque characteristic.
Figure 1.16 Induced voltage profile.
Figure 1.17 Control strategies for AC brushless PM machines.
Figure 1.18 Torque variation in time for US06 drive cycle.
Figure 1.19 Shaft power variation in time for US06 drive cycle.
Figure 1.20 Efficiency maps in motoring and generating modes with superimposed US06 drive cycle points (blue dots).
Figure 1.21 Estimated electrical machine parameters for rated operation point in Motor-CAD / LAB module (100 Nm at 7500 rpm).
Figure 1.22 Three-dimensional view of the stator assembly and cooling jacket.
Figure 1.23 Cooling fluid flow and properties definition in Motor-CAD.
Figure 1.24 Radial view of the steady-state temperature distribution at 100 Nm at 7500 rpm operation point—based on thermal network.
Figure 1.25 Axial view of the steady-state temperature distribution at 100 Nm at 7500 rpm operation point—based on thermal network.
Figure 1.26 Radial view of the steady-state temperature distribution at 100 Nm at 7500 rpm operation point—finite-element thermal analysis.
Figure 1.27 Axial view of the steady-state temperature distribution at 100 Nm at 7500 rpm operation point—finite-element thermal analysis.
Figure 1.28 Continuous torque at average winding temperature <180°C and magnet temperature <140°C.
Figure 1.29 Continuous power at average winding temperature <180°C and magnet temperature <140°C.
Figure 1.30 Continuous current at average winding temperature <180°C and magnet temperature <140°C.
Figure 1.31 Average winding temperature variation with speed at steady-state thermal conditions.
Figure 1.32 Magnet temperature variation with speed at steady-state thermal conditions.
Figure 1.33 ANSYS workflow on robust design analysis based on ANSYS Maxwell as finite element analysis solution.
Figure 1.34 Electric motor design topology: (a) IPM 2D-cross section; (b) stator slot dimensions; (c). IPM rotor dimensions.
Figure 1.35 A whole-coiled winding topology.
Figure 1.36 Design of Experiments creates a reduced parametric table for further robust design analysis.
Figure 1.37 Goodness of Fit representation based on Response Surface analysis.
Figure 1.38 Local sensitivities for selected design variables with respect to defined output parameters.
Figure 1.39 Three-dimensional surface plots: output parameters versus input design variables. (a) Average torque-P6 versus P2 and P4; (b) Ripple torque-P7 versus P2 and P4
Figure 1.40 Three-dimensional surface plots: output parameters versus input design variables. (a) Average torque-P6 versus P3 and P4; (b) Ripple torque-P7 versus P3 and P4
Figure 1.41 Pareto front search.
Figure 1.42 Normal distribution parameters and their graphical representation.
Figure 1.43 Running torque distribution on 10%-10%-10% design variable tolerances.
Figure 1.44 Torque ripple distribution on 10%-10%-10% design variable tolerances.
Figure 1.45 Sensitivity results for 10%-10%-10% design variable tolerances.
Figure 1.46 Running torque distribution on 10%-10%-1% design variable tolerances.
Figure 1.47 Torque ripple distribution on 10%-10%-1% design variable tolerances.
Figure 1.48 Sensitivity results for 10%-10%-1% design variable tolerances.

Chapter 2

Figure 2.1 Flowchart of Newton–Raphson iteration sequence with the use of under-relaxation factor.
Figure 2.2 Search for the optimal under-relaxation factor.
Figure 2.3 One-phase winding and its cutting domain.
Figure 2.4 Induced eddy currents in the bar at t = 0.004 s.
Figure 2.5 Flux density plot in the stator, induced eddy current plot in the bar, and the source currents in the windings at t = 0.0132 s.
Figure 2.6 Induced eddy currents in the bar at t = 0.0088 s.
Figure 2.7 Flux density plot in the stator and rotor, induced eddy current plot in the bar and the source currents in the windings at t = 0.0464 s.
Figure 2.8 Spindle motor fed by inverter with chopped-current control.
Figure 2.9 Hysteresis-controlled phase currents with adaptive time step.
Figure 2.10 The geometry layout of the sample PMDC motor.
Figure 2.11 The flat expanded view of the sample PMDC motor.
Figure 2.12 The commutating circuit of the sample PMDC motor.
Figure 2.13 Contact conductance as a periodic function of rotor position.
Figure 2.14 Different conducting position when WidB > WidC.
Figure 2.15 Different conducting position when WidB < WidC.
Figure 2.16 Load coil commutating current at applied voltage 10.8 V and load speed 3162 rpm compared with that of 10° brush shift.
Figure 2.17 Voltage across the insulation between seg0 and seg1 at applied voltage 10.8 V and load speed 3162 rpm.
Figure 2.18 The 3D FEA model of three-limb three-phase power transformer with 1/4 symmetry.
Figure 2.19 Applied three-phase voltage waveforms without using AF model.
Figure 2.20 Simulated three-phase flux linkage waveforms without using AF model.
Figure 2.21 Simulated three-phase current waveforms without using AF model.
Figure 2.22 Applied three-phase voltage waveforms using the AF model.
Figure 2.23 Simulated three-phase flux linkage waveforms using the AF model.
Figure 2.24 Simulated three-phase current waveforms using the AF model.
Figure 2.25 The 2D FEA model of the 125 MVA non-salient synchronous generator.
Figure 2.26 Coupling circuit for three-phase armature windings.
Figure 2.27 Simulated three-phase current waveforms without using AF algorithm.
Figure 2.28 Simulated torque waveforms without using AF algorithm.
Figure 2.29 Simulated three-phase flux linkage waveforms with the AF model.
Figure 2.30 Simulated three-phase current waveforms with the AF model.
Figure 2.31 Simulated torque waveforms with the AF model.
Figure 2.32 Two subdomains are overlapped.
Figure 2.33 Direct block triangular matrix solver.
Figure 2.34 Flowchart of nonlinear iteration for the time decomposition method.
Figure 2.35 FEA solution domain of 1/8 geometry of 5.5 kW, 4-pole three phase induction motor.
Figure 2.36 8 kW, 10-pole three phase BLDC motor.
Figure 2.37 Three-phase abc axes and dq axes.
Figure 2.38 Circuit model for three-phase PM rotating machines.
Figure 2.39 Circuit model for actuators.
Figure 2.40 Circuit model for three-phase transformers.
Figure 2.41 2D FEA model for the switched reluctance motor with only one-phase winding for current sweep.
Figure 2.42 Circuit connection for four separate one-phase ROMs of 8/6 pole SRM.
Figure 2.43 Simulated current waveforms of the switched reluctance motor for all phases. (a) FEA simulated waveforms; (b) circuit simulated waveforms.
Figure 2.44 Simulated torque waveforms of the switched reluctance motor. (a) FEA simulated waveforms; (b) circuit simulated waveforms.
Figure 2.45 One-pole 2D FEA model for the 4-pole PM generator.
Figure 2.46 Simulated no-load induced voltage waveforms of the PM generator. (a) FEA simulated waveforms; (b) circuit simulated waveforms.
Figure 2.47 Simulated load current waveforms of the PM generator. (a) FEA simulated waveforms; (b) circuit simulated waveforms.
Figure 2.48 Simulated load torque waveforms of the PM generator. (a) FEA simulated waveform; (b) circuit simulated waveform.

Chapter 3

Figure 3.1 A typical B–H curve and its derivative.
Figure 3.2 Piecewise linear interpolant and its derivative.
Figure 3.3 Quadratic interpolation B–H curve and its derivative.
Figure 3.4 Quadratic extrapolation B–H curve and its derivative (the green line for linear extrapolation, the red line for μ₀ extrapolation).
Figure 3.5 Permeability and W--H curve for axis x (or y or z by replacing x by y or z).
Figure 3.6 The iterative process to determine k_μz.
Figure 3.7 Numerical experiment results applying a magnetic field at 0°, 55°, and 90° in M-15 non-oriented steel. T and N denote the tangential and normal directions, respectively.
Figure 3.8 Comparison between the numerical experiment and the computed results by the proposed model and the elliptical model.
Figure 3.9 The structure of a 400 W 4-pole synchronous reluctance motor with axially laminated anisotropic rotor.
Figure 3.10 Three-phase current waveforms when the rotor rotates with a slip of 150 rpm.
Figure 3.11 The d-axis magnetic field distribution.
Figure 3.12 Computed air-gap torque varying with the torque angle. The torque angle varies from 0 to 180 electric degrees as time changes from 0.1 to 0.2 s.
Figure 3.13 The equivalent elliptical loop (EEL) having the same area as the original hysteresis loop.
Figure 3.14 Eddy current produced by the normal component of flux density.
Figure 3.15 Eddy current produced by the tangential components of flux density.
Figure 3.16 Iterative process considering the lamination core loss effects.
Figure 3.17 Instantaneous core loss distribution of the 250 kVA three-phase amorphous metal power transformer.
Figure 3.18 Increase of input electrical power due to the core loss effects compared with the computed core loss (k_h = 21.08, k_c = 0, k_e = 0).
Figure 3.19 Meshed 3D model (without the stator windings) of the 165 W, 4-pole BLDC motor.
Figure 3.20 Input mechanical power increase due to the core loss effects compared with the computed core loss (k_h = 260.4, k_c = 0.822, k_e = 40.54).
Figure 3.21 Play operator for the ordinary play model.
Figure 3.22 Magnetizing curves (H_rev, H_asd, and H_dsc represent anhysteretic, ascending, and descending curves, respectively).
Figure 3.23 Vector diagram for vector play operator.
Figure 3.24 Play operator for the improved play model.
Figure 3.25 Vector diagram to derive h from b.
Figure 3.26 Simulated major loop by the improved play model compared with the measured one.
Figure 3.27 Rotational loss varying with the magnitude of the rotating flux density.
Figure 3.28 Iterative function: (a) H(B) curve; (b) B-correction scheme based on (3.116) and (3.117).
Figure 3.29 Iterative function: (a) B-correction scheme based on (3.116) and (3.118); (b) H-correction scheme based on (3.119) and (3.120).
Figure 3.30 The flowchart for the proposed adaptive fixed point iteration (NR is the Newton–Raphson iteration [35]).
Figure 3.31 The 2D geometry of the inductor with hysteresis core.
Figure 3.32 Simulated hysteresis loop compared with the input major loop.
Figure 3.33 Irreversible demagnetization due to working point below knee point.
Figure 3.34 New working point determination in linearized demagnetization model: (a) B-correction scheme; (b) new scheme.
Figure 3.35 Comparison of Bi-H and B–H characteristics between (a) copied from vendor datasheet, and (b) derived based on the proposed temperature-dependent demagnetization model (α₁ = −0.1, α₂ = 0, β₁ = −0.6, and β₂ = 0).
Figure 3.36 Irrevesible demagnetization when tempurature changes.
Figure 3.37 Three-parameter algorithm.
Figure 3.38 Constructed demagnetization curve for Alnico 5 based on three parameters (B_r = 1.27 T, H_c = 50.9 kA/m, (BH)_max = 43.4 kJ/m³) compared with the original curve.
Figure 3.39 Four-parameter algorithm.
Figure 3.40 Constructed demagnetization curve for Alnico 5 based on four parameters (B_r = 1.27 T, H_c = 50.9 kA/m, (BH)_max = 43.4 kJ/m³, μ_r = 2.8) compared with the original curve.
Figure 3.41 Constructed demagnetization curve for Ceramic 5 based on four parameters (B_r = 0.395 T, H_c = 191 kA/m, B_k = 0.145 T, μ_r = 1.068) compared with the original curve.

Chapter 4

Figure 4.1 Pressed coil for high slot fill factor.
Figure 4.2 Profiled and standard coils [6]. Source: http://www.ifam.fraunhofer.de/
Figure 4.3 Turbulent flow in [8]. Reproduced with permission of Royal Society.
Figure 4.4 Convection heat transfer coefficient variation with velocity [9–11]
Figure 4.5 (a) Servomotor with optimized fins for natural convection, (b) influence of the gap between fins on motor thermal response [19,21].
Figure 4.6 In-wheel electric motor, air cooled and axle water jacket [66]. Reproduced with permission of Protean Electric Ltd.
Figure 4.7 (a) Nissan leaf electric motor; (b) CFD analysis of the cooling system [67].
Figure 4.8 High torque density BPM for electric racing cars [69].
Figure 4.9 Axial cross-section for high torque density motor with dual cooling system: oil through the slots and air through the rotor [69]. Source: Reproduced with permission of IEEE.
Figure 4.10 YASA motor cooling system (a) axial cross-section, (b) 3D view of the motor—US 20130187492 A1 [71].
Figure 4.11 Spray cooled motor.
Figure 4.12 Nissan leaf traction motor—water jacket with oil and air induction system [70].
Figure 4.13 Oil cooled motor/generator for an automotive powertrain—US 8169110 B2 [72].
Figure 4.14 (a) Tesla S induction motor, (b) co-axially cooled stator and rotor—US 2014/0368064 A1 [73].
Figure 4.15 Example of heat transfer network for an electric motor [7].
Figure 4.16 Radial and axial cross-sections of a synchronous wound field machine [7].
Figure 4.17 1 ×1 (top) and 10 × 10 (bottom) rotor winding models [7].
Figure 4.18 Rotor winding discretization results up to 11 × 11 [7].
Figure 4.19 Steady-state 2D finite element analysis of the temperature rise in a slot with rectangular copper conductors.
Figure 4.20 Magnetic flux distribution across machine radial cross-section [7].
Figure 4.21 Model of the fluid flow around end winding in ANSYS Fluent.
Figure 4.22 Model of the fluid flow around stator pack in ANSYS Icepak.
Figure 4.23 Fluid flow and temperature gradients in a three-parallel channels housing jacket (ANSYS Fluent).
Figure 4.24 Demagnetization curves for NdFeB type magnet [1,2].
Figure 4.25 Skin and proximity effects in a bundles coils and strands-in-hand random wound coils in slot.
Figure 4.26 Instantaneous AC losses per slot variation with rotor position at high speed.
Figure 4.27 Current density in a high speed BPM with spoke magnets.
Figure 4.28 Ratio of the AC/DC losses with frequency for slot configuration given in Figure 4.27.
Figure 4.29 Measured relative permeability curves versus peak induction for different frequency levels—fully processed material M43.
Figure 4.30 Dynamic hysteresis loops for peak induction 1 T—fully processed material M43.
Figure 4.31 Iron losses distribution in a BPM under load conditions.
Figure 4.32 Distribution of the induced eddy current density in the magnets for one instance in time.
Figure 4.33 Distribution of specific magnet losses in a BPM under load conditions; losses are averaged for a full mechanical cycle.
Figure 4.34 Effect of segmentation on the magnet losses.
Figure 4.35 Slot equivalent thermal conductivity variation with slot volume fill ratio [26].
Figure 4.36 Housing configurations for BPMs cooling systems.
Figure 4.37 BPM for traction application with S1 operation results [14].
Figure 4.38 Measured and predicted transient thermal data for a BPM servomotor.
Figure 4.39 BPM outer rotor traction motor—steady-state and transient thermal results [18].

Chapter 5

Figure 5.1 Pareto front in a multi-objective optimization problem. Finite element analysis is used for the performance calculation [2].
Figure 5.2 Comparison of machine topologies over a wide range of designs leading to the conclusion that Topology A has the minimum cost per loss among those compared. Parallel computing power is used to solve this large-scale optimization problem.
Figure 5.3 Independent variables in an example interior PM machine [4].
Figure 5.4 Two major approaches for automated design optimization. Combinations between the two are possible.
Figure 5.5 Algorithm for DOE and RS methods.
Figure 5.6 Example response surface plot.
Figure 5.7 Sensitivity analysis: first-order coefficients of the response surface, which can be used to identify which independent input variables affect the output. This applies to the machine topology illustrated in Figure 5.3 [28].
Figure 5.8 Algorithm of differential evolution. In this case, electromagnetic CEFEA is used for fast evaluations of the objective functions [33].
Figure 5.9 Gradual population of the Pareto set. At the start of the process, the Pareto set is sparse, and gets gradually more populated as the solution progresses [31].
Figure 5.10 Independent design variables of the spoke type IPM machine [35].
Figure 5.11 Results of the optimization study. D1, D2, and D3 represent best compromise designs selected out of 3400 candidates [35].
Figure 5.12 (a) The distribution of values of the split ratio, and (b) the ratio of PM thickness to pole pitch for 100 designs on the Pareto front [35].
Figure 5.13 (a) Cross section of the 6-pole PMSM considered for study and (b) sensitivity of the objective function to the different independent input variables [3].
Figure 5.14 Results of the optimization study. Optimum values of design variables such as rotor pole arc, height of stator yoke, and width of stator tooth vary nonlinearly with the power rating [3].
Figure 5.15 Cross section of the studied machine without and with PMs in the barriers. Nine independent variables which relate to the shape of the rotor flux barriers are considered [38].
Figure 5.16 Flux lines in (a) synchronous reluctance and (b) PM assisted synchronous reluctance machine [37, 38].
Figure 5.17 Results of the optimization study showing the performance of the best compromise designs for the (a) constrained and (b) unconstrained optimization problems. The optimal designs have torque ripple below 20%, and a power factor of 0.5–0.6, with badness of 0.6. The lower index of the best compromise design in the unconstrained optimization problem indicates that the solution was arrived at faster [37].
Figure 5.18 Pareto surface for the unconstrained problem [37].
Figure 5.19 Pareto sets for the SynRel and the PMaSynRel including 10,000 candidate designs. The PMaSynRel designs can achieve lower badness and higher power factor [38].
Figure 5.20 Cross sections of the studied IPM machines: (a) Spoke PM, (b) V-type, and (c) flat bar [33].
Figure 5.21 Pareto fronts for the studied machines. It is seen that all designs with NdFeB PMs, irrespective of rotor topology have similar performance, while the spoke-PM designs with ferrites have lower cost [33].

Chapter 6

Figure 6.1 General electrical drive system schematic for electric machines.
Figure 6.2 Classification of power electronic devices.
Figure 6.3 Typical structure of a PiN diode (cross section) and its electrical symbol [8].
Figure 6.4 Real characteristic of a PiN power diode.
Figure 6.5 Electric field and carrier distributions in off state and on state of a PiN diode [8].
Figure 6.6 Typical switching waveforms for a PiN diode.
Figure 6.7 Cross section of a horizontal, N-channel power MOSFET's cell (a), cross section of a trench power MOSFET structure (b), and electrical symbol of a power MOSFET (c) [8].
Figure 6.8 Output characteristics of a power MOSFET.
Figure 6.9 Switching waveforms of a power MOSFET (a) and equivalent circuit diagram of a MOSFET during switching (b).
Figure 6.10 Cross section of a horizontal, N-channel IGBT cell (a), cross section of a trench IGBT structure (b), electrical symbol of an IGBT (c) [8].
Figure 6.11 Typical flowchart of a circuit-based simulation.
Figure 6.12 Screenshot of the Simplorer schematic editor environment.
Figure 6.13 Screenshot of the Simplorer simulation results viewer.
Figure 6.14 Three-phase, full-bridge diode rectifier with finite inductance on the AC side.
Figure 6.15 View of the three-phase diode rectifier exported from Simplorer schematic editor.
Figure 6.16 Simulation output: voltage waveforms (top), current waveforms (bottom).
Figure 6.17 View of the three-phase inverter exported from the Simplorer schematic editor.
Figure 6.18 A pulse-width modulation technique (a) and circuit schematic of a gate drive (b).
Figure 6.19 Simplorer simulation output for a three-phase sinusoidal PWM.
Figure 6.20 Simulation output: phase current waveforms (top) and phase voltage waveforms (bottom).
Figure 6.21 View of the boost-type PFC exported from Simplorer schematic editor.
Figure 6.22 Simulation output: filtered phase current and voltage waveforms after the insertion of a PFC.
Figure 6.23 Simulation output: phase current and voltage waveforms after the insertion of a PFC.
Figure 6.24 Typical cross section of a semiconductor power module.
Figure 6.25 Equivalent thermal network for the structure presented in Figure 6.24.
Figure 6.26 View of the three-phase inverter exported from the Simplorer schematic editor.
Figure 6.27 Typical voltage and current waveforms across an inverter IGBT switch.
Figure 6.28 Instantaneous power loss waveform across an inverter IGBT switch.
Figure 6.29 ANSYS SpaceClaim CAD structure for the inverter with IGBT power modules.
Figure 6.30 Simulated structure in ANSYS Icepak environment with finite volume mesh.
Figure 6.31 Junction temperature trnasient profile.
Figure 6.32 Temperature distribution for the simulated IGBT package structure at t = 0.3 s.

Preface

Electric machines are being used in wide and novel applications throughout the world, driven by the need for greater power efficiency in the transportation, aerospace and defense, and industrial automation markets. The automotive sector is driven by the need for hybrid and electric vehicle technology to meet ever-stringent miles-per-gallon standards. The aerospace and defense sectors are focused on replacing existing power transfer technologies in an aircraft such as the central hydraulic system, with fault-tolerant electric power, where major subsystems such as engine starting, primary flight control actuation, pumps, and braking would be controlled and driven electronically. In the US industrial sector, over 40 million electric motors convert electricity into useful work in manufacturing operations. Industry spends over $30 billion (US) annually on electricity, dedicated to electric motor-driven systems that drive pumps, fan and blower systems, air compression, and motion control. Globally, 42% of all electricity is used in power industries, where two-thirds of this is consumed by electric motors. There is a clear global demand for a comprehensive design methodology to support these new applications and satisfy power efficiency requirements.

With the present trend of global industrial automation, the application of electric drive systems (including power electronics and drive control) is expected to grow rapidly in the next decade. In the automotive sector, the utilization of power electronics and their control to drive electric motors can significantly contribute to control environmental pollution. In addition, intensive environmentally clean photovoltaic and wind energy resources also show a bright future.

As part of electric drive systems, the power semiconductor devices at the heart of modern power electronics are under continuous development. The improved technology in semiconductor processing, device fabrication, and packaging to produce high-density, high-performance, high-reliability, and high-yielding microelectronic chips, together with new semiconductor material discovery, made possible significant reduction in energy consumption, driving these systems to an incredible economical price.

Without doubt, these achievements force the control strategy techniques to evolve rapidly to the newly created drive conditions and adapt to the overall systems performance requirements. In recent years, soft switching converters became the center of interest when compared with more conventional hard switching converters due to their major advantages such as:

Minimization of switching loss
Improved efficiency
Improved reliability due to soft stress
Reduced electromagnetic emission

The continuous growing interest in the electric drive area relates to the intelligent power electronics modules, where the power and the control are embedded in the same package and interface directly with logic signals. For variable frequency drives, the converter modules and control are mounted directly on the machine for the low and medium power applications.

Readers' Advantage

The book is mainly addressed to design engineers, application engineers, technical professionals, and graduate engineering students with a strong interest in electric machines and drives.

The comprehensive design approach described in this book supports new applications required by technologies, sustaining high drive efficiency. The highlighted framework considers the electric machine at the heart of the entire electric drive. The book delivers the multiphysics know-how based on practical electric machine design methodologies. Simulation by design concept elevated in the book constitutes the new paradigm that frames the entire highlighted design methodology, which is described and illustrated by various advanced simulation technologies.

Which Design Problems Are We Trying to Solve?

Throughout this book, we apply knowledge of design best practices into multiphysics and multidomain simulation processes to address a complete electrical machine and drive design.

In the face of global competition, electric machine manufacturers, like manufacturers in most industries, are searching for ways to reduce cost, optimize designs, and deliver them quickly to market. Companies able to achieve these objectives hold a competitive advantage in the marketplace. The ability to predict design performance with simulation software without the time and expense of constructing prototypes plays a significant role in creating this competitive advantage.

Several computation approaches are available to predict electric machine performance, including classical closed-form analytical analysis, lumped parameter models based on the determination of detailed parameters from finite element analysis, and nonlinear time-domain finite element analysis. Each method has advantages and disadvantages. Selecting the best method may not be straightforward because it requires the user to understand the differences among the calculation methods. The fundamental issue differentiating these methods is the trade-off among model complexity, accuracy, and computing time. Engineers use a combination of these calculation techniques as the optimal solution to simulate electric machine performances.

What Motivated Us to Write This Book?

This collaborative work brought together a group of experts from both academia and software industry with strong expertise on electrical machine design and manufacturing. The main idea that fueled our initiative and commitment to make this project a reality was to bring back to the engineering and academic communities a comprehensive expertise and validated know-how on designing electrical machines by simulation.

Why Simulation by Design?

The advancements in modern digital computers brought CAD (computer-aided design) and CAE (computer-aided engineering) tools at the heart of virtual prototyping, reducing the time to design and market and saving cost by reducing and eliminating the physical prototyping need. The embedded 3D-physics design into drive system coupling with the power electronics and control algorithms enables the electric drive community to accurately predict the efficiency and performance of the electrical machine at the heart of the entire drive system.

Without doubt, the design of a simulation model—a virtual prototype—can help tremendously the engineers to build confidence on validating the required technical specifications making critical decisions on design realization and understanding the level of design complexity considering inter-dependencies and design parameter variations, and collaboratively to examine strategic choices for optimization and robustness.

Chapter Description

Chapter 1: Basics of Electrical Machines Design and Manufacturing Tolerances

This chapter discusses fundamental aspects of the state-of-the-art design process and includes examples from industrial practice and case studies to introduce basic concepts and methods. This chapter emphasizes the basic steps in designing a typical electrical machine using power traction application as an example. The chapter starts with magnetic sizing steps and it extends the basic design to thermal constraints. Typical electric motor characteristics used in traction applications such as efficiency map of standardized driving cycles are considered to highlight the electric motor sustainability on dynamic performance. The chapter concludes with the robust design analysis framing a methodology that applies stochastic analysis to study manufacturing tolerances.

Chapter 2: FEM-Based Analysis Techniques for Electrical Machine Design

In this chapter, a detailed description of finite element method (FEM) employed in ANSYS Maxwell software is presented. The numerical technique developed to account for eddy currents in conductive domains on configurations that involve rigid motion is presented, the numerical technique related to multiply connected regions is highlighted, and it also presents the algorithms used for nonlinear iterations and strategies to accelerate the nonlinear convergence. Filed-circuit coupling technology is explained and specific algorithms used to reduce the computation time to reach steady-state conditions are described. High-performance computing (HPC) is a key technology, increasing the capacity of solving large design spaces and reducing significantly the total time computation by solving the time steps on magnetic transient problem simultaneously rather than sequentially. All technologies highlighted in this chapter are explained through sets of case studies.

Chapter 3: Magnetic Material Modeling

This chapter introduces advanced magnetic material modeling capabilities employed in numerical computation. From isotropic nonlinear characteristics to anisotropic behavior corresponding to grain-oriented magnetic materials, the chapter describes the implementation aspects and detailed modeling techniques. Lamination topologies are considered based on special modeling technique with emphasis on core loss computation. Advanced magnetic modeling on vector magnetic hysteresis is presented and specific case studies are used to highlight the computational merits.

Chapter 4: Thermal Problems in Electrical Machines

In this chapter, the heat generation and extraction in electrical motors are investigated. Using the three thermal paths—conduction, convection, and radiation—an electromagnetic device can be cooled within the acceptable limits for the environment and corresponding application. A highly efficient electrical machine is required in most industrial fields, but the high efficiency is not telling us the full story of a good motor performance. The losses—electromagnetic and mechanical—must be dissipated from the machine into the ambient and the mode in which the cooling system manages to do that represents the key in a reliable and high-performance electrical machine. Within the chapter, the theoretical aspects of thermal management are illustrated with a state-of-the-art collection of practical examples for cooling electrical machines published in the literature.

Chapter 5: Automated Optimization for Electric Machines

This chapter discusses optimization as applied to electrical machine design. Some commonly used optimization methods are explained. Case studies illustrating the utility of systematic design optimization to compare different machine topologies, to develop design rules, and to quantify the effect of different design features are included.

Chapter 6: Power Electronics and Drive Systems

This chapter describes the entire drive system from semiconductor as the main component of any modern power electronics circuit to more complex topologies that include active components to rectify the energy, reduce harmonic distortions, and correct power factors in various drive systems. Electrical machines need drive systems to be correctly controlled if they need to be operated at variable speed. This can be achieved by modulating the energy flow to/from them. The chapter also highlights the need of multiphysics studies for such designs to account for thermal analysis under certain cooling conditions. For instance, inverter modules need a careful design approach as losses vary continuously during normal operation. Poor thermal management can lead to overheating and thus degrade the reliability of the components.

Framing the Multiphysics Design Methodology

The electric machine is a very complex device, being multidomain by nature involving electromagnetics, thermal, and mechanical aspects. The multiphysics methodology built around the core of electric machine design encompasses a systematic approach to develop a platform where comprehensive analysis is the key to understand and design a complex drive system to predict their performances and analyze their robustness. The multiphysics simulation technology enables users to design, analyze, and deliver efficient, optimized electric machine and drive designs.

As shown in Figure 1, the first step in the overall workflow is to develop design requirements. Those requirements may be created within a particular design organization, or they may be provided from a purchaser of the electric machine. Requirements may include machine speed, output power, input power, torque, efficiency, thermal properties, weight, size, etc. At this stage, motor sizing and model creation take place, where many motor configurations may be considered. Often, engineers will use classical, closed-form analytical methods to select appropriate motor configurations that will meet requirements. In a similar manner, both magnetic and thermal designs can be evaluated using template-based approaches that are using such closed-form analytical methods. At the end of this stage, the designer acquires knowledge about the most suitable motor topology to fulfill the level of technical specifications with a degree of confidence on practical realization.

Flow diagram shows labels for stage I, stage II, stage III, and stage IV where design requirements (template-based, spreadsheet) leads to performance (electromagnetics), leads to cooling (thermal), and finally leads to manufacturability (structural). — **Figure 1** Multiphysics design methodology.

At the second stage of the workflow with the set of knowledge already acquired, accurate and detailed motor studies using 2D and 3D finite element analysis are performed. This important step in the design process further qualifies electromagnetically the topologies selected by the magnetothermal sizing analysis. Various design characteristics are numerically evaluated employing cutting-edge techniques, for example, permanent magnet demagnetization due to irreversible temperature effect, power electronics switching loss effect on electric motor core-loss, efficiency and power loss maps.

The thermal study can be developed similar to electromagnetic analysis as a separate design simulation or in connection with electromagnetic solution. With feedback from electromagnetic losses from either template-based solution or finite element analyses, the thermal study can be migrated from simple temperature rise computation based on thermal conduction to more complex studies involving computational fluid dynamics (CFD), where convection and radiation are considered. In such configurations, detailed cooling systems can be evaluated and optimized. Speeding up the entire thermal profile prediction is a key in the overall design process, that provides with design alternatives as,

– Creating an equivalent thermal model (ETM) for motor topology to be used within CFD environment to build around it the physical cooling system configuration with focus on outside enclosure optimization analysis or,
– Coupling detailed finite element-based integrated losses with thermal simulation model to accurately predict temperature profile, fluid flow, and its velocity within various parts of the entire assembly.

The final stage of the proposed workflow relates to mechanical design and manufacturability. Although this step follows electromagnetic and thermal analyses, the structural simulation can be performed any time during the multiphysics design process with emphasis on specific mechanical analyses, for example, deformation studies, noise vibration, and structural dynamics analysis, to more complex induced thermal stress and magnetostrictive analyses.

This simulation framework allows the engineer to understand the electrical, thermal, structural, and acoustical behavior of the design, considering electric motor as an independent component or part of an electric drive system including power electronics. Finally, the motor design is considered in the broader context of its power control unit and integration with other systems.

The flexibility of such a design flow is provided by the data exchange among all physics involved, providing various design adoption alternatives.

The multiphysics design flow can be further detailed at each and every individual stage. In spite of this granularity, the Chapter 1 will focus on Stage I regarding the generic design flow for topology selection during a motor design to examine the process of basic design.

Marius Rosu
Ping Zhou
Dingsheng Lin
Dan Ionel
Mircea Popescu
Frede Blaabjerg
Vandana Rallabandi
David Staton

Giancarlo Fortino	Xiaoou Li	Ray Perez
Dmitry Goldgof	Andreas Molisch	Linda Shafer
Don Heirman	Saeid Nahavandi	Mohammad Shahidehpour
Ekram Hossain	Jeffrey Nanzer	Zidong Wang