BLDC Motor Drive Controller for Electric Vehicles

Transcription

1 BLDC Motor Drive Controller for Electric Vehicles Alireza Tashakori Abkenar Faculty of Science, Engineering and Technology Swinburne University of Technology A thesis submitted for the degree of Doctor of Philosophy May 2014

2 I would like to dedicate this thesis to my loving parents.

3 Abstract Electric vehicles are the best solution for green transportation due to their high efficiency and zero greenhouse gas emissions. Various electric motors have been used as the propulsion system of electric vehicles. Performance of brushed Direct Current (DC) motors, induction motors, switched reluctance motors, and permanent magnet Brushless DC (BLDC) motors are compared according to the in-wheel motor technology requirements under normal and critical conditions through simulation. This study shows that BLDC motors are the most suitable electric motor for the high performance electric vehicles. An accurate model of a BLDC motor is needed to investigate the motor performance for different control algorithms. Therefore a BLDC motor with an ideal back-electro Motive Force (EMF) voltage and its control drive are modelled in Simulink. Correct performance of the BLDC motor drive model is validated through experimental data. Direct torque control technique is a type of flux linkage based sensorless control methods in the BLDC motors. In this thesis, direct torque control switching technique of the BLDC motor is discussed. Results of this study show effective torque control, reduction of torque ripples and improved performance of the BLDC motor compared to the conventional switching control techniques. An optimized back-emf zero crossing detection based sensorless technique of the BLDC motor is presented in this thesis. The proposed sensorless algorithm generates commutation signals of the BLDC motor according to back-emf zero crossing detection points of only one phase of the motor. This algorithm is simple and remarkably reduces sensing circuitry, noise susceptibility and cost of the sensorless BLDC motor drives. A digital pulse width modulation (PWM) switching technique is implemented to control the speed of the BLDC motor. Stability of the proposed sensorless BLDC motor drive using a digital PWM speed controller is analysed by Lyapunov s second method. A novel condition for duty cycle of the PWM speed controller is introduced for stability analysis of the BLDC motor drive. Effectiveness of the proposed sensorless algorithm and correctness of the introduced PWM controller stability condition are verified ii

4 Abstract through simulation and experimental results. Robust performance of the in-wheel BLDC motor drives is an important factor in safety of the electric vehicles. Effect of inverter switch faults of an in-wheel BLDC motor on performance of the four wheel drive electric vehicle is studied through simulation. Results show unstable performance of the electric vehicle after fault occurrence and demonstrate need of the fault tolerant control system for the in-wheel motors. This thesis presents two novel fault tolerant control systems for inverter switch faults and position detection sensor faults in the BLDC motor drives. Performance of the BLDC motor is studied under various fault conditions through a validated simulation model. Knowledge based tables were developed to diagnose the inverter switch and Hall Effect sensor faults based on discrete Fourier transform analysis of the BLDC motor line voltages. The developed fault diagnosis algorithms are simple and capable of detecting the fault occurrence, identify fault type and the faulty switch or position sensor of the BLDC motor drive. Simulation results and the proposed knowledge based fault diagnosis tables are validated through experimental data. The proposed fault diagnosis algorithms do not need massive computational effort and can be implemented as a subroutine of the main control algorithm of the BLDC motor. iii

5 Acknowledgement First and foremost my deepest gratitude goes to my supervisor Dr. Mehran Motamed Ektesabi for accepting me as a PhD student. I would like to thank for his guidance and support not only on the research topic but also in my personal life throughout these years. Our regular meetings and discussions helped me a lot through my research during my PhD candidature. I gratefully acknowledge the financial, academic and technical support of the Faculty of Engineering and Industrial Science, Swinburne University of Technology and its staff that made my PhD research work possible. Lastly, I would like to thank my family specially my parents, to whom I dedicate this thesis. Words can not express how grateful I am to my mother, father and my sister for their love, encouragement and all of the sacrifices that they have made on my behalf. Alireza Tashakori Abkenar iv

6 Declaration I hereby declare that this Ph.D. thesis entitled BLDC Motor Drive Controller for Electric Vehicles has been compiled by me under the supervision of Dr. Mehran Motamed Ektesabi at Faculty of Engineering and Industrial science, Swinburne University of Technology, Melbourne, Australia. This thesis contains no material which has been accepted for the award of any other degree or diploma, except where due reference is made. To the best of my knowledge, this thesis contains no material previously published or written by another person except where due reference is made in the text of the thesis. Alireza Tashakori Abkenar Place: Melbourne Date: v

7 Publication Portions of the material in this thesis have previously appeared in the following publications: Book Chapter: 1. A. Tashakori and M. Ektesabi, Direct torque control of in-wheel bldc motor used in electric vehicle, In Gi-Chul Yang, Sio-long Ao, and Len Gelman, editors, IAENG Transactions on Engineering Technologies, volume 229 of Lecture Notes in Electrical Engineering, pp , Springer Netherlands, Journals: 2. A. Tashakori and M. Ektesabi, Position sensors fault tolerant control system in BLDC motors, Engineering Letters, Volume 22 Issue 1, pp , Feb A. Tashakori and M. Ektesabi, Comparison of different PWM switching modes of BLDC motor as drive train of electric vehicles, World Academy of Science, Journal of Engineering and Technology 2012, Vol. 67, pp Peer Reviewed Conference Papers: 4. A. Tashakori and M. Ektesabi, Fault Diagnosis of In-wheel BLDC Motor Drive for Electric Vehicle Application, Proceeding of the 2013 IEEE Intelligent Vehicles Symposium, pp , June 2013, Gold Coast Australia. 5. A. Tashakori and M. Ektesabi, A simple fault tolerant control system for Hall Effect sensors failure of BLDC motor, Proceeding of the 8th IEEE Conference on Industrial Electronics and Applications (ICIEA 2013), pp , June 2013, Melbourne Australia. 6. A. Tashakori and M. Ektesabi, Stability analysis of sensorless BLDC motor drive using digital PWM technique for electric vehicles, Proceeding of 38th Annual Conference on IEEE Industrial Electronics Society (IECON 2012), pp , October 2012, Montreal Canada. vi

8 Publication 7. A. Tashakori, M. Ektesabi, Direct torque controlled drive train for electric vehicle, Lecturer notes in engineering and computer science: Proceeding of the world congress on engineering 2012 (WCE 2012), pp , July 2012, London UK. 8. A. Tashakori, M. Ektesabi, and N. Hosseinzadeh, Characteristics of suitable drive train for electric vehicle, in Proceeding of the International Conference on Instrumentation, Measurement, Circuits and Systems (ICIMCS 2011), Vol. 2, pp , ASME, A. Tashakori, M. Ektesabi and N. Hosseinzadeh, Modelling of BLDC motor with ideal back-emf for automation application, Lecture Notes in Engineering and Computer Science: Proceedings of The World Congress on Engineering 2011 (WCE 2011), Vol. 2, pp , July 2011, London UK. vii

9 Contents Contents Nomenclature List of Figures List of Tables viii xiii xiv xix 1 Introduction 1 2 Selection of a Suitable Motor for Electric Vehicles Chapter Overview Introduction The Drive Train of Electric Vehicles Conventional AC and DC Motors Switched Reluctance Motors BLDC Motors Motor Comparison Performance Comparison of the Motors under Normal Condition Performance Comparison of the Motors under Critical Condition Transient Electric Faults Vibration and Mechanical Shocks Conclusion viii

10 CONTENTS 3 Modelling of the BLDC Motor Drive for EV Application Chapter Overview Introduction Overall View of the BLDC Motor Drive Modelling of the BLDC Motor Modelling of the BLDC Motor Drive Simulation Results and Discussion Simulation Model Validation Conclusion Direct Torque Control Drive of BLDC Motor for EV Application Chapter Overview Introduction Direct Torque Control of the BLDC Motor Using Three Phase Conduction Mode Simulation Results and Discussion Experimental Results Conclusion Stability Analysis of a Novel Sensorless Drive of BLDC Motor Chapter Overview Introduction Proposed Sensorless Technique for BLDC Motor Stability Analysis of Digital PWM Controller Simulation Results and Discussion Experiment Results Conclusion Fault Diagnosis of the BLDC Motor Drive for EV Application Chapter Overview Introduction Inverter Open Circuit Switch Faults EV Dynamics Analysis under Inverter Open Circuit Switch Fault ix

11 CONTENTS No Fault Condition VSI Open Circuit Fault Fault Diagnosis Fault Detection Fault Identification Experimental Results Remedial Strategy Position Detection Sensors Failure Performance of the BLDC Motor under Position Sensor Faults Hall Effect Signal is Constant Zero Hall Effect Signal is Constant One Fault Diagnosis Experimental Results Remedial Strategy Conclusion Conclusion 137 References 141 Appendix A 154 A Reference Links of Table B Details of the motor models in Chapter C State Space Equation of BLDC Motor D Clarke Transformation E Lyapunov s Second Method for Stability F Comaprison of Different PWM Switching Techniques of The BLDC Motor F.1 Normal Condition F.2 Critical Condition F.2.1 Mechanical Shocks F.2.2 Inverter Switch Faults G EV Model Simulation Results under Inverter Open Circuit Switch Fault x

12 CONTENTS Appendix B 170 xi

13 Nomenclature Roman Symbols ω ref ω m Θ S θ e θ m ϕ rα ϕ rβ ϕ Sα ϕ Sβ E e α e β Reference speed of the controller Angular speed of the rotor Stator flux angle Electrical angle of the rotor Mechanical angle of the rotor α-axis rotor flux vector β-axis rotor flux vector α-axis stator flux vector β-axis stator flux vector Back-EMF voltage α-axis back-emf β-axis back-emf F (θ e ) Reference back-emf signals of the BLDC motor with respect to the electrical angle of the rotor i i α Current α-axis current vector xii

14 Nomenclature i β K e K L K t L M P R T e T l V DC V V α V β β-axis current vector Back-EMF constant flux linkage of the BLDC motor Torque constant Inductance Mutual inductance Number of poles Resistance Electric torque Load torque Voltage of the inverter DC link Voltage α-axis voltage vector β-axis voltage vector xiii

15 List of Figures 2.1 Four wheel drive train of an IECEV Various switched reluctance motor geometries Schematic diagram of a two pole BLDC motor drive The internal view of a BLDC motor Ideal current, back-emf and commutation signals of BLDC motor Transient speed responses of the motors under normal condition Transient torque responses of the motors under normal condition Transient torque/speed characteristics of the motors Speed responses of the motors under same transient electrical fault Torque responses of the motors under same transient electrical fault Speed responses of the motors under same mechanical shocks Torque responses of the motors under same mechanical shocks Overall structure of the BLDC motor drive Ideal reference back-emf waveforms of the BLDC motor model BLDC motor simulation model Schematic diagram of a 3 phase, 4 poles, star connected BLDC motor drive Three phase VSI simulation model Speed characteristics of the BLDC motor simulation model Torque characteristics of the BLDC motor simulation model Voltage, Current and Hall Effect signal of phase A of the BLDC motor Back-EMF signals of the BLDC motor model Experimental test set-up of the BLDC motor xiv

16 LIST OF FIGURES 3.11 Line voltage and Hall Effect signal of phase A of the BLDC motor Overall structure of DTC drive of the BLDC motor Speed and torque responses of the direct torque controlled BLDC motor drive Pulsating torque of the BLDC motor for different hysteresis band limits Calculated stator flux magnitude and flux angle of the BLDC motor Stator flux linkage trajectory of the BLDC motor for 5 and 10 N.m loads Speed response of the BLDC motor under sudden increase of load Torque response of the BLDC motor under sudden increase of load Experimental set-up of the BLDC motor Torque characteristics of the experimental BLDC motor Equivalent electrical circuit of the BLDC motor drive Ideal commutation signals, terminal and back-emf voltages of the BLDC motor Schematic diagram of the proposed BLDC motor sensorless drive Line voltage, Back-EMF and ZCD points of phase A of BLDC motor Zero crossing points and the commutation signal of phase A Current, commutation signal and ZCD points of phase A Speed response of the BLDC motor and duty cycle values selected by PI controller State plane of digital PWM speed controller Speed and torque characteristics of the BLDC motor during brake Duty cycle values during the brake condition State plane of digital PWM speed controller during the brake Experimental speed response of the sensorless BLDC motor drive Experimental speed response of the BLDC motor drive using sensors Generated commutation signals by sensorless drive of BLDC motor PWM switching signals applied to the upper side switches of VSI Line voltage and commutation signal of the phase C of BLDC motor The in-wheel BLDC motor set-up in a light weight EV xv

17 LIST OF FIGURES 5.18 Line voltage and commutation signal of the in-wheel BLDC motor at different operating condition of the light weight EV Overall BLDC motor drive model BLDC motor output characteristics and VSI switching steps Schematic diagram of the four in-wheel drive EV model EV speed under no fault condition Normal tire forces under no fault condition Speed responses of the BLDC motors under no fault condition Torque responses of the BLDC motors under no fault condition EV speed under open circuit fault of switch S Normal tire forces under open circuit fault of switch S Torque responses of the BLDC motors under open circuit fault of switch S Speed responses of the BLDC motors under open circuit fault of switch S Line voltage and Hall Effect signal of phase A of BLDC motor Line voltages of BLDC motor during open circuit fault of switch S Line voltages of BLDC motor during open circuit fault of switch S The modified LV development board control drive of BLDC motor Line voltages of BLDC motor under open circuit fault of switch S Line voltages of BLDC motor under open circuit fault of switch S Schematic diagram of the proposed four switches topology inverter Schematic diagram of the proposed fault tolerant inverter with a redundant leg Speed and torque responses of BLDC motor under H a = 0 fault condition Line voltages of BLDC motor under H a = 0 fault condition Line voltages of BLDC motor under H a = 1 fault condition Amplitude spectrum of the phase A line voltage of BLDC motor Half-bridge gate driver and inverter of LV development board xvi

18 LIST OF FIGURES 6.25 Corresponding switching LED lights on the control board under position sensor faults of phase A: (a) Open circuit fault (b) Short circuit fault Line voltages of the experimental BLDC motor under H a = 0 fault Line voltages of the experimental BLDC motor under H a = 1 fault Amplitude spectrum of the phase A line voltage of experimental BLDC motor Speed response of the fault tolerant controlled BLDC motor drive 135 B1 Block diagram of the induction motor drive model B2 Block diagram of the DC motor drive model B3 Block diagram of the switched reluctance motor drive model B4 Block diagram of the BLDC motor drive model F1 Speed responses of BLDC motor for different PWM switching modes160 F2 Torque responses of BLDC motor for different PWM switching modes F3 Torque responses of BLDC motor for different PWM switching modes F4 Line voltage of BLDC motor for different PWM switching modes. 162 F5 Duty cycle chosen by PI controller for different PWM switching modes F6 Torque responses of the BLDC motor under mechanical shocks for different PWM switching modes F7 Duty cycle chosen by PI controller under mechanical shocks for different PWM switching modes F8 Speed responses of the BLDC motor under inverter switch faults for different PWM switching modes F9 Duty cycle chosen by PI controller under inverter switch faults for different PWM switching modes G1 EV speed characteristics under open circuit fault of switch S G2 Normal tire forces under open circuit fault of switch S G3 Torque characteristics of the BLDC motors under open circuit fault of switch S xvii

19 LIST OF FIGURES G4 Speed characteristics of the BLDC motors under open circuit fault of switch S xviii

20 List of Tables 2.1 Drive Train Specifications of the Electric Vehicles Available in the World Market Brushed DC Motor Specifications Induction Motor Specifications Switched Reluctance Motor Specifications BLDC Motor Specifications Motors Comparison According to the In-wheel Motor Specifications Hall Effect Signals and Inverter Switches Status of the BLDC Motor Specifications of the BLDC Motor Model BLK423S Specifications of the Experimental In-wheel BLDC Motor Three Phase Conduction Switching Mode for DTC of the BLDC Motor Specification of BLDC Motor Used in Simulation Model Specifications of the Experimental BLDC Motor Common Faults in the BLDC Motor Drive Specification of the Vehicle s Body Used in the EV Model Simulation PSD Values for Open Circuit of S Simulation PSD Values for Open Circuit of S Proposed Knowledge Based Table for Inverter Switches Faults Diagnosis Experimental PSD Values for Open Circuit of S Experimental PSD Values for Open Circuit of S xix

21 LIST OF TABLES 6.8 Effect of the Various Sensor Faults on the Switching Signals of the VSI PSD Values for H a = 0 Fault Condition PSD Values for H a = 1 Fault Condition Proposed Knowledge Based Table for Position Sensor Faults Diagnosis PSD Values for Experimental H a = 0 Fault Condition PSD Values for Experimental H a = 1 Fault Condition xx

22 Chapter 1 Introduction The idea of employing electric power instead of fossil fuels as motive energy of vehicles is not new. Scientists and manufacturers have attempted to design an Electric Vehicle (EV) since long time ago. Rodert Anderson had built the first electric carriage in 1839 and David Salomon developed an electric car using a light electric motor in 1870 [1]. Since then, the heavy electric batteries and poor performance electric motors were the main concern. Interest on electric vehicles reduced due to development of electric self-starters for the gasoline vehicles and low price of oil, until early 1980 s when environmental concerns raised up [2]. Nowadays, hybrid electric vehicles are more popular than pure electric vehicles due to the better range and lack of enough infrastructures for charging battery. Conventional electric vehicles have a central electric motor that actuates two or all four wheels of the vehicle [3]. In-wheel motor technology is of interest for high performance electric vehicles by researchers and auto-mobile manufacturers in recent years. However the in-wheel motor idea first introduced in 1884 by Wellington Adams who have built and attached an electric motor directly in the vehicle s wheel through complicated gearings. In-wheel motor electric vehicles employ motors embedded inside each wheel. Since in an in-wheel motor EV individual control of each wheel is possible; better vehicle speed, torque and acceleration control can be achieved. Using in-wheel motor technology improves drive train efficiency, dynamic stability control and safety of electric vehicles [4][5]. 1

23 1. Introduction As mentioned earlier, poor performance of the electric motors has been of concern by researchers and various electric motor types have been used in electric vehicles so far. There is always an important question, what is the most suitable electric motor for electric vehicles? The answer to this question depends highly on the type of the EV application. The scope of this thesis is on high performance pure electric vehicles comparable with other gasoline vehicles. As there is no comprehensive comparison on electric motors for the high performance electric vehicle application; in this thesis various common motors such as brushed DC, induction, switched reluctance and permanent magnet BLDC motor are compared in the context of an in-wheel motor vehicle. In Chapter 2, advantages and disadvantages of each motor are discussed according to in-wheel motor requirements and their output characteristics such as speed and torque are compared under same operating condition. As a result of this study, the BLDC motor is introduced as the most suitable in-wheel motor for high performance electric vehicles. BLDC motors were first introduced by T.G. Wilson and P.H. Trickey in 1962 for some specific low power applications and named as a DC machine with solid state commutation [6]. Higher power BLDC motors came on the market after the development of the high power transistors and permanent magnet materials. The first high power BLDC motor (50 horsepower or more) was designed by Robert E. Lordo at Powertec Industrial Corporation in the late 1980s [6]. This thesis focuses on the three phases, star connected BLDC motors. Control of the BLDC motor depends on position of the permanent magnet rotor. Electronic commutation increases complexity of the BLDC motor drives compared to the other motors. Precise simulation model of the BLDC motor is required to study behaviour of the motor for different control algorithms. Therefore a model of the 3 phases, star connected BLDC motor drive with ideal trapezoidal back-emf waveforms is presented in Chapter 3. The mathematical model of the BLDC motor and principle of its operation are also discussed in details. To control speed of the BLDC motor, a digital Pulse width modulation controller is implemented in the model. The BLDC motor drive model is validated through experimental results. 2

24 1. Introduction There are two major commutation techniques for the BLDC motors based on the rotor position detection method. Hall Effect sensors are generally mounted inside the BLDC motor to detect the rotor position in sensor mode. Sensorless control schemes are generally based on back-emf detection of the unexcited phase and flux linkage trajectory of the BLDC motor [7]. Simple BLDC motor construction, low manufacturing cost and less maintenance need are main advantages of sensorless control techniques. However sensorless control algorithms of the BLDC motor are much more complex than the conventional switching techniques [8]. Torque ripple is one of the main limitations of the BLDC motor in EV application. Cogging torque, reluctance torque and mutual torque are various electric torque components in the BLDC motor [9]. Cogging torque is the result of interaction between the permanent magnet rotor magnetic flux and variable permeance of the air gap due to the geometry of stator slots. Cogging torque, distortion of the trapezoidal distribution of the magnetic flux in the air gap and differences between permeances of the air gap in the d and q axes are the main sources of the torque ripples in the BLDC motor [10]. Cogging torque is a more dominant component at low speeds and fortunately its effect is filtered by the motor inertia at high speeds. In Chapter 4, direct torque control switching technique of the BLDC motor in constant torque region below the rated speed is discussed. Direct torque control technique is a flux linkage based sensorless method with no position sensors used to detect the rotor position. In this technique, hysteresis controllers are implemented to limit the torque error level. The simulation results show effective control of the produced electric torque of the BLDC motor. Hysteresis controller effectively limits the torque ripple amplitude of the BLDC motor compared to the conventional Hall Effect switching techniques. Direct toque controlled BLDC motor is also tested under sudden change of the load torque. Dynamic torque response of the motor is much faster than the conventional reported controllers. Direct torque control of the in-wheel motors increase efficiency and safety of electric vehicles. 3

25 1. Introduction Although commutation of the BLDC motor is much simpler by using Hall Effect position sensors but it has some critical drawbacks such as regular need of the motor maintenance, high electromagnetic interference radiation and limitations due to the temperature sensitivity of the in-built sensors [11]. Back-EMF based sensorless drives of the BLDC motor are widely used in industrial applications. Back-EMF zero crossing detection, back-emf integration, back-emf harmonic analysis are examples of the back-emf based sensorless technique of the BLDC motor [7]. In Chapter 5, various back-emf based sensorless drives of the BLDC motor are discussed in details and their advantages and disadvantages are highlighted. In this thesis a novel back-emf based sensorless control algorithm of the BLDC motor is proposed. BLDC motor is commutated through back-emf zero crossing detection of one phase of the motor. Sensing circuitry, noise susceptibility and cost of the sensorless BLDC motor drives are reduced by the proposed technique. Digital pulse width modulation technique (PWM) using a Proportional Integral (PI) controller is employed to control the speed of the BLDC motor. Details of the PWM speed controller are presented in the Chapter 5. Stable performance of in-wheel motors is significant in overall safety of the electric vehicles. Stability of the proposed back-emf based sensorless BLDC motor drive using digital PWM technique is studied by Lyapunov second method. This analysis results in deriving a new equation to calculate the ideal duty cycle value of the PWM controller that keeps the BLDC motor stable at the desired speed. Effect of the load torque is also considered in stability analysis. Accuracy of the proposed sensorless technique to control the BLDC motor and correctness of the introduced novel equation for stability analysis of the motor drive are investigated through simulation and experiment. Good agreements between simulation and experimental results validate correctness of the proposed sensorless technique and stability analysis condition of the BLDC motor. Safety is the most important factor in automotive applications. Safety of electric vehicles is highly dependant on the reliability and robustness of the in-wheel motors as any malfunction or fault in drive train of electric vehicles may result in a fatal accident. Fault tolerant control systems (FTCS s) are one of the effective solutions to increase robustness of the electric motors. FTCS s are designed to 4

26 1. Introduction detect and isolate various faults and apply appropriate remedial actions to keep the stable performance of the motor in post-fault condition [12]. Hazard conditions in the drive train of an electric vehicle can be divided into the electrical and the mechanical faults. In a BLDC motor drive faults may happen in the stator, the rotor or the inverter. Common faults in a BLDC motor drive are analysed and two fault diagnosis systems are proposed in Chapter 6; one for the inverter open circuit switch faults and the other for Hall Effect position sensors failure in the BLDC motor drives. A four in-wheel drive electric vehicle using BLDC motors is modelled in Simulink to analyse the effect of inverter open circuit switch faults on the EV performance. Dynamic parameters of the electric vehicle such as speed, vertical force on tires due to the vehicle s body, speed and torque characteristics of each in-wheel motor are compared and discussed in details under healthy and faulty conditions. Simulation results show that the EV performance is not stable and proves the need of FTCS s for in-wheel motors. Signal analysis based, model based and knowledge based methods are three main fault diagnosis algorithms for electric motors [13]. Advantages and disadvantages of each fault diagnosis method are discussed in Chapter 6. The BLDC motor behaviour is also studied under inverter open circuit as well as Hall Effect sensors faults through a validated simulation model. Inverter open circuit switch faults and position sensors failure effect directly on the output voltage of the variable source inverter (VSI). In Chapter 6, reported fault tolerant control systems for the mentioned faults of the BLDC motor are presented and their merits and demerits are also highlighted. Three phase line voltages of the BLDC motor are analysed and discussed in details under fault condition. Two fault diagnosis systems are proposed based on Discrete Fourier Transform (DFT) analysis of line voltages of the BLDC motor. The proposed fault diagnosis systems are categorised in knowledge based systems where the knowledge is gathered by analysing the line voltages under fault condition through the validated simulation model of the BLDC motor. The proposed fault diagnosis algorithms are not only capable of detecting inverter switch and position sensor faults, but also can identify faulty switches and faulty sensors. The developed knowledge based fault diagnosis systems are validated through experimental data too. 5

27 1. Introduction In this study, suitable fault tolerant inverter drives of the BLDC motor for EV applications are discussed and a fault tolerant control VSI with a redundant leg is recommended for inverter open circuit faults. A novel technique is introduced to generate the commutation signal of the faulty position sensor based on electrical delays between commutation signals in the BLDC motors. The proposed fault tolerant control systems of the BLDC motor are simple, fast and do not need complex calculations. At a glance, this thesis is focused on improving control drives of a three phase BLDC motor for electric vehicle application with novelty in controllability, safety and fault tolerance. First of all advantages of the BLDC motor over other motor types for in-wheel motor application are discussed in Chapter 2. Modelling of the BLDC motor with ideal trapezoidal back-emf and principle of the motor operation are presented in Chapter 3. Various sensorless control algorithms of the BLDC motor, direct torque and back-emf based sensorless control techniques, are proposed in Chapters 4 and 5. A novel stability analysis condition for PWM speed controllers of the BLDC motor is also presented in Chapter 5. Finally in Chapter 6, simple fault tolerant control techniques for inverter switch faults and position sensor faults in the BLDC motor drives are proposed. This chapter presents a very short introduction of the thesis and complete literature reviews are given inside the chapters. 6

28 Chapter 2 Selection of a Suitable Motor for Electric Vehicles 2.1 Chapter Overview One way to limit the emission of greenhouse gases to the atmosphere is to use Electric Vehicles. Electric vehicles are of interest to most of the automotive manufacturers due to their high efficiency and zero greenhouse gas emissions. Different types of electrical motors have been used as the propulsion system of electric vehicles so far. However there is not an overall comparison study that answers clearly which electric motor is the most suitable choice for electric vehicle s drive train. In this chapter a brushed DC motor, an Induction Motor (IM), a Switched Reluctance Motor (SRM) and a permanent magnet Brushless DC motor (BLDC) are simulated and their output characteristics are compared under normal and critical conditions with respect to in-wheel motor technology requirements. Merits and demerits of each electric motor are highlighted, and BLDC motor is recommended as the most suitable electric motor for high performance electric vehicles. 7

29 2. Selection of a Suitable Motor for Electric Vehicles 2.2 Introduction Vehicles with an Internal Combustion Engine (ICE) and conventionally transformed/retrofitted electrical vehicles have a central drive train propelling two rear, front or all four wheels of the vehicle [3]. In-wheel motor technology uses separate motors mounted inside the tire to propel an EV. In-wheel motors have been a focus for research in the last decade. Applying in-wheel motor technology increases the overall safety and efficiency of electric vehicles [5]. Better dynamic stability control of electric vehicles is possible by using four in-wheel motors [4]. This approach improves controllability of each individual wheel and decreases the total chassis weight [14]. It is possible to achieve better acceleration, torque control and regenerative braking in electric vehicles by applying the in-wheel motor technology. Some of the major requirements of a high performance electric vehicle are summarized as follows [15]: being safe and causing no environmental hazards; being autonomous; having a good mileage (a minimum range between charges of at least 50 miles when loaded with two 166-pound occupants and operated at a constant 45 mph 1 ); having a quick charging time (The battery charger shall be capable of recharging the main propulsion battery to a state of full charge from any possible state of discharge in less than 12 hours 1 ); having acceleration of seconds for the speed range of 0 to 100 Km/h; being able to be driven up a 5 to 10 percent ramp at the legal speed under full load condition (a minimum payload of 400 pounds 1 ). Nowadays conventional hydraulic, pneumatic and mechanical control systems are being replaced by electronic control systems, by-wire technologies, electromechanical actuators, and human machine interfaces in the automotive industry 1 EV America Technical Specification, Effective from 1 Oct 1999, is given in appendix B. 8

30 2. Selection of a Suitable Motor for Electric Vehicles [16]. An Intelligent Electronically Controlled Electric Vehicle (IECEV) is being targeted by implementing By-Wire Steering (BWS) system, Brake by-wire (BbW) system, Dynamic Radar Cruise Control (DRCC) system, Pre-Collision Safety System (PCS), Intelligent Parking Assist System (IPAS), Electronic Stability Control (ESC), Traction Control (TRAC) etc., in an in-wheel motor electric vehicle. The reputed car companies such as BMW, Toyota, Lexus, Mercedes Benz, Land Rover, Volkswagen and General Motors have used various by-wire systems in their vehicles. Mercedes Benz and Toyota are using BbW systems in their vehicles. The BWS systems are also currently used in electric forklifts, stock pickers and some tractors [17]. A schematic diagram of a four-wheel drive IECEV is shown in Figure 2.1. Integration of an in-wheel motor and its intelligent controller results in a drive train for the electric vehicles which is safer, more efficient and reliable [18]. Figure 2.1: Four wheel drive train of an IECEV 9

31 2. Selection of a Suitable Motor for Electric Vehicles In-wheel motor requirements are discussed in the next section. Advantages and disadvantages of brushed DC motors, induction motors, switched reluctance motors and BLDC motors are discussed according to the in-wheel motor requirements in following sections. Simulation models of the motors are tested under various (normal and critical) operating conditions. Presented comparison simulation results in this chapter have been published by Tashakori et al. [3][5]. 2.3 The Drive Train of Electric Vehicles Drive train specifications of the electric vehicles available in the world market are given in alphabetical order in Table As can be seen, BLDC and induction motors are the most popular from the manufacturer s point of view. Companies such as Mercedes-Benz, Lightning Car and ECOmove have designed in-wheel motor electric vehicles in the recent years. In-wheel motor technology is considered the most suitable solution for the high performance electric vehicles driving force system nowadays. It is important to choose the correct in-wheel motor to build an efficient and reliable IECEV [3]. An overall comparison of electric motors is needed to select an appropriate machine to fulfil the in-wheel motor technology requirements. Some of the most important requirements of the in-wheel motors are [19]: high torque at low speeds; high torque/power to size ratio; constant power in wide speed range; high efficiency; high dynamic response (fast torque and speed response); accurate electronic controllability; robustness and reliability of the motor and its drive; low Electro Magnetic Interface (EMI) noise susceptibility reasonable cost of production. 1 Reference links are given in Appendix A. 10

32 2. Selection of a Suitable Motor for Electric Vehicles Table 2.1: Drive Train Specifications of the Electric Vehicles Available in the World Market No. EV name Manufacturer Passenger Electric Power Top speed Country/ company capacity motor (KW) (Km/h) Release year 1 BMW MiniE BMW 2 Induction Germany/ Buddy Buddy Electric 3 DC Norway/ BYD E6 BYD Auto 5 BLDC China/ C1 ev ie Citroen 4 Induction France/ Electron Ross Blade 4 Induction Australia/ Lightning Lightning 2 2 in-wheel UK/2013 GT Car synchronous 7 Mitsubishi Mitsubishi 4 BLDC Japan/ 2009 i-miev 8 Morgan Morgan motors 2 BLDC UK Plus E 9 MyCar EuAuto 2 BLDC 64 Hong Kong/2003 Technology 10 Nissan Leaf Nissan 5 BLDC Japan/ QBEAK ECOmove 2 2 in-wheel Denmark/2012 PMAC 12 REVAi REVA Electric 2 Induction India/ SLS AMG Mercedes-Benz 2 4 in-wheel Germany/2013 Eletric synchronous 14 Smart Smart Automobile 2 BLDC Germany/ Tesla Tesla Motors 5 Induction USA/ Think City Think Global 2 Induction Norway/ ZeCar Stevens Vehicles 5 Induction UK/

33 2. Selection of a Suitable Motor for Electric Vehicles There are three main noise sources in electrical motors: 1- Mechanical noise due to shaft misalignment, rotor imbalance or bearing problems; 2- Aerodynamic noise due to internal or external fans; 3- Electromagnetic noise produced by the air gap magnetic flux waves [20]. Most of the electronic control systems (such as motor control drive, electronic stability control system and so on) are compacted and placed near the tire due to the confined space in in-wheel motor electric vehicles. Therefore in-wheel motor EMI noise may cause malfunction or performance degradation in the adjacent electronic systems on board and nearby vehicles, for example in a traffic jam. Common mode currents noise, differential noise, radiated noise and bearing noise are various types of EMI noise which are generated by high frequency pulse width modulation (PWM) switching and surge voltage appearing on motor terminals [21]. Implementation of noise control methods increases complexity of motor controllers and is quite difficult in the electric vehicles, because defining of the EMI noise propagation route is complicated due to high density packaging [22]. Therefore noise susceptibility of the in-wheel motors is a critical factor in overall performance of the EV drive train. In-wheel motors must also be capable of the frequent start, stop and reverse rotation with maximum output electric torque. A high performance electric vehicle should be able to start from halt position and repeatedly accelerate in a short time to overcome the inertia of the load [5]. However the average operational efficiency of a torque converter in vehicles during city traffic conditions is less than 60% [23] Conventional AC and DC Motors Selection of a suitable in-wheel motor for the high performance EV drive train demands considerations on power, voltage and current handling, torque/speed characteristics, power to size ratio, noise susceptibility, maintenance and controllability of motor. Since conventional AC and DC motors are discussed enough in the literature, there is no need to discuss their structural and operational characteristics in this chapter. However in this section, their merits and demerits are discussed according to the requirements of in-wheel motors. 12

34 2. Selection of a Suitable Motor for Electric Vehicles Angular velocity difference between the produced flux of stator and flux of rotor causes slip in the conventional AC (squirrel cage induction) motors. The rotor speed always lags the angular velocity of the stator magnetic field by slip speed. Slip is directly proportional to load torque in the induction motors. Slip causes vibrations of the induction motor at the starting time which is not suitable for the in-wheel motor technology. AC induction motors generally produce lower torque, draw higher initial current and have slip as compared to the conventional DC (brushed DC) motors that experience no slip [3]. As the speed of the induction motor approaches the rated speed, the current and slip decrease and the electrical torque increases. On the other hand, in the DC motor torque is inversely proportional to the angular velocity of the rotor. Therefore DC motors produce higher electric torque at low speeds that is essential for the in-wheel motors. The output power to size ratio of the in-wheel motors is a significant factor due to the space limitations inside the tire. The heat produced by armature winding of the DC motors is dissipated in the air gap and increases the air gap temperature. Therefore DC motors have a moderate or a low output power to size ratio. Since both the stator and the rotor of induction motors have windings, size of the motor is large and output power to size ratio of the motor is low [3]. Therefore both conventional DC and AC motors do not have a suitable output power to size ratio. Extended speed range of the in-wheel motors with constant power is an important factor in the EV application. Torque of DC motors is decreased effectively over the base speed; therefore they have a limited extended speed range. Break down flux weakening speed of the induction motors is almost two times of their rated speed [23]. A specific design of a spindle induction motor with a field orientation control drive can be run up to five times of the rated speed [24], however construction complexity and size of the motor is increased which is not suitable for the in-wheel motor application. Noise in the DC motors is mostly due to PWM switching, therefore filters are used to smooth the average voltage and reduce motor noise. PWM switching, surge voltage and aerodynamic noise, due to the internal fan, are the main noise sources in induction motors. Modulation techniques of the two level inverter have 13

35 2. Selection of a Suitable Motor for Electric Vehicles also a crucial effect in noise emission of the induction motors. Randomize Space Vector Modulation (RSVM) technique increases acoustic noise, whereas the Offline Optimized Pulse Pattern (OOPP) method minimizes the current harmonics and reduce noise emission [25]. Controllers of the DC motors are much simpler and cheaper compared to that of the induction motor controllers. Complex control techniques and poor dynamic characteristics of the induction motors at starting time make them an unsuitable choice for the EV drive train application. DC motors show better dynamic characteristics at starting time but the existence of brushes increase the need of motor maintenance, reduces efficiency, reliability and the Ingress Protection (IP) rating of the in-wheel motors [3] Switched Reluctance Motors Switched reluctance motors, also known as the variable reluctance motors, are type of synchronous motors. However in a comparison to the regular synchronous motors it has no field winding, slip ring and brushes [26]. Reluctance motors were first built nearly 200 years ago. Davidson s motor, one of the most well known reluctance motors, was built in 1839 [27]. Structure of the SR motors is similar to the BLDC motors, though it has a ferromagnetic rotor instead of a permanent magnet rotor. Therefore a SR motor construction cost is cheaper than that of a BLDC motor. As shown in Figure 2.2, different switched reluctance geometries are possible by changing the number of stator phases, number of stator poles and number rotor poles [26]. Switched reluctance motors have electronic commutation control system which provides sequential pulses to the stator windings [3]. Each phase of the SR motor is independent physically and electrically from the other motor phases. Therefore direction of the produced torque is independent of the current direction and depends on the rotor position and the sequence of energized phases [26]. By energizing the stator windings, the rotor moves into the alignment with the stator poles to minimize the reluctance in the air gap. Inductance of the stator windings increases when the stator and the rotor poles are aligned. Positive electric torque is produced when the gradient of the inductance is positive [27]. 14

36 2. Selection of a Suitable Motor for Electric Vehicles Figure 2.2: Various switched reluctance motor geometries Desirable features of the switched reluctance motor that make them attractive for traction applications are: simple and rugged construction, high speed operation, wide speed range with constant power, hazard free operation, high reliability and low manufacturing cost [23]. The major drawbacks of the SR motors are large torque ripples, rotor position detection, low power factor and acoustic noise [28]. Using PWM control technique reduces acoustic noise of the switched reluctance motors compared to the hysteresis current controllers [29]. Five and six phase switched reluctance motors produce lesser torque ripples, however their control techniques are more complex. High amplitude torque ripples and noise susceptibility of the switched reluctance motor drives are not suitable for the inwheel motor applications. Efficiency of the switched reluctance motors is similar to the induction motors and is lower than that of the BLDC motors [30]. 15

37 2. Selection of a Suitable Motor for Electric Vehicles BLDC Motors Permanent magnet synchronous motors have received a considerable attention in the industrial application since 1970 s. Nowadays they are used in various applications such as automotive, aerospace, medical equipment, industrial automation and instrumentation. Permanent magnet synchronous motors are mainly divided into two various types based on their back-emf waveform; the one with a sinusoidal-wave back-emf that is called Permanent Magnet Synchronous AC Motor (PMSM) and the other with a trapezoidal-wave back-emf that is called Permanent Magnet Brushless DC (BLDC) Motors. A BLDC motor with the trapezoidal back-emf produces larger torque compared to a PMSM with the sinusoidal back-emf [31]. The focus of this thesis is on the three phase star connected BLDC motors. A schematic diagram of a two pole BLDC motor and its drive system is shown in Figure 2.3 [12]. BLDC motors are a novel type of the conventional DC motors where commutation is done electronically, not by brushes. Therefore a BLDC motor needs less maintenance, has lower noise susceptibility and lesser power dissipation in the air gap compared to a brushed DC motor due to absence of the brushes. Permanent magnet rotors can vary from two pole pairs to eight pole pairs [32]. Magnet material is chosen with respect to the required magnetic field density in the rotor. Ferrite magnets are usually used to make the permanent magnet rotor of the BLDC motor, however they have the disadvantage of low flux density. In contrast, alloy materials such as Neodymium (Nd), Samarium Cobalt (SmCo), Ferrite and Boron (NdFeB) have higher magnetic density. Hence these alloy magnets produce more torque for the same volume compared to the ferrite magnets; therefore they improve power to size ratio of the BLDC motor which is more suitable for the in-wheel motors [32]. BLDC motor needs a complex control algorithm due to the electronic commutation that is done according to the exact position of the permanent magnet rotor. There are two algorithms for rotor detection; one method that uses sensors and the other does not that is called sensorless [33]. Hall Effect sensors are normally mounted on the non-rotating end inside the BLDC motor with 120 electrical degree phase difference at the constant position to detect rotor angle. 16

38 2. Selection of a Suitable Motor for Electric Vehicles Figure 2.3: Schematic diagram of a two pole BLDC motor drive 17

39 2. Selection of a Suitable Motor for Electric Vehicles Optical encoders are used as position sensors for high resolution applications. The internal view of a BLDC motor is shown in Figure 2.4 [32]. Figure 2.4: The internal view of a BLDC motor Hall Effect signals are generated according to the permanent magnet rotor position. These signals are decoded in controller to choose the correct voltage space vector that must be fed to the three phase Voltage Source Inverter drive of the BLDC motor. Ideal back-emf voltage, current, commutation signals and on switches of the VSI drive of the BLDC motor are shown in Figure 2.5 [34]. It is a fact that noise susceptibility of the BLDC motors is less than the other motor types, specifically the SR motors. Sound pressure (acoustic noise) of a BLDC motor and a SR motor are measured experimentally and compared for the same working conditions in the context of electric brakes [29]. Results show that acoustic noise of the SR motor is 6 db-a more than the BLDC motor at 1000 RPM speed under 0.65 N.m load torque. The sound pressure levels of the BLDC and SR motors at 5000 RPM speed under 0.2 N.m load are measured 48 db-a and 69 db-a respectively [29]. Therefore acoustic noise of the BLDC motor is much higher than the SR motor at high speed operating condition. 18

40 2. Selection of a Suitable Motor for Electric Vehicles Figure 2.5: Ideal current, back-emf and commutation signals of BLDC motor Manufacturing costs of the BLDC motor are higher than the other motor types due to the permanent magnet material price in the world market. The other disadvantage of the BLDC motors is that their extended speed range with constant power is less than twice the synchronous speed due to the limited field weakening capability [35]. An additional field winding can be used to solve this problem in a way that the field produced by the permanent magnet rotor is weakened in the extended constant-power speed region by controlling the DC 19

41 2. Selection of a Suitable Motor for Electric Vehicles field current. These motors are called permanent magnet hybrid motor and their maximum speed is up to four times of the synchronous speed [23]. However, low efficiency of these motors at high speeds and complex structure are their main drawbacks. Using a multi-gear transmission can solve the extended constant power speed range limitation of the BLDC motors. High efficiency, high speed ranges and high dynamic response due to a permanent magnet (low inertia) rotor are the immediate advantages of the BLDC motor for in-wheel motor technology application [5]. The high output power to size ratio of the BLDC motor, due to absence of the field windings, makes it suitable as an in-wheel motor where the space and the weight are significant considerations. The absence of brushes also effectively reduces the maintenance needs of the BLDC motors that is an advantage for the EV applications. Noiseless operation of the BLDC motor also makes it more convenient to design the integral in-wheel motors [18]. 2.4 Motor Comparison Choosing a suitable electric motor for the in-wheel drive train of electric vehicles is an important parameter which affects overall performance of the vehicle. Appreciated research works have been reported on the motor selection for hybrid and electric vehicles [19][23][30][31][35]. Some of the reported research works have suggested that the switched reluctance motor is a better choice for the hybrid electric vehicle (HEV) and EV applications [19][23][30]. Wider speed range with constant power of the SR motors compared to the BLDC motors is the most important discussed reason for recommending the SR motors for HEV and EV applications [23][30]. Brushed DC, induction, BLDC and switched reluctance motors are compared based on efficiency, weight and manufacturing cost and the SR motors are recommended by Xue et al. [19] due to the high cost and difficulties in accessing magnetic materials. Torque ripples reduction control techniques are suggested to overcome the main drawback of the SR motors. Although the extended speed range of the BLDC motors is less than SR motors, applying a multi-gear transmission can solve the problem. Brushed DC, induction, BLDC and switched reluctance motors are compared based on power 20

42 2. Selection of a Suitable Motor for Electric Vehicles density, efficiency, controllability, reliability, technological maturity and cost and the induction motors are recommended for HEV application [35]. The BLDC motor is recommended for the EV drives due to its high power density, efficiency and smooth torque response [31]. All the reported comparisons are based on literature review of the motor specifications and there is no performance analysis and comparison based on simulation or experimental results. In this chapter a DC motor, an induction motor, a switched reluctance motor, a BLDC motor and their controllers are modelled. Details of the motor models are presented in Appendix A. Their output characteristics such as speed and torque are compared under the same operation conditions (the same input power, load torque and reference speed of the controllers). Simulation results are discussed under normal and critical operating conditions. Critical condition analysis is important with respect to safety of the electric vehicle. The term normal condition is defined as the normal operation of the EV with constant speed on a flat, uphill or downhill roads (load torque is constant). Critical conditions are considered as the operation of the EV under electrical faults and mechanical shocks. The electrical faults may happen in the electric motor or its controller and the mechanical shocks on the in-wheel motors may occur due to various road conditions, sudden braking, or sudden change of vehicle direction [5]. Specifications of various electric motors used in the simulation models are presented in Tables 2.2, 2.3, 2.4, 2.5. Table 2.2: Brushed DC Motor Specifications Description Value Unit DC Voltage 400 V Resistance 1.78 Ω Inductance 0.21 H Inertia 0.08 kg-m 2 Damping Ratio N.m.s 21

43 2. Selection of a Suitable Motor for Electric Vehicles Table 2.3: Induction Motor Specifications Description Value Unit DC Voltage 400 V Phase Resistance 0.73 Ω Phase Inductance H Inertia kg-m 2 Damping Ratio N.m.s Poles 4 - Table 2.4: Switched Reluctance Motor Specifications Description Value Unit DC Voltage 400 V Phase Resistance 2 Ω Unaligned Inductance H Aligned Inductance H Inertia kg-m 2 Damping Ratio 0.01 N.m.s Poles 6/4 - Table 2.5: BLDC Motor Specifications Description Value Unit DC Voltage 400 V Phase Resistance 2 Ω Phase Inductance H Inertia kg-m 2 Damping Ratio N.m.s Torque constant 1.4 N.m/A Poles 8-22

44 2. Selection of a Suitable Motor for Electric Vehicles Performance Comparison of the Motors under Normal Condition Simulation models of the brushed DC motor, the induction motor, the switched reluctance motor and the BLDC motor are tested for 1500 RPM reference speed of the controller under 10 N.m load torque. Transient speed responses of the motors are plotted and shown in Figure 2.6 [5]. Figure 2.6: Transient speed responses of the motors under normal condition As can be seen in the figure, speed response of the BLDC motor is much faster than the other motor types. Higher dynamic response of the BLDC motor is due to its permanent magnet (low inertia) rotor. Fast dynamic response is one of the most important requirements of the in-wheel motors. Simulation results show that the DC motor has the second fastest dynamic response and the switched reluctance motor has the slowest dynamic response among the motors. The induction motor has the most speed oscillations in transient time though it has an acceptable dynamic response [5]. Transient torque responses of the motors are plotted and shown in Figure 2.7 [5]. The DC motor has the highest initial torque value and the BLDC has the 23

45 2. Selection of a Suitable Motor for Electric Vehicles fastest torque response. Torque response of the BLDC motor is also approached the final value, the load torque, much faster than the other motors. Therefore as can be seen the BLDC motor has a wider speed range with constant torque below the rated speed. Torque fluctuation of the induction motor in transient time can be seen in the figure. Slip of the induction motor at low speeds acts an essential role in output characteristics of the motor in transient condition. Slip is dependent on supply voltage frequency, rotor resistance and torque load. Change of voltage frequency results in slip variations and torque oscillation in the induction motor in transient condition. As can be seen in the figure, torque ripple is one of the major drawbacks of a switched reluctance motor. Torque ripple results in fluctuation of delivered output power from the motor to the tires which is not acceptable for an in-wheel motor EV. Low efficiency and low speed ranges are the major drawbacks of the conventional DC motors for the in-wheel motor application even though it has the highest initial torque value and high dynamic speed response. Therefore the BLDC motor is the most suitable choice as an in-wheel motor according to torque response analysis [3][5]. Figure 2.7: Transient torque responses of the motors under normal condition 24

46 2. Selection of a Suitable Motor for Electric Vehicles Transient torque/speed characteristics of the brushed DC motor, the induction motor, the switched reluctance motor and the BLDC motor from the halt position up to the controller reference speed (1500 RPM) are shown in Figure 2.8 [5]. It is shown that the BLDC motor has the minimum torque oscillation and the switched reluctance motor has the maximum torque oscillation in the transient time. Torque fluctuation of the induction and the switched reluctance motors during transient condition can be seen in the figure. Output electric torque of the BLDC motor reaches the load torque when the speed of the motor passes the 53.3 percent of its final value. Therefore the BLDC motor has a better torque/speed characteristics for in-wheel application compared to the other motors [3][5]. The BLDC motor has the best overall output characteristics with respect to the inwheel technology requirements during the normal operating condition. Figure 2.8: Transient torque/speed characteristics of the motors Performance Comparison of the Motors under Critical Condition Safety of the vehicle s passengers is the most significant issue in automobile industry. The most of research topics in the automotive industry are concentrated 25

47 2. Selection of a Suitable Motor for Electric Vehicles on developing intelligent systems to improve safety, efficiency and convenience of driver and passengers in the vehicle. Since electric motors are used as the propulsion system for electric vehicles, therefore robustness and reliability of the electric motor drive in abnormal conditions play a critical role in overall safety and performance of the EV. Electrical motors are subjected to various types of fault inside the motor or its drive. In-wheel motors are also subjected to the mechanical shocks (sudden change of load torque and vibration) due to the operational and environmental conditions. Therefore an in-wheel motor and its controller must be reliable during the mechanical shocks, the transient electric faults and at initial times of major electric faults until fault tolerant control systems make the appropriate decisions. In this chapter different fault conditions are modelled and applied to the respective electric motor models and the behaviour of the motors are compared during the critical operating condition Transient Electric Faults Various electric faults may happen in an electric motor or its control drive. In this section behaviour of the induction, switched reluctance and BLDC motors is studied during a transient three phase to ground fault of the line voltages. A single phase to ground electric fault is applied for the DC motor. The fault is applied at t = 0.4 s for duration of 0.1 s while the motors are running in the normal condition at 1500 RPM reference speed under 10 N.m load torque. Speed characteristics of the motors during the transient electrical fault condition are plotted and shown in Figure 2.9 [5]. Induction and switched reluctance motors have more stable speed responses during the transient electric fault. The DC motor becomes completely unstable during the fault. The BLDC motor has a fast speed response to the fault due to the permanent magnet rotor and high dynamic response characteristics. BLDC motor speed is also reduced remarkably during transient electric fault. Torque characteristics of the motors during the transient electrical fault condition are plotted and shown in Figure 2.10 [5]. 26

48 2. Selection of a Suitable Motor for Electric Vehicles Figure 2.9: Speed responses of the motors under same transient electrical fault Figure 2.10: Torque responses of the motors under same transient electrical fault 27

49 2. Selection of a Suitable Motor for Electric Vehicles As can be seen from the figure, the DC motor behaves as a generator during the transient fault. Torque ripples amplitude of the switched reluctance motor increases drastically during fault condition. Induction motors and BLDC motors have the least torque fluctuation of all. Since delivered power to the wheels is directly proportional to the speed response and the produced electric torque of the motor, the induction motor and the BLDC motor have desired behaviour during transient electric fault in order Vibration and Mechanical Shocks Abrupt changes of the load torque on an electric motor are called mechanical shocks. Mechanical shocks may be applied to the in-wheel motor due to sudden changes of the road condition, brakes or changes of the vehicle direction by driver. Sudden 30% changes of the load torque are applied to the simulation models of motors to study their behaviour during the mechanical shocks. Speed responses of the motors under same mechanical shocks are plotted and shown in Figure 2.11 [5]. Figure 2.11: Speed responses of the motors under same mechanical shocks 28

50 2. Selection of a Suitable Motor for Electric Vehicles As can be seen from the figure, the induction motor has the least speed variations where the BLDC motor has the most speed changes due to high dynamic response especially at the exact times of load change. However the BLDC motor follows the reference speed after the load change much faster than other motors. The DC and switched reluctance motors have almost the same speed response to the mechanical shocks. However low dynamic response of the switched reluctance motor prevents the motor from following the reference speed quickly enough after the load change. Torque responses of the motors under same mechanical shocks are plotted and shown in Figure 2.12 [5]. Figure 2.12: Torque responses of the motors under same mechanical shocks As can be seen torque ripples of the switched reluctance motor are increased at the times of load change. Torque of the BLDC motor approaches the load torque much faster than the other motors. In contrast, the DC motor shows the slowest torque response among all. Induction motor torque ripples also are increased due to the mechanical shocks; however its dynamic response is not quick enough. Comparison discussions according to the in-wheel technology and motor specifications are summarized in Table 2.6. Numeric values from 1 to 5 are assigned in order to the terms very bad, bad, moderate, good and very good 29

51 2. Selection of a Suitable Motor for Electric Vehicles according to the EV application requirements. As it can be seen in the table, the BLDC motor got the highest points and is the well suitable choice for drive train of high performance electric vehicles. Table 2.6: Motors Comparison According to the In-wheel Motor Specifications Features BLDC motor SR motor Induction motor DC motor Commutation electronic electronic - brushes Slip - - applicable - Efficiency High rated speed Extended constant power speed range Control complexity Torque/Speed Dynamic response Power/size ratio Operation life time Maintenance needs Noise susceptibility Speed during fault Torque during fault Speed during mechanical shocks Torque during mechanical shocks Manufacturing Cost Total

52 2. Selection of a Suitable Motor for Electric Vehicles 2.5 Conclusion There is a growing interest in electric vehicles for future transportation due to its zero carbon emissions and high efficiency. Correct electric motor selection for propulsion system of the high performance electric vehicle is essential to attain the maximum safety and efficiency. In-wheel motor technology as a propulsion system in electric vehicles is one of the main research interests in automotive industry. In this chapter behaviour of the brushed DC, induction, BLDC and switched reluctance motors are studied and compared with respect to the inwheel motor requirements under normal and critical (the electric fault and the abrupt mechanical shocks) conditions in order to select the most suitable electric motor for electric vehicles. The BLDC motor has the most suitable characteristics during normal condition operation according to the in-wheel motor requirements. Better torque/speed characteristics, higher efficiency, higher output power to size ratio, higher dynamic response, higher operating life, lower maintenance, noiseless operation and higher speed ranges are advantages of the BLDC motor in normal operation over all the other motors according to the discussed comparison results. Comparison results show that the induction motor is the most robust among all the other motors during the critical condition; however speed range limitations, low efficiency of motor at high speed, slow dynamic response and slip of the motor at low speed make it a poor choice for high performance electric vehicles. The switched reluctance motor has also similar speed characteristics to induction motors during critical condition; however its torque ripple amplitude shows a remarkable increase. Low efficiency, high amplitude torque ripples and noise susceptibility in the switched reluctance motor drive are its main drawbacks which make it unsuitable for the in-wheel motor application. Output characteristics of the DC motors during electric fault are the worst among all the motors; however the DC motor behaviour is more robust during mechanical shocks. Low efficiency, low speed ranges and periodic need of maintenance are factors which limit the use of the DC motor for high performance electric vehicles. Although the BLDC speed response has sharp notches at the time of the load change, but its fast dynamic response is adequate to follow the reference speed. BLDC motors also have a more desirable torque response in critical condition 31

53 2. Selection of a Suitable Motor for Electric Vehicles than switched reluctance motors. Implementing a fault tolerant control system will increase reliability of the BLDC motor drives during critical conditions. Fault diagnosis systems of the BLDC motor drives for some specific faults are presented in Chapter 6. Finally with respect to the presented comparison discussions in this chapter, the BLDC motor is recommended as the most suitable choice as drive train of the high performance electric vehicles. 32

54 Chapter 3 Modelling of the BLDC Motor Drive for EV Application 3.1 Chapter Overview Electric motors play a significant role in electric vehicles. In-wheel motor technology improves efficiency and safety of the high performance electric vehicles. BLDC motors are recommended as the propulsion system in the electric vehicles due to their high efficiency, desired torque versus speed characteristics, high power density and low maintenance cost. An accurate and precise model of the BLDC motor is required to study different control algorithms of an EV drive train. Therefore in this chapter, a BLDC motor drive with an ideal back-emf is modelled in Simulink. Correct performance of the simulation model of a BLDC motor drive are validated through experimental data. 3.2 Introduction BLDC motors are in the category of synchronous motors. Principle of their operation is similar to the brushed DC motors, however BLDC motors are commutated electronically and have a permanent magnet rotor. Electronic commutation increases the control drive complexity of the BLDC motor. As discussed in the previous chapter, control techniques of the BLDC motors are divided into 33

55 3. Modelling of the BLDC Motor Drive for EV Application two categories; control drives using sensors and sensorless drives. An accurate model of the BLDC motor that gives the precise values of torque, speed, current and back-emf is required to study the various control schemes of the motor [36]. This chapter focuses on modelling of a three phase, star connected BLDC motor using three Hall Effect sensors for rotor position detection. Sensorless control techniques of the BLDC motors are discussed in the next chapters. Various BLDC motor drive models have been reported for different applications in the last decade [37-45]. Although the previously reported research works contributed to the BLDC motor modelling, there is not a simple BLDC motor model with the ideal trapezoidal back-emf appropriate for the EV application [33]. In this chapter a three phase, star connected BLDC motor with ideal trapezoidal back-emf waveforms is simulated in Simulink. Overall view of the BLDC motor drives, modelling of the motor, BLDC motor controllers, simulation and experimental results are discussed and presented in next sections respectively. Presented results in this chapter have been published by Tashakori et al. [33]. 3.3 Overall View of the BLDC Motor Drive A BLDC motor, a three phase voltage source inverter and a closed loop control algorithm are the main sections of the BLDC motor drives. A BLDC motor includes two separate electrical and mechanical parts. Three Hall Effect sensors (with 120 electrical degree phase difference) detect the rotor position of the motor. Hall Effect signals are decoded in the controller and the appropriate voltage space vectors are chosen to supply the motor. Corresponding switching signals are fed to the three phase VSI to supply voltages to the windings of BLDC motor. In this section speed of the BLDC motor is adjusted by a digital PWM speed controller in a closed loop scheme. Overall structure of the BLDC motor drive is shown in Figure 3.1. Each part of the BLDC motor drive model is modelled separately and integrated in the overall simulation model [33]. 34

56 3. Modelling of the BLDC Motor Drive for EV Application Figure 3.1: Overall structure of the BLDC motor drive 3.4 Modelling of the BLDC Motor This section presents modelling of a three phase, four poles, star connected permanent magnet synchronous motor with the trapezoidal back-emf. The trapezoidal back-emf implies that the mutual inductance between stator and rotor is non-sinusoidal, thus an abc phase variable model is more applicable than a d-q axis model [42]. Following assumptions are made to simplify the mathematical equations and the overall BLDC motor model. magnetic circuit saturation is ignored; stator resistance, self and mutual inductances of all three phases are equal and constant; hysteresis and eddy current losses are eliminated; inverter semiconductor switches are ideal. The simplified electrical and mechanical mathematical equations of the BLDC motor can be written as below, V a = Ri a + (L M) di a dt + E a (3.1) 35

57 3. Modelling of the BLDC Motor Drive for EV Application V b = Ri b + (L M) di b dt + E b (3.2) V c = Ri c + (L M) di c dt + E c (3.3) E a = K e ω m F (θ m ) E b = K e ω m F (θ m 2π) (3.4) 3 E c = K e ω m F (θ m + 2π) 3 T ea = K t i a F (θ m ) T eb = K t i a F (θ m 2π) (3.5) 3 T ec = K t i c F (θ m + 2π) 3 T e = T ea + T eb + T ec (3.6) T e T l = J d2 θ m dt 2 + dθ m dt (3.7) θ e = P 2 θ m (3.8) ω m = dθ m dt (3.9) Where V a,b,c is voltage of phase a, b, c that is applied from inverter to the BLDC motor; i a,b,c is current of phase a, b, c; E a,b,c is back-emf voltage of phase a, b, c and T e(a,b,c) is produced electric torque in phase a, b, c. An embedded program has been written to generate the ideal back-emf reference signal, F (θ e ), with respect to the electrical degree angle of the permanent magnet rotor. Since phase windings are distributed symmetrically in the stator, back-emf signals have 120 electrical degrees phase shift with respect to each other. Ideal output characteristics of the BLDC motor drive are shown in Figure 2.5 on Page 19. Ideal reference back-emf waveforms of the BLDC motor model with respect to the rotor electrical degree are shown in Figure 3.2 [33]. 36

58 3. Modelling of the BLDC Motor Drive for EV Application Figure 3.2: Ideal reference back-emf waveforms of the BLDC motor model Since the neutral point of the BLDC motors is not stable and most of the times is also not provided by manufacturers, phase voltage differences are used to generate state space equations (refer to the equations (C5) and (C6) on page 158). Although the neutral point of the BLDC motor is not stable, it is possible to estimate it with zero crossing detection of the unexcited phase back-emf voltage. State space to Laplace transform and reverse can be done for linear and zero initial condition systems. Therefore two simple Laplace equations of electrical and mechanical systems of the BLDC motor supplied by phase to neutral voltages are derived (refer to the equations (3.12) and (3.15)) and used to model the BLDC motor instead of state space equations [33]. 37

59 3. Modelling of the BLDC Motor Drive for EV Application v a,b,c (t) = Ri a,b,c (t) + L di a,b,c dt Laplace transform of the equation (3.10) is, + k a,b,c ω m (t) (3.10) V a,b,c (s) = RI a,b,c (s) + Ls[I a,b,c (s) i a,b,c (0)] + k a,b,c ω m (s) (3.11) By solving the equation (3.11) for I(s) if initial condition of system is zero (i(0) = 0), the electrical system Laplace equation of each phase is derived as, I(s) V (s) kω m (s) = 1 R + Ls (3.12) Electric torque of each phase and total electric torque produced by the BLDC motor are calculated from equations (3.5) and (3.6) respectively. Total electric torque is applied to the mechanical system of BLDC motor. T e (t) T l = J dω m(t) dt Laplace transform of the equation (3.13) is + βω m (t) (3.13) T e (s) T l = Js[sω m (s) ω m (0)] + βω m (s) (3.14) By assuming that the initial speed of motor is zero and solving the equation (3.14) for ω m (s), mechanical system Laplace equation of the BLDC motor is derived as, ω m (s) T e (s) T l = 1 β + Js (3.15) Back-EMF signals of the BLDC motor are generated according to the electrical degree of rotor for each phase and applied as a negative feedback to the input voltages. This approach makes the BLDC motor model simpler and more convenient for various control technique implementation. BLDC motor simulation model is shown in Figure 3.3 [33]. 38

60 3. Modelling of the BLDC Motor Drive for EV Application Figure 3.3: BLDC motor simulation model 39

61 3. Modelling of the BLDC Motor Drive for EV Application 3.5 Modelling of the BLDC Motor Drive The schematic diagram of a three phase, four poles, star connected BLDC motor drive is shown in Figure 3.4. A three phase inverter is used to supply voltage to the BLDC motor windings. Metal Oxide Semiconductor Field Effect Transistors (MOSFET) are used to model the three phase VSI in Simulink. The simulation model of three phase VSI is shown in Figure 3.5. Two phase conduction mode voltage space vectors of the VSI are selected based on Hall Effect position sensor signals. Three Hall Effect sensors are detecting permanent rotor position of the BLDC motor. In this model Hall Effect signals of the BLDC motor are generated through an embedded Matlab code file according to the electrical degree rotation of the motor. Electrical degree sections, corresponding Hall Effect signals and inverter switches status of the BLDC motor are shown in Table 3.1. Table 3.1: Hall Effect Signals and Inverter Switches Status of the BLDC Motor Electrical Hall A Hall B Hall C Inverter switches status degree S 1 S 2 S 3 S 4 S 5 S On Off Off Off Off On On Off Off On Off Off Off Off Off On On Off Off On Off Off On Off Off On On Off Off Off Off Off On Off Off On Speed is directly proportional to the average value of applied voltages in the BLDC motor. Variable DC link inverters and pulse width modulation switching techniques are the two basic methods to control the average applied voltage to the BLDC motor. Variable DC link inverters have a poor harmonic control and extra conversion systems compared to the PWM controlled inverters. One direction power flow, high stresses of the components and high peak currents that cause EMI problems are the main drawbacks of variable DC link inverters. The main disadvantages of PWM inverters are complexity of controller and high frequency switching losses [46]. Performance comparison of PWM inverter and variable DC link inverter for high-speed (up to RPM) sensorless control of the BLDC 40

62 3. Modelling of the BLDC Motor Drive for EV Application Figure 3.4: Schematic diagram of a 3 phase, 4 poles, star connected BLDC motor drive 41

63 3. Modelling of the BLDC Motor Drive for EV Application Figure 3.5: Three phase VSI simulation model motor states that more stable sensorless operation can be obtained using the variable DC link inverters at high speeds [47]. A regenerative brake system is an essential factor to increase the battery life or mileage of the electric vehicle. One direction power flow characteristics of the variable DC inverters is not suitable for the regenerative brake system in the electric vehicles. Therefore PWM switching technique is more suitable to control the average output voltage of VSI. Details of the PWM speed controller of the BLDC motor are discussed in Chapter Simulation Results and Discussion A three phase star connection BLDC motor with the ideal trapezoidal back- EMF and its control drive are modelled in Simulink. The BLDC motor model BLK423S specifications manufactured by Anaheim Automation Company are used in model. The BLDC motor model BLK423S specifications are summarized in Table 3.2. Specification of the power MOSFET model IRFR2407 (refer to 42

64 3. Modelling of the BLDC Motor Drive for EV Application Appendix B) is used to model the three phase VSI. Simulation model is tested for 2000 RPM controller reference speed and 5.9 N.m torque load. PWM speed control signals are applied to the upper switches in each leg of VSI. Speed characteristic of the BLDC motor simulation model is shown in Figure 3.6. As can be seen speed of the BLDC motor follows the reference speed of the PWM speed controller. Table 3.2: Specifications of the BLDC Motor Model BLK423S Description Value Unit DC voltage 310 V Rated speed 3000 RPM Phase resistance 0.38 Ω Phase inductance H Inertia kg-m 2 Damping ratio N.m.s Poles 8 - Figure 3.6: Speed characteristics of the BLDC motor simulation model 43

65 3. Modelling of the BLDC Motor Drive for EV Application Figure 3.7: Torque characteristics of the BLDC motor simulation model The torque characteristic of the BLDC motor simulation model is shown in Figure 3.7. Initial produced electrical torque of the BLDC motor simulation model is 92.6 N.m. Produced electric torque of the BLDC motor in steady state condition pulsating around the load torque value. Line voltage, Current and corresponding Hall Effect signal of the phase A of BLDC motor model are shown in Figure 3.8. The line voltage of the motor is measured with respect to the negative terminal of the DC link of inverter (and scaled down by factor 0.1) to have a comprehensive view of all three signals in one figure. As can be seen in Figure 3.8, the BLDC motor current is maximum (10.4 Amperes) when the Hall Effect signal rises to logic one and is minimum when the Hall Effect signal falls to logic zero. The trapezoidal back-emf signals of the BLDC motor model with respect to electrical degree rotation of the rotor are shown in Figure 3.9. As can be seen in figure, there are exact 120 electrical degree phase difference between back-emf signals of the BLDC motor. 44

66 3. Modelling of the BLDC Motor Drive for EV Application Figure 3.8: Voltage, Current and Hall Effect signal of phase A of the BLDC motor Figure 3.9: Back-EMF signals of the BLDC motor model 45

67 3. Modelling of the BLDC Motor Drive for EV Application 3.7 Simulation Model Validation A three phase in-wheel BLDC motor hub designed for electric motor cycle application is used as a practical test motor to validate the BLDC motor simulation model. The experimental set-up of the BLDC motor is shown in Figure The BLDC motor simulation model is developed based on specifications of the experimental test motor. Specifications of the three phase in-wheel BLDC motor hub are given in Table 3.3. Figure 3.10: Experimental test set-up of the BLDC motor Table 3.3: Specifications of the Experimental In-wheel BLDC Motor Description Value Unit DC voltage 48 V Rated speed 600 RPM Phase resistance 0.4 Ω Phase inductance H Inertia kg-m 2 Damping ratio N.m.s Poles 8-46

68 3. Modelling of the BLDC Motor Drive for EV Application Figure 3.11: Line voltage and Hall Effect signal of phase A of the BLDC motor 47

69 3. Modelling of the BLDC Motor Drive for EV Application The experimental in-wheel BLDC motor and simulation model are tested under the same operating conditions for 600 RPM reference speed. The inbuilt drum brake of the in-wheel motor hub is used to apply load to the test motor. The applied load torque to the motor is 1.54 N.m according to manufacturer test datasheet. The PWM switching signal is applied to the upper switches of VSI. The line voltage and corresponding Hall Effect signal of phase A of the test BLDC motor and simulated motor model are shown in Figure The simulation results and the test data are not 100% match. The pattern of the voltage and duration of electrical degrees are almost same; however duty cycle of the control PWM signal is different. Good agreements between the simulation results and experiment results validate the simulation model of the in-wheel BLDC motor. 3.8 Conclusion Improving control strategies of the in-wheel motors result in improving overall performance of the electric vehicle. An accurate model of the in-wheel motor which provides precise information of produced torque and motor speed values is needed to study the behaviour of electric vehicles under different control algorithms and working conditions. In this chapter, the simulation model of a three phase star connected BLDC motor model with the ideal back-emf is presented. The proposed model is simulated in Matlab/Simulink. Simulation results under load conditions show proper performance of the BLDC motor model. A three phase in-wheel BLDC motor hub designed for electric motor cycle application is used as an experimental test motor to validate the BLDC motor model. Experimental results prove correct performance of the simulation model of the BLDC motor. The presented simulation model is simple and based on Laplace transform of mathematical equations of the BLDC motor. Simplicity and discussed specifications of the proposed model make it useful in the design of the BLDC motor drives with different control algorithms for the EV application. 48

70 Chapter 4 Direct Torque Control Drive of BLDC Motor for EV Application 4.1 Chapter Overview Two BLDC motor control methods exists based on using sensors for permanent magnet rotor position detection or not. Simpler motor construction, manufacturing cost reduction, less maintenance needs and no possibility of the motor malfunctions due to unbalanced positioning or failure of the position sensors are immediate advantages of sensorless control techniques. Various back-emf monitoring algorithms and flux linkage based techniques are discussed to commutate the BLDC motor in the sensorless mode. Direct torque control technique (DTC) is a flux linkage based sensorless control method. It does not use any sensors for detecting permanent magnet rotor position. Correct speed and torque control of an in-wheel motor results in controlling of drive train output power in the electric vehicle. In this chapter direct torque control switching technique of the BLDC motor for the EV propulsion application is discussed. Results show effective control of the torque and remarkable reduction of the torque ripples amplitude compared to conventional reported switching techniques. Improving torque control of the EV drive train results in more efficient and safer vehicles. 49

71 4. Direct Torque Control Drive of BLDC Motor for EV Application 4.2 Introduction Generally three Hall Effect sensors are mounted inside the BLDC motor with 120 electrical degrees phase difference to detect permanent magnet rotor position in the sensor mode control scheme. Eliminating rotor position detection sensors in the BLDC motor reduces the cost and construction complexity of the motor. However the BLDC motor control algorithm will be more complicated by implementing the sensorless control methods. In the sensorless control mode, rotor position is detected through output parameters of the motor such as voltage and current. Back-EMF sensing, back-emf integration, freewheeling diode conduction of unexcited phase, flux linkage based, speed independent position function and third-harmonic analysis of back-emf are sensorless techniques for commutation of the BLDC motor [7]. Back-EMF sensing at low speeds and transient time and discontinuous response due to high commutation rates are the main disadvantages of the sensorless techniques [48]. Valuable research works have been published on different sensorless control algorithms of the BLDC motor [7][14][49][50]. A DSP-controlled PWM chopper with a C-dump converter drive has been presented for BLDC motors [49]. dual speed and current closed-loop control are used to keep a constant voltage to frequency ratio to maintain constant torque operation of the BLDC motor. Forced commutation RC circuits and the effect of snubber circuits to control commutation and dv rating on switches have been discussed. Simulation results dt show a number of current spikes that increase torque ripples of the BLDC motor that is not suitable for the in-wheel application. A A current controlled PWM technique with a four switch inverter drive has been reported for BLDC motors [14]. Difficulties in generating 120 conducting current profiles for the three phase BLDC motor with four switch inverters and current distortion of two phases due to back-emf of the silent phase are main drawbacks of the proposed technique [48]. Kim and Ehsani have discussed a sensorless control technique with a new flux linkage function for BLDC motors [7]. A speed-independent flux linkage position function, G(θ), has been defined according to the rotor mechanical angle. This technique provides a precise commutation pulse even in transient state and is able to detect position of the rotor at around 1.5% of nominal speed. Therefore 50

72 4. Direct Torque Control Drive of BLDC Motor for EV Application problems of sensorless control techniques at low speeds have been improved by the proposed approach. It is suitable for in-wheel BLDC motors where it is needed to control the EV at low speeds, for example in a traffic jam. A BLDC motor control drive with two modes of conduction angle control and current control operations has been introduced by Rodriguez and Emadi [50]. Torque and current are directly proportional in electric motors, therefore current control results to torque control of the BLDC motor. Speed oscillations of the BLDC motor is reduced up to maximum of 3.4% by the proposed digital controller. Implementing a torque ripple reduction techniques to the proposed digital controller makes it more suitable for the EV application [8]. Cogging torque due to the stator slots interacting with the rotor magnetic field, reluctance torque due to the variation in phase inductance and mutual torque due to the mutual coupling between the stator winding current and rotor magnetic field are the main electric torque production sources in BLDC motors [9]. Skewing of rotor magnets with respect to the rotor axis, skewing of the stator and coordinating the number of stator sluts in the motor design are techniques which remarkably reduce the effect of the first two torque production sources [51]. In general there are three types of permanent magnet rotors for the BLDC motor; polar magnets or surface mounted magnets, sub-polar (inset) magnets and buried (interior) magnets [51]. The surface-mounted magnet rotors enlarge the effective air gap and minimize armature effect on the rotor magnetic field. Torque ripples are reduced due to the larger effective air gap and smoother flux density distribution in the air gap. Therefore they are widely used in the high performance BLDC motors [9]. DTC technique is a sensorless control technique because it does not use any sensors for detecting position of the permanent magnet rotor. In electric motors, output power is directly proportional to the produced electric torque and speed of the motor. Therefore simultaneous torque and speed control is important factor in the drive train of electric vehicles. In-wheel motors need to operate at high speed that is inappropriate for frequent start, stop and low speed operation of electric vehicles. Therefore a gear box is used to reduce electric motor speed and increase produced torque when the electric vehicle is operating at low speeds. Small speed fluctuations are damped by the gear box due to the large mechanical time 51

73 4. Direct Torque Control Drive of BLDC Motor for EV Application constant, but torque oscillations are more significant. Improving performance of the in-wheel motors increases the safety of electric vehicles. Safety and efficiency of the electric vehicles increase by delivering the desired ripple free torque to the wheels of an electric vehicle in various operating conditions. Therefore direct torque control switching technique is a suitable choice for the high performance electric vehicles [8]. Presented simulation and experimental results in this chapter have been published by Tashakori et al. [8][48]. 4.3 Direct Torque Control of the BLDC Motor Using Three Phase Conduction Mode Direct torque control technique for induction motors was introduced for the first time by Takahashi and Noguchi in 1986 [52], and later by Depenbrock in 1988 [53]. Many research works on DTC of BLDC motors have been reported for various applications that need precise torque control in the last decade [54][55][56][57][58][59]. Direct torque control of BLDC motor as a drive train of hybrid electric vehicles have been reported by Gupta et al. [58]. A schematic diagram of DTC drive of the BLDC motor is shown in Figure 4.1. Torque error, stator flux error and stator flux angle must be calculated to select the correct voltage space vector for switching in DTC drive of the BLDC motor. Flux linkage error is eliminated in the DTC model presented in this chapter due to variations of stator flux magnitude by changes in resistance, current and voltage and specifically sharp dips at every commutation interval [54]. In-wheel BLDC motors should operate in both the constant torque region and the extended constant power region. Back-EMF of motor is less than DC link voltage of the VSI in constant torque region (below rated speed) and is more than DC link voltage value above the base speed. Stator inductance avoids an abrupt increase of phase current in constant power region and distorts the output torque of the BLDC motor. Therefore in this chapter, it is considered that the BLDC motor is operating in constant torque region below the rated speed. Accurate estimation of flux linkage magnitude and produced electrical torque is important in DTC drive of the in-wheel motors. In some techniques, current 52

74 4. Direct Torque Control Drive of BLDC Motor for EV Application Figure 4.1: Overall structure of DTC drive of the BLDC motor sensors have been used to determine flux linkage and estimate voltage from the DC bus of inverter [56][58][59]. This method is too sensitive to voltage errors caused by dead-time effects of the inverter switches, voltage drop of power electronic devices and fluctuation of the DC link voltage [51]. In this chapter both current and voltage signals are measured for accurate estimation of flux linkage magnitude and produced electric torque [57]. Precise estimation of electric torque and flux angle of the BLDC motor mainly depend on accurate sensing of phase currents and voltages of the motor. Variations of stator winding resistance due to changes in temperature cause errors in the stator flux estimation. Analogue integrators also produce DC offset to the signal that causes errors in torque estimation. As it is assumed that the BLDC motor is operating in constant torque region below the rated speed, therefore the 53

75 4. Direct Torque Control Drive of BLDC Motor for EV Application stator flux magnitude does not change during operation. The second algorithm proposed by Hu et al. [60] with limiting level of 2K L π/(3 3) (where K L is the flux linkage of motor) is used to eliminate analogue integrator DC drift error [48]. The balanced three phase system (voltages and currents) is converted to the αβ-axis references by applying Clarke transformation (refer to the equations (D1) and (D2) on the page (158)). Stator flux linkage magnitude, stator flux angle and electrical torque of motor can be estimated by [8], ϕ Sα = (V Sα i Sα ).dt (4.1) ϕ Sβ = (V Sβ i Sβ ).dt (4.2) ϕ rα = ϕ Sα Li Sα (4.3) ϕ rβ = ϕ Sβ Li Sβ (4.4) ϕ S = ϕ 2 Sα + ϕ2 Sβ (4.5) ( ) Θ S = tan 1 ϕsβ ϕ Sα (4.6) e α = dϕ rα dt (4.7) e β = dϕ rβ dt (4.8) T e = 3 P 2 2 [ eα ω i Sα + e ] β ω i Sβ (4.9) After calculating the values of stator flux linkage in the stationary αβ-axis from equations (4.1) and (4.2), flux linkage magnitude and angle are determined from equations (4.5) and (4.6). By substituting the αβ-axis rotor flux vectors values calculated from equations (4.3) and (4.4), produced electric torque of the 54

76 4. Direct Torque Control Drive of BLDC Motor for EV Application BLDC motor is estimated from equation (4.9). Hysteresis controller generates a square wave pulse if the torque error is over the predefined band limits (T error = T ref T estimated ). Output of the hysteresis controller is 1 if the electric torque produced by the BLDC motor is more than reference torque input of the controller and is 0 if the produced electric torque is less than the reference torque. In this chapter the simulation model is tested for hysteresis band limits of 1, 0.1 and to show the torque ripples reduction capability of the DTC. The maximum switching frequency for minimum value of hysteresis band limits is around 10KHz. motor. The three phase conduction mode is used for switching of VSI drive of BLDC Six non zero voltage vectors that have been used to switch VSI are V 1 (100), V 2 (110), V 3 (010), V 4 (011), V 5 (001), V 6 (101). The estimated stator flux angle of the BLDC motor is divided into six equal sectors (each sector is 60 degrees) starting from 30 degrees. The correct voltage space vector is chosen according to the hysteresis torque controller output and flux angle sectors of the BLDC motor [61]. Switching look-up table for choosing voltage space vectors of VSI is shown in Table 4.1. Table 4.1: Three Phase Conduction Switching Mode for DTC of the BLDC Motor Torque error Flux angle sectors error V 6 (101) V 1 (100) V 2 (110) V 3 (010) V 4 (011) V 5 (001) 0 V 2 (110) V 3 (010) V 4 (011) V 5 (001) V 6 (101) V 1 (100) 4.4 Simulation Results and Discussion The proposed DTC drive of the BLDC motor is simulated in Simulink. Specification and parameters of BLDC motor used in the simulation model are listed in Table 4.2. Simulation results show that the DTC algorithm precisely estimates torque, flux linkage magnitude and angle of the BLDC motor. Torque ripples of the BLDC motor for various hysteresis controller band limits of DTC technique are compared with the conventional Hall Effect switching control technique. 55

77 4. Direct Torque Control Drive of BLDC Motor for EV Application Table 4.2: Specification of BLDC Motor Used in Simulation Model Description Value Unit DC voltage 400 V Phase resistance Ω Phase inductance H Inertia kg-m 2 Damping ratio N.m.s Flux linkage Wb Poles 4 - DTC drive of the BLDC motor is run at 1500 RPM under 10 N.m torque load (hysteresis band limits are set to 0.01). Speed and torque responses of the direct torque controlled BLDC motor drive are shown in Figure 4.2 [48]. Fast and smooth torque response of BLDC motor is remarkable in Figure 4.2. Figure 4.2: Speed and torque responses of the direct torque controlled BLDC motor drive 56

78 4. Direct Torque Control Drive of BLDC Motor for EV Application DTC drive is also tested for various hysteresis band limits and results are compared with the conventional Hall Effect switching control technique of the BLDC motor. Torque responses of the conventional control drive and DTC drive of the BLDC motor for various hysteresis band limits under same speed and load condition are shown in Figure 4.3 [48]. The electric torque ripple amplitude decreases as the hysteresis band limits are reduced. Figure 4.3: Pulsating torque of the BLDC motor for different hysteresis band limits Amplitude of the torque ripples is decreased up to four percent of the reference torque (0.4 N.m) in the proposed DTC model. That is ten times smaller than the conventional Hall Effect switching control technique. The proposed DTC drive of the BLDC motor has a better torque characteristics compared to the presented 57

79 4. Direct Torque Control Drive of BLDC Motor for EV Application model in [58]. High frequency switching in the VSI drive of the BLDC motor is the main disadvantage of small hysteresis band limits. Switching frequency is directly proportional to the switching loss in the inverters and practically hysteresis band limits can not be less than a particular threshold [48]. The stator flux magnitude and flux angle of the BLDC motor calculated by DTC technique are shown in Figure 4.4 [48]. As can be seen the stator flux magnitude is almost constant, around 0.22 Wb, in constant torque region below the rated speed. The calculated flux magnitude is almost the same as the limiting level of integration algorithm (2K L π/(3 3)). The change of the stator flux angle from 0 to 360 degrees shows a full electric rotation of the rotor. Figure 4.4: Calculated stator flux magnitude and flux angle of the BLDC motor 58

80 4. Direct Torque Control Drive of BLDC Motor for EV Application The direct torque controlled BLDC motor drive is tested at 1500 RPM under 5 and 10 N.m load torques. Stator flux linkage trajectory of the BLDC for 5 and 10 N.m load torques are shown in Figure 4.5. Six equal flux angle sectors discussed in the switching table of the DTC can be seen in the figure. Stator flux linkage locus of 10 N.m load torque has higher flux magnitude and sharper changes at the boundaries of the flux angle sectors [48]. Figure 4.5: Stator flux linkage trajectory of the BLDC motor for 5 and 10 N.m loads Robustness and fault tolerance of the in-wheel motor and its controller are the most significant parameter with respect to the electric vehicle s safety. Therefore in this section, behaviour of the proposed DTC drive and the conventional control drive of the BLDC motor are compared under the same sudden change of the load torque. An abrupt fifty percent increase to the load is applied at t=0.4 s while the BLDC motor is running at 1500 RPM under 10 N.m load torque. Speed response of DTC and conventional switching control drives of the BLDC motor under sudden increase of the load torque are shown in Figure

81 4. Direct Torque Control Drive of BLDC Motor for EV Application Figure 4.6: Speed response of the BLDC motor under sudden increase of load Figure 4.7: Torque response of the BLDC motor under sudden increase of load 60

82 4. Direct Torque Control Drive of BLDC Motor for EV Application As can be seen in the figure, the speed response of the BLDC motor for DTC drive follows the controller reference speed almost fifteen times faster than conventional switching technique after abrupt change of the load. However speed fluctuation of the DTC drive is more than the conventional switching technique. Torque response of DTC and conventional switching control drives of the BLDC motor under sudden increase of the load torque are shown in Figure 4.7. As can be seen the torque response of the DTC drive is much faster than the Hall Effects switching technique of the BLDC motor. Dynamic torque response of the electric motor plays an important role in the overall stability of the EV when it is subjected to frequent changes of the load torque. Therefore the DTC drive is more suitable for in-wheel BLDC motors compared to the conventional control drive in high performance electric vehicles. 4.5 Experimental Results Performance of the proposed direct torque controlled BLDC motor is investigated through experiment. A low voltage development board of the microchip using PIC18F4231 microcontroller is programmed to test the proposed DTC drive control system on a 24 volts experimental test BLDC motor. The experimental set-up of the BLDC motor is shown in Figure 4.8. MOSFET switches are used in VSI drive of the BLDC motor. Specifications of the experimental BLDC motor is given in Table 4.3 (refer to the hurst motor data-sheet in Appendix B). Direct torque control drive of the experimental BLDC motor is tested at 2000 RPM (below rated speed of the experimental BLDC motor) under 0.1 N.m reference torque and for 0.01 hysteresis band limits. Torque characteristics of the experimental BLDC motor are shown in Figure 4.9. As can be seen, produced electric torque of the BLDC motor pulsates around 0.1 N.m (load torque) with the maximum torque ripples amplitude of 0.12 N.m. 61

83 4. Direct Torque Control Drive of BLDC Motor for EV Application Figure 4.8: Experimental set-up of the BLDC motor Table 4.3: Specifications of the Experimental BLDC Motor Description Value Unit DC voltage 24 V Rated Speed 3000 RPM Rated torque 0.28 N.m Phase resistance Ω Phase inductance H Inertia kg-m 2 Poles 8-62

84 4. Direct Torque Control Drive of BLDC Motor for EV Application Figure 4.9: Torque characteristics of the experimental BLDC motor 4.6 Conclusion Electric vehicles are a viable alternative for the future transportation that does not emit greenhouses gases into the atmosphere. BLDC motors are commonly used by auto-mobile manufacturers as a propulsion system of the electric vehicles. Torque control of the in-wheel BLDC motors is an important factor in overall safety of the electric vehicles. In this chapter direct torque control switching technique of the BLDC motor is introduced as a suitable choice for the EV drive train application. DTC drive model of the BLDC motor is simplified, flux linkage observation is eliminated, for the constant torque region operation. The proposed DTC model of the BLDC motor is simulated in Simulink. Simulation results show that the estimated torque by state observer is as same as the produced electric torque of the BLDC motor. It is also possible to control the torque ripples of the BLDC motors by adjusting the hysteresis band limit. Simulation results show effective reduction of torque ripple amplitude by DTC drive compared to the conventional control system of the BLDC motor. The proposed 63

85 4. Direct Torque Control Drive of BLDC Motor for EV Application DTC drive is tested on a low voltage BLDC motor through the experimental setup. Experimental results show effective control of torque and correct performance of the proposed drive. Developed DTC switching technique of BLDC motor is capable of minimizing the torque ripples and delivering the smoother mechanical power to the wheels. 64

86 Chapter 5 Stability Analysis of a Novel Sensorless Drive of BLDC Motor 5.1 Chapter Overview Permanent magnet brushless DC motors have been widely used in traction applications such as propulsion system of electric vehicles in the last decade. Sensorless control drives of the BLDC motor have been extensively used in industrial application in recent years. In this chapter a novel sensorless technique based on back-emf zero crossing detection (ZCD) of one phase of the BLDC motor is proposed for EV application. The presented sensorless algorithm is simple and remarkably reduces sensing circuitry, noise susceptibility and cost of the BLDC motor control drives. Speed controller of the BLDC motor is a digital pulse width modulation technique using a Proportional Integral controller. Stability of the proposed sensorless BLDC motor drive using digital PWM technique is analysed using Lyapunov stability method. Based on Lyapunov stability criterion a novel condition for stability analysis of the PWM speed controller is derived. Effectiveness of the proposed sensorless technique and precision of the introduced stability analysis condition of the PWM speed controller are proved through simulation and experiment. 65

87 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor 5.2 Introduction BLDC motors have been used extensively in various industrial applications such as auto-mobile and aerospace industries in the 21st century. As it is discussed in previous chapters, BLDC motor is a type of conventional DC motor that is commutated electronically instead of using brushes. Schematic diagram of a 3 phases, 4 poles, star connected BLDC motor drive is shown in Figure 3.4 on Page 41. Detecting position of the permanent magnet rotor is the key point for electronic commutation of the BLDC motor. Position of rotor is normally detected by three Hall Effect sensors mounted in non-rotating end of motor with 120 electrical degree phase difference. Signal of the Hall Effect sensor is high or low as the rotor magnetic poles N or S passes near the sensor [32]. Therefore output of each sensor is high for 180 electrical degree and is low for the next 180 degree according to the rotor position. Correct voltage space vectors for switching inverter drive of the BLDC motor are chosen by decoding Hall Effect signals (refer to Table 3.1 on the page 3.1). Commutation of the BLDC motors using position sensors are much easier than sensorless methods. However increase of the motor manufacturing cost, problems due to the sensors breakdown, need of sensors to be mounted accurately that increase complexity of the manufacturing process, regular need of maintenance, extra wiring and limited operation of the motor due to temperature sensitivity of sensors are the main drawbacks of using position sensors to commutate BLDC motor [11]. Sensorless control drive of the BLDC motors have became popular in some specific applications such as EV in the last decades. Reduction complexity of the BLDC motor construction, cost and need of maintenance are immediate advantages of sensorless control. Various sensorless control techniques of the BLDC motor are reported so far. Back-EMF zero crossing detection, back-emf integration, back-emf harmonic analysis, freewheeling diode conduction of the unexcited phase and flux linkage based methods are examples of reported sensorless technique of the BLDC motor [7]. However transient time response and high commutation rates are the main drawbacks of the BLDC motor sensorless drives [48]. Since the trapezoidal back-emf permanent magnet synchronous mo- 66

88 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor tors require accurate phase current control and any phase errors in commutation signals cause significant pulsating torque and increase copper losses in the BLDC motor [62]. Back-EMF based sensorless methods mainly rely on zero crossing detection of the phase voltages. There are 120 electrical degree conduction period and 180 electrical degree conduction period switching techniques for VSI drive of the BLDC motors. There is an unexcited phase, silent phase, winding during each step in 120 conduction electrical degree mode that is used for Back-EMF ZCD and rotor positioning in sensorless drives of the BLDC motor [63]. Phase voltages of the BLDC motor during the silent period is the same as back-emf voltage [64]. Since neutral point of the BLDC motor windings is neither provided by the manufacturer nor stable during high frequency PWM switching; the back-emf of each phase is detected through the line voltage of the same phase of motor with respect to the negative terminal of VSI DC power supply [34]. This method also eliminates unwanted common mode noises on the voltage signal. Therefore the measured back-emf voltages does not need filtering and are less susceptible for switching noise [65]. Filters produce phase delays to the measured voltages that is dependent to the frequency or the motor speed. Reduction of torque per ampere capability of the BLDC motor, increase of torque ripples, additional copper loss, limited speed range operation of the motor, poor signal to noise ratio during starting time and severe commutation delays at high speeds are the main disadvantages of filtering phase delays [62][66]. Sensing line voltages of the BLDC motor with respect to the negative terminal of DC link of VSI also avoids filtering delay problems [67]; however operation of the BLDC motor at low speeds is the main drawback of this method [7]. Difficulties of back-emf sensing at low speeds (proper rotor positioning is not possible below 20% of the rated speed) and position detection errors during quick acceleration or deceleration of the BLDC motor are the general problems in ZCD of the back-emf sensing based sensorless methods [65]. Sensorless techniques based on zero crossing detection of line voltage differences of the BLDC motor is introduced [11][68]. Three phase line voltages of the BLDC motor have to be measured separately in the proposed methods that increase the measuring circuitry. There is 30 electrical degree phase delay be- 67

89 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor tween ZCD points and commutation points at each phase of the BLDC motor. The proposed method by Kim et al. [68] is much simpler because it does not need implementation of electrical degree phase delay after ZCD points. However the BLDC motor operation at low speeds is a drawback for the both proposed sensorless drives. Digital PWM control and hysteresis current control techniques are more common in the BLDC motor speed controllers than sliding mode control technique [69]. Pulse width modulation technique is used for speed control in this chapter. High frequency PWM signal is superimposed on the neutral point of star connected windings and induce noise to the measured line voltages. Back-EMF integration and third harmonic voltage integration are introduced to reduce the inverter switching noise effects and avoid using filters in the BLDC motor sensorless drives [66]. In the first method, commutation points are detected through back-emf ZCD of the unexcited phase of the motor as soon as the voltage integral pass a predefined threshold [64]. However accurate control and poor performance of the BLDC motor at low speeds are weaknesses of back-emf integration approach [11][66]. The other main problem of this technique is DC drift errors produced by analogue integrators, as it is discussed in Chapter 4, specifically at low speeds [64]. Addition of three phase voltages of the BLDC motor results in the third harmonic and multiples of the third harmonic components due to the symmetric three phase star connected windings. Zero crossing of integrated third harmonic signal occurs at the exact current commutation points of the BLDC motor [69]. This technique has a wider speed range compare to the back-emf integration method. Need of the neutral point voltage of the motor, variation of the winding inductance due to the rotation of permanent magnet motor (which needs to be constant in this method), low amplitude of the third harmonic components at low speeds, variation of magnitude and phase of harmonic components due to magnetic saturation and unbalanced situation in the surface mounted permanent magnet machines are the main limitations of the BLDC motor third harmonic back-emf based sensorless drives [7][70]. An application-specific integrated circuit (ASIC) based controller using third harmonic integration method with a phase locked-loop (PLL) is proposed to improve the BLDC motor performance at 68

90 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor ultra-high speeds [71]. The current phase delay is eliminated and the BLDC motor commutation is improved by the proposed technique. Sensorless commutation of the BLDC motor based on conducting status of anti-parallel connected freewheeling diode of the unexcited phase has a wide speed range and simple starting procedure [65]. Position detection errors during transient state, additional independent voltage sources, need for the proper isolation mechanism, complexity of algorithm and high cost of the controller drive are the main drawbacks of this method [66][72]. Switching signals in flux linkage based sensorless methods, same as the direct torque control discussed in Chapter 4, are extracted through flux linkage magnitude and flux angle of the BLDC motor. Flux linkage magnitude and flux angle values are calculated from the motor voltage and current signals. Flux based sensorless methods have significant flux estimation errors at low speeds [34]. Kim and Ehsani are reported a speed independent method to solve the flux estimation errors of the flux based sensorless methods of the BLDC motor at low speed [7]. The proposed method is based on the derivative of phase currents and needs digital implementation, thus it is susceptible to noise [11]. In this chapter, a simple sensorless commutation technique is proposed for the BLDC motor based on the ZCD of back-emf of one phase of motor instead of measuring back-emf of all the three phases. Back-EMF sensing circuitry and cost of the sensorless drive of BLDC motor is effectively reduced in the proposed method. Back-EMF of BLDC motor is sensed through line voltage of one phase of motor with respect to the negative terminal of DC bus of VSI. Therefore the sensed back-emf voltage does not need filtering and is less susceptible to the noise. The proposed commutation technique also can be used as a remedial strategy in the Hall Effect sensors fault tolerant control system (refer to Chapter 6, Section 6.5). The proposed sensorless technique is suitable to design integral in-wheel BLDC motor for electric vehicles due to its simplicity and low electromagnetic interference (EMI) [34]. Back-EMF of the BLDC motor is directly proportional to the rotor speed, therefore back-emf based sensorless methods have a poor performance at low speeds. The main problem raises up at start up of the BLDC motor where there is no back-emf at stand still situation. Therefore a starting procedure is needed 69

91 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor to speed up the BLDC motor to a point that is possible to measure the back- EMF voltage and implement the sensorless algorithm [65][73]. Various start-up methods are reported for the sensorless BLDC motor drives [73][74][75][76]. Start up techniques of the sensorless BLDC motor drives are not within the scope of this chapter. Starting algorithm reported by Iizuka et al. [74] is used in the proposed method in this chapter. The discussed sensorless commutation technique in this chapter is applicable after prepositioning and start up process. Digital pulse width modulation technique is used for speed control of the BLDC motor. Stable performance of the motor and its drive is critical and important in applications such as the electric vehicles that any instability of the system may put life of the passengers in danger. Up to the knowledge of the author, there are not much reported research works on stability analysis of the PWM control drives of BLDC motors so far. Milivojevic et al. [77] have discussed stability analysis of FPGA based PWM controller drive of the BLDC motor according to the Lyapunov stability method. Merits and demerits of the proposed stability analysis method are discussed in details in section 5.4. The introduced stability analysis method is improved and a new equation is introduced to analyse the stability of the BLDC motor drives using PWM speed controller for EV application. The discussed stability analysis method is validated through simulation and experimental results. Some of the presented simulation and experimental results in this chapter have been published by Tashakori et al. [34]. 5.3 Proposed Sensorless Technique for BLDC Motor A comprehensive knowledge of the BLDC motor performance such as exact information about sensors output signals with respect to the permanent magnet rotor position and correlation between the back-emf, line voltages and commutation points of the BLDC motor is needed to develop a sensorless algorithms. Equivalent electrical circuit of a three phase star connected BLDC motor and VSI drive are shown in Figure

92 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor Figure 5.1: Equivalent electrical circuit of the BLDC motor drive Terms V a, V b and V c are referred to the line voltages of the BLDC motor, V n is the neutral point voltage and V dc is the DC bus voltage of the inverter. Ideal commutation signals, terminal and back-emf voltages of the BLDC motor are shown in Figure 5.2. As can be seen in the figure, at each instant of time only two switches of inverter is conducting according to the permanent magnet rotor position (refer to the Table 3.1 on the page 40). It means that two phases are conducting and current of one phase is zero, therefore it is possible to measure the back-emf voltage through the unexcited phase. Consider phase B and C are conducting and phase A is the silent phase. Therefore, i a + i b + i c = 0 i b = i c and i a = 0 (5.1) Resistance and inductance of stator windings are assumed to be constant. Magnetic circuit saturation and losses are also ignored in calculations. Then 71

93 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor Figure 5.2: Ideal commutation signals, terminal and back-emf voltages of the BLDC motor terminal voltage equations of the BLDC motor can be expressed as [34], V a = E a + V n (5.2) V b = Ri b + L di b dt + E b + V n (5.3) V c = Ri c + L di c dt + E c + V n (5.4) where E b = E c and V a + V b = V dc, therefore the neutral point voltage can be 72

94 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor calculated by adding equations (5.3) and (5.4), V n = V DC 2 (5.5) Substituting the calculated neutral voltage in equation (5.2), terminal voltage of the unexcited phase can be written as, V a = E a + V DC 2 (5.6) Zero crossing points of back-emf can be detected from terminal voltage of the floating phase through equation (5.6). Thus zero crossing points of back-emf voltage (E a = 0) occur when the corresponding floating terminal voltage is [72], V a = V DC 2 (5.7) Therefore as it is also shown in the Figure 5.2, zero crossing points of back- EMF voltage of the unexcited phase happen when terminal voltage of the corresponding phase of the BLDC motor is equal to half of the DC power supply of inverter. On the other words as can be seen in the figure, commutation instants of the BLDC motor occur 30 electrical degree after the ZCD points and there is the exact 120 electrical phase shift between commutation signals of different phases in the BLDC motor. Knowing the exact back-emf zero crossing points of one phase is enough to generate commutation signals of the other two phases of the BLDC motor. Therefore in this chapter an optimized sensorless commutation technique of the BLDC motor is introduced based on back-emf zero crossing detection of one phase of motor. In the proposed method line voltage of only one phase of the BLDC motor is sensed with respect to the VSI DC link instead of measuring all three terminal voltages [34]. A reference commutation signal is generated based on ZCD of the line voltage. The reference commutation signal is set to logic one at zero crossing points of rising edge and it is set to logic zero at zero crossing points of falling edge of the measured line voltage. Then commutation signals of all three phases are generated according to their correlated electrical degree delays from the reference signal. 73

95 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor A simple formula is introduced to calculate time of the electrical degree delay with respect to the speed of the BLDC motor. If speed of the motor is constant during the commutation intervals (controller keeps the motor speed constant) and the number of motor pole pairs (Ratio of electrical degree to mechanical degree rotation of rotor) is known; therefore the time needed for one electrical degree rotation of the rotor can be calculated in seconds as below [34], T one electrical degree = 60 P (360 ω 2 ref) (5.8) Where P is the number of the motor poles and ω ref is the reference speed in RPM. Practically it is easy and convenient to calculate and implement the electrical degree delays as time delays in microcontrollers. Schematic diagram of the proposed BLDC motor sensorless drive is shown in Figure 5.3. Figure 5.3: Schematic diagram of the proposed BLDC motor sensorless drive 74

96 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor 5.4 Stability Analysis of Digital PWM Controller Hysteresis current control and pulse width modulation control techniques are the most widely used methods in the BLDC motor speed control drives [69][78]. Speed of the BLDC motor is directly proportional to the applied terminal voltages of the motor. A three phase voltage source inverter is used to supply the BLDC motor in the six-step algorithm. Variable DC link inverter and PWM technique are commonly used techniques for adjusting the average output voltage of the VSI. In the first method voltage of the DC power supply of the inverter changes according to the motor speed. In PWM technique, a high frequency duty cycle controlled signal is added to switching signals of the VSI. By adjusting the duty cycle of the PWM signal according to the motor speed, it is possible to the applied voltages to the BLDC motor and consequently the speed of the motor. The PWM signal can either be multiplied to the switching signal of upper switches, lower switches, or all six switches of the VSI. Pattern of the applied line voltage to the BLDC motor varies for different PWM switching mode [5]. In this chapter, PWM signals are applied to the upper switches of the VSI in the simulation model and experimental set-ups. A digital control scheme for the BLDC motor drives is reported by Sathyan et al. [78] based on the two predetermined duty cycle values (state high, D H, and state low, D L, PWM duty cycles) for PWM signal. Controller switches between state high and state low according to the BLDC motor speed. Predefined duty cycle values limit the functionality of the controller for the variable speed applications such as electric vehicle. A proportional and integral controller is used to adjust the duty cycle of PWM signal with respect to the speed error. Ideally one duty cycle is chosen by PI controller, (0 D 1), at any particular reference speed. However practically the controller adjusts duty cycle values in close boundaries of the ideal duty cycle instead of having two predefined states [34]. Therefore there is no need to know desired duty cycle states for different speed operations and it is suitable for the application with the frequent change of speed such as the in-wheel motors. Stable performance of the BLDC motor control drive is critical in application such as EV with respect to the safety point of views. In this chapter stability 75

97 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor of the proposed sensorless BLDC motor drive using digital PWM controller is analysed through Lyapunov s second method stability criterion (refer to the Appendix A, Section Four). Milivojevic et al. [77] are discussed stability analysis of a FPGA-based controller of the BLDC motor using PWM technique based on Lyapunov stability method. Stability analysis results in introducing limit conditions for D Hmax and D Lmin (two predefined duty cycle states) that define the stable operation range of the BLDC motor drive. Effect of the load torque and rate of change of load torque are not considered in the analyses that are important in EV application due to the frequent change of load on tires. The effect of load torque is also considered in the discussed stability analysis in this chapter. Speed error is considered as the structural variable or switching surface of the PWM controller to apply the Lyapunov stability condition. s = ω ref ω (5.9) Candidate Lyapanuv function introduced in [77] is used for stability analysis of digital PWM controller. C(s(x)) = 1 2 st (x)s(x) (5.10) According to the Lyapunov stability criterion, PWM controller is stable equilibrium if the Lyapunov candidate function is locally positive at s(x) = 0, then its derivative should be locally negative ( dc < 0 C s < 0). In the other word dt s t the control system is Lyapunov stable if s and its first derivative ṡ, have opposite signs [34]. Mathematical equations of electrical and mechanical systems of the BLDC motor can be expressed as, DV DC = Ri + L di dt + K eω (5.11) K t i T l = j dω dt + βω (5.12) From now on it is assumed that torque constant and back-emf constant are equal (K) in this chapter. A second order differential equation with ω as the 76

98 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor state variable can be derived by substituting the current value from equation (5.12) into the equation (5.11). d 2 ω dt + a dω 1 dt + a dt l 2ω b 1 D + b 2 T l + b 3 dt = 0 (5.13) where constants a 1, a 2, b 1, b 2, b 3 are defined as below to simplify the equation (5.13). a 1 = ( a 2 = ( βl + jr ) (5.14) jl βr + K2 ) (5.15) jl b 1 = KV DC jl (5.16) b 2 = R jl (5.17) b 3 = 1 j (5.18) State variable x in the candidate function represents speed of the motor [77]. Therefore derivative of switching surface function is, ṡ = s dω ω dt = dω dt (5.19) Solving equation (5.13) for derivative of speed and substitute the result in equation (5.19) results in, ṡ = a 1 ω + a 2 ω.dt b 1 D.dt + b 2 T l.dt + b 3 T l (5.20) According to the Lyapunov stability criterion, if s is negative (actual speed is more than reference speed) then ṡ should be positive and if s is positive (actual speed below reference speed) ṡ should be negative. Therefore limit conditions of the digital PWM speed controller s duty cycle for system to be Lyapunov stable 77

99 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor are [34], a 1 dω b 1 dt + a 2 ω + b 2 T l + b 3 dt l b 1 b 1 b 1 dt a 1 dω b 1 dt + a 2 ω + b 2 T l + b 3 dt l b 1 b 1 b 1 dt > D ε if s < 0 (5.21) < D + ε if s > 0 (5.22) Ideal performance of the digital PWM controller concludes to have a constant duty cycle if the actual speed of the BLDC motor is exactly as same as the reference speed ( dω = 0). Furthermore if it is assumed that there is no load dt torque change during operation of the motor ( dt l = 0); then equations (5.21) and dt (5.22) can be merged as, βr + K2 R D = ( )ω ref + ( )T l (5.23) KV DC KV DC Duty cycle condition for stability analysis of the digital PWM controller for the constant torque and speed applications is expressed by equation (5.23). As the controller does not work ideal in practice and the motor speed fluctuates around the reference speed; therefore the duty cycle values chosen by the PI controller oscillate around the ideal duty cycle value expressed by equation (5.23). However change of the speed and load torque must be considered for stability analysis of the BLDC motor in the electric vehicle that torque and speed parameters changes continuously due to frequent start, stop, acceleration and deceleration. 5.5 Simulation Results and Discussion The proposed back-emf based sensorless commutation technique of the BLDC motor is simulated in Simulink. The BLDC motor specifications used in the simulation model are given in Table 2.5 on page 22. The reference commutation signal is generated based on the back-emf ZCD of the measured line voltage of phase A of the BLDC motor. The line voltage of phase A is measured with respect to the negative terminal of the VSI DC link. An embedded Matlab code has been written to detect zero crossing points of the back-emf based on the discussed method. The reference commutation signal is set to logic one at the 78

100 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor rising ZCD point and is set to logic zero at the falling ZCD point. The main three commutation signals of the BLDC motor are generated by implementing the electrical degree delays calculated from the equation (5.8). Commutation signal of the measured phase has 30 electrical degree delay and the other two commutation signals have 150 and 270 electrical degree delays from the reference signal respectively. The generated commutation signals are decoded and the correct switching signals are applied to the inverter. MOSFET switches are used to model the three phase voltage source inverter. The inverter is switched in a six-step sequence to direct the current to the three-phase BLDC motor. Digital PWM speed controller is implemented to control speed of the motor. A Matlab file is embedded in Simulink model to generate a duty cycle controlled PWM signal. Duty cycle of the PWM signal is determined by a PI controller based on the speed error. Duty cycle controlled PWM signals is multiplied to switching signal of the upper switches in each phase of the VSI. The start-up algorithm proposed by Iizuka et al. [74] is used to run the BLDC motor up to the point that the proposed sensorless drive is able to detect the back-emf voltage and control the motor. The proposed sensorless drive simulation model of the BLDC motor is tested for 2000 RPM reference speed of the PWM controller under 5 N.m torque load. Line voltage of phase A, corresponding back-emf of phase A and zero crossing detected points by controller are shown in Figure 5.4 [34]. As can be seen zero crossing points of the back-emf occurs exactly when the line voltage value pass through half of the inverter DC supply. There are two zero crossing points during each electrical cycle of phase voltage; one at rising edge and the other at the falling edge of the line voltage. These ZCD points are the reference for the commutation signal to set high and low respectively. Zero crossing points and the commutation signal of phase A of the BLDC motor are shown with respect to the electrical degree in Figure 5.5 [34]. As it is magnified in the figure, the commutation signal of phase A is delayed exactly 30 electrical degree from back-emf zero crossing points. Commutation signal of each phase is logic one for 180 electrical degree and logic zero for the other 180 electrical degree during one full electric rotation of the BLDC motor. 79

101 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor Figure 5.4: Line voltage, Back-EMF and ZCD points of phase A of BLDC motor Figure 5.5: Zero crossing points and the commutation signal of phase A 80

102 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor Current and commutation signal of phase A of the BLDC motor are shown in Figure 5.6. As can be seen in the figure, current flows at the same time of rising edge of the corresponding commutation signal. At each commutation sequence one phase is connected to the positive terminal of inverter DC link (current entering the phase winding), and one phase is connected to the negative terminal of inverter DC link (current exiting the phase winding) and the third phase is not excited (current is zero). Therefore as can be seen in the figure, phase current has one positive cycle, one negative cycle and two unexcited periods (zero current) during one full electrical rotation of the BLDC motor. Zero crossing points of back-emf occur in the period that phase current is zero. Figure 5.6: Current, commutation signal and ZCD points of phase A Speed response of the BLDC motor sensorless drive and chosen duty cycles by PI controller are shown in Figure 5.7 [34]. By considering the operation of the BLDC motor under constant speed and load torque, the ideal PWM duty cycle percentage calculated from the equation (5.23) according the motor parameters and 2000 RPM reference speed is 75%. As can be seen PI controller has chosen duty cycles around the ideal value. 81

103 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor Figure 5.7: Speed response of the BLDC motor and duty cycle values selected by PI controller Figure 5.8: State plane of digital PWM speed controller 82

104 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor State plane of the PWM speed controller is plotted in Figure 5.8. It can be seen that the PWM controller keeps the BLDC motor speed at the stable equilibrium around the switching surface at 2000 RPM. Therefore the proposed sensorless drive of the BLDC motor using digital PWM speed controller is Lyapunov stable. The BLDC motor model is tested in brake condition of an electric vehicle from constant speed and torque operation to full stop position. Brake condition is chosen to study the proposed sensorless drive using PWM speed controller performance under variable speed and load torque conditions. The BLDC motor model is run at 2000 RPM under 5 N.m torque load (constant speed and torque operation) and a soft brake is applied at t = 1 s to the motor and vehicle is supposed to stop at t = 3 s. Brake duration is two seconds for the electric vehicle to stop completely (zero speed). It is assumed that brake is applied both electrically, decreasing reference speed of controller to zero and mechanically, increasing the load torque on the wheels. Speed and torque characteristics of the BLDC motor during brake condition are shown in Figure 5.9 [34]. Figure 5.9: Speed and torque characteristics of the BLDC motor during brake 83

105 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor The BLDC motor has a stable speed response and speed follows the reference speed of the controller. Produced electric toque and torque ripples of the BLDC motor is increased after the mechanical brake is applied to the motor. Maximum torque ripples occurs when the BLDC motor speed is around 1300 RPM. The produced electric torque of the motor is zero when vehicle is fully stopped. Ideal, estimated and chosen duty cycle values by the PI controller during the brake condition are shown in Figure 5.10 [34]. The first graph shows the ideal duty cycle values calculated from equation (5.23). Changes of speed and load torque of the BLDC motor are not considered in the first graph. The second graph shows the estimated duty cycle values calculated from equations (5.21) and (5.22). Changes of speed and load torque of the BLDC motor are considered in the second graph. The third graph shows the simulation results for duty cycle values chosen by the PI controller. As can be seen in the figure, duty cycle values of the simulation results are following pattern of the estimated duty cycle values. Therefore simulation results validate correctness of the equations (5.21) and (5.22) for variable speed and torque operation condition of the BLDC motor. Figure 5.10: Duty cycle values during the brake condition 84

106 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor State plane of the PWM speed controller during the brake condition is plotted in Figure As can be seen in the figure, the system remains at the stable equilibrium point around the reference speed at each instant of time during the brake condition. Simulation results show that the PI controller has selected the correct duty cycle values to keep the BLDC motor drive stable during the brake condition. Figure 5.11: State plane of digital PWM speed controller during the brake 5.6 Experiment Results Effectiveness of the proposed sensorless drive of the BLDC motor using digital PWM speed controller is investigated through experiment. Experimental test rig of the BLDC motor is same as reported in Chapter 4, Section 5. A low voltage development board of microchip using PIC18F4231 microcontroller is programmed to test the proposed sensorless BLDC motor drive. PIC microcontroller is also programmed to implement the digital PWM speed controller of the BLDC motor in a closed loop scheme. One of the in-built Hall Effect sensors of the motor is 85

107 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor used to estimate the actual speed of the motor. MOSFET switches are employed in the three phase voltage source inverter to supply the BLDC motor. Experimental speed response of the BLDC motor from the halt position up to 2000 RPM reference speed of the PWM controller under 0.1 N.m torque load is shown in the Figure 5.12 [34]. It takes around two seconds till the proposed sensorless commutation algorithm to be able to sense the back-emf and speed up the motor to the reference speed. Experimental speed response of the BLDC motor drive using sensors is also shown in Figure 5.13 to highlight the starting delay in sensorless BLDC motor drives. The speed response of the sensorless BLDC motor drive oscillates around 2027 RPM. Speed error is 1.35% of the speed controller reference speed that is acceptable. Figure 5.12: Experimental speed response of the sensorless BLDC motor drive Three phase commutation signals of the BLDC motor that are generated by the proposed sensorless method are shown in Figure 5.14 [34]. As shown in the figure, they are logic for 180 electrical degree and there is 120 electrical degree phase differences between commutation signals. Phase differences between commutation signals are implemented based on the electrical degree time delays (refer to the equation (5.8) on the page 74). 86

108 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor Figure 5.13: Experimental speed response of the BLDC motor drive using sensors Figure 5.14: Generated commutation signals by sensorless drive of BLDC motor 87

109 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor PWM switching signals that are applied to the upper side switches of the VSI are shown in Figure 5.15 [34]. The Ideal duty cycle percentage calculated from the equation (5.23) based on the experimental motor specification and operation condition is 60.31%. As shown in figure, duty cycle values chosen by controller vary from 47% to 67% that are in the close boundary of the ideal duty cycle. Figure 5.15: PWM switching signals applied to the upper side switches of VSI Line voltage and corresponding generated commutation signal of the phase C of sensorless drive of the BLDC motor are shown in Figure As can be seen phase C line voltage of the motor is exactly in the same phase with the corresponding generated commutation signal. Experimental results prove effectiveness and stable performance of the sensorless drive of the BLDC motor using digital PWM speed controller. Performance of the digital PWM speed controller is also studied in the context of a light weight electric vehicle. The proposed PWM speed controller is applied to the in-wheel BLDC motors of the concept four wheel drive electric vehicle. The in-wheel BLDC motor set-up and the concept light weight four wheel drive EV are shown in Figure

110 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor Figure 5.16: Line voltage and commutation signal of the phase C of BLDC motor Figure 5.17: The in-wheel BLDC motor set-up in a light weight EV 89

111 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor Four similar three phase in-wheel BLDC motors designed for electric motor cycle application are used as the drive train of the concept EV. Specifications of the in-wheel BLDC motors are given in Table 3.3 on the page 46. Permanent magnet rotor position detection of the in-wheel motors are based on inbuilt Hall Effect sensors. Each in-wheel motor has its own individual control drive; however all of the motors operate with the same speed controlled by the acceleration pedal of the vehicle. Acceleration pedal provide a control analogue voltage signal to the speed controller drive of in-wheel BLDC motors. The reference speed control input voltage varies from 0.7 volts for zero speed to 3.6 volts for the full speed rotation. In-built drum brakes of the motor hubs are used as mechanical brake of the electric vehicle. Four 12 volts lead-acid batteries are used to supply inverters of the in-wheel motors in the vehicle. The concept light weight four in-wheel drive EV is tested for three different speed operating condition of the in-wheel BLDC motors on a flat road. Line voltage and corresponding commutation signal of one of the in-wheel BLDC motors at different operating condition are shown in Figures Levels of the reference input voltage to the speed controllers provided by the acceleration pedal are 1.7 volts at low speed, 2.8 volts at moderate speed and 3.6 volts at full load speed operating conditions. In-wheel BLDC motors drive 6 Amps at full speed operating condition. Therefore according to the manufacturer datasheet, the in-wheel motors are producing 2.93 N.m torque at RPM. The BLDC motor line voltages are measured with respect to the negative terminal of the inverter DC power supply. As can be seen in the figure, duty cycle of the PWM signal increases based on the input reference speed of the controller. At full speed operating condition duty cycle of the PWM signal is 100% and the maximum possible voltage is applied to the in-wheel motors. Experimental results confirm the correct and stable performance of the proposed digital PWM speed controller of the BLDC motor. 90

112 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor Figure 5.18: Line voltage and commutation signal of the in-wheel BLDC motor at different operating condition of the light weight EV 91

113 5. Stability Analysis of a Novel Sensorless Drive of BLDC Motor 5.7 Conclusion Sensorless commutation techniques of the BLDC motors have been widely used in industrial applications in recent years. An optimized sensorless commutation technique based on zero crossing detection of back-emf voltage of only one phase of the BLDC motor is discussed in this chapter. In this method back-emf zero crossing points are detected through the measured line voltage of one phase of the motor with respect to the negative terminal of the VSI DC link. Measuring only one phase voltage of the BLDC motor instead of all three phase voltages, remarkably reduce the cost, noise susceptibility, sensing components and wiring of the motor drive. The proposed sensorless method is simulated and tested through the experiment. Simulation and experimental results prove correct and stable performance of the proposed sensorless drive of the BLDC motor. Simplicity, low cost, low noise susceptibility and ease of implementation of the control technique on a single chip microcontroller or a digital signal processor are advantages of the proposed sensorless commutation method of the BLDC motor as the drive train of high performance electric vehicles. A digital PWM switching technique is implemented to control speed of the proposed sensorless BLDC motor drive in a closed loop scheme. A PI controller is utilized to select the duty cycle of the PWM controller instead of setting two predefined duty cycle values. Stability of the proposed sensorless drive of the BLDC motor using digital PWM speed controller is analysed through Lyapunov second method. Stability analysis results in deriving a novel condition for duty cycle of the PWM signal based on the motor parameters and operating condition of the motor such as speed and load torque. Validity of the presented stability analysis condition is verified through simulation and experiment. Effective performance of the digital PWM speed controller is also tested in a light weight electric vehicle using four in-wheel BLDC motors. Experimental results show the stable performance of the electric vehicle using digital PWM speed controllers in different operating condition. 92

114 Chapter 6 Fault Diagnosis of the BLDC Motor Drive for EV Application 6.1 Chapter Overview Safe operation of electric vehicles is of the prime concerns in automotive industry. Various Fault Tolerant Control Systems have been developed for electric motors to diagnose and handle the motor faults and maintain the motor performance in post-fault condition in the last decades. Implementing FTCS s in control drives of the in-wheel BLDC motors increase reliability, robustness and safety of the electric vehicles. In this chapter, two fault tolerant control systems are proposed to handle inverter switch faults and position sensors (Hall Effect sensors) failure in the BLDC motor drives. Fault diagnosis in both proposed FTCS s are based on Discrete Fourier Transform (DFT) analysis of line voltages of the BLDC motor. A four wheel drive EV using in-wheel BLDC motors is modelled to study and analyse the EV performance under various drive train fault conditions. Simulation results show instability of the electric vehicle immediately after inverter open circuit switch fault occurrence. A BLDC motor drive is modelled in Simulink to study the motor performance under fault conditions. The BLDC motor model was validated by experimental data under no fault condition. Various VSI switch faults and Hall Effect sensor faults are applied to the validated BLDC motor model. Expert systems are de- 93

115 6. Fault Diagnosis of the BLDC Motor Drive for EV Application signed to detect and identify various VSI switch faults and Hall Effect position sensors failure. Multidimensional fault diagnosis knowledge based tables are developed by analysing simulation results of the BLDC motor model under various fault conditions. The proposed fault diagnosis systems are capable of detecting fault occurrence and identify faulty switch or faulty sensor. Simple strategies are recommended to remove the faults and keep the BLDC motor drive operation in post-fault condition. Simulation results and developed knowledge based tables are validated through experimental data. The proposed fault tolerant control systems are simple, do not need excessive computations and can be executed with the main control program of the BLDC motor. 6.2 Introduction BLDC motors are popular as drive train in traction applications such as hybrid electric vehicles and pure electric vehicles. Stable motor operation is important on the overall EV drive train performance and directly effects on safety of the vehicle. Control of the BLDC motor mainly depends on the accurate detection of the permanent magnet rotor position that leads to choose the correct voltage space vectors to switch voltage source inverter and supply the BLDC motor. Therefore any malfunction of the position detecting sensors or the VSI switches degrade the BLDC motor performance. Three Hall Effect sensors with 120 electrical degree phase difference are used to detect rotor position of the BLDC motor in simulation model and experimental test rig in this chapter. This chapter presents two novel fault tolerant control systems for inverter switch faults and position sensors failure in the BLDC motor drives. The presented FTCS s are focused on the three phase star connected BLDC motors. Fault tolerant control systems for the four-phase, five-phase and six-phase BLDC motors are the prospective potential research studies. Overall model of a 3 phases, 4 poles star connected BLDC motor and its VSI drive are shown in Figure 6.1 [79]. As shown in the figure, control system and variable voltage source inverter are main sections of the BLDC motor drives. Control system is responsible for decoding and choosing the three phase inverter switching signals to commutate 94

116 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Figure 6.1: Overall BLDC motor drive model 95

117 6. Fault Diagnosis of the BLDC Motor Drive for EV Application the BLDC motor. Six step conducting algorithm is used to switch inverter and supply voltage to the BLDC motor. Commutation signals, phase currents, phase back-emf and line voltages of the BLDC motor and on state switches of VSI with respect to the electrical degree of the rotor are shown in Figure 6.2 [80]. Accurate rotor positioning and correct performance of the inverter switches are the key factors on the BLDC motor performance [12]. Figure 6.2: BLDC motor output characteristics and VSI switching steps 96

118 6. Fault Diagnosis of the BLDC Motor Drive for EV Application A digital PWM controller is implemented to control speed of the BLDC motor. A duty cycle controlled PWM signal is multiplied by the switching signals of the inverter to control the average output voltage of the VSI. Behaviour of the BLDC motor are compared for various PWM switching modes under normal and critical condition (refer to Appendix A, Section E). The BLDC motor shows the most robust performance when PWM signals are applied to all inverter switches [5]. In this chapter duty cycle controlled PWM signal is applied to all switches of the inverter in simulation and experimental set-up. In applications where safety is critical such as an electric vehicle, any fault or failure in propulsion system results in an accident or a hazard. Therefore implementing various electric motor fault tolerant control systems is necessary in electric vehicles [81]. Generally a FTCS is responsible to do the following tasks, Fault diagnosis (fault detection and identification); Fault isolation; Remedial action. Fault diagnosis systems are responsible to detect and identify the fault. The faulty section must be isolated immediately to avoid any further damage to the system after fault detection. Finally, appropriate remedial actions should be taken to keep the system working with the maximum possible efficiency in the post-fault condition [12]. Various faults may occur in stator, rotor, position sensors or voltage source inverter of the BLDC motor drives. Possible common faults in each section of the BLDC motor are summarized in Table 6.1 [79]. Faults have various effects on the BLDC motor performance; some faults degrade the motor performance and cause severe damage if they last longer and some others cause the motor failure and stop operation in few seconds after fault occurrence. Therefore various fault diagnosis algorithms must be implemented in the BLDC motor drive, however there is a priority on applying fault isolation and remedial strategies if number of successive faults occur at the short time intervals [79]. 97

119 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Table 6.1: Common Faults in the BLDC Motor Drive Section Fault type Description Stator Short circuit of Three phase Rotor windings Open circuit of windings Change of resistance Eccentricity Asymmetry Rotor unbalanced Rotor magnet damage Misalignment Bearing fault Three phase to ground Two phase Two phase to ground One phase to ground Turn to turn fault It may happen by some inverter faults Overheating Overloading Inverter Switch faults Open circuit fault DC link fault Short circuit fault Short circuit to ground Capacitor bank fault Position Sensors breakdown Flaws in the core sensors Unbalanced positioning Change in the bias current Change in core magnetic property Change in induced magnetic field 98

120 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Fault diagnostic algorithms of the BLDC motor are classified by employing signal analysis, model based and knowledge based methods [13]. In the signal analysis based methods fault is detected through comparison of the extracted features of the motor signals with the ideal signals at normal operating condition. The main advantage of this method is that it does not need the BLDC motor model for fault detection, however fault diagnosis is not fast compared to the other two methods [81]. Parameter estimation techniques are normally used to diagnose the fault in the model based techniques. Fault diagnosis in this method is quite fast and can be used for online fault monitoring, however an exact dynamic model of the BLDC motor is needed. Reported research works by Liu et al. [13] and Moseler et al. [82] are good examples of a model-based fault diagnosis system for the BLDC motors. Expert systems using fuzzy logic or neural network are developed for fault diagnosis in the knowledge based methods [13]. The knowledge can be gathered either through an experienced engineer with a thorough understanding of the BLDC motor system, or via a comprehensive study of the BLDC motor dynamics through simulation model of the motor [81]. Fault diagnosis systems presented in this chapter are knowledge based expert systems that use signal analysis methods. A fault tolerant control system for inverter open circuit switch faults of the BLDC motor for EV application is discussed in the next section. Inverter short circuit switch faults are removed by six fast acting fuses that are connected in series with inverter switches. Therefore a short circuit fault is treated as an open circuit fault by the proposed FTCS [79]. Dynamic parameters of a four wheel drive EV using in-wheel BLDC motors are analysed under inverter open switch fault of the motor. Fault diagnosis technique and remedial strategies are discussed. A fault tolerant control system for position sensors breakdown in the BLDC motors is also presented in the fourth section of this chapter. The BLDC motor behaviour is analysed under various position sensor faults. Position sensor fault diagnosis algorithm and the remedial strategy to rectify the fault are discussed. Effectiveness of the both proposed fault diagnosis techniques are investigated through experimental results. 99

121 6. Fault Diagnosis of the BLDC Motor Drive for EV Application 6.3 Inverter Open Circuit Switch Faults A fault tolerant control system of the BLDC motor drive for open circuit fault of the inverter switches is discussed in this section. It is assumed that short circuit faults of the inverter switches are removed by series fast acting fuses. Therefore a short circuit fault effects on the BLDC motor as an open circuit fault. Precious research works are published on FTCS s for VSI switch faults in the BLDC motor drives [81-85]. A fault diagnosis algorithm based on wavelet analysis of the inverter DC link current are introduced for the BLDC motors [83]. Wavelet analysis needs massive computational efforts and increase complexity of the FTCS. Open circuit fault diagnosis of the inverter switches based on the BLDC motor stator current is reported by Park et al. [84]. The proposed method is simple and does not need massive computations. However current based fault diagnosis methods are not capable of distinguishing the fault occurrence either is inside the motor or the inverter [85]. Four various VSI switch faults diagnosis methods based on different voltage sensing points of the BLDC motor are discussed [85]. Voltage errors are used for fault detection in the presented techniques. Fault diagnosis time is significantly reduced; however the proposed methods have major limitations. Neutral point voltage of the BLDC motor is required for two of the presented techniques. Since neutral point of the BLDC motor is not stable during high frequency PWM switching therefore the proposed fault diagnosis techniques based on neutral voltage sensing are not consistent in a closed loop control scheme. Pattern of the BLDC motor line voltages change continuously in applications such as electric vehicles with the frequent changes of the speed and load torque. Therefore the ideal reference voltages should also change dynamically to find the correct voltage errors of the BLDC motor [81]. A fault diagnosis system based on the voltage of lower switches in each phase of the VSI is proposed for voltage fed PWM inverter systems by Yu et al. [86]. Noise susceptibility of sensors used inside the inverter due to high frequency PWM switching signals is the main limitation of the proposed method [79]. A fault diagnosis technique based on the neural network system is proposed to detect the most of common inverter faults in induction motor drives for EV and hybrid EV applications [87]. Features used for 100

122 6. Fault Diagnosis of the BLDC Motor Drive for EV Application fault detection in neural network are extracted from torque, voltage and current signals of the induction motor. The proposed fault diagnosis technique is fast and accurate. Complex algorithm, high number of used sensors and need of sensing the neutral voltage point of the BLDC motor are main drawbacks of the proposed method [81]. Effect of the inverter switch faults on dynamic parameters of a four in-wheel drive EV are analysed in this section. A knowledge based fault diagnosis technique is proposed to detect and identify the inverter open circuit fault in the following. The proposed fault diagnosis algorithm is validated through experimental data. Some of the presented simulation and experimental results in this section have been published by Tashakori et al. [79][81] EV Dynamics Analysis under Inverter Open Circuit Switch Fault Simulation models are mainly developed to decrease the cost and length of the design process of advanced systems. They can be used to study behaviour of the systems under abnormal condition. Modelling of the hybrid electric vehicles has been grown since 1970s [88]. Simulation models are used to study various aspects of the vehicle; for instance vibration, vehicle handling, noise, vehicle performance, safety, stability, component testing and etc. [89]. However there are few simulation models of pure electric vehicle to study effect of the in-wheel motor faults on the EV performance. A four in-wheel drive EV using four BLDC motors is modelled in Simulink using Simscape library to analyse the EV performance during inverter open circuit switch faults in the in-wheel motors. Schematic diagram of the four in-wheel drive EV model is shown in Figure 6.3 [79]. Four BLDC motors using digital PWM speed controllers are modelled as drive train of the EV. Specifications of the BLDC motor model BLK423S manufactured by Anaheim Automation Company are used in the EV model (refer to Table 3.2 on the page 43). The duty cycle PWM control signal is applied to all six switches of the VSI. A vehicle body in longitudinal motion from SimDriveline library is used as vehicle body model. Specifications of the vehicle s body used in the EV model are given in Table

123 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Figure 6.3: Schematic diagram of the four in-wheel drive EV model Table 6.2: Specification of the Vehicle s Body Used in the EV Model Description Value Unit Mass 1200 Kg Number of in-wheel motors 4 - Horizontal distance from 1.4 m CG to front wheels Horizontal distance from 1.6 m CG to rear wheels CG height above ground 0.5 m Frontal area 3 m 2 102

124 6. Fault Diagnosis of the BLDC Motor Drive for EV Application No Fault Condition The electric vehicle model is tested to run from stall position up to 2000 RPM reference speed of the in-wheel BLDC motors on a flat road (zero inclination) under no fault condition. Speed response of the electric vehicle is shown in Km/h in Figure 6.4 [79]. Speed of the vehicle is increased gradually up to 113 Km/h in 50 seconds after start moving. Figure 6.4: EV speed under no fault condition Normal forces applied to the tires from the electric vehicle s body are shown in Figure 6.5 [79]. As can be seen in the figure, force on tires changes fast during the first few seconds when the EV has the highest acceleration rate. The force on the rear tires are more than the front tires at starting time that is due to central gravity (CG) position of the vehicle s body. As the vehicle approaches to the constant speed region, tire forces are almost constant but forces on the front tires are more than those on the rear tires. 103

125 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Figure 6.5: Normal tire forces under no fault condition In-wheel BLDC motors start from stall position up to 2000 RPM reference speed of digital PWM controllers. Speed responses of the in-wheel BLDC motors are shown in Figure 6.6 [79]. As can be seen the in-wheel electric motors operate at the same speed to keep the EV moving stable in a straight line. Speed fluctuation of the front wheels at starting point is due to the slip of front tires. Electric torques produced by in-wheel motors are shown in Figure 6.7 [79]. As can be seen the in-wheel motors deliver the same electric torques to the wheels that keep the electric vehicle moving stable. Effect of the front tires slip can also be seen in the produced torques of the in-wheel BLDC motors A and B. In-wheel motors are produced the initial torque of 350 N.m to overcome the high inertia of the vehicle s body at stall position. Produced electric toques by the in-wheel motors are reduced to 30 N.m as the speed of the vehicle reaches to 113 km/h. 104

126 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Figure 6.6: Speed responses of the BLDC motors under no fault condition Figure 6.7: Torque responses of the BLDC motors under no fault condition 105

127 6. Fault Diagnosis of the BLDC Motor Drive for EV Application VSI Open Circuit Fault The electric vehicle model is run from stall position up to 2000 RPM reference speed of the in-wheel BLDC motors. Open circuit fault switch S 1 (refer to Figure 6.1) of the inverter is applied to the in-wheel BLDC motor A at t = 40 s while the EV speed is reached to 110 km/h. Other three in-wheel motors were operating under no fault condition. No fault protection system is implemented to the in-wheel BLDC motor drive to see the maximum fault effect on the EV performance. Speed characteristic of the electric vehicle under open circuit fault of switch S 1 is shown in Figure 6.8 [79]. As can be seen in the figure speed of the EV is suddenly decreased and the vehicle is unstable after fault occurrence. Figure 6.8: EV speed under open circuit fault of switch S 1 Normal forces applied to the tires from the electric vehicle body under open circuit fault of the inverter switch S 1 in the BLDC motor A are shown in Figure 6.9 [79]. Applied Vertical forces to the tires oscillate with high amplitude ripples after the fault occurrence that is due to the unbalanced load torques on the wheels in post-fault condition. 106

128 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Figure 6.9: Normal tire forces under open circuit fault of switch S 1 Speed and torque responses of the in-wheel BLDC motors under open circuit fault of the inverter switch S 1 in the BLDC motor A are shown in Figure 6.10 and Figure 6.11 respectively [79]. High amplitude notches can be seen in speed and torque responses of the faulty in-wheel motor (the BLDC motor A). The faulty in-wheel motor is totally unstable and speed and torque responses of other three in-wheel BLDC motors are also deteriorated after fault. Other three motors are effected due to unbalanced distribution of the vehicle force on tires, load torques on the motors, after fault occurrence (refer to the Figure 6.9). The four wheel drive EV model is also tested for inverter open circuit fault of switch S 2 of the BLDC motor A. Simulation results (refer to the Appendix A) show unstable performance of the EV and almost are same as the inverter open circuit fault of switch S 1. Therefore performance of the electric vehicle is not safe under inverter open circuit faults. Sudden VSI switch fault occurrence put life of the passengers at risk and makes the electric vehicle a hazard to other nearby vehicles or people on the road. 107

129 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Figure 6.10: Torque responses of the BLDC motors under open circuit fault of switch S 1 Figure 6.11: Speed responses of the BLDC motors under open circuit fault of switch S 1 108

130 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Comparison of the EV performance under faulty condition with no fault condition demonstrates the need of FTCS s for the in-wheel motors to improve safety of the electric vehicles Fault Diagnosis In this section fault detection and identification algorithms of the inverter open circuit switch faults of the three phase BLDC motors are presented. Any switch fault of the VSI effects directly on the applied voltages to the BLDC motor. Therefore pattern change of the BLDC motor line voltages are used for fault diagnosis. Line voltages of the BLDC motor are measured with respect to the negative terminal of the VSI DC link Fault Detection Any pattern change of the line voltages of the BLDC motor for constant speed and torque load condition is the signature of fault occurrence. Variations of the motor speed and load torque should be considered in fault diagnosis. Discrete Fourier transform analysis is used to detect pattern changes of the BLDC motor line voltages. Frequency spectrum of the measured voltages are extracted from equation (6.1) for the specific intervals of time. Power Spectral Density (PSD) of the calculated frequency spectrum of the measured voltages of each phase of the BLDC motor are calculated from equation (6.2) for each time interval. Successive PSD values are compared to find the PSD errors of each phase from equation (6.3). V (f) = N 1 n=0 V n e j2πk n N k = 0, 1,..., N (6.1) E m (f) = V (f) 2 (6.2) ε m = E m (f) E m 1 (f) (6.3) In practice, line voltages of the BLDC motor are sensed for specific intervals 109

131 6. Fault Diagnosis of the BLDC Motor Drive for EV Application of time continuously while the motor is operating in constant speed and torque load condition. The minimum time interval for correct fault detection is one electrical rotation of the BLDC motor. The time of one electrical degree rotation is inversely proportional to the BLDC motor speed. Fault occurrence is detected if calculated PSD errors of the BLDC motor line voltages exceeds the predefined limits. Five percent of the calculated PSD value of the BLDC motor line voltages under no fault condition are set as the limit to avoid short term disturbance detections Fault Identification An expert system is developed to identify the faulty switch of inverter based on studying the BLDC motor performance under faulty condition through the simulation model. The BLDC motor model is validated through the experimental set-up. A three phase low voltage BLDC motor is used as an experimental test motor. A Low voltage (LV) development board of microchip using PIC18F4231 micro-controller is programmed to control the experimental BLDC motor. Experimental test rig of the BLDC motor and specifications of the experimental test motor used in simulation model are given in Chapter 4 (refer to Figure 4.8 and Table 4.3). Digital PWM controller is implemented to control speed of the BLDC motor. An embedded code is written in Simulink model to produce a duty cycle controlled PWM signal. High frequency PWM signal is applied to all switches of the inverter. The experimental test BLDC motor and its simulation model are tested at 2000 RPM under 0.1 N.m load torque. Line voltage and Hall Effect signal of phase A of the experimental set-up and the simulation model of the BLDC motor are shown in Figure 6.12 [12]. Agreements between simulation and experimental results validate the BLDC motor model. The validated BLDC motor model is also used to develop fault diagnosis system for Hall Effect sensors failure in Section Five. Inverter open circuit switch faults are applied to the validated BLDC motor model and results are analysed to develop the fault diagnosis system. Power spectral density errors of the line voltages of the BLDC motor model are calculated under healthy operating condition and after fault occurrence. A knowledge based 110

132 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Figure 6.12: Line voltage and Hall Effect signal of phase A of BLDC motor table is developed to identify the fault by analysing the calculated PSD errors of the motor line voltages in various fault conditions. Open Circuit Fault of Switch S 1 is applied at t = 0.5 to the BLDC motor model. Line voltages of the BLDC motor during the open circuit fault of switch S 1 are shown in Figure 6.13 [79]. Positive amplitude spikes can be seen in line voltage of the faulty phase of the motor. The line voltage of phase A has totally distorted after the fault occurrence. Line voltages of phase B and phase C 111

133 6. Fault Diagnosis of the BLDC Motor Drive for EV Application are also changed, though the voltage of phase A has the most variations. Therefore the PSD error of the phase A voltage should be the maximum. Calculated Power spectral density errors of the BLDC motor line voltages for open circuit fault of switch S 1 are given in Table 6.3 [79]. Figure 6.13: Line voltages of BLDC motor during open circuit fault of switch S 1 Table 6.3: Simulation PSD Values for Open Circuit of S 1 Description Phase A Phase B Phase C PSD before fault [E m 1 (f)] PSD after fault [E m (f)] e PSD error [ε m ] Open Circuit Fault of Switch S 2 is applied at t = 0.5 to the BLDC motor model. Line voltages of the BLDC motor during the open circuit fault of switch S 2 are shown in Figure 6.13 [79]. As can be seen the line voltage of 112

134 6. Fault Diagnosis of the BLDC Motor Drive for EV Application phase A, the faulty phase, has negative amplitude spikes. Voltage of phase B and C are also significantly distorted, however variations of the phase A voltage is much more than other two phases. Calculated power spectral density errors of the BLDC motor line voltages for open circuit fault of switch S 2 are given in Table 6.4 [79]. Figure 6.14: Line voltages of BLDC motor during open circuit fault of switch S 2 Table 6.4: Simulation PSD Values for Open Circuit of S 2 Description Phase A Phase B Phase C PSD before fault [E m 1 (f)] PSD after fault [E m (f)] e PSD error [ε m ] Line voltages of the BLDC motor are also studied under open circuit faults of other inverter switches through the validated simulation model. The calculated 113

135 6. Fault Diagnosis of the BLDC Motor Drive for EV Application PSD errors for open circuit faults of phase A switches of the inverter can be generalized for the other two phases due to symmetry of the BLDC motor [12]. Two flags are defined for fault diagnosis; Switch Fault Flag (SFF) for each phase of the motor to identify the faulty switch and Fault Phase Flag (FPF) to identify the faulty phase. Numeric values are assigned to SFF of each phase and FPF according to the linguistic variables based on the calculated PSD errors as below [79], SFF is -1 if PSD error is over the limits and negative; SFF is 0 if PSD error is in the limits; SFF is 1 if PSD error is over the limits and positive; FPF is 0 if no fault is detected; FPF is 1 if maximum PSD error related to phase A; FPF is 2 if maximum PSD error related to phase B; FPF is 3 if maximum PSD error related to phase C. A multidimensional knowledge based table are developed to identify the faulty switch based on simple quasi-fuzzy if-then rules according to the assigned numeric values of the flags. Developed knowledge based if-then rules for inverter switches faults diagnosis are shown in Table 6.5 [79]. Table 6.5: Proposed Knowledge Based Table for Inverter Switches Faults Diagnosis Fault type SFF SFF SFF FPF phase A phase B phase C No fault Open circuit S Open circuit S Open circuit S Open circuit S Open circuit S Open circuit S

136 6. Fault Diagnosis of the BLDC Motor Drive for EV Application As the fault diagnosis is based on the PSD errors of line voltages, there is no need to know the exact patterns of the BLDC motor line voltages in advance. This is the most advantage of the proposed fault diagnosis system compared to the previously reported systems for the EV application Experimental Results The low voltage development board of the microchip is modified to test the open circuit fault of the inverter switches and Hall Effect sensors. The modified control board of the BLDC motor is shown in Figure 6.15 [79]. There is an in-built over current protection circuit in the control board that avoids phase currents to exceed a predefined limit. Figure 6.15: The modified LV development board control drive of BLDC motor Open circuit faults of phase A switches of the VSI is applied to the BLDC motor while it is running at 2000 RPM under 0.1 N.m load torque. Line voltages of the motor under open circuit faults of switches S 1 and S 2 are shown in Figure 6.16 and Figure 6.17 respectively [79]. As can be seen the voltage of faulty phase, phase A, is totally deteriorated for both open circuit faults of switches of 115

137 6. Fault Diagnosis of the BLDC Motor Drive for EV Application S 1 and S 2. Voltages of phase B and phase C are also distorted, though the phase A voltage has always the most variations same as the simulation results. Figure 6.16: Line voltages of BLDC motor under open circuit fault of switch S 1 116

138 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Figure 6.17: Line voltages of BLDC motor under open circuit fault of switch S 2 117

139 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Power spectral density errors of the BLDC motor line voltages for experimental open circuit faults of switches S 1 and S 2 are given in Tables 6.6 and 6.7 respectively. AS can be seen PSD errors of the line voltage of phase A is maximum for both open circuit faults of switches of S 1 and S 2 compared to the other two phases. However the PSD errors of experimental results are not as large as the PSD errors of simulation results due to the effect of over current protection circuit of the control drive. Table 6.6: Experimental PSD Values for Open Circuit of S 1 Description Phase A Phase B Phase C PSD before fault [E m 1 (f)] PSD after fault [E m (f)] PSD error [ε m ] Table 6.7: Experimental PSD Values for Open Circuit of S 2 Description Phase A Phase B Phase C PSD before fault [E m 1 (f)] PSD after fault [E m (f)] PSD error [ε m ] The experimental BLDC motor drive is also tested under open circuit switch faults of other legs of the inverter. Experimental results are similar to the open circuit switch faults of leg A of inverter. The most pattern changes always belong to the line voltage of faulty phase. Discussed experimental results validate the fault diagnosis algorithm developed by simulation results for inverter switches open circuit faults. 118

140 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Remedial Strategy EV drive train faults must be isolated and rectified in a mean time to maintain the maximum possible vehicle performance and safety of the passengers. Therefore after the faulty switch is identified, the corresponding faulty leg is disconnected from the inverter by implemented controlled switches in each leg of the VSI. There are various inverter reconfiguration topologies to maintain the BLDC motor operation in post-fault condition. Inverter topologies are not in the scope of this chapter; however two simple inverter topologies are discussed for postfault condition with respect to the electric vehicle application. Reconfiguration of the three phase voltage source inverter to the four switches topology inverter for the BLDC motors in post-fault condition is recommended by Lee et al. [90]. Schematic diagram of the proposed four switches topology inverter is shown in Figure 6.18 [81]. Figure 6.18: Schematic diagram of the proposed four switches topology inverter After isolating the faulty leg of the inverter, the corresponding phase of the BLDC motor is connected to the midpoint of DC link of the VSI through control switches S a ; S b ; S c. Reconfiguring to the four switch mode operation of the 119

141 6. Fault Diagnosis of the BLDC Motor Drive for EV Application inverter avoids further major faults inside the in-wheel BLDC motor drive. However performance of the in-wheel BLDC motor is degraded. This method can be used to increase the reliability of the in-wheel BLDC motors just for a short time in post-fault condition until the vehicle gets the proper service [81]. A modular and easy controlled fault tolerant VSI using a redundant leg is proposed for the BLDC motor drives by Errabelli et al. [91]. The faulty leg of the inverter is replaced by the redundant leg in post-fault condition. Schematic diagram of the proposed fault tolerant inverter of the BLDC motor with a redundant leg is shown in Figure 6.19 [81]. The corresponding faulty phase of the BLDC motor is connected to the redundant leg of the inverter through independent control switches S ra ; S rb ; S rc. In this technique performance of the in-wheel BLDC motor is not degraded compared to the four switches topology inverter; however its manufacturing cost is due to the redundant leg. Since the reliability and safety of the in-wheel motor drive is more important than the cost in EV application, therefore the proposed fault tolerant VSI with a redundant leg is recommended for in-wheel motors. Figure 6.19: Schematic diagram of the proposed fault tolerant inverter with a redundant leg 120

142 6. Fault Diagnosis of the BLDC Motor Drive for EV Application 6.4 Position Detection Sensors Failure In this section a fault tolerant control system for position sensors failure of a three phase BLDC motor are discussed. Commutation of the BLDC motor is done with respect to the position sensor signals. Therefore failure or malfunction of the position sensors effects directly on the motor performance. On the other hand faults in position sensor may result in immediate over current of the BLDC motor drive under high load torques [92]. Unbalanced positioning of the Hall Effect sensors by manufacturer is not in the sensor failure categories and scope of this study. However it increases low-frequency harmonics in the torque ripples and degrades performance of the BLDC motor [93]. The main faults that may result in failure of a Hall Effect sensor in BLDC motors are [94]: 1. Flaws in the sensor s core due to corrosion, cracks, residual magnetic fields and core breakage; 2. Effect of temperature variations on the magnetic properties of the ferrite core; 3. Effect of mechanical shocks on the orientation of the induced magnetic field in the sensor; 4. Changes in the bias current of the sensor. There are few reported research works on fault tolerant control system of position sensors failure in the BLDC motors. Major sensor faults of an Interior Permanent Magnet Motor (IPMM) as the propulsion system of an electric vehicle are discussed [92]. Position sensor faults are detected through difference between the calculated rotor angle and the actual rotor angle. Permanent magnet rotor position angle is calculated by a sensorless algorithm based on extended EMF in rotating reference frame [95]. Sensorless mode control of the permanent magnet motor is recommended as a remedial strategy. Complexity of the BLDC motor sensorless control drives and transition algorithms to the sensorless mode are the main drawbacks of the proposed method [12]. Performance of the BLDC motor is analysed under Hall Effect sensors faults for lunar rover wheel application [96]. Effects of the position sensor faults on inverter switching signals and phase currents are shown only through a simulation model; however simulation results 121

143 6. Fault Diagnosis of the BLDC Motor Drive for EV Application are not discussed. There is also no discussion on fault diagnosis techniques and remedial strategies to rectify the fault in the paper [80]. Fault diagnosis technique and a novel remedial strategy for Hall Effect sensor failure in the BLDC motors is discussed in this section. Performance of the BLDC motor under position sensors fault are analysed through a validated simulation model. A knowledge based fault diagnosis technique is developed to detect the faulty position sensor based on simulation results analysis. Finally effectiveness of the proposed fault diagnosis algorithm is proved through experimental data. Some of the presented simulation and experimental results in this section have been published by Tashakori et al. [12][80] Performance of the BLDC Motor under Position Sensor Faults Behaviour of the BLDC motor drive under various position sensor faults are studied through a validated simulation model (refer to Figure 6.12 on the page 111) to develop the fault diagnosis algorithm. The BLDC motor experimental set-up is the same as the one that is used and explained for inverter switch faults in the previous section. Position sensor faults are applied to the validated simulation model of the BLDC motor while the motor is running under stable and healthy condition. Then output characteristics of the motor such as line voltages, phase currents, speed and torque characteristics are analysed. Hall Effect sensor failure is divided into two categories based on the output signal of the sensors. Output signal of the sensor is constant high (logic 1, H a = 1), sensor is short circuit to it power supply, or it is constant low (logic 0, H a = 0), sensor is open circuit [12]. Behaviour of the BLDC motor is discussed under both fault conditions of the corresponding Hall Effect position sensor of phase A Hall Effect Signal is Constant Zero Hall Effect sensor fault H a = 0 is applied to the validated BLDC motor model at t = 0.5 s while the motor is ruining at 2000 RPM under 0.1 N.m load torque. Speed and torque responses of the BLDC motor under H a = 0 fault condition are 122

144 6. Fault Diagnosis of the BLDC Motor Drive for EV Application shown in Figure 6.20 [12]. As can be seen in the figure, speed and torque of the BLDC motor is unstable and out of the control under H a = 0 fault condition. Figure 6.20: Speed and torque responses of BLDC motor under H a = 0 fault condition Line voltages of the BLDC motor under H a = 0 fault condition are shown in Figure 6.21 [12]. Line voltages of the BLDC motor are measured with respect to the negative terminal of DC link of VSI. As shown in figure line voltages of all phases are changed due to the direct effect of Hall Effect sensor faults on the switching signal of the inverter. Hall Effect sensor faults also increase the BLDC motor torque ripples and degrade its performance. Effect of the various position sensor faults on the switching signals of the VSI are summarized in Table 6.8 [80]. Each position sensor fault changes two switching signals of the inverter to constant zero. This effect can be considered as an open circuit fault of two inverter switches at the same time [80]. As given in Table 6.8 switching signals of the switches S 1 and S 6 are constant zero (switches S 1 and S 6 are open circuit) under H a = 0 fault condition. 123

145 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Figure 6.21: Line voltages of BLDC motor under H a = 0 fault condition Hall Effect Signal is Constant One Hall Effect sensor fault H a = 1 is applied to the validated BLDC motor model at t = 0.5 s while the motor is ruining at 2000 RPM under 0.1 N.m load torque. Speed and torque responses of the BLDC motor under H a = 1 fault condition are exactly similar to the H a = 0 fault condition. Torque ripples of the BLDC motor is increased and the motor is not stable after fault occurrence. However switching signals of the switches S 2 and S 5 remain open circuit after H a = 1 fault occurrence that is not the same as H a = 0 fault condition. Therefore line voltages of the BLDC motor should be different in H a = 1 fault condition compared to H a = 0 fault condition. Line voltages of the BLDC motor under H a = 1 fault condition are shown in Figure 6.21 [12]. 124

146 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Table 6.8: Effect of the Various Sensor Faults on the Switching Signals of the VSI Fault type Effected switches of VSI Switches status H a = 0 S 1, S 6 Open H a = 1 S 5, S 2 Open H b = 0 S 3, S 2 Open H b = 1 S 1, S 4 Open H c = 0 S 5, S 4 Open H c = 1 S 3, S 6 Open Fault Diagnosis Correct space voltage vectors are chosen according to the electrical position of the permanent magnet rotor by decoding the commutation signals (refer to Table 3.1). As can be seen there is no electrical position of the rotor that all sensor signals are one or zero. Therefore addition of the commutation signals, H f, are either one or two under healthy operating condition of the BLDC motor. H f = H a + H b + H c (6.4) Hall Effect sensor faults change the value of H f during one electrical rotation of the rotor. If H f = 3, it shows that one of the position sensor signals is constant one and if H f = 0, it means that one of the position sensor signals is constant zero. The minimum required time for fault detection is time of one electrical rotation of the rotor that is quite fast. Hall Effect sensors Fault detection Flag (HFF) is introduced to detect the fault occurrence as below [80], HFF is set 0 if H f = 1 or H f = 2; HFF is set 1 if H f = 3; HFF is set -1 if H f = 0. Position sensors faults are detected if the Fault Detection Flag is not zero. Detection of the fault and fault type are possible through HFF; however faulty sensor cannot be identified through fault detection flag. DFT analysis of the line voltages of the BLDC motor is used to identify the faulty sensor. Same as 125

147 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Figure 6.22: Line voltages of BLDC motor under H a = 1 fault condition discussed for the inverter switch faults, line voltages of the BLDC motor are measured and saved for specific intervals of time continuously. Frequency spectrum of the measured line voltages are calculated from equation (6.1). Power spectral density of the line voltages frequency spectrum are calculated from equation (6.2). PSD errors of the sensed line voltages between successive time intervals under constant speed and torque load are signature to identify the faulty sensor [12]. Single-sided amplitude spectrum of line voltage of phase A of the BLDC motor under no fault and Hall Effect sensor faults of phase A (H a = 0 and H a = 1) are shown in Figure 6.23 [80]. As can be seen High amplitude harmonics are added to frequency spectrum of the phase A line voltage of the BLDC motor under position sensor faults condition. Therefore energy density of the line voltage frequency spectrum under position sensor fault is not same as the healthy operating condition of the BLDC motor. PSD errors of all three phase line voltages of the BLDC motor under Hall Effect sensor faults of H a = 0 and H a = 1 are calculated 126

148 6. Fault Diagnosis of the BLDC Motor Drive for EV Application and given in Tables 6.9 and 6.10 respectively [80]. Figure 6.23: Amplitude spectrum of the phase A line voltage of BLDC motor Table 6.9: PSD Values for H a = 0 Fault Condition Description Phase A Phase B Phase C PSD before fault [E m 1 (f)] PSD after fault [E m (f)] PSD error [ε m ] Table 6.10: PSD Values for H a = 1 Fault Condition Description Phase A Phase B Phase C PSD before fault [E m 1 (f)] PSD after fault [E m (f)] PSD error [ε m ]

149 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Hall Effect sensors fault Identification Flag (HIF) is introduced for each phase of the BLDC motor. Numeric values are assigned to the fault identification flag of each phase based on the sign of the calculated PSD errors. FIF numeric values are assigned as below [80], HIF is set -1 if PSD error is negative; HIF is set 1 if PSD error is positive. A multidimensional table is developed to diagnose the position sensor faults of the BLDC motor based on fault detection and identification flags. Fault diagnosis algorithm is developed based on the gathered knowledge from characteristics of the BLDC motor model under various Hall Effect sensor faults. Simplicity and no need of knowing the exact pattern of the BLDC motor line voltages for different speed and torque loads are the main advantages of the proposed fault diagnosis method. The proposed multidimensional knowledge based table for Hall Effect sensors fault diagnosis in the BLDC motors is shown in Table 6.11 [12]. Table 6.11: Proposed Knowledge Based Table for Position Sensor Faults Diagnosis Fault type HIF HIF HIF HFF phase A phase B phase C No fault X X X 0 H a = H a = H b = H b = H c = H c = Hall Effect sensor fault occurrence and fault type are detected through the fault detection flag, when HFF is not zero, and faulty sensor is identified based on the fault identification flag of each phase of the BLDC motor according to the Table

150 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Experimental Results Correctness of the proposed fault diagnosis system for Hall Effect sensors failure of the BLDC motor is investigated through experiment. Low voltage development board of microchip has been modified to test Hall Effect sensors faults of the experimental BLDC motor. Three half-bridge gate drivers using MOSFETs as shown in Figure 6.24 are used as the VSI drive of BLDC motor [97]. Signal PWM0 is applied switching signal of the switch Q 0 that is equivalent of the switch S 1 in Figure 6.1. Figure 6.24: Half-bridge gate driver and inverter of LV development board Open circuit, H a = 0, and short circuit, H a = 1, faults of the corresponding Hall Effect sensors of phase A are applied to the experimental BLDC motor while it is running at 2000 RPM under 0.1 N.m torque load. Speed oscillations and high acoustic noise are the first observations in the BLDC motor drive after fault occurrence. Six LED lights are provided on the control board that are on when the corresponding switching signals are logic high. Status of the LED lights for open circuit and short circuit faults the corresponding Hall Effect sensors of phase A are shown in Figure 6.25 [80]. Equivalent switching signals of S 1 and S 6 (PWM1 and PWM4) of VSI drive of the experimental BLDC motor are constant 129

151 6. Fault Diagnosis of the BLDC Motor Drive for EV Application zero for H a = 0 fault. Equivalent switching signals of S 5 and S 2 (PWM5 and PWM0) of inverter are also constant zero for H a = 1 fault. Figure 6.25: Corresponding switching LED lights on the control board under position sensor faults of phase A: (a) Open circuit fault (b) Short circuit fault Therefore simulation results presented in Table 6.8 on the page 125 are validated through experimental results. Line voltages of the experimental BLDC motor under H a = 0 and H a = 1 fault conditions are shown in Figure 6.26 and Figure 6.27 respectively [80]. As is shown in the figures, line voltages of the experimental BLDC motor are changed during faults compared to the healthy operating condition. Single-sided amplitude spectrum of the experimental BLDC motor line voltage of phase A under no fault and Hall Effect sensor faults of phase A are shown in Figure 3.3 [80]. High amplitude harmonics can be seen on the phase A line voltage of the experimental BLDC motor under position sensor fault conditions. However amplitude and frequency of the added harmonics are different for H a = 0 fault compared to H a = 1 fault. Therefore power spectral density of the line voltages are different according the fault types in post-fault condition. 130

152 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Figure 6.26: Line voltages of the experimental BLDC motor under H a = 0 fault 131

153 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Figure 6.27: Line voltages of the experimental BLDC motor under H a = 1 fault 132

154 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Figure 6.28: Amplitude spectrum of the phase A line voltage of experimental BLDC motor Power spectral density (PSD) errors of the experimental BLDC motor line voltages under H a = 0 and H a = 1 fault conditions are calculated and given in Table 6.12 and Table Calculated PSD errors of the experimental line voltages under Hall Effect sensor fault condition validate the calculated PSD errors of the simulation line voltages. However experimental PSD errors are not as large as simulation ones that is due to the inbuilt over current protection circuit of the controller board. The experimental BLDC motor is also tested under the corresponding Hall Effect sensors faults of other phases. Calculated experimental PSD errors of the line voltages under Hall Effect sensors faults of other two phases are similar to the presented experimental results for phase A of the BLDC motor. Experimental results validate the proposed knowledge based fault diagnosis table for position 133

155 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Table 6.12: PSD Values for Experimental H a = 0 Fault Condition Description Phase A Phase B Phase C PSD before fault [E m 1 (f)] PSD after fault [E m (f)] PSD error [ε m ] Table 6.13: PSD Values for Experimental H a = 1 Fault Condition Description Phase A Phase B Phase C PSD before fault [E m 1 (f)] PSD after fault [E m (f)] PSD error [ε m ] sensors breakdown of the BLDC motors Remedial Strategy The signal of the faulty Hall Effect sensor should be disconnected from the BLDC motor drive after fault detection. Signal of the faulty sensor is generated by the controller by implementing 120 electrical degree delay to one of the other available Hall Effect signals. Electrical degree delays are calculated in time as it is discussed in Chapter 5. If the BLDC motor speed does not change during commutation intervals, the time for one full electrical rotation of the permanent magnet rotor of the BLDC motor can be calculated from equation (6.5). T one electrical degree = 60 P (360 ω 2 ref) (6.5) where P is number of the motor poles and ω ref is there reference speed of the controller. Effectiveness of the proposed equation to calculate the correct time of electrical delays of the BLDC motors is proved through simulation and experimental results in the previous chapter [34]. An embedded Matlab code is implemented in the BLDC motor simulation model to test performance of the proposed FTCS for Hall Effect sensors fault. H a = 0 fault is applied to the BLDC motor model at t = 0.5 s while the motor 134

156 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Figure 6.29: Speed response of the fault tolerant controlled BLDC motor drive is running at 2000 RPM under 0.1 N.m torque load. Speed characteristics of the fault tolerant controlled BLDC motor drive is shown in Figure 6.29 [12]. As can be seen in the figure, fault tolerant control drive of the BLDC motor detect, identifies and maintains performance of the motor in post-fault condition. The commutation signal of phase A is generated correctly by implementing the calculated the time delays. Fault diagnosis time in simulation model is about second which is fast and acceptable for the BLDC motor drives. 6.5 Conclusion In this chapter two fault tolerant control systems are presented for inverter switch faults and Hall Effect position sensors failure of the BLDC motor drives. Performance of a four in-wheel drive EV are studied under open circuit faults of inverter switches S 1 and S 2 in one of the front in-wheel BLDC motors through simulation. Results show that EV becomes unstable immediately after VSI switch faults occurrence. Behaviour of the BLDC motor is analysed under various inverter switches and position sensor faults through a validated simulation model. 135

157 6. Fault Diagnosis of the BLDC Motor Drive for EV Application Results show that these faults effect directly on the applied voltages and cause the BLDC motor to be unstable. Therefore necessities of implementing FTCS s for the in-wheel BLDC motors are inevitable to increase safety and reliability of the electric vehicles. The proposed fault diagnosis systems are based on the DFT analysis of the line voltages of the BLDC motor. Knowledge based tables are developed to diagnose the faulty switch of inverter and the faulty sensor by analysing power spectral density errors of the BLDC motor line voltages. Since the PSD errors of the line voltages are signature of fault detection and identification in both proposed fault diagnosis systems; the exact pre-knowledge of the line voltages pattern for various reference speed or torque loads is not needed. Fault tolerant three phase VSI with a redundant leg is recommended to isolate and rectify the inverter switch faults. A simple method is also introduced to generate the signal of faulty position sensor by implementing the time delays between Hall Effect signals in post fault condition. Effectiveness of the proposed fault diagnosis algorithms and the knowledge based tables are validated through the experiment results. 136

158 Chapter 7 Conclusion Electric vehicles have been considered for the green transportation since 1980 s. Electric motors are the main propulsion system in the electric vehicles. Different electric motors are used as drive train of electric vehicles. Induction and BLDC motors are the most popular drive trains used in the electric vehicles available in the world market by the manufacturers. In-wheel motor technology has been one of the research interests in automotive industry in the last decade. Important technical requirements of an in-wheel motor are: high torque at low speeds; high torque/power to size ratio; high efficiency; high dynamic response; robustness; low noise susceptibility. Performance of the brushed DC, induction, BLDC and switched reluctance motors are compared according to the in-wheel technology requirements through simulation. Simulation models of the motors are tested under the same normal and critical conditions. Simulation results show better torque/speed characteristics and faster dynamic response of the BLDC motor compared to the other motors in the normal condition. Brushed DC motor also has high torque at low speeds and fast dynamic response in the normal condition. Low efficiency, low speed ranges and periodic need of maintenance are the main disadvantages of the brushed DC motors that are not suitable for the EV application. On the other hand, the induction motor has better performance compared to the other motors in the critical condition. Low efficiency at high speeds, slow dynamic response and slip are the major drawbacks of the induction motor for the in-wheel motor application. The switched reluctance motor has also a good performance in the critical 137

159 7. Conclusion condition, however high amplitude torque ripples and noise susceptibility are its main limitations. The BLDC motor performance is not suitable for in-wheel motor application specifically during electrical faults. Implementing fault tolerant control systems improve reliability and robustness of the BLDC motors. Comparison results and discussions of the motors presented in Chapter 2 conclude to the point that the BLDC motor is the most suitable motor for the in-wheel motor electric vehicles. Precise electronic control of the in-wheel motors improves safety, efficiency and overall performance of the electric vehicles. An accurate simulation model of the in-wheel motors are needed to study performance of the motor for different control algorithms. Therefore a three phase star connected BLDC motor model with an ideal back-emf has been modelled in the Chapter 3. The presented model is based on Laplace transform of the mathematical equations of the BLDC motor. Simulation results of the motor model are validated through the experimental data of a three phase in-wheel BLDC motor hub designed for electric motor cycle application. Simplicity and the ideal back-emf waveforms of the proposed model make it useful for performance analysis of the various BLDC motor drives. Mechanical output power of the electric motors is dependant on speed and produced electric torque. Therefore, precise torque control of the in-wheel motors improves performance of the electric vehicles. Direct torque control switching technique of the BLDC motor is discussed in Chapter 4. DTC drive of the BLDC motor is simulated in Simulink and results are compared with those of the conventional switching technique of the motor. Comparison results show effective control of the BLDC motor torque and lower torque ripples by the DTC switching technique compared to the conventional switching technique. The proposed DTC drive is implemented to test the experimental low voltage BLDC motor. Experimental results show capability of the DTC drive in effective control of the torque in various conditions. Sensorless drives of the BLDC motor are widely used in various applications in recent years. The novel back-emf based sensorless drive of the BLDC motor presented in Chapter 5, is based on back-emf zero crossing detection of only one phase of the BLDC motor. Commutation signal of one phase is generated by 30 electrical degree delays from ZCD points of the back-emf voltage. Com- 138

160 7. Conclusion mutation signals of the other phases are generated based on 120 electrical degree phase delays from the first commutation signal. Cost, sensing circuitry and noise susceptibility of the BLDC motor sensorless drives are decreased by the proposed method. The proposed sensorless technique of the BLDC motor is simulated, built practically and tested experimentally. Good agreements between simulation and experimental results justify the correct performance of the proposed sensorless BLDC motor drive. A digital PWM speed controller using a PI controller to select the duty cycle of the PWM signal is implemented in the proposed sensorless drive of the BLDC motor. Stability of the BLDC motor drive using digital PWM speed controller is analysed through Lyapunov method. A novel condition is presented to calculate the ideal duty cycle of PWM signal based on the motor parameters to keep the BLDC motor stable at the reference speed. Correctness of the introduced stability analysis condition is investigated through simulation and experiment. Experimental and simulation results validate the introduced stability analysis condition for PWM speed controller of the BLDC motor. Fault tolerant control systems effectively improve performance of the electric motors under fault condition. Performance of the electric motors in post-fault condition in applications such as drive train of electric vehicles where safety is the most concern is critical. Fault tolerant control systems of the BLDC motors used in EV application are discussed in Chapter 6. A four in-wheel drive EV using BLDC motors are modelled in Simulink dynamic parameters of the modelled EV is studied under inverter open circuit switches faults in one of the front in-wheel BLDC motors. Results show immediate instability of the electric vehicle after faults occurrence. Results of this study show that using fault tolerant control systems for the in-wheel motors is inevitable. Performance of the BLDC motor is studied under various inverter switches and position sensors faults through a validated simulation model. Two fault diagnosis systems are proposed for inverter open circuit switch faults and Hall Effect position sensors failure based on DFT analysis of line voltages of the BLDC motor. Multidimensional knowledge based tables are introduced to diagnose the faulty switch of inverter and the faulty position sensor through spectral energy density errors of the motor line voltages. One of the advantages of these fault 139

161 7. Conclusion diagnosis systems is that the exact line voltages pattern of the BLDC motor for various reference speed or torque loads is not needed. Fault tolerant inverters with a redundant leg are recommended to isolate and rectify the inverter switch faults of the in-wheel BLDC motor drives. A simple and reliable method, same as the method introduced for sensorless drive of the BLDC motor in Chapter 5, is also recommended to generate the signal of faulty sensor by implementing the time delays from the other healthy sensors. Inverter open circuit switch faults and position sensor faults are tested on a low voltage BLDC motor through experiment. The proposed fault diagnosis algorithms and the knowledge based tables are validated through the experimental results too. 140

162 References [1] K. Bergsson, Hybrid vehicle history more than a century of evolution and refinement, accessed 13 July [2] K. Rajashekara, History of electric vehicles in general motors, IEEE Transactions on Industrial Application, vol. 30, pp , August [3] A. Tashakori, M. Ektesabi, and N. Hosseinzadeh, Characteristics of suitable drive train for electric vehicle, in Proceeding of the International Conference on Instrumentation, Measurement, Circuits and Systems (ICIMCS 2011), (Hong Kong,China.), pp , Dec , 8, 10, 13, 14, 24, 25 [4] F. A. Barata, J. C. Quadrado, and J. F. Silva, Brushless dc motor: position linear control simulation, in Proceedings of the 9th WSEAS International Conference on Systems, ICS 05, (Stevens Point, Wisconsin, USA), pp. 48:1 48:6, World Scientific and Engineering Academy and Society (WSEAS), , 8 [5] A. Tashakori and M. Ektesabi, Comparison of different pwm switching modes of bldc motor as drive train of electric vehicles, World Academy of Science, Engineering and Technology, vol. 67, pp , , 8, 10, 12, 20, 21, 23, 24, 25, 26, 28, 29, 75, 97, 159, 160, 162, 163, 165 [6] E. C. Lee, Application of brushless dc drives in blow molding, Published by PowerTech Industrial Motors

163 REFERENCES [7] T.-H. Kim and M. Ehsani, Sensorless control of the bldc motors from nearzero to high speeds, IEEE Transactions on Power Electronics, vol. 19, no. 6, pp , , 4, 50, 66, 67, 68, 69 [8] A. Tashakori and M. Ektesabi, Direct torque controlled drive train for electric vehicle, in Proceeding of the World Congress on Engineering 2012 (WCE 2012), vol. Vol. 2, (London, UK.), p , July Lecture notes in engineering and computer science. 3, 51, 52, 54 [9] S. Park, H. Park, M. Lee, and F. Harashima, A new approach for minimumtorque-ripple maximum-efficiency control of bldc motor, IEEE Transactions on Industrial Electronics, vol. 47, no. 1, pp , , 51 [10] J. Gieras, Analytical approach to cogging torque calculation of pm brushless motors, IEEE Transactions on Industry Applications, vol. 40, no. 5, pp , [11] P. Damodharan and K. Vasudevan, Sensorless brushless dc motor drive based on the zero-crossing detection of back electromotive force (emf) from the line voltage difference, IEEE Transactions on Energy Conversion, vol. 25, no. 3, pp , , 66, 67, 68, 69 [12] A. Tashakori and M. Ektesabi, A simple fault tolerant control system for hall effect sensors failure of bldc motor, in Proceeding of the 8th IEEE Conference on Industrial Electronics and Applications (ICIEA 2013), (Melbourne, VIC), pp , , 16, 96, 97, 110, 114, 121, 122, 123, 124, 126, 128, 135 [13] X.-Q. Liu, H.-Y. Zhang, J. Liu, and J. Yang, Fault detection and diagnosis of permanent-magnet dc motor based on parameter estimation and neural network, IEEE Transactions on Industrial Electronics, vol. 47, no. 5, pp , , 99 [14] B. Lee and M. Ehsani, Advanced bldc motor drive for low cost and high performance propulsion system in electric and hybrid vehicle, in IEEE International Electric Machines and Drives Conference, (IEMDC), ,

164 REFERENCES [15] C. Lungoci, M. Georgescu, and M. Calin, Electrical motor types for vehicle propulsion, in Proceedings of the International Conference on Optimisation of Electrical and Electronic Equipment, OPTIM, 2012, pp , [16] R. Hoseinnezhad and A. Bab-Hadiashar, Missing data compensation for safety-critical components in a drive-by-wire system, IEEE Transactions on Vehicular Technology, vol. 54, no. 4, pp , [17] J. Fuller, How drive by wire technology works, accessed 15 Feb [18] M. Ektesabi, Design of integral motor and its embedded system, in Proceeding of the 8th International Power Engineering Conference, IPEC 2007, pp , , 20 [19] X. Xue, K. W. E. Cheng, and N. C. Cheung, Selection of electric motor drives for electric vehicles, in Australian Universities Power Engineering Conference, AUPEC 08., , 20 [20] H. Tischmacher, I. Tsoumas, B. Eichinger, and U. Werner, Case studies of acoustic noise emission from inverter-fed asynchronous machines, IEEE Transactions on Industry Applications, vol. 47, no. 5, pp , cited By (since 1996)3. 12 [21] N. Mutoh, M. Nakanishi, M. Kanesaki, and J. Nakashima, Emi noise control methods suitable for electric vehicle drive systems, IEEE Transactions on Electromagnetic Compatibility, vol. 47, no. 4, pp , cited By (since 1996) [22] N. Mutoh and M. Kanesaki, A suitable method for ecovehicles to control surge voltage occurring at motor terminals connected to pwm inverters and to control induced emi noise, IEEE Transactions on Vehicular Technology, vol. 57, no. 4, pp , cited By (since 1996)

165 REFERENCES [23] M. Ehsani, Y. Gao, and S. Gay, Characterization of electric motor drives for traction applications, in IECON Proceedings (Industrial Electronics Conference), 2003, vol. 1, (Roanoke, VA), pp , , 13, 15, 20 [24] A. Boglietti, P. Ferraris, M. Lazzari, and F. Profumo, New design criteria for spindles induction motors controlled by field oriented technique, Electric Machines and Power Systems, vol. 21, no. 2, pp , [25] I. Tsoumas and H. Tischmacher, Influence of the inverter s modulation technique on the audible noise of electric motors, in In proceedings of the 20th International Conference on Electrical Machines, ICEM 2012, (Marseille), pp , [26] M. T. DiRenzo, Switched reluctance motor control basic operation and example using the tms320f240, tech. rep., Texas Instruments, [27] T. J. E. Miller, Electronic control of switched reluctance machines. Reed Educational and Professional Publishing Ltd, [28] O. Ichinokura, T. Onda, M. Kimura, T. Watanabe, T. Yanada, and H. Guo, Analysis of dynamic characteristics of switched reluctance motor based on spice, IEEE Transactions on Magnetics, vol. 34, pp , [29] A. Omekanda, S. Gopalakrishnan, and H. Klode, Acoustic noise of switched reluctance and permanent magnet motors: A comparison in the context of electric brakes, (New Orleans, LA), pp , Conference Record - IAS Annual Meeting (IEEE Industry Applications Society). 15, 18 [30] S. Gay, H. Gao, and M. Ehsani, Fuel cell hybrid drive train configurations and motor drive selection, in IEEE Vehicular Technology Conference, 2002, vol. 56, pp , , 20 [31] M. Jain and S. Williamson, Suitability analysis of in-wheel motor direct drives for electric and hybrid electric vehicles, in Proceeding of the 2009 IEEE Electrical Power and Energy Conference, EPEC 2009, , 20,

166 REFERENCES [32] P. Yedamale, Brushless dc (bldc) motor fundamentals, tech. rep., Microchip Technology Inc., , 18, 66 [33] A. Tashakori, M. Ektesabi, and N. Hosseinzadeh, Modeling of bldc motor with ideal back-emf for automotive applications, in Proceeding of the World Congress on Engineering 2011 (WCE 2011), vol. Vol. 2, (London), pp , July Lecturer notes in engineering and computer science. 16, 34, 36, 37, 38 [34] A. Tashakori and M. Ektesabi, Stability analysis of sensorless bldc motor drive using digital pwm technique for electric vehicles, in Proceeding of 38th Annual Conference on IEEE Industrial Electronics Society, IECON 2012, pp , October , 67, 69, 70, 72, 73, 74, 75, 76, 78, 79, 81, 83, 84, 86, 88, 134 [35] M. Zeraoulia, M. Benbouzid, and D. Diallo, Electric motor drive selection issues for hev propulsion systems: A comparative study, IEEE Transactions on Vehicular Technology, vol. 55, no. 6, pp , , 20, 21 [36] Y. Jeon, H. Mok, G. Choe, D. Kim, and J. Ryu, A new simulation model of bldc motor with real back emf waveform, in Proceeeding of the IEEE Conference On Computer and Power Electronics, COMPEL 2000, pp , [37] U. Vinatha, S. Pola, and K. Vittal, Simulation of four quadrant operation & speed control of bldc motor on matlab / simulink, in Proceeding of the IEEE Region 10 Annual International Conference, TENCON,, [38] B. K. Bose, Modern Power Electronics and AC Drives. Prentice Hall, ISBN [39] K.-H. Kim and M.-J. Youn, Performance comparison of pwm inverter and variable dc link inverter schemes for high-speed sensorless control of bldc motor, Electronics Letters, vol. 38, no. 21, pp , [40] A. Tashakori and M. Ektesabi, Direct torque control of in-wheel bldc motor used in electric vehicle, in IAENG Transactions on Engineering Technolo- 145

167 REFERENCES gies (G.-C. Yang, S.-l. Ao, and L. Gelman, eds.), vol. 229 of Lecture Notes in Electrical Engineering, pp , Springer Netherlands, , 52, 54, 56, 57, 58, 59, 66 [41] F. L. Luo and H. G. Yeo, Advanced pm brushless dc motor control and system for electric vehicles, in Conference Record - IAS Annual Meeting (IEEE Industry Applications Society),, vol. 2, pp , [42] F. Rodriguez and A. Emadi, A novel digital control technique for brushless dc motor drives: Steady state and dynamics, in IECON Proceedings (Industrial Electronics Conference), 2006,, pp , , 51 [43] P. Vas, Sensorless Vector and Direct Torque Control. Oxford University Press, , 53 [44] I. Takahashi and T. Noguchi, A new quick-response and highefficiency control strategies of an induction motor, IEEE Transactions on Industrial Application, vol. 22, pp , [45] M. Depenbrock, Direct self-control (dsc) of inverter-fed induction machine, IEEE Transactions on Power Electronics, vol. 3, no. 4, pp , [46] S. Ozturk and H. Toliyat, Direct torque control of brushless dc motor with non-sinusoidal back-emf, in Proceeding of the IEEE International Electric Machines and Drives Conference, IEMDC 2007,, vol. 1, pp , [47] S. Ozturk and H. Toliyat, Sensorless direct torque and indirect flux control of brushless dc motor with non-sinusoidal back-emf, in Proceeding of the 34th Annual Conference of the IEEE Industrial Electronics Society, IECON 2008,, pp , [48] J. Yang, Y. Hu, W. Huang, J. Chu, and J. Gao, Direct torque control of brushless dc motor without flux linkage observation, in Proceeding of the 2009 IEEE 6th International Power Electronics and Motion Control Conference, IPEMC 09,, pp , ,

168 REFERENCES [49] W.-S. Yan, H. Lin, H. Li, and Y. Wei, Sensorless direct torque controlled drive of brushless dc motor based on fuzzy logic, in Proceeding of the th IEEE Conference on Industrial Electronics and Applications, ICIEA 2009,, pp , , 53 [50] A. Gupta, T. Kim, T. Park, and C. Lee, Intelligent direct torque control of brushless dc motors for hybrid electric vehicles, in Proceeding of the 5th IEEE Vehicle Power and Propulsion Conference, VPPC 09,, pp , , 53, 58 [51] S. Ozturk and H. Toliyat, Direct torque and indirect flux control of brushless dc motor, IEEE/ASME Transactions on Mechatronics, vol. 16, no. 2, pp , , 53 [52] J. Hu and B. b. b. Wu, New integration algorithms for estimating motor flux over a wide speed range, IEEE Transactions on Power Electronics, vol. 13, no. 5, pp , [53] L. Zhong, M. Rahman, W. Hu, and K. Lim, Analysis of direct torque control in permanent magnet synchronous motor drives, IEEE Transactions on Power Electronics, vol. 12, no. 3, pp , [54] G.-J. Su and J. McKeever, Low-cost sensorless control of brushless dc motors with improved speed range, IEEE Transactions on Power Electronics, vol. 19, no. 2, pp , [55] H.-C. Chen, T.-Y. Tsai, and C.-K. b. Huang, Low-speed performance comparisons of back-emf detection circuits with position-dependent load torque, IET Electric Power Applications, vol. 3, no. 2, pp , [56] Q. Jiang, C. b. Bi, and R. Huang, A new phase-delay-free method to detect back emf zero-crossing points for sensorless control of spindle motors, IEEE Transactions on Magnetics, vol. 41, no. 7, pp , , 68 [57] J. Gamazo-Real, E. Vzquez-Snchez, and J. Gmez-Gil, Position and speed control of brushless dc motors using sensorless techniques and application trends, Sensors, vol. 10, no. 7, pp , , 69,

169 REFERENCES [58] J. Shao, D. Nolan, M. Teissier, and D. Swanson, A novel microcontrollerbased sensorless brushless dc (bldc) motor drive for automotive fuel pumps, IEEE Transactions on Industry Applications, vol. 39, no. 6, pp , , 68, 69 [59] J. Shao, D. Nolan, and T. Hopkins, A novel direct back-emf detection for sensorless brushless dc (bldc) motor drives, in Annual IEEE Applied Power Electronics Conference, vol. 1, pp , March [60] K. C. L. J. Kim, T., A new sensorless drive scheme for a bldc motor based on the terminal voltage difference, in Proceeding of the 37th Annual Conference of the IEEE Industrial Electronics Society, IECON 2011, (Melbourne, VIC), pp , Nov Conference of 37th Annual Conference of the IEEE Industrial Electronics Society, IECON , 68 [61] F. Rodriguez and A. Emadi, A novel digital control technique for brushless dc motor drives, IEEE Transactions on Industrial Electronics, vol. 54, no. 5, pp , , 75 [62] J. Shen, Z. Zhu, and D. Howe, Practical issues in sensorless control of pm brushless machines using third-harmonic back-emf, vol. 2, pp , [63] J. Shen and S. Iwasaki, Sensorless control of ultrahigh-speed pm brushless motor using pll and third harmonic back emf, IEEE Transactions on Industrial Electronics, vol. 53, no. 2, pp , [64] Y.-S. Lai and Y.-K. Lin, Novel back-emf detection technique of brushless dc motor drives for wide range control without using current and position sensors, IEEE Transactions on Power Electronics, vol. 23, no. 2, pp , , 73 [65] P. Damodharan, R. Sandeep, and K. Vasudevan, Simple position sensorless starting method for brushless dc motor, IET Electric Power Applications, vol. 2, no. 1, pp ,

170 REFERENCES [66] K. Iizuka, H. Uzuhashi, M. Kano, T. Endo, and K. Mohri, Microcomputer control for sensorless brushless motor, IEEE Transactions on Industry Applications, vol. IA-21, no. 3, pp , , 79 [67] L. Mingyao, Z. Zhiyao, and L. Keman, A novel and easy-realizing initial rotor position detection method and speedup algorithm for sensorless bldc motor drives, pp , [68] A. Ungurean, V. Coroban-Schramel, and I. Boldea, Sensorless control of a bldc pm motor based on i-f starting and back-emf zero-crossing detection, pp , [69] N. Milivojevic, M. Krishnamurthy, Y. Gurkaynak, A. Sathyan, Y.-J. Lee, and A. Emadi, Stability analysis of fpga-based control of brushless dc motors and generators using digital pwm technique, IEEE Transactions on Industrial Electronics, vol. 59, no. 1, pp , , 76, 77 [70] A. b. Sathyan, N. Milivojevic, Y.-J. Lee, M. Krishnamurthy, and A. Emadi, An fpga-based novel digital pwm control scheme for bldc motor drives, IEEE Transactions on Industrial Electronics, vol. 56, no. 8, pp , [71] A. Tashakori and M. Ektesabi, Fault diagnosis technique for vsi drive of bldc motors in ev application. IEEE Trans. on Power Electronics. (Submitted and under revision), , 97, 99, 100, 101, 103, 104, 106, 107, 111, 112, 113, 114, 115 [72] A. Tashakori and M. Ektesabi, Position sensors fault tolerant control system in bldc motors. Engineering Letters Journal. (Submitted and under revision), , 122, 123, 125, 126, 127, 128, 129, 130 [73] A. Tashakori and M. Ektesabi, Fault diagnosis of in-wheel bldc motor drive for electric vehicle application, in In proceeding of the 2013 IEEE Intelligent Vehicles Symposium, (Gold Coast, Australia), pp , June , 99, 100, 101, 119,

171 REFERENCES [74] O. Moseler and R. Isermann, Application of model-based fault detection to a brushless dc motor, IEEE Transactions on Industrial Electronics, vol. 47, no. 5, pp , [75] M. Awadallah and M. Morcos, Automatic diagnosis and location of openswitch fault in brushless dc motor drives using wavelets and neuro-fuzzy systems, IEEE Transactions on Energy Conversion, vol. 21, no. 1, pp , [76] B.-G. Park, K.-J. Lee, R.-Y. Kim, T.-S. Kim, J.-S. Ryu, and D.-S. Hyun, Simple fault diagnosis based on operating characteristic of brushless directcurrent motor drives, IEEE Transactions on Industrial Electronics, vol. 58, no. 5, pp , [77] R. De Araujo Ribeiro, C. Jacobina, E. Da Silva, and A. Lima, Fault detection of open-switch damage in voltage-fed pwm motor drive systems, IEEE Transactions on Power Electronics, vol. 18, no. 2, pp , [78] O.-S. Yu, N.-J. Park, and D.-S. Hyun, A novel fault detection scheme for voltage fed pwm inverter, in Conference of IECON nd Annual Conference on IEEE Industrial Electronics, (Paris), pp , 6-10 November [79] M. Abul Masrur, Z. Chen, and Y. Murphey, Intelligent diagnosis of open and short circuit faults in electric drive inverters for real-time applications, IET Power Electronics, vol. 3, no. 2, pp , [80] K. Butler, M. Ehsani, and P. Kamath, A matlab-based modeling and simulation package for electric and hybrid electric vehicle design, IEEE Transactions on Vehicular Technology, vol. 48, no. 6, pp , [81] D. W. Gao, C. Mi, and A. Emadi, Modeling and simulation of electric and hybrid vehicles, Proceedings of the IEEE, vol. 95, Issue 4, pp , [82] B.-K. Lee, T.-H. Kim, and M. Ehsani, On the feasibility of four-switch three-phase bldc motor drives for low cost commercial applications: Topol- 150

172 REFERENCES ogy and control, IEEE Transactions on Power Electronics, vol. 18, no. 1, pp , [83] R. Errabelli and P. Mutschler, Fault-tolerant voltage source inverter for permanent magnet drives, IEEE Transactions on Power Electronics, vol. 27, no. 2, pp , [84] Y.-S. Jeong, S.-K. Sul, S. Schulz, and N. Patel, Fault detection and faulttolerant control of interior permanent-magnet motor drive system for electric vehicle, IEEE Transactions on Industry Applications, vol. 41, no. 1, pp , [85] N. B. Samoylenko, Q. C. Han, and J. Jatskevich, Dynamic performance of brushless dc motors with unbalanced hall sensors, IEEE Transactions on Energy Conversion, vol. 23, no. 3, pp , [86] E. Balaban, A. Saxena, P. Bansal, K. Goebel, and S. Curran, Modeling, detection, and disambiguation of sensor faults for aerospace applications, IEEE Sensors Journal, vol. 9, no. 12, pp , [87] S. b. Morimoto, K. c. Kawamoto, M. b. Sanada, and Y. b. Takeda, Sensorless control strategy for salient-pole pmsm based on extended emf in rotating reference frame, IEEE Transactions on Industry Applications, vol. 38, no. 4, pp , [88] L. Wang, J. Liu, and X. Wu, Fault analysis on driving motors of lunar rover wheels, in Proceeding of the International Conference on Electrical Machines and Systems, ICEMS 2011, (Beijing, China), Aug [89] Picdem mc lv development board users guide, tech. rep., Microchip, [90] A. M. Liapunov, The general problem of stability of motion. PhD thesis, University of Kharkov,

173 Appendix A A Reference Links of Table 2.1 Reference links of Table 2.1 Drive train specification of the electric vehicles available in the world market, on page 11 are given in order of vehicle numbers as below. All links are accessed on 10 August

174 Appendix A B Details of the motor models in Chapter 2 All motors are simulated using machine blocks of the SimPowerSystems library in Matlab/Simulink. Block diagram of the motor drive models are shown below, Figure B1: Block diagram of the induction motor drive model 153

175 Appendix A Figure B2: Block diagram of the DC motor drive model Figure B3: Block diagram of the switched reluctance motor drive model 154

176 Appendix A Figure B4: Block diagram of the BLDC motor drive model C State Space Equation of BLDC Motor Phase voltage difference state space equations of BLDC motor can be derived as below from equations (3.1), (3.2), (3.3), (3.7) on page 35. V ab = R(i a i b ) + (L M) d dt (i a i b ) + E ab (C1) V bc = R(i b i c ) + (L M) d dt (i b i c ) + E bc (C2) where i a + i b + i c = 0, therefore by neglecting mutual inductance and rearranging equations (C1) and (C2), di a dt = R L i a + 2 3L (V ab E ab ) + 1 3L (V bc E bc ) di b dt = R L i b + 2 3L (V ab E ab ) + 1 3L (V bc E bc ) (C3) (C4) 155

177 Appendix A Then state space equations of BLDC motor are: i ȧ i ḃ ω m = R 0 0 L 0 R 0 L 0 0 R L i a i b i c ω m i a i b ω m + 2 3L 1 3L 1 3L 0 1 3L J = i a i b ω m V ab Eab V bc Ebc T e T l (C5) (C6) D Clarke Transformation Clarke transformation is a mathematical transformation employed to simplify the analysis of three-phase circuits. i α i β i γ = i a i b i c (D1) In a balanced system where i a + i b + i c = 0, i γ = 0. Therefore two of currents are enough to calculate α and β components. The transform simplifies to, [ i α i β ] = 2 3 [ ] [ i a i b ] (D2) E Lyapunov s Second Method for Stability Alexandr Mikhailovich Liapunov, in his PhD thesis titled as The general problem of stability of motion in 1982, proposed two methods for stability analysis [98]. The second stability analysis method, that is universally used nowadays, is introduced as follow: 156

178 Appendix A For a system having a point of equilibrium at x = 0, consider a Lyapunov candidate function V (x) which has an analogy to the potential function of classical dynamics. Therefore considering the function V (x) : R n R, System is stable in the sense of Lyapunov, if V (x) 0 if and only if x = 0 (positive definite), then V (x) = and only if x = 0 (negative definite). dv (x) dt 0 if F Comaprison of Different PWM Switching Techniques of The BLDC Motor A BLDC motor drive using a digital PWM speed controller is modelled in Simulink. A PI controller dynamically chooses the duty cycle of the PWM signal based on the speed error of the motor. Specifications of the BLDC motor used in simulation model are given in Table 3.3 on the page 46. In this section the BLDC motor drive performance for various PWM switching modes is compared under healthy and inverter switch faults conditions. To distinguish various PWM switching modes numbers are assigned to them as below, 1. Mode one : PWM signal is applied to upper side switches the inverter; 2. Mode Two : PWM signal is applied to lower side switches of the inverter; 3. Mode Three : PWM signal is applied to all of the inverter switches. F.1 Normal Condition BLDC motor model is tested for 1500 RPM reference speed and 10 N.m load torque under no fault condition. Speed characteristics of the BLDC motor for various PWM switching modes are shown in Figure F1 [5]. As can be seen speed response of the BLDC motor for switching mode three has the lowest peak overshoot where the highest one belongs to the mode one. High peak overshoot speed response is not suitable for the in-wheel motors. Speed oscillation of the 157

179 Appendix A BLDC motor around the reference speed of the controller is almost same for all switching modes in the steady state condition. Figure F1: Speed responses of BLDC motor for different PWM switching modes Torque characteristics of the BLDC motor for various PWM switching modes are shown in Figure F3 [5]. The BLDC motor produces same initial peak torque for all switching modes. Torque ripples amplitude of the BLDC motor for mode three is higher than other modes in the steady state situation. High amplitude torque ripples is not desirable for the in-wheel motors. Phase A terminal voltage of the BLDC motor for various PWM switching modes are shown in Figure F4 [5]. Terminal voltage of the motor is measured with respect to the negative terminal of the DC link of VSI. As is shown the DC voltage of the inverter is chopped by PWM signal during the upper switch conduction period and is zero during conduction of the lower switch for switching mode one. As can be seen the conducting condition of switches in the mode Two is exactly opposite of the mode one. DC voltage of the inverter is chopped by PWM signal during both conduction periods of the upper and lower switches in the mode Three. The line voltage pattern of the BLDC motor is a good signature to recognise the applied PWM switching mode of the speed controller. 158

180 Appendix A Figure F2: Torque responses of BLDC motor for different PWM switching modes Figure F3: Torque responses of BLDC motor for different PWM switching modes 159

181 Appendix A Figure F4: Line voltage of BLDC motor for different PWM switching modes Duty cycle values chosen by the PI controller for various PWM switching modes are shown in Figure F5 [5]. As can be seen the PI controller is chosen various duty cycle values for each PWM switching mode. Duty cycle changes limit in the PWM switching mode three is smaller compared to the other two modes in the steady state condition. F.2 Critical Condition Critical condition is considered as mechanical shocks and inverter switch faults. Mechanical faults are implemented as 30% sudden changes of load torque of the BLDC motor. The speed control algorithm is same and the only difference is that various switches of inverter are chosen for PWM switching in different modes. In this section behaviour of the BLDC motor for open and short circuit faults of the upper side switch of phase A in inverter is analysed. 160

182 Appendix A Figure F5: Duty cycle chosen by PI controller for different PWM switching modes F.2.1 Mechanical Shocks Thirty percent changes of the load torque are applied at t = 0.4 s (increase of the load), t = 0.5 s (decrease of the load) and t = 0.6 s (decrease of the load) while the BLDC motor is running at 1500 RPM under initial 10 N.m load toque. Digital PWM speed controller keeps the BLDC motor stable during mechanical shocks for all PWM switching modes. Torque characteristics of the BLDC motor under mechanical shocks for various PWM switching modes are shown in Figure F6 [5]. Torque ripples amplitude of the BLDC motor in the PWM switching mode three is almost same as other two methods for the higher load torque, however at the low load torque it is remarkable more than the other two modes. Duty cycle values chosen by the PI controller under mechanical shocks for various PWM switching modes are shown in Figure F7 [5]. Changes limit of duty cycles chosen by the controller for the switching modes one and two are much more than the mode three under higher load torque and it is vice versa under the lower load torque. Since the in-wheel motors should perform under high load 161

183 Appendix A Figure F6: Torque responses of the BLDC motor under mechanical shocks for different PWM switching modes Figure F7: Duty cycle chosen by PI controller under mechanical shocks for different PWM switching modes torque, therefore behaviour of the BLDC motor and its speed controller are more robust during mechanical shocks for the PWM switching mode three. 162

184 Appendix A F.2.2 Inverter Switch Faults Open and short circuit faults of the upper side switch of phase A of the inverter is applied to the BLDC motor at t = 0.2 s while the motor is running at 1500 RPM under 10 N.m load torque. Speed responses of the BLDC motor for various PWM switching modes under inverter switch faults are shown in Figure F8 [5]. Speed response of the motor are almost similar for all the PWM switching modes during open circuit fault. The BLDC motor is lost the operating point and its speed starts oscillating. Speed of the BLDC motor for the PWM switching mode one during short circuit fault is much higher that the other two modes. Duty cycle values chosen by the PI controller for various PWM switching modes under inverter switch faults are shown in Figure F9 [5]. Duty cycle values of the mode two and three are constant 100% during open circuit fault where in the mode one it toggle between 0 and 100%. The PI controller for the PWM switching mode one does not work during short circuit fault and the duty cycle value is zero where in the other two modes duty cycle values changes from 0 to 100%. Although the BLDC motor is not stable during the inverter switch faults for all the PWM switching modes but speed controller in the PWM switching mode three shows more robust response during short circuit switch fault. G EV Model Simulation Results under Inverter Open Circuit Switch Fault The four wheel drive electric vehicle model is run from stall position up to 2000 RPM reference speed of the in-wheel BLDC motors. Open circuit fault switch S 2 (refer to Figure 6.1 on page 95) of the inverter is applied to the inwheel BLDC motor A at t = 40 s while the EV speed is reached to 110 km/h. Simulation results and output characteristics of the electric vehicle and in-wheel motors under inverter open circuit fault condition are shown in Figures G1, G2, G3 and G4. 163

185 Appendix A Figure F8: Speed responses of the BLDC motor under inverter switch faults for different PWM switching modes 164

186 Appendix A Figure F9: Duty cycle chosen by PI controller under inverter switch faults for different PWM switching modes 165

187 Appendix A Figure G1: EV speed characteristics under open circuit fault of switch S 2 Figure G2: Normal tire forces under open circuit fault of switch S 2 166

188 Appendix A Figure G3: Torque characteristics of the BLDC motors under open circuit fault of switch S 2 Figure G4: Speed characteristics of the BLDC motors under open circuit fault of switch S 2 167

189 Appendix B 168

190 1999 EV AMERICA TECHNICAL SPECIFICATIONS Effective October 1, 1999 Prepared by ElectricTransportationApplications

191 1999 EV AMERICA TECHNICAL SPECIFICATIONS MINIMUM VEHICLE REQUIREMENTS For a vehicle to be considered qualified as an EV America-USDOE Production level vehicle, it must meet the minimum criteria defined by shall terminology utilized in the Specification. [For clarity, the use of the word Shall defines minimum requirements, whereas the use of the word Should defines design and performance objectives.] Vehicles which cannot meet all of the Shall requirements will be considered Prototypes, and will not be considered as having passed EV America. The following requirements shall be met by any vehicle before it can receive EV America Production level status: (1) Vehicles shall have a minimum payload of 400 pounds. (2) For Conversion vehicles, OEM GVWR shall not be increased. Suppliers shall provide the OEMs Gross Vehicle Weight Rating (GVWR). (3) For conversion vehicles, OEM Gross Vehicle Axle Weight Ratings (GAWR) shall not be increased. Suppliers shall provide axle weights for the vehicle as delivered, and at full rated payload. (4) Seating capacity shall be a minimum of 2, (one driver and at least one passenger). Suppliers shall provide seating capacity (available seat belt positions) for their vehicle. (5) Suppliers shall provide information on their selected battery manufacturer s recycling plan, including how it has been implemented. (6) For conversion vehicles, the OEM passenger space shall not be intruded upon by the battery, battery box or other conversion materials. (7) Vehicles shall comply with the requirements of 49 CFR S5.2.1, or alternatively, 49 CFR S5.2.2 for parking mechanisms. (8) Vehicles shall have a minimum range between charges of at least 50 miles when loaded with two 166-pound occupants and operated at a constant 45 mph. (9) Vehicles shall comply with Federal Motor Vehicle Safety Standards applicable on the date of manufacture and such compliance shall be certified by the manufacturer in accordance with 49 CFR 567. Suppliers shall provide a completed copy of Appendix B with their submittal, indicating the method of compliance with each section of 49 CFR 571. If certification includes exemption, the exemption number issued by the National Highway Transportation Safety Administration (NHTSA), the date of it s publication in the Federal Register and the page number(s) of the Federal Register acknowledging issuance of the exemption shall be provided along with Appendix B. Only exemptions for nonapplicable requirements shall be allowed. 2

192 1999 EV AMERICA TECHNICAL SPECIFICATIONS (10) Batteries and/or battery enclosures shall be designed and constructed in accordance with the requirements of SAE J1766 FEB96. Further, batteries and electrolyte will not intrude into the passenger compartment during or following FMVSS frontal barrier, rear barrier and side impact collisions, and roll-over requirements of 49 CFR Suppliers shall provide verification of conformance to this requirement. (11) Batteries shall comply with the requirements of SAE J1718 APR97, and at a minimum shall meet the requirements of NEC 625 for charging in enclosed spaces without a vent fan. (12) Concentrations of explosive gases shall not be allowed to exceed 25% of the LEL (Lower Explosive Limit) in the battery enclosure. Suppliers shall describe how battery boxes will be vented, to ensure any battery gases escape safely to atmosphere during and following normal or abnormal charging and operation of the vehicle. (13) The battery charger shall be capable of recharging the main propulsion battery to a state of full charge from any possible state of discharge in less than 12 hours, at temperatures noted in Section 5.6. (14) Chargers shall have the capability of accepting input voltages of 208V and 240V single phase 60 Hertz alternating current service, with a tolerance of ±10% of rated voltage. Charger input current shall be compatible with the requirements for Level II chargers, and shall comply with the requirements of SAE J1772 OCT96 and/or SAE J1773 JAN95. Personnel protection systems shall be in accordance with the requirements of UL Standard 2202, Published (15) Chargers shall have a true power factor of.95 or greater and a harmonic distortion rated at 20% (current at rated load). (16) The charger shall be fully automatic, determining when end of charge conditions are met and transitioning into a mode that maintains the main propulsion battery at a full state of charge while not overcharging it, if continuously left on charge. (17) Vehicles shall not contain exposed conductors, terminals, contact blocks or devices of any type that create the potential for personnel to be exposed to 50 volts or greater (the distinction between low-voltage and high voltage, as specified in SAE J1127 JAN95, J1128 JAN95, et al.). (18) Vehicles being tested shall be accompanied by non-proprietary manuals for parts, service, operation and maintenance, interconnection wiring diagrams and schematics, (with pricing for optional manuals). These documents shall either be provided or available to the end user. 3

193 1999 EV AMERICA TECHNICAL SPECIFICATIONS (19) The vehicle shall include a state of charge indicator for the main propulsion batteries. 20) Propulsion power shall be isolated from the vehicle chassis such that battery leakage current is less than 0.5 MIU in accordance with UL Standard 2202, Published (21) Charging circuits shall be isolated from the vehicle chassis such that ground current from the grounded chassis at any time while the vehicle is on charge or the charger is connected to an off-board power supply does not exceed 5 ma, in accordance with UL Standards 2202, Published (22) Replacement tires shall be commercially available to the end user in sufficient quantities to support the purchaser s needs, (23) The vehicle shall prevented from being driven with the key turned on and the drive selector in the DRIVE or REVERSE position while the vehicle s charge cord is attached. Additionally, the following interlocks shall be present: The controller shall not initially energize to move the vehicle with the gear selector in any position other than PARK or NEUTRAL; The start key shall be removable only when the ignition switch is in the Off position, with the drive selector in PARK; With a pre-existing accelerator input, the controller shall not energize or excite such that the vehicle can move under its own power from this condition. (24) All vehicles shall comply with the FCC requirements for unintentional emitted electromagnetic radiation, as identified in 47 CFR 15, Subpart B, Unintentional Radiators. (25) Failure of a battery or battery pack shall be determined through a discharge test. The discharge test shall be performed with the discharge current regulated to achieve a C/1 discharge rate based on the ampere hour capacity of the battery specified by the Supplier as required in Section 6.1 and with a battery temperature of at least 77º F. Subsequent to receiving a full charge and equalization, the battery shall be discharged at such current and temperature until the terminal voltage of any cell in the battery drops below the voltage specified by the Supplier as required in Section 6.3. The ampere hours delivered by the battery to that point shall be calculated and shall become the actual battery capacity. Failure of the battery shall be deemed to have occurred if the actual battery capacity is not at least 80% of the nominal ampere hour capacity specified by the Supplier as required by Section 6.1. (26) Vehicles shall be equipped with an automatic disconnect for the main propulsion batteries. They shall also have a manual service disconnect. These disconnects shall be clearly labeled. [See Section 7.3] 4

194 1999 EV AMERICA TECHNICAL SPECIFICATIONS (27) Any conductive or inductive type charging systems shall be compatible with the Personnel Protection requirements of SAE J1772 or J1773, as appropriate. (28) Suppliers shall provide Material Safety Data Sheets (MSDS) for all batteries. (29) Suppliers shall indicate the level of charge below which the batteries should not be discharged and how the controller automatically limits battery discharge below this level. (30) Suppliers shall verify that the method(s) of charging the propulsion batteries and the charging algorithm have been reviewed and approved by the battery manufacturer. (31) Regardless of the charger type used, the charger shall be capable of meeting the requirements of Section 625 of the National Electric Code (NEC). (32) If the vehicle is equipped with fuel fired heaters, the vehicle shall comply with the requirements of 49 CFR (33) The vehicle shall have an on-board Battery Management System (BMS). The following sections constitute the Technical Requirements of the Specification. Information has been categorized according to component and/or function. These sections provide an overview of the requirements and recommendations for Suppliers to use. This Technical Specification establishes the minimum requirements for Production level electric vehicles, as well as identifying design and performance objectives. Suppliers shall clearly describe the vehicle they are proposing by completing a copy of Appendix A. Drawings should be provided showing the installation, location and layout of the conversion components including the batteries, motor and controller, and powered accessories. The drive line should also be described, i.e., direct drive transmission, reduction gear ratio, etc.. Suppliers should include any other information required to describe the vehicle. No inference should be drawn by Suppliers or any other person that the measures listed in this specification are sufficient to make the vehicle safe, and each Supplier shall acknowledge in writing that 1) it is solely responsible for determining whether each vehicle offered for sale is safe, and 2) it is not relying on EV America, Electric Vehicle Market Development Group (EVMDG), the Procurement Management Board (PMB), or any of the EV America participants, their Consultants, or the U.S. Government as having, by this specification and its requirements, established minimally sufficient safety standards. This written statement shall be provided in the Supplier s proposal. 5

195 BLK42 Series - Brushless DC Motors FEATURES NEMA Size 42 BLDC Motors IP65 Rating Complete Protection from Dust Particles Can be Subjected to Wet Enviornments Long Life and Highly Reliable Cost Effective Replacement for Brush DC Motors Hall Sensor Feedback Available in Three Different Stack Lengths SPECIFICATIONS DESCRIPTION The powerful BLK42 Series are NEMA 42 Brushless DC Motors that are IP65 rated which meet the splashproof requirements for most applications. The BLK42 Series has complete protection from debris and dust particles. The sealed shafts further protect the motor providing longer life cycles. The BLK42 Series are square-bodied motors including an aluminum square-mounting flange to allow for easy installation, includes Class F insulation for higher temperature operation and has a maximum rated torque of 840 oz-in. The BLK42 Series is available in three different stack lengths with varying power levels, has a rated speed of 3000 RPM and utilizes Hall Sensor Feedback. Model # FRAME Size Rated Voltage (VDC) Rated Speed (RPM) Rated Power (W) Rated Torque (oz-in) Peak Torque (oz-in) Peak Current (A) BLK421S-160V BLK421S-310V BLK422S-170V BLK422S-310V BLK423S-310V Line to Line Resistance (ohms) Line to Line Inductance (mh) Torque Constant (oz-in/a) Back EMF Voltage (V/k RPM) Rotor Inertia (oz-insec2) Weight (lb) "L" Length (in) L East Orangefair Ln. Anaheim, CA Tel. (714) Fax. (714)

196 DIMENSIONS *Note: All units are mm Motor Line Feedback Line Y ellow - (A) Red - (VCC) R ed - (B) Blue - (GND) B lack - (C) Gray - (Hall A) Y elgrn - (GND) Green - (Hall B) White - (Hall C) Winding Type: Hall Effect Angle: Insulation Class: Dielectric Strength: Insulation Resistance: Star, 8 Poles 120 Degree Electrical Angle Class F 1500VDC for the minut e >=100MOhm, 500VDC WIRING INFORMATION SPECIFICATIONS 910 East Orangefair Ln. Anaheim, CA Tel. (714) Fax. (714)

197 l Surface Mount (IRFR2407) l Straight Lead (IRFU2407) l Advanced Process Technology l Dynamic dv/dt Rating l Fast Switching l Fully Avalanche Rated Description Seventh Generation HEXFET Power MOSFETs from International Rectifier utilize advanced processing techniques to achieve extremely low on-resistance per silicon area. This benefit, combined with the fast switching speed and ruggedized device design that HEXFET power MOSFETs are well known for, provides the designer with an extremely efficient and reliable device for use in a wide variety of applications. G IRFR2407 IRFU2407 HEXFET Power MOSFET D S PD V DSS = 75V R DS(on) = 0.026Ω I D = 42A The D-Pak is designed for surface mounting using vapor phase, infrared, or wave soldering techniques. The straight lead version (IRFU series) is for throughhole mounting applications. Power dissipation levels D-Pak I-Pak up to 1.5 watts are possible in typical surface mount IRFR2407 IRFU2407 applications. Absolute Maximum Ratings Parameter Max. Units I T C = 25 C Continuous Drain Current, V 10V 42 I T C = 100 C Continuous Drain Current, V 10V 29 A I DM Pulsed Drain Current 170 P C = 25 C Power Dissipation 110 W Linear Derating Factor 0.71 W/ C V GS Gate-to-Source Voltage ± 20 V E AS Single Pulse Avalanche Energy 130 mj I AR Avalanche Current 25 A E AR Repetitive Avalanche Energy 11 mj dv/dt Peak Diode Recovery dv/dt ƒ 5.0 V/ns T J Operating Junction and -55 to T STG Storage Temperature Range C Soldering Temperature, for 10 seconds 300 (1.6mm from case ) Mounting Torque, 6-32 or M3 screw 10 lbf in (1.1N m) Thermal Resistance Parameter Typ. Max. Units R θjc Junction-to-Case 1.4 R θja Junction-to-Ambient (PCB mount)* 50 C/W R θja Junction-to-Ambient 110 * When mounted on 1" square PCB (FR-4 or G-10 Material). For recommended footprint and soldering techniques refer to application note #AN /1/00

198 IRFR/U2407 Electrical T J = 25 C (unless otherwise specified) Parameter Min. Typ. Max. Units Conditions V (BR)DSS Drain-to-Source Breakdown Voltage 75 V V GS = 0V, I D = 250µA V (BR)DSS / T J Breakdown Voltage Temp. Coefficient V/ C Reference to 25 C, I D = 1mA R DS(on) Static Drain-to-Source On-Resistance Ω V GS = 10V, I D = 25A V GS(th) Gate Threshold Voltage V V DS = 10V, I D = 250µA g fs Forward Transconductance 27 S V DS = 25V, I D = 25A I DSS Drain-to-Source Leakage Current 20 V µa DS = 75V, V GS = 0V 250 V DS = 60V, V GS = 0V, T J = 150 C I GSS Gate-to-Source Forward Leakage 200 V GS = 20V na Gate-to-Source Reverse Leakage -200 V GS = -20V Q g Total Gate Charge I D = 25A Q gs Gate-to-Source Charge nc V DS = 60V Q gd Gate-to-Drain ("Miller") Charge V GS = 10V t d(on) Turn-On Delay Time 16 V DD = 38V t r Rise Time 90 I D = 25A ns t d(off) Turn-Off Delay Time 65 R G = 6.8Ω t f Fall Time 66 V GS = 10V Between lead, D L D Internal Drain Inductance 4.5 6mm (0.25in.) nh G from package L S Internal Source Inductance 7.5 and center of die contact S C iss Input Capacitance 2400 V GS = 0V C oss Output Capacitance 340 pf V DS = 25V C rss Reverse Transfer Capacitance 77 ƒ = 1.0MHz, See Fig. 5 C oss Output Capacitance V GS = 0V, V DS = 1.0V, ƒ = 1.0MHz C oss Output Capacitance 220 V GS = 0V, V DS = 60V, ƒ = 1.0MHz C oss eff. Effective Output Capacitance 220 V GS = 0V, V DS = 0V to 60V Source-Drain Ratings and Characteristics Parameter Min. Typ. Max. Units Conditions D I S Continuous Source Current MOSFET symbol 42 (Body Diode) showing the A G I SM Pulsed Source Current integral reverse 170 (Body Diode) p-n junction diode. S V SD Diode Forward Voltage 1.3 V T J = 25 C, I S = 25A, V GS = 0V t rr Reverse Recovery Time ns T J = 25 C, I F = 25A Q rr Reverse RecoveryCharge nc di/dt = 100A/µs t on Forward Turn-On Time Intrinsic turn-on time is negligible (turn-on is dominated by L S +L D ) Notes: Repetitive rating; pulse width limited by max. junction temperature. Starting T J = 25 C, L = 0.42mH R G = 25Ω, I AS = 25A. ƒ I SD 25A, di/dt 290A/µs, V DD V (BR)DSS, T J 175 C Pulse width 300µs; duty cycle 2%. C oss eff. is a fixed capacitance that gives the same charging time as C oss while V DS is rising from 0 to 80% V DSS Calculated continuous current based on maximum allowable junction temperature. Package limitation current is 30A 2

199 Customer Sample Motor Data Sheet Microchip Date: Model Number Serial # 2/14/02 DMB0224C Notes: L-L Resistance (R tm ) Ohms : 4.03 Electrical Time Constant (t e ) msec. : 1.14 L-L Inductance (L tm ) mh at 1Khz : 4.60 Mechanical Time Constant (t m ) msec. : 3.74 Torque Constant (K t ) oz.in./amp : 9.79 Thermal Resistance (R th ) C/watt 4.78 Voltage Constant (K e ) V peak /K RPM : 7.24 Thermal Time Constant (t th ) min. : 16 Amb. Temp. ( ºC ) : 22.7 Rotor Inertia (Jr) oz-in-s² : Stack Length: Speed / Torque Test Data -Control Input set at 100% duty cycle. Load Volts (DC) Amps (DC) Watts (DC) Speed (RPM) Torque (oz.in.) Output (watts) Output (HP) Max Continuous Rating Special Load Points 1 2 Sample Motor Test Data Eff. (%) System Watts (InPut) 100 System Eff. ( % ) 100 System Amps (DC) 4 Motor Speed (RPM) Torque (oz.in.) RPM vs Tq. Amps vs Tq. Eff. vs Tq. Watts vs Tq. This motor is intended for sampling and customer approval only. No application fitness approval is implied, as that can only be determined by the customer. These data represent performance of a single sample motor. These values are not to be construed as guaranteed values. Hurst Mfg. Company Confidential Motor Test Data 2/18/03