Analysis of fault conditions in permanent- magnet in-wheel motors
by
Yonghui Liao
Master Thesis
Royal Institute of Technology Department of Electrical Engineering Electrical Machines and Power Electronics
Stockholm 2011-06-14
XR – EE – EME 2011: 014
Analysis of fault conditions in permanent- magnet in-wheel motors
Abstract
The use of light in-wheel motors is an attractive alternative for smaller passenger vehicles in urban areas. An in-wheel motor design has already been developed in a previous master thesis project. In this project, the proposed design is further analysed during electrical faults using a FEM software package (JMAG).
A failure modes and effects analysis (FMEA) is carried out to identify the important faults. The faults simulated in this project are: phase short circuit, phase open circuit, high impedance, inter-turn short circuit, magnetization faults and inverter faults. Each of these faults is then further divided into the second level detailed faults.
Four new JMAG models are developed based on the JMAG model in the previous master thesis. Each fault is simulated by using a suitable JMAG model. The simulation results under fault conditions and those under normal conditions of the corresponding JMAG models are then compared and analysed.
A demagnetization check and a simple thermal check during fault conditions are carried out. At last a torque-speed model is discussed for further study based on the obtained torque under fault conditions.
Keywords
In-wheel motor, permanent-magnet synchronous machine, electrical fault analysis, finite element analysis
Sammanfattning
Användningen av lätta navmonterade elmotorer (hjulmotorer) är ett attraktivt alternativ för mindre personbilar i tätort. En hjulmotor-design har utvecklats i ett tidigare examensarbete. I detta projekt har den föreslagna konstruktionen analyseras ytterligare med avseende på elektriska fel med hjälp av ett FEM-programpaket som heter JMAG.
En feleffektsanalys (FMEA) har utförts för att identifiera viktiga fel. Felen som simuleras i detta projekt är: fas-kortslutning, avbrott i matning, hög impedans, varvkortslutning, fel magnetisering och omriktare fel. Var och en av dessa fel är sedan vidare indelade i en andra nivå av detaljerade fel beskrivningar.
Fyra nya JMAG modeller har utvecklas genom att bygga vidare på JMAG modeller från tidigare examensarbeten. Varje fel simuleras med hjälp av en lämplig JMAG modell. Simuleringsresultat från fel fallen har jämförts med motsvarande JMAG modeller under normala driftsförhållanden.
En kontroll av förhållanden som permanent avmagnetiserar magneterna och en enkel termisk kontroll vid de simulerade felen har genomförs. Slutligen har en moment-varvtals modell diskuterats för vidare analys, baserad på de erhållna resultaten under felförhållanden.
Nyckelord
Hjulmotor, permanent magnetiserade synkronmaskin, feleffektsanalys, finit element analys
1
Acknowledgements
This master thesis is carried out at the division of Electrical Machines and Power Electronics, Royal Institute of Technology, from January 2011 until June 2011. This is definitely an unforgettable period of time in my life, not only for the technical knowledge I gained, but also for the joyful experience I had during these five months. I would like to express my gratitude to all those that helped me during the writhing of my thesis.
First and foremost, my utmost gratitude goes to my supervisor Anders Malmquist for his continuous guidance and support and for the technical knowledge he provided me. I appreciate the time he spared to help me even though he is extremely busy.
I am heartily thankful to Oskar Wallmark for providing great suggestions, pointing out the mistakes and coordinating a lot of things for all the people related to this project from the initial to the final level of this thesis.
Many thanks to Juliette Soulard for correcting my report and for her important feedback which will definitely be very helpful in my future career.
I would also like to thank Daniel Wanner for his concerning and advice to my thesis. Thanks to Sture Eriksson for his participation in FMEA meetings and his great advice on the electrical faults rating based on his rich experience. Thanks to Mats Leksell, Andreas Krings, Shuang Zhao, KashifKhan, Shafigh Nategh and Ara Bissal for their academic supports. Thanks to Eva Pettersson and Peter Lönn for removing all the obstacles concerning administrative and computer issues. And thanks to all the staff in the department for creating the wonderful working atmosphere.
Apart from the school side, I want to thank my dear friends Angus, Bai, Cheng, Guillaume, Lindsay, Leah, Lei, Meng, Tyler... for so much fun they gave me after work. Love you all!
At last but not least, I would like to express my gratitude to my family especially to my parents for their endless love.
Yonghui Liao Stockholm, Sweden July, 2011
2
Contents
1 2
Introduction ...................................................................... 1 Failure modes and effects analysis (FMEA) ................. 3
2.1 FMEA .................................................................................................... 3 2.2 FMEA results ......................................................................................... 4
3 JMAG models ................................................................... 6
3.1 Previous designed model ....................................................................... 6 3.2 Current source supplied model with end-windings ............................... 7
3.2.1 Grounding .......................................................................................................... 7
3.2.2 End winding calculation .................................................................................... 7
3.3 Voltage source model .......................................................................... 10 3.4 Model with 26 poles ............................................................................ 11 3.5 Torque comparison of three new models ............................................ 13
4 Faults simulation ............................................................ 14
4.1 Motor speed for simulation ................................................................. 14 4.2 Phase short circuit ................................................................................ 15
4.2.1 Ground fault ..................................................................................................... 15 4.2.2 Two-phase short circuit ................................................................................... 18 4.2.3 Three-phase short circuit ................................................................................. 19 4.2.4 Torque comparison .......................................................................................... 20
4.3 Phase open circuit ................................................................................ 20 4.4 High impedance ................................................................................... 23 4.5 Inter-turn short circuit .......................................................................... 25
4.5.1 JMAG simulation ............................................................................................ 25 4.5.2 Simulation results ............................................................................................ 27
4.6 Magnets falling off .............................................................................. 28
4.6.1 JMAG Simulation ............................................................................................ 29 4.6.2 Comparison and conclusion ............................................................................ 30
4.7 Demagnetization .................................................................................. 30
4.7.1 JMAG simulation ............................................................................................ 31
4.8 Faulty inverter...................................................................................... 35
4.8.1 PWM voltage source ....................................................................................... 35
1
4.8.2 Normal condition with PWM voltage source supply ...................................... 38 4.8.3 One upper switch open circuit ......................................................................... 38 4.8.4 One upper switch short circuit ......................................................................... 40 4.8.5 One lower switch faults ................................................................................... 41 4.8.6 Inverter shuts down ......................................................................................... 42 4.8.7 Inverter fault results comparison ..................................................................... 45
4.9 Lost current in field-weakening range................................................. 45
4.9.1 Field-weakening .............................................................................................. 46 4.9.2 JMAG simulation ............................................................................................ 47 4.9.3 Torque comparison .......................................................................................... 50
5 Thermal effects ............................................................... 52
5.1 Demagnetization check ....................................................................... 52
5.1.1 Normal condition ............................................................................................. 54
5.1.2 Phase short circuit ............................................................................................ 55 5.1.3 Inverter faults ................................................................................................... 62 5.1.4 Inter-turn short circuit ...................................................................................... 63
5.2 Thermal check ..................................................................................... 65
5.2.1 Phase short circuit ............................................................................................ 65 5.2.2 Results of inverter faults and inter-turn short circuit ....................................... 66
6 Conclusions and further work ...................................... 67
6.1 Conclusions ......................................................................................... 67 6.2 Recommendations for further work .................................................... 68
References .............................................................................. 69
2
Chapter 1. Introduction
1 Introduction
Applying light in-wheel motors with individual steering actuators in smaller passenger vehicles can bring the vehicles a lot of advantages. For example, it is possible to freely generate forces as closely as possible to the tire contact patch. The tire forces can be allocated freely to each wheel. [1]
However, there are also challenges for vehicles designed with these light in-wheel motors. For example, a new complex control system is required to keep balance among different in-wheel motors in the vehicle. The introduction of drivelines containing power electronics and electric machines also raises important questions concerning reliability and behaviour during electrical faults which, indirectly, affects passenger safety. This master thesis project is part of the PhD project “Fault-Tolerant Over-Actuated Hybrid Electric Vehicles”, whose goal is to analyse the impact of failure modes and the degree of in-built fault-tolerance for different vehicle control strategies in electric and hybrid-electric drivelines. The goal of this master thesis project is to analyse the electric motor under different fault conditions and then help to develop suitable transient motor fault models in vehicle dynamical simulations by using the motor output data under fault conditions.
Three in-wheel motor models have been designed in a previous master thesis project, called motor 1, motor 2 and motor 3 respectively. Due to the relatively better performance of motor 2, it is selected in this project. [2]
Winding type Weight[kg] Pole number Torque at maximum
speed[Nm] Copper losses[kW] Iron losses[kW] Efficiency Thermal analysis
Motor 1 Motor 2
distributed 17 12.2 16 26 59 69 1.09 0.76 0.07 0.09 0.93 0.95 No thermal problems
Motor 3 concentrated
12.6 28 90
1.21 0.07 0.92
Close to the magnet limit(Risk of
demagnetization)
Table 1.1 Comparison of the three motor models
The motor is simulated by using a FEM software package (JMAG1). A description of JMAG is shown in App. A. The task is done in the following steps:
1
JMAG is a registered trademark of Japan Research Institute, Limited
1
Chapter 1. Introduction
a. Select the faults to be analyzed
b. Simulate the selected different types of faults
c. Analyze the resulting thermal effects during the fault conditions
The fault models use the motor output torque as input to the models. Thus the torque is the most interested output in this project for developing the transient motor fault models used in vehicle dynamical simulations. Some other data during faults like the currents and the voltages are also noted. Since they could be used in further steps, for example, fault detection.
Notice that in this thesis project, the torque control system is assumed to be broken down, which means that the motor will not maintain the same torque during faults. The motor is analysed as an independent system without current control loop.
2
Chapter 2. Failure modes and effects analysis (FMEA)
2 Failure modes and effects analysis (FMEA)
Many faults could happen when the motor is operating. However, the importance of all the faults differs since they have different possibilities to happen, different severities when happening, and different detectability. It is important to identify the major faults and neglect those do not have important impacts. Thus a FMEA was carried out to rank the faults.
2.1 FMEA
A failure modes and effects analysis (FMEA) is a procedure in product development and operations management for analysis of potential failure modes within a system for
classification by the severity and likelihood of the failures. Failure modes are any errors or defects in a process, design, or item, especially those that affect the customer, and can be potential or actual. Effects analysis refers to studying the consequences of those failures. [3]
Here the faults are evaluated by RPN (risk priority number), which is the product of three evaluation parameters: S (Severity rating), O (occurrence rating) and D (detection rating). [3]
(2.1) Severity rating (S): This parameter is used to evaluate the severity of the faults. It is given from 1 (no danger) to 10 (critical).
Rating 1 2 3 4/5/6 7/8 9/10
Meaning
No effect
Very minor (only noticed by discriminating customers)
Minor (affects very little of the system, noticed by average customer) Moderate (most customers are annoyed)
High (causes a loss of primary function; customers are dissatisfied)
Very high and hazardous (product becomes inoperative; customers angered; the failure may result unsafe operation and possible injury)
Table 2.1 Meaning of different severity ratings [4]
Occurrence rating (O): It is given from 1 to 10 to evaluate the occurrence frequency of the faults.
3
Chapter 2. Failure modes and effects analysis (FMEA)
Rating 1 2/3 4/5/6 7/8
Meaning
No effect
Low (relatively few failures) Moderate (occasional failures) High (repeated failures)
9/10 Very high (failure is almost inevitable)
Table 2.2 Meaning of different occurrence ratings [4]
Detection rating (D): It is given from 1 to 10 to describe the difficulties of detecting different faults.
Rating 1 2 3 4/5/6 7/8 9/10
Meaning
Almost certain High Moderate
Moderate - most customers are annoyed Low
Very remote to absolute uncertainty
Table 2.3 Meaning of different detection ratings [4]
2.2 FMEA results
An FMEA meeting was held to rank the faults. The discussion result is shown in Table 2.4.
Failure mode
S O D (severity (occurrence (detection rating) rating) rating)
8 8
8 4 8 6 8 5 3 4 2 1 2 1 RPN (risk priority number) 6 240 6 3 8 8 2 3 144 96 64 64 24 24 Too high magnetization - Lost current in field wakening range Short circuit Open circuit High impedance
Rotor body - Magnets(s) fall off Too low magnetization - Demagnetized Shaft - Overload (torque) Table 2.4 Result of the first FMEA meeting
After further discussions in the second FMEA meeting, the fault “Shaft-Overload (torque)” has been removed because this fault cannot be simulated in JMAG and is not in the scope of this thesis project. But another fault “uneven magnetization” was added in due to its substantial influence on the behaviour of the motor.
All the faults that should be analysed in this thesis project are now selected and shown in Table 2.5.
4
Chapter 2. Failure modes and effects analysis (FMEA)
Fault
Phase short circuit
Detailed fault (2nd level)
Ground fault 2-phase short circuit 3-phase short circuit 1-phase open circuit 2-phase open circuit 3-phase open circuit Broken strands
High impedance faults due to other reasons, for
example, loose connection
Different number of winding turns inter-turn short
circuit Magnets fall off
Demagnetization (uneven magnetization) Lost current in field-weakening range
One switch open circuit One switch short circuit Inverter shuts down
Phase open circuit
High impedance
Inter-turn short circuit Magnetization faults
Inverter faults
Table 2.5 All the faults to be analysed in this project
5
Chapter 3. JMAG models
3 JMAG models
The previously designed JMAG model is a 1-pole model supplied with current source, shown in Fig 3.1. This model only contains the machine geometry of one pole part, and it is repeated by the boundary conditions. In JMAG, this pole part has a symmetrical boundary with another pole part with opposite polarity. And then these two parts together will be repeated 13 times (this design has 13 pole pairs). However, this model is not working for some faults, for example unsymmetrical faults or short circuit faults. For short circuit faults, a model with current sources would not show the transients when the fault happens, see chapter 4.2.1.1. And for the unsymmetrical faults like one magnet falls off, a 1-pole model is not able to do the simulation since the conditions of all poles are defined as symmetrical in the 1-pole model.
In addition, the previous model does not take the end-windings into consideration. As a result, a new model that contains end-windings, a model supplied with voltage source and a 26-pole full model are developed first. In chapter 4, a model with PWM voltage source supply is developed afterwards.
3.1 Previous designed model
(a) (b)
c) d)
Fig 3.1 a) Previous JMAG FEM model b) JMAG circuit of previous model c) The current source vector d) Circuit inside “Star connection” of b) [2]
6
Chapter 3. JMAG models
The previously designed JMAG motor model is shown in Fig 3.1 This model is a current source supplied model, with the three phase windings grounded, shown in Fig 3.1 d). The current source shown in Fig 3.1 b) is a 3-phase AC current source with the peak amplitude of 75A. The vector diagram is shown in Fig 3.1 c). Here the phase angle θ is set to be 0, while the rotor initial angle is set to be 2.31˚ to obtain maximum torque. [2]
3.2 Current source supplied model with end-windings
In this project, end-winding is added to the circuit, since it is an important factor which should be taken into account. The new circuit is shown in Fig 3.2. L1, R1, L2, R2, L3 R3 are the end-winding inductors and resistors of the three phases respectively.
Fig 3.2 JMAG circuit with end-windings
3.2.1 Grounding
Notice that in Fig 3.2, there is R0 connected to the ground. In reality, the motor three phase winding is connected with a floating neutral instead of grounded neutral. However, a ground connection is needed for the simulation with current source. As a result, a resistor R0 with high resistance of 100kΩ is chosen to be connected between the neutral point and the ground.
3.2.2 End winding calculation
The normal way of doing end-winding calculation is to ask the machine manufacturer or an experienced winder for estimation [5]. However, the selected person could not supply the requested information in due time. Thus a theoretical calculation was carried out instead.
The end winding length is calculated as follows: [5]
The pitch length (in number of slots) ,
The slot pitch is:
7
Chapter 3. JMAG models
(3.1)
Where is the air gap radius (shown in Fig 3.3), q is the number of slots per pole and phase, p is the number of poles.
The inner radius of end winding is:
(3.2)
Where is the pitch length in number of slots.
The outer radius of end winding is:
(3.3)
Thus the length of end winding of one turn at one side is:
( ) (3.4)
The end winding inductance is calculated as follows: [5]
The pole pitch is:
( )
(3.5)
Where is the air gap length.
The specific permeance coefficient (given in H/m) is:
(3.6)
Where is an empirical value in the range 0.15- 0.2, here it is chosen as 0.17.
The total inductance of end winding in one phase
() ()
(3.7)
Where is the number of conductors per slot, is the connection factor. since the stator winding is series Y connected. The number of conductors per slot is:
8
Chapter 3. JMAG models
* + [ ] (3.8)
Where the number of equivalent turns per pole pair is:
̂√
√ ( ) (3.9)
̂Where and are the phase voltage, peak air gap flux density and supply current
frequency when motor is running at the speed of 100rpm. D is the air gap diameter. The winding resistance is calculated as follows:
The copper area is:
(
) (3.10)
Where is the copper diameter.
The phase resistance of the active part of the winding is:
()
(3.11)
Where is resistivity, is winding active length, N is the number of turns per slot.
The total end winding resistance of one phase is
(
(3.12)
)
Fig 3.3 Machine geometrical parameters [2]
9
Chapter 3. JMAG models
3.3 Voltage source model
Since the phase voltage amplitude will change if the rotor rotating speed changes, while the phase current amplitude keeps constant at different speeds. Thus normally it is more convenient to simulate with the current source model.
However, the motor is supplied by a voltage source in reality. The current source model cannot fulfil the requirements of some fault simulations, for example, ground fault. If current source is applied to simulate this fault, the transient during the fault would not appear due to the stable current in the two healthy phases. See further discussion in Chapter 4.2.1.1.
A model supplied by voltage source should be developed. It is more convenient to compare the results if the output of this new model has the same output as the current source model. As a result, the phase voltage of the current source model (in normal condition running at 850rpm) calculated by JMAG is used as the 3-phase voltage source supply in the voltage source model, shown in Fig 3.4 a).
200 VaVbVcVoltage amplitude[V]1000-100-200 012
(a)
1503Time[s]45x 106-3
Harmonic components of voltage source (phase A)Amplitude[V]100500051015 (b)
Fig 3.4 a) Supply voltage source at 850rpm b) Harmonic components of phase A voltage source
10
2025Harmonic order30354045Chapter 3. JMAG models
The voltage source contains substantial harmonics due to the back emf harmonics. The harmonic components are shown in Fig 3.4 b). The fundamental amplitude is 150V (adapted to the motor emf at 850rpm). Other important harmonics are 3rd, 5th, 7th, 11th and 13rd harmonics, with the amplitude of 6.5V, 18.5V, 9.7V, 4.3V and 5.7V respectively.
Whether this new model has the same performance as the current source supplied model should be verified. Since the output torque is the most interested result, it will be used here for comparison.
Fig 3.5 Torque comparison of the current source model and the voltage source mo
400300Torque of current source modelTorque of voltage source model Torque[Nm]2001000-100 00.0050.010.015Time[s]0.020.0250.03del
The output torques of the current source model and the voltage source model are both shown in Fig 3.5. The torque of the voltage source model reaches its steady state at around 0.015s. The transient time before steady state is long because the phase voltage does not begin with 0 at the first simulation step. When JMAG solves the differential equation at the first step, there will be very large transient value.
In steady state, two torque waveforms overlap. As a result, it is verified that the voltage source model can be used in further fault simulations.
3.4 Model with 26 poles
When unsymmetrical faults happen (for example, one magnet falls off), there is no symmetry among all the poles. In order to simulate the unsymmetrical faults, a full model with 26 poles has been developed, shown in Fig 3.6.
Whether this new model has the same outputs as the former model with one pole should be verified first. The torque comparison is shown in Fig 3.7. From the result it can be seen that the torque waveforms of the full model with 26 poles and 1-pole model almost overlap in both cases of current source supply and voltage source supply. Thus the model with 26 poles can be used for the further simulation.
11
Chapter 3. JMAG models
200150100500 0
Fig 3.6 Full model with 26 poles
Torque[Nm]Torque of 1 pole modelTorque of 26-pole model12
3Time[s]45x 106-3 (a)
400300
Torque of 1 pole modelTorque of 26-pole modelTorque[Nm]2001000-100 00.0050.010.0150.020.025Time[s]
(b)
Fig 3.7 Torque comparison of full model and the model with one pole at 850rpm with a) current source supply b) voltage source supply
12
Chapter 3. JMAG models
3.5 Torque comparison of three new models
Steady state torque [Nm]
Maximum Minimum Average value value value 189.4 139.4 168.4 Current source model
190.3 138.2 168.4 Voltage source model
189.6 138.5 168.3 Model with 26 poles
Table 3.1 Torque comparison of three models
All the three newly developed models here take end-winding into account. Table 3.1 only compare the torque in steady state.
The average torque of the 3 new models is approximately 0.9% lower than 170Nm of the previous designed JMAG model without end-winding, due to the losses in end-windings.
The torque waveform of the voltage source model fluctuates more than the current source model due to the inductor. In the voltage source model, the current in the inductor (same as the phase current) is changing slowly until it reaches steady state. While in the current source model, the current in the inductor is forced to change rapidly to steady state because of the stable phase current supply.
13
Chapter 4. Faults simulation
4 Faults simulation
With the new models, the selected faults can now be simulated in JMAG. The faults are separated into six parts: Phase short circuit, phase open circuit, high impedance, inter-turn short circuit, magnetization faults (except field-weakening) and inverter faults. Field-weakening is actually not a fault, which will be discussed at the last part separately. Here it is considered as a fault and discussed in FMEA meetings because it has large impact on the motor output torque.
4.1 Motor speed for simulation
Assume that field-weakening does not happen, which means that the motor speed is lower than the base speed (see chapter 4.9). The output torque has the same amplitude but different frequency at different speeds in most cases. Because the motor winding phase current amplitude remains the same at different speeds. While the flux linkage is also the same since there is no field-weakening.
The torque is calculated as:
Where p is the number of poles, and are stator flux and stator current in αβ frame respectively.
However, in field weakening range, at different speed the flux is weakened at different level, thus the torque will be reduced in different level. See further discussion in chapter 4.9.
The analysis above is verified by JMAG. The simulations of current source model under normal condition at the speed of 100rpm, 624rpm, 850rpm and 1123rpm are done. The torque in these cases are compared and shown in Fig 4.1.
850 rpm is the motor rated speed, and 100rpm is chosen as a low speed case. 624rpm and 1123rpm are two typical speeds in urban driving condition. All these speeds here are lower than base speed.
Notice that in Fig 4.1 the abscissa is the simulation step instead of time, the comparison result shows that the motor has the same torque at different speeds lower than the base speed. Thus in further simulation, mostly only one case at the motor rated speed 850rpm will be simulated.
̅ { } [6] (4.1)
14
Chapter 4. Faults simulation
200180 Torque[Nm]160140120 050100150200250Simulation step100rpm624rpm300850rpm3501123rpm400
Fig 4.1 Comparison of torqus at four different the speeds
However, in some cases, the current is different at different speeds, for example, phase short circuit faults, since there is no current symmetry in three phases. Thus, the torque amplitude will be different at different speeds. But in order to analyse different faults and compare the output, 850rpm is also used to simulate these faults. In case the data is needed at other speeds, the simulation can be run with the required rotor speeds.
4.2 Phase short circuit
Phase short circuit could happen in the connection box or other connection lines. It contains ground fault (one phase grounded), two-phase short circuit and three-phase short circuit.
4.2.1 Ground fault
4.2.1.1 Simulation with current source model
It is mentioned in chapter 3 that the current source may not be sufficient for simulation of short circuit faults. This is verified by JMAG simulation. The output torque of the model during ground fault is compared with the torque under normal condition. The comparison result is shown in Fig 4.2.
200150 Torque[Nm]100500-50 00.0020.0040.006Time[s]Normal conditionGround fault0.0080.010.012
Fig 4.2 Comparison of torque under ground fault condition and under normal condition
15
Chapter 4. Faults simulation
It can be seen that there is no obvious oscillation when fault happens at 0.0216 s. The reason is that when the ground fault occurs, the induced current of the inductor in phase A remains the same as the current before the fault happens because the two healthy phases has the same current as before fault. The phase currents and voltages are shown in Fig 4.3. The 3-phase current is stable sinusoidal. In reality when the motor is supplied by a voltage source, there could be large oscillation during short circuit faults. Thus it is necessary to simulate this fault with voltage source supply.
100 Phase APhase BPhase CPhase current[A]500-50-100 00.0020.004
(a)
4000.006Time[s]0.0080.010.012
Phase voltage[V]2000-200-400 0Phase APhase BPhase C0.0020.004 (b)
Fig 4.3 a)Phase currents and b)voltages of current source model under ground fault condition
0.006Time[s]0.0080.010.0124.2.1.2 Simulation with voltage source model
Assume that ground fault happens at 0.0216 s (the end of four electrical periods), the output torque is shown in Fig 4.4 (a). When the fault happens, the torque has a sudden drop, until it goes to the maximum torque in the negative direction. The oscillation during fault now is obvious. Negative torque is braking torque, which means that the motor is working as a generator.
The simulation circuit is shown in Fig 4.5. When the ground fault happens, the induced current generated by L1 flows through R0 and the faulty ground. The resistance of R0 is
16
Chapter 4. Faults simulation
very high, thus the voltage at VP0 increases to a high value. The voltages are shown in Fig 4.4 (c).
From Fig 4.4 (c) it can be seen that the voltage at VP0 is almost in phase with the phase C voltage (V3, thick line in the figure). As a result, the voltage drop on phase C is reduced to a low value, and this is the reason why the amplitude of the current in phase C (high lighed in Fig 4.4 (b)) is low. However, the current in phase C reduces slowly before it reaches the steady state due to the effect of inductor L3.
4002000-200-400Torque[Nm]00.010.02
(a)
2000.03Time[s]0.040.050.06
Phase current[A]1000-100-200 0Phase APhase B0.05Phase C0.010.02
(b)
2001000.03Time[s]0.040.06
Voltage[V]0-100-200 0Phase APhase BPhase CNeutral point0.010.02voltage (c)
Fig 4.4 Voltage surce model outputs during ground fault a)Torque b)Phase currents c)Phase voltages and neutral point voltage
17
0.03Time[s]0.040.050.06Chapter 4. Faults simulation
Fig 4.5 Voltage source model JMAG simulation circuit of ground fault
4.2.2 Two-phase short circuit
The output torque during two-phase short circuit is shown in Fig 4.6. The average torque during steady state is negative (-39.1 Nm), which means that the motor mostly works as a generator during the fault.
Fig 4.6 Motor output torque during two-phase short circuit fault
400Phase current[A]2000-200-400 0Phase APhase BPhase C 0.010.02
Fig 4.7 Currents during two-phase short circuit fault at 850rpm
0.03Time[s]0.040.050.06
18
Chapter 4. Faults simulation
Fig 4.8 JAMG circuit of 2-phase short circuit
When fault happens, the current in phase C (shown in Fig 4.8) is almost 0 (shown in Fig 4.7). This is because when short circuit happens, the current goes only through R0 to the ground instead of phase A and phase B. While R0 has very high value, phase C is similar to open circuit. This is the same as winding connection with floating neutral, which is the reality.
4.2.3 Three-phase short circuit
The output torque during 3-phase short circuit is shown in Fig 4.9. In this case there will only be induced currents generated by the inductors when the fault happens. The phase current is shown in Fig 4.10. The motor now works mostly as a generator.
Fig 4.9 Motor output torque during 3-phase short circuit fault
19
Chapter 4. Faults simulation
Fig 4.10 Phase currents during 3-phase short circuit fault
4.2.4 Torque comparison
Torque[Nm]
Maximum value during
fault
___ 376.1 306.1 283.0
Minimum value during
fault
___ -302.9 -466.1 -517.2
Maximum value in steady state after fault 190.3 258.2 190.5 107.8
Minimum value in steady state after fault 138.2 -135.5 -245.9 -197.5
Average value in steady state 168.4 86.8 -39.1 -59.8
Normal condition Ground fault 2-phase short circuit 3-phase short circuit
Table 4.1 Torque under phase short-circuit fault conditions compared with normal condition
When phase short circuit faults happen, the torque will have overshoots due to the high transient currents. In the steady states after fault, the torque fluctuates more than that under normal condition, and the average torque is reduced. During two- and three- phase short circuit, the average torque is negative which means that the motor will work as a generator mostly.
4.3 Phase open circuit
Phase open circuit could happen in the connection box or the connection lines between the inverter and the motor. It could also happen when the fuse is blown. This fault contains one-, two- and three- phase open circuit. The voltage source model will be used here.
20
Chapter 4. Faults simulation
(a) (b)
(b) (d)
(e) (f)
Figure 4.11 a) Output torque during one-phase open circuit b) Phase currents during one-phase open circuit c) Output torque during two-phase open circuit d) Phase currents during two-phase open circuit e) Output torque during three-phase open circuit f) Phase currents during three-phase open circuit
21
Chapter 4. Faults simulation
The torque during phase open circuit faults are shown in Fig 4.11.
In case of a single-phase open circuit fault, the faulty phase current goes to 0, highlighted in Fig 4.11 (b). While two healthy phases have the same current in opposite direction, which means that the same current is flowing through two healthy phases, similar to a single phase machine.
When a two-phase open circuit fault happens, the current in the healthy Phase C flows through R0 (circuit shown in Fig 4.12). But due to the high resistance of R0, the current in Phase C will be close to 0, similar to three-phase open circuit. This is the reason why two-phase and three-phase open circuit faults show almost the same torque.
Fig 4.12 Circuit of 2-phase open circuit fault
Torque[Nm]
Maximum value during
fault
Minimum value during
fault
___ -2.6 -0.3 -0.3
Maximum value in steady state after fault 190.3 187.4 3.1 3.3
Minimum value in steady state after fault 138.2
-2.6 -3.3 -3.4
Average value in steady state 168.4 90.1 -0.1 -0.1
___ Normal
condition
188.5 1-phase open
circuit
2-phase open 188.5 circuit
188.5 3-phase open
circuit
Table 4.2 Torque under open-circuit fault conditions compared with normal condition
During two- and three-phase open circuit fault, the steady state torque after fault is the same as cogging torque, since the winding currents are 0, only the mechanical cogging torque exists. The negative minimum torque in single-phase open circuit is also because of the cogging torque when the current in two healthy phases is very low, and the torque generated by the current is lower than the cogging torque amplitude.
22
Chapter 4. Faults simulation
4.4 High impedance
High impedance faults can occur due to bad connections (for example, a loose connector), corrosion, poor tightening, etc.
Since for high impedance faults there could be a lot of combinations, like different levels of resistance increase in different phases, and also due to the low fault occurrence rating (in Table 2.4) of this fault, only one phase high impedance fault in some levels will be studied here, the faults happen in more than one phase will be shortly discussed at the end. In addition, since different kinds of high impedance faults will have the same effect on the simulation circuit: increasing the resistance, only one typical fault- broken strands - will be analysed here.
Broken strand fault: When some strands of the wire are broken for some reason like vibration, the conductor area will be reduced and thus the resistance will be increased.
In this thesis project, the high impedance fault discussed here only contains high resistance faults, since it is the most common case. During broken strand fault, the inductance will change due to the reduced number of turns, but only the resistance increase will be considered here so that the result can represent other high resistance fault like loose connection.
This fault can be simulated in JMAG by series connecting an extra resistor to one phase. The voltage source model will be used here. Assume that ς is the percentage of the broken strands. The extra resistance is calculated as
(4.1)
(a) (b)
Fig 4.13 Output torque with a) 35% strands broken b) 80% strands broken
Two typical torque waveforms during the fault is shown in Fig 4.13, the data comparison is shown in Table 4.3.
23
Chapter 4. Faults simulation
Compare with the normal condition, the torque during broken strand fault fluctuates more and the average torque is lowered. The more broken strands, the lower the average torque of the motor will be.
Steady state torque after Maximum fault [Nm] value Normal condition 20% broken strands 35% broken strands 65% broken strands 80% broken strands
190.3 183.6 184.5 186.1 187.3
Minimum value 138.2 133.6 131.5 118.8 99.2
Average value [ ] (Percentage of normal condition torque) 168.4 (100.0%) ---- 162.0 (96.2%) 0.02493 161.3 (95.8%) 0.05369 156.5 (92.9%) 0.1852 151.3 (89.8%) 0.3988
Table 4.3 Torque and extra resistance during different level of broken strand faults
When the fault happens only in one phase, in the six-torque-ripple periods of one electrical period, only two (1/3) of the six ripples are lowered while other four ripple periods stay almost the same as under normal condition. This can be seen in Fig 4.13 (b).
As a result, it can be inferred that when the fault happens in different phases, the corresponding torque ripple periods will be lowered substantially, while the remaining ripple periods will only be reduced a little (see “maximum value in Table 4.3”). Because different phase currents are placed in different slots, one slot will bring one ripple in the torque waveform of one period. There are six slots in the mechanical angle corresponding to one electrical period.
The output torque of 80% broken strands in two phases is shown in Fig 4.14 (a), and the output torque of 80% broken strands in three phases is shown in Fig 4.14 (b). From these two figures it can be seen that the inference is verified.
(a) (b)
Fig 4.14 Torque of broken strand fault in a) Two phases b) Three phases
24
Chapter 4. Faults simulation
4.5 Inter-turn short circuit
Inter-turn fault denotes insulation failures between two windings in the same phase of the stator. [7] However, in order to make the simulation results as close as possible to reality, the inter-turn faults discussed in this chapter could be between different phases.
The inter-turn fault is mostly caused by mechanical stress, moisture and partial discharge. The insulation failure is modelled by a resistance where its value depends on the fault severity. [7]
4.5.1 JMAG simulation
Here the JMAG model with 26 poles supplied by voltage source will be used. The JMAG model of the faulty part is shown in Fig 4.15 (a), where Rh_1 is the insulation resistance. Here the insulation resistance is assumed to be a certain value R_insulation shown in Fig 4.15 (b). When the fault happens, R_insulation will change from a very high value to 0 in a very short time. The value of R_insulation before fault does not matter because the switch S1 in Fig 4.15 (a) is open. [8]
Rinsulation10000Resistance[ohm]500000(a) (b)
5001000JMAG simulation step
Fig 4.15 a) Faulty part of JMAG simulation circuit in inter-turn short circuit fault b) Insulation resistance in JMAG
Four different cases of inter-turn will be discussed here:
(1) Windings in three slots under one pole inter-turn short circuited, the JMAG circuit is shown in Fig 4.16. The faulty FEM coil is coil 1, coil 14 and coil 27.
(2) Windings in three slots not under one pole inter-turn short circuited. Since when the faulty slots locate in different relative position, the result may be different, this case should be discussed and compare with case (1). The faulty slots location is shown in Fig 4.17 (a), the corresponding faulty FEM coil is coil 1, coil 26 and coil 39 in JMAG.
25
Chapter 4. Faults simulation
Fig 4.16 JMAG circuit of windings in 3 slots inter-turn short circuit
(a) (b)
Fig 4.17 a) Faulty slots location in case (2) b) Faulty part JMAG circuit of inter-turn fault happening in only one slot
(3) Windings in six slots inter-turn short circuited. It is the same as case (1) but three more FEM coils short circuit in the same way.
(4) Six turns in one slot short circuited (Fig 4.18). This case will be different from the above cases. The faulty coil is now separated into the healthy part (Coil 1, four turns) and the faulty part (Coil 40, six turns), shown in Fig 4.17 (b).
Fig 4.18 6 turns in one slot short circuit JMAG circuit
26
Chapter 4. Faults simulation
4.5.2 Simulation results
The torque waveforms in above four cases are shown in Fig 4.19, and the torque data is shown in Table 4.4.
(a) (b)
(c) (d)
Fig 4.19 Motor torque under inter-turn fault conditions a) Three slots under one pole inter-turn fault b) Three slots not under one pole inter-turn fault c) Six slots inter-turn fault d) Six turns in one slot inter-turn fault
When inter-turn short circuit happens, the inductor will generate induce current through the insulation resistor, which will result in the phase current rise in a short instant. While after this instant, the current increase is due to the reduction of impedance. For example, the currents under 3 slots inter-turn short circuit are shown in Fig 4.20. Thus the torque will increase due to the phase current increase.
From Table 4.4 it can be seen that the more turns that are inter-turn short circuited, the higher the torque will be, and the more the torque waveform will fluctuate. If the faulty slots are located under different poles, the torque will be slightly lower than that when faulty slots are located under one pole.
27
Chapter 4. Faults simulation
Fig 4.20 Currents under 3 slots inter-turn short circuit fault
Steady state torque after fault [Nm]
Normal condition 6 turns in one slot 3 slots under one pole 3 slots not under one pole 6 slots
Maximum value 189.6 193.5 198.2 196.8 207.6
Minimum value 138.5 140.5 143.2 142.2 148.6
Average value (Percentage of normal condition torque) 168.3 (100.0%) 171.0 (101.6%) 174.4 (103.6%) 173.8 (103.3%) 181.4 (107.8%)
Table 4.4 Torque of different level of inter-turn faults compare with full model normal condition
4.6 Magnets falling off
Magnets falling off is a mechanical fault, but this fault will have impact on the magnetization of the motor. It is unlikely that a lot of magnets will fall off, so simulations of one magnet and two magnets falling off are sufficient for analysing this fault and to understand the behaviour of the fault. Here the mechanical impact (for example, when magnets fall off, the rotor could be stuck) due to magnets falling off is not considered since this thesis project is aiming at analyse the electrical faults.
The simulation of this fault requires the full model since it is an unsymmetrical fault. The current source is used here and it is assumed that the fault happens from the beginning. The reason is that for this fault, only the magnetization is of interest, not the transient when fault happens since the transient caused by reducing parts of the magnetization will not result in obvious oscillation.
Here four cases are discussed: one magnet falls off, two adjacent magnets fall off (one N-pole and one S-pole), two adjacent S poles and two nonadjacent S poles.
28
Chapter 4. Faults simulation
4.6.1 JMAG Simulation
The locations of the falling off magnets with flux lines in four different cases are shown from Fig 4.21.
(a) (b)
(c) (d)
Fig 4.21 a) 1 magnet falls off b) 2 adjacent magnets fall off c)2 adjacent S-pole magnets fall off d) 2 non-adjacent S-pole magnets fall off
The area where magnets fall off is replaced by air. The resulting torque has the same shape but a different value range. For example, the torque in case (1) is shown in Fig 4.22.
29
Chapter 4. Faults simulation
200150100500Torque[Nm]012x 10
Fig 4.22 Motor output torque when 1 magnet falls off
34Time[s]567-3
4.6.2 Comparison and conclusion
Steady state torque [Nm]
Normal condition fed by current source
One magnet falls off
Two adjacent magnets(one N pole, one S pole) fall off Two adjacent S poles
Two non-adjacent S poles
Maximum Minimum Average (Percentage of
normal condition torque)
189.4 139.4 168.4 (100.0%) 181.4 175.0 173.9 173.2
133.0 127.7 128.1 127.4
161.1 (95.7%) 155.3 (92.2%) 155.2 (92.2%) 154.4 (91.7%)
Table 4.5 Torque comparison of magnets falling off faults
When magnets fall off, the magnetic flux is weakened, and the torque will then be reduced since the current remains the same. It can be found that the average torque is almost proportional to the number of healthy magnets. When 2 magnets fall off, the longer the distance between the 2 falling off magnets is, the lower the average torque will be.
4.7 Demagnetization
Here in this project, model VACODYM 854 TP [9] is selected as the permanent magnet material. The typical demagnetization curves of this model at different temperatures are shown in Fig 4.23. It can be seen that at each temperature, when the current loading is higher, the flux density of the magnet will be lower.
There are some points called the knees on these curves, marked with dots in Fig 4.24. When the current loading is so high that the operating points go beyond these knees, the magnet will be demagnetized. Once the magnet is demagnetized, for example, the operating point goes to the triangle in Fig 4.24, the operating points will not follow the original demagnetization curve anymore but follow the new curve shown as the thick line in Fig 4.24. [10]
30
Chapter 4. Faults simulation
Demagnetization could happen when there is too high current loading, or normal current loading but high temperature.
Here the magnets are assumed to be demagnetized at the beginning to check how much the torque will be if the magnets are demagnetized.
Fig 4.23 Typical demagnetization curves of VACODYM 854 TP at different temperatures [9]
Fig 4.24 Knee points and the new demagnetization curve at 120˚C after the magnet is demagnetized (thick line) [9]
4.7.1 JMAG simulation
Four cases (all magnets, one magnet, three magnets and six magnets demagnetized) at different demagnetization levels will be discussed.
31
Chapter 4. Faults simulation
4.7.1.1 All magnets demagnetized
Here the JMAG model with one pole supplied by current source will be used. The residual flux density before demagnetization happens is 1.32T, six different demagnetization levels will be discussed here:
1. When is reduced to 1.26T, the magnets are demagnetized by 4.5%. 2. When , the magnets are demagnetized by 7.6%. 3. When , the magnets are demagnetized by 10.6%. 4. When , the magnets are demagnetized by 24.2% . 5. When , the magnets are demagnetized by 43.6% . 6. When , the magnets are demagnetized by 61.7%. The torque waveforms at different demagnetization levels are shown in Fig 4.25.
(a) (b)
(c) (d)
32
Chapter 4. Faults simulation
(e) (f)
Fig 4.25 Output torque when all magnets a) Br=1.26T b) Br=1.22T c) Br=1.18T d) Br=1.00T e) Br=0.744T f) Br=0.506T
All the magnets demagnetized by a percentage of 4.5% 7.6% 10.6% 24.2% 43.6% 61.7%
Maximum Torque [Nm] (Steady-state) 181.7 177.7 172.0 147.9 110.7 75.3
Minimum Torque[Nm] (Steady-state) 135.8 134.2 131.0 117.3 90.3 61.8
Torque ripple Peak-to-peak Value [Nm] 45.9 43.5 41.0 30.6 20.4 13.5
Average torque [Nm] (percentage of normal condition torque) 162.8 (96.7%) 160.0 (95.0%) 155.8 (92.5%) 137.1 (81.4% 104.2 (61.9%) 71.0 (42.2%)
Table 4.6 Output torque when all magnets are demagnetized at different levels
The torque ripples in Fig 4.25 (d) and (e) seem to be much lower than the rest. This is due to the different starting points selected by JMAG. The actual torque ripple peak-to-peak values are shown in Table 4.6. It can be seen that the torque ripple will be lower if the magnets are demagnetized in a higher percentage or in other words, higher severity level.
4.7.1.2 One magnet, three magnets and six magnets demagnetized
Now the full model supplied with current source is used since it is an unsymmetrical fault.
If only a few magnets are demagnetized, the fault influence will be very small. The toque waveforms have the same shape.
33
Chapter 4. Faults simulation
(a) (b)
(c) (d)
Fig 4.26 Torque when a) One magnet demagnetized b) Three magnets demagnetized c) Six magnets demagnetized by 10.6% d) Six magnets demagnetized by 61.7%
One magnet demagnetized by a percentage of 4.5% 10.6% 61.7%
Maximum Torque [Nm] (Steady-state) 189.3 188.9 184.9 Maximum Torque [Nm] (Steady-state) 187.6 184.8 180.6 176.4
Minimum Torque[Nm] (Steady-state) 138.2 138.0 135.5 Minimum Torque[Nm] (Steady-state) 137.7 136.1 133.0 129.8
Torque ripple Peak-to-peak Value [Nm] 51.1 50.9 49.4
Torque ripple Peak-to-peak Value [Nm] 49.9 48.7 47.6 46.6
Average torque [Nm] (percentage of normal condition torque) 168.0 (99.8%) 167.8 (99.6%) 164.3 (97.6%)
Average torque [Nm] (percentage of normal condition torque) 166.9 (99.1%) 164.7 (97.8%) 160.9 (95.5%) 157.2 (93.3%)
Table 4.7 Output torque when 1 magnet is demagnetized
Three magnets demagnetized by a percentage of 10.6% 24.2% 43.6% 61.7%
Table 4.8 Output torque when 3 magnets are demagnetized
34
Chapter 4. Faults simulation
Six magnets demagnetized by a percentage of 10.6% 24.2% 43.6% 61.7%
Maximum Torque [Nm] (Steady-state) 186.3 179.9 171.7 163.4
Minimum Torque[Nm] (Steady-state) 137.1 133.5 127.7 121.1
Torque ripple Peak-to-peak Value [Nm] 49.2 46.4 44.0 42.3
Average torque [Nm] (percentage of normal condition torque) 165.9 (98.5%) 161.0 (95.6%) 154.0 (91.4%) 146.0 (86.7%)
Table 4.9 Output torque when 6 magnets are demagnetized
From Table 4.6 – 4.9, it can be found that at a higher severity level of demagnetization (higher number of magnets and higher demagnetization percentage), the average torque as well as the torque ripple will be lower.
4.8 Faulty inverter
Since the motor in reality is fed by a PWM inverter, the simulation of inverter faults has to be done. The inverter is a carrier based PWM inverter with DC link voltage of 350V, shown in Fig 4.27. [11]
Fig 4.27 Inverter topology
4.8.1 PWM voltage source
The voltage source shown in Fig 3.4 (a) is used as the reference. The carrier has a frequency of approximately 4600 Hz. The frequency modulation ratio is 25. Fig 4.28 shows the reference and the carrier. [12]
35
Chapter 4. Faults simulation
carrier based PWM of phase A300200carrier referenceVoltage[V]1000-100-200-300 050100150200step250300350400
Fig 4.28 The reference of phase A and the carrier
The obtained PWM voltage source of phase A is shown in Fig 4.29. Notice that it only shows one period. With the same way, the voltage sources of phase B and phase C are also obtained, and then implemented in JMAG model. Now a new JMAG model is obtained. It is basically the same as the voltage source model, the only difference is that the voltage source supply is changed to the obtained PWM voltage source.
The harmonic components of the obtained phase A PWM source are shown in Fig 4.30. The amplitude of the fundamental is 141V. The 25th harmonic (carrier harmonic) has the amplitude of 142V, with high amplitude of sidebands (43V for 23rd harmonic and 35V for 27th harmonic).
Obtained PWM voltage source of phase A in 1 period200100Voltage[V]0-100-2000123Time[s]45x 10-3
Fig 4.29 Obtained PWM voltage source of phase A in 1 period
36
Chapter 4. Faults simulation
Harmonic components of PWM source150Amplitude[V]10050001020 Fig 4.30 Harmonic components of the obtained PWM source of phase A
While in this JMAG circuit model, the most important voltage harmonics are the line-to-line voltage harmonics due to the floating neutral. The line-to-line voltage and its harmonic components are shown in Fig 4.31. It can be seen that the high carrier harmonic is removed. However, there are still some harmonics with high amplitude. This means that there will be a lot of harmonic losses and due to these losses the average torque will be lowered.
Line to line voltage Vab in 1 period4002003040Harmonic order506070Voltage[V]0-200-400012
(a)
3003Time[s]45x 10-3
Line to line voltage Vab harmonic componentsAmplitude[V]200100001020 (b)
Fig 4.31 a) Line-to-line voltage of the PWM source and b) its harmonic components
37
3040Harmonic order5060Chapter 4. Faults simulation
4.8.2 Normal condition with PWM voltage source supply
Feed the PWM voltage source above in JMAG, the torque is then obtained and shown in Fig 4.32.
Steady state Torque [Nm] PWM voltage source
Continuous voltage source
Maximum value 198.3 190.3
Minimum value 65.2 138.2
Average value 142.2 168.4
Table 4.10 Torque comparison between model with PWM voltage source and continuous voltage source
From the torque comparison in Table 4.10 it can be seen that when fed by PWM voltage source, the output torque of the model is lowered to an average value of 142.2Nm. This is due to the harmonics discussed in 4.7.1. Thus when analysing the inverter faults, the normal condition torque with PWM voltage source instead of the continuous voltage source will be used to compare with the torque under fault conditions.
Fig 4.32 Normal condition torque when fed by PWM voltage source
4.8.3 One upper switch open circuit
Assume that Sa+ in Fig 4.27 is open circuit, while Sa- is still working properly. The JMAG circuit is shown in Fig 4.33. Before fault happens, S2 is on, S1 and S3 is off. V1 is the phase A PWM voltage source under normal condition. The V2 connected to D1 is a constant voltage +175V and the V3 connected to D2 is -175V.
When fault happens, S2 is off, S1 and S3 are on. If Sa- is switched on (S4 is switched on), the voltage at point “A” in Fig 4.27 (point “N” in Fig 4.33) will be -175V. When Sa- is switched off, the inductor will generate follow current flowing through D1, thus the voltage at point N will be higher than +175V. Part of the voltage of point N and the
38
Chapter 4. Faults simulation
current through D1 during the fault are shown in Fig 4.34. It can be seen that every time the voltage of point N jumps from negative to positive, there will be current through D1. The positive voltage at point N is around 175.5V, slightly higher than 175V.
The output torque under this fault condition is shown in Fig 4.35.
Fig 4.33 JMAG circuit for 1 upper switch open circuit fault
200
Voltage&Current Amplitude1000-100-200 0.0220.0225
0.023Tims[s]iD1[A]0.0235VN[V]0.024
Fig 4.34 The voltage at point N and the current through D1 during one upper switch open circuit fault
39
Chapter 4. Faults simulation
Fig 4.35 Torque under 1 upper switch open circuit fault
4.8.4 One upper switch short circuit
During this fault, the voltage of point will always be +175V. Assume fault happens at 0.0216 s, the torque is obtained and shown in Fig 4.36.
Fig 4.36 Torque under 1 upper switch short circuit fault
When this fault happens, phase A will be supplied by a DC voltage 175V, thus there will be dominant DC component in three phases. The three phase currents are shown in Fig 4.37. Since the current reaches around 1500A, components like other switches and capacitors will probably break down in reality.
40
Chapter 4. Faults simulation
Fig 4.37 Phase currents under 1 upper switch short circuit fault Due to the large DC components in the currents, the torque waveform fluctuates approximately from -1000Nm to 1000Nm, while the average value is only around 100Nm.
4.8.5 One lower switch faults
With the similar way above, the torques during one lower switch open circuit and one lower switch short circuit are obtained and shown in Fig 4.38 and 4.39 respectively.
Fig 4.38 Torque during one lower switch open circuit fault
41
Chapter 4. Faults simulation
Fig 4.39 Torque during one lower switch short circuit fault
The torque waveform in Fig 4.39 is almost the same as in Fig 4.36. High DC component is the same reason as when fault happens in the upper switch. The torque values are shown in Table 4.12.
However, there is some difference between the fault happening at the upper switch and at the lower switch. This is because the average value of PWM voltage source shown in Fig 4.29 is not 0. In the simulated 15 electrical periods (0.081s), there are 1504 negative points and 1496 positive points out of 3000 points in total in the PWM voltage source.
4.8.6 Inverter shuts down
Inverter shuts down means that due to some reason, the inverter does not work anymore, all the switches open circuit, only six diodes are functioning The JMAG circuit is shown in Fig 4.40.
Fig 4.40 JMAG circuit of inverter shuts down fault
Before the fault happens, switch S6, S7 and S8 are on, while S1, S2, S3, S4 and S5 are off. The motor is supplied by original PWM voltage source. When fault happens, S6, S7
42
Chapter 4. Faults simulation
and S8 are off, while S1~ S5 are on. V1= -175V, V2= 175V. The output torque is shown in Fig 4.41 (a).
(a)
(b)
Fig 4.41 a) Motor torque b) Phase currents during inverter shuts down fault at 850 rpm
The simulation above is done at rated speed 850 rpm. When the fault happens, there is no PWM voltage source supply, the back emf will supply the circuit. While at this speed the back emf is not high enough to turn on the diodes. All the six diodes are reverse bias, there will be no current in the phases (shown in Fig 4.41 (b)). As a result, the output torque will be just cogging torque.
However, at high speed, the back emf could be high enough to turn on the diodes. At 1373rpm, the output torque is shown in Fig 4.42 (a). The torque now is a negative torque, because there is no voltage supply from outside, the back emf is supplying the circuit. The motor now is working as a generator, the torque is braking torque.
43
Chapter 4. Faults simulation
The phase currents are shown in Fig 4.42 (b). The average torque at other speeds during inverter shuts down fault is shown in Fig 4.42 (c), notice that the torque value shown in this curve is the absolute value, the actual value is listed in Table 4.11.
(a)
200Phase APhase BPhase C
Current amplitude[A]1000-100-200 00.0050.01
Absolute value of average torque[Nm]2502001501005000.0150.02Time[s]0.0250.030.035
(b)
05001000150020002500Speed[rpm]300035004000 (c)
Fig 4.42 a) Motor torque b) Phase currents during inverter shuts down fault at 1373 rpm c) Average torque – speed curve
44
Chapter 4. Faults simulation
Speed [rpm] Torque [Nm] Speed [rpm] Torque [Nm] 0 0 1872 -206.8 850 0 1900 -207.3 1123 -0.1 1984 -208.8 1162 -1.9 2500 -200.0 1248 -11.2 3000 -189.4 1373 -75.3 3500 -166.5 1498 -130.7 4000 -160.8 1623 -171.7 Table 4.11 Average torque – speed during inverter shuts down fault
4.8.7 Inverter fault results comparison
Condition
Maximum torque[Nm] during fault 288.7 295.3 1026.2 1022.9 8.3 -5.3
Minimum torque[Nm] during fault -54.1 -65.6 -1011.7 -1010.0 -12.9 -111.6
Torque ripple Steady state average Peak-to-peak torque[Nm] Value [Nm] (Percentage of normal
condition torque)
342.8 128.3 (90.2%) 360.9 124.6 (87.6%) 2037.9 103.1 (72.5%) 2032.9 102.0 (71.7%) 21.2 -0.0 (0.0%) 106.3 -80.1 (56.3%)
One switch
open circuit One switch short circuit Inverter shuts down Upper Lower Upper Lower 850rpm 1373rpm
Table 4.12 Torque comparison between normal condition and during inverter faults
Table 4.12 shows the output torque of all the inverter faults discussed here. It can be seen that the inverter faults are very severe faults, which cause high torque pulsation and high negative torque.
4.9 Lost current in field-weakening range
Field-weakening is actually not a fault. But due to its substantial influence on the motor output torque, it is considered in FMEA. It is necessary to be analysed here also because the importance of finding out the base speed.
In the faults analysed above, only the case of 850rpm is simulated except in the inverter shuts down fault where high speed case needs to be examined. This is based on the conclusion in 4.1 that within the base speed, the torque will be the same at different speeds. However, in field-weakening range, the torque will change when the speed is changing.
In order to better focus on each fault and analyse the faults thoroughly, multiple faults are not considered in this thesis. Thus other faults analysed above are considered not happening in field weakening range. In this chapter, it is assumed that the motor works in the field-weakening range but no other faults happen.
45
Chapter 4. Faults simulation
4.9.1 Field-weakening
At normal condition before the speed goes beyond the base speed, the current lies on the q axis, shown in Fig 4.43 (a), no field weakening happened. Ψm is the magnet flux. The current only generates flux on q-axis, the d-axis flux Ψd generated by the current is 0. [13]
The induced voltage is:
(4.2)
Where Ψ is the resulting flux of Ψm and Ψd, ω is the speed.
Before going to field weakening range, when speed increases, the induced voltage will increase. However, the induced voltage cannot exceed the supply voltage, which means that is limited due to the dc-link voltage limit175V.
As a result, when induced voltage reaches its maximum value while the speed is still increasing, the resulting flux Ψ has to be decreased which means that Ψd ≠ 0, the current angle is changed shown in Fig 4.43 (b).
The speed where the induced voltage reaches its maximum value while field-weakening still has not happened yet is called the base speed. When speed goes beyond the base speed, field-weakening happens, the motor goes into the field-weakening range.
(a) (b)
Fig 4.43 Current and flux linkage a) in normal condition b) in field-weakening range
46
Chapter 4. Faults simulation
4.9.2 JMAG simulation
4.9.2.1 Base speed
The base speed is limited by the supply voltage. Keep the same current amplitude and phase angle to ensure that no field-weakening happens. Then increase the speed thus the phase voltage will go up, until the phase voltage reaches its limit, the base speed is found.
Assume that there is no over modulation, since over modulation will cause a lot of losses. Because the windings are connected with floating neutral, the peak fundamental phase voltage can then reach the maximum value of
̂ (4.3)
√ √
First take a guess of the base speed. The peak fundamental phase voltage at 850rpm is around 150V, the base speed should be around
̂
̂
(4.4)
Because the back emf is proportional to the speed, the phase voltage will increase almost at this ratio.
Apply 1145 rpm in the JMAG model with current source, the phase voltage is obtained in Fig 4.44 along with its harmonic components. The peak fundamental phase voltage is 199V lower than 202.1V.
400200Phase APhase BPhase C Voltage[V]0-200-400 00.511.5
(a)
2Time[s]2.533.5x 104-3
47
Chapter 4. Faults simulation
Harmonic components of phase voltage200X = 2Y = 199Peak value[V]1501005000510
1520Harmonic order2530 (b)
Fig 4.44 (a) Phase voltages at 1145rpm (b) Phase voltage harmonic components
Increase the speed and do the same as above, then the base speed is found at 1162rpm. The phase voltage and its harmonic components at this speed are shown in Fig 4.45. Thus the base speed is 1162rpm which is lower than the base speed in [2]. There are two reasons for this: 1. in analytical calculation, the DC link voltage limit is not taken into account; 2. the inductance value is different in JMAG compared with in the analytical calculation due to the end-winding.
400200Phase APhase BPhase C Voltage[V]0-200-400 00.511.5
(a)
250200X = 2Y = 2022Time[s]2.533.5x 104-3
Harmonic components of phase voltagePeak value[V]15010050005101520Harmonic order2530
(b)
48
Chapter 4. Faults simulation
200150100500Torque[Nm]0123x 10
(c)
Fig 4.45 (a) Phase voltages at 1162rpm (b) Phase voltage harmonic components c) Output torque at 1162rpm
The output torque at base speed is shown in Fig 4.45 (c), the value and comparison is shown in Table 4.13.
4Time[s]5678-34.9.2.2 1373rpm
At the speed of 1373rpm, the motor works in the field-weakening range. Keep the same current amplitude while changing the current phase angle, until the phase voltage reaches the limit. This current phase angle is then found to be 33˚ (electrical angle). The phase voltage is shown in Fig 4.46. The peak fundamental is 202V, does not go beyond the limit. The field-weakening is achieved. The torque is shown in Fig 4.47.
400200Phase APhase BPhase C Voltage[V]0-200-400 00.511.5
(a)
2Time[s]2.533.5x 104-3
49
Chapter 4. Faults simulation
Harmonic components of phase voltage250200X = 2Y = 202Peak value[V]15010050005101520Harmonic order2530
(b)
Fig 4.46 (a) Phase voltages at 1373rpm (b) Phase voltage harmonic components
200Torque[Nm]1000-100012
34Time[s]56x 107-3
Fig 4.47 Torque at 1373rpm
4.9.3 Torque comparison
Steady state torque[]Nm] 850rpm 1162rpm 1373rpm
Maximum value 189.4 189.3 162.3
Minimum value 139.4 139.5 111.3
Average value 168.4 168.5 149.6
Current phase angle (˚) 0
0
33
Table 4.13 Steady state torque of normal condition and torques in field-weakening range
From Table 4.13 it can be seen that at high speed in field weakening range, the torque reduces substantially when the speed increases. The base speed of this motor is found at 1162 rpm.
The average torque at other speeds is listed in Table 4.14 along with an average torque – speed curve in Fig 4.48.
50
Chapter 4. Faults simulation Speed[rpm] Current phase Average torque [Nm] (Percentage of angle[˚] normal condition torque) 100 0 168.8 (100.2%) 624 0 168.6 (100.1%) 850 0 168.4 (100.0%) 1162 0 168.5 (100.1%) 1248 16 166.9 (99.1%) 1373 33 149.6 (88.8%) 1498 45 127.7 (75.8%) 1623 55 104.2 (61.9%)
1872 74 50.2 (29.8%) 1984 90 -0.3 (-0.2%)
Table 4.14 Motor average torque – speed list
200]m150N[euq100rot ega50revA0-500500100015002000Speed[rpm]
Fig 4.48 Torque - speed curve
51
Chapter 5. Thermal effects
5 Thermal effects
When discussing the faults above, high temperature is not considered. While in reality, the faults could result in high temperature which could bring damages to the motor. Thus a thermal analysis during fault should be done here to check if damages will happen during above faults. The two most important parts that should be protected from high temperature are the magnets and the winding insulation.
From Fig 4.24 it can be seen that at high temperature, the magnets are easier to be demagnetized. Thus it should be checked that if demagnetization will happen at high temperature during faults.
When some faults happen, overheat could happen to some parts of the motor. For example, when short circuit happens, the current increase will generate higher copper losses which could overheat the windings. As a result, it is necessary to check if the temperature will reach the limit during the faults.
5.1 Demagnetization check
When simulating all the faults above, the magnet temperature is assumed to be 20˚C. At 20˚C, the magnet demagnetization curves are straight lines, the operating points will not go beyond the “knees”. Thus the faults in Chapter 4 can be analysed independently without being influenced by demagnetization.
At higher temperature, the magnets are easier to be demagnetized when there is high current loading. As a result, the current loading should be checked to see if the operating points have gone beyond the knees in Fig 4.24.
Eight probes are now put on the magnet to measure the magnetic field (current loading) and the flux density, shown in Fig 5.1. From the peak value of the magnetic field it can be found out if the working point has gone beyond the knees of the demagnetization curves.
Fig 5.1 Eight probes to detect the Magnetic field and flux density
Now when simulating the faults at higher temperature, the irreversible demagnetization curves shown in Fig 5.2 (a) is implemented in JMAG. The way of defining these
52
Chapter 5. Thermal effects
demagnetization curves which result in the difference with the curves in Fig 4.23 is shown in Fig 5.2 (b). The JMAG demagnetization curves are determined by Residual flux density, Coercive force, Relative recoil permeability and Drooping permeability. The obtained irreversible demagnetization curves in JMAG are simplified curves of the
ones shown in Fig 4.23, but they can fulfil the requirements of demagnetization check.
(a)
(b)
Fig 5.2 a) Irreversible demagnetization curve implemented in JMAG b) JMAG demagnetization curve parameters [14]
The dot points in Fig 5.2 (a) are the knees of the magnet at different temperature. If the operating points go beyond the knees, the magnet will be demagnetized.
53
Chapter 5. Thermal effects
The diamonds in the figure are the points where if the operating points go beyond, the magnet is magnetized in the negative direction, which means that the magnet is completely destroyed.
5.1.1 Normal condition
The current loading under normal condition is shown in Fig 5.3 (a). The maximum current loading is around 560kA/m, which means that demagnetization could happen when the temperature is above 150˚C according to Fig 4.24. Fig 5.3 (b) shows that at 150˚C, there is no demagnetization while at 180˚C demagnetization happens.
(a)
Normal condition B-H curve10.8150 degree C180 degree C
Flux density[T]0.60.40.20 -600-500-400-300-200Magnetic field[kA/m]-1000
(b)
54
Chapter 5. Thermal effects
1 150 degree C180 degree CFlux density[T]0.80.60.40.2 -400-350 (c)
Fig 5.3 a) Current loading under normal condition b) B-H curve in normal condition c) B-H curve in normal condition in steady state
There is high current loading at the beginning of the simulation because there is sudden phase voltage jump at the first step of simulation. See Chapter 3.3. This is not the same as reality. In [2] it is shown that demagnetization will not happen at 180˚C in normal condition.
Thus, when checking the B-H curve, the transient part at the beginning before reaching steady state should be removed. This new B-H curve in steady state is shown in Fig 5.3 (c), no demagnetization happens at 180˚C.
When analysing the faults, only those that have current loadings of higher than the normal condition will be discussed here since no demagnetization will happen during normal condition. After checking the current loading during each fault, they are found to be: phase short circuit faults, inverter faults (except the “inverter shuts down” fault) and inter-turn short circuit faults. One set of typical demagnetization images are shown in App. B.
-300-250-200Magnetic field[kA/m]-150-1005.1.2 Phase short circuit
5.1.2.1 Ground fault
55
Chapter 5. Thermal effects
Fig 5.5 Magnetic field during ground fault at a) 120˚C b) 150˚C c) 180˚C
(a)
(b)
(c)
56
Chapter 5. Thermal effects
The magnetic field during ground fault at 120˚C is shown in Fig 5.5 (a). The maximum current loading is around 700kA/m. The highest current loading appears at probe “point 5”, which means that point 5 is most probably to be demagnetized first.
The cases of 100˚C, 120 ˚C, 150 ˚C and 180 ˚C during ground fault are simulated. The resulted B-H curves of point 5 at different temperature are shown in Fig 5.6. At 100˚C and 120˚C, no obvious demagnetization happened. While at 150˚C, the operating points go beyond the knee and then go back through a demagnetized straight line when the current loading is lower. Thus point 5 is demagnetized.
Notice the circled part of dashed line (180˚C). In this part it seems like the magnet is magnetized again. This is because the B and H values obtained from JMAG are all absolute values. When plotting this B-H curve, H values are changed into negative values while B values remain the absolute value. The circled part is the part where the operating points go beyond the diamond points in Fig 5.2. As a result, the circled part shown in Fig 5.6 means that the magnet will be completely destroyed when ground fault happens at 180˚C. This can also be found out by checking Fig 5.5 (c) and Fig 5.2.
Fig 5.6 B-H curve of “point 5” during ground fault with irreversible demagnetization
The output torque at different temperatures during ground fault is shown in Table 5.1, the torque curves are shown in Fig 5.7.
Steady state torque after ground fault [Nm] 100˚C 120˚C 150˚C
Maximum 333.6 326.8 233.1
Minimum -258.9 -252.4 -223.6
Average 81.3 80.6 67.8
Table 5.1 Output torque during ground fault at different temperatures
57
Chapter 5. Thermal effects
(a) (b)
(c)
Fig 5.7 Torque during ground fault at a) 100˚C b) 120˚C c) 150˚C
At 100˚C and 120˚C, there is no demagnetization and the torque has the same shape as at 20˚C, because the maximum current loading does not reach the knee point. Demagnetization happens at 150˚C and 180˚C, and there will not be output torque at 180˚C since the magnets are completely destroyed.
5.1.2.2 Two-phase short circuit
The cases of 80˚C, 100˚C, 120˚C, 150 ˚C and 180 ˚C under two-phase short circuit fault condition are simulated. The current loading at 100˚C, 120˚C, 150 ˚C and 180 ˚C is shown in Fig 5.8.
58
Chapter 5. Thermal effects
(a)
(b)
(c)
59
Chapter 5. Thermal effects
(d)
Fig 5.8 Magnetic field during ground fault at a) 100˚C b) 120˚C c) 150˚C d) 180˚C
There is no demagnetization at 100˚C. Thus there will not be demagnetization at 80˚C either. At 120˚C the working points go beyond the knee and demagnetization happens while at 150 ˚C and 180 ˚C, the magnet is completely destroyed according to Fig 5.2.
The torque of different cases is shown in Table 5.2 and the torque waveforms are shown in Fig 5.9.
Steady state torque after fault [Nm] 80˚C 100˚C 120˚C
Maximum 254 .0 241.7 208.8
Minimum -432.9 -410.8 -401.9
Average -34.9 -32.3 -25.3
Table 5.2 Output torque under two-phase short circuit at different temperatures
(a) (b)
60
Chapter 5. Thermal effects
(c)
Fig 5.9 Torque under two-phase short circuit at a) 80˚C b) 100˚C c) 120 ˚C
5.1.2.3 Three-phase short circuit
The cases of 60˚C, 80˚C, 100˚C, 120˚C, 150 ˚C and 180 ˚C under three-phase short circuit fault condition are simulated. The current loading at 60˚C is shown in Fig 5.10.
Fig 5.10 Current loading during three-phase short circuit fault at 60˚C
According to Fig 5.2, the magnet is completely destroyed at 60˚C. Thus the magnet will be also destroyed at higher temperatures. There will be no output torque in all these cases.
61
Chapter 5. Thermal effects
5.1.3 Inverter faults
5.1.3.1 One switch open circuit
The current loadings during “one switch open circuit” fault at 150 ˚C and 180 ˚C are shown in Fig 5.12. By checking Fig 5.2, it can be seen that there is no demagnetization at 150 ˚C, while at 180 ˚C, the magnet is demagnetized.
(a)
(b)
Fig 5.12 Current loading during one switch open circuit fault at a) 150˚C b) 180˚C
62
Chapter 5. Thermal effects
5.1.3.2 One switch short circuit
The current loading during one switch short circuit fault at 60˚C is shown in Fig 5.13. It can be seen that at 60˚C, the magnet will be destroyed since the maximum value goes beyond the diamond point in Fig 5.2. At higher temperature, the magnet will also be destroyed.
Fig 5.13 Current loading during one switch short circuit fault at 60˚C
During inverter shuts down fault, no high current is generated, the current loading does not go beyond the current loading under normal condition, and thus it is not necessary to do the demagnetization check here.
5.1.4 Inter-turn short circuit
Since inter-turn short circuit faults will only result in a slight increase of current. The demagnetization check will be an estimation based on the current loading during fault at 20˚C. No extra simulations with irreversible JMAG demagnetization curves will be done. Fig 5.14 shows the current loading during three level of inter-turn short circuit fault.
From Fig 5.14 it can be seen that after 0.0216s when faults happen, the maximum current loadings at these three levels of the fault are approximately 530kA/m, 500kA/m and 440kA/m. According to Fig 4.23, some parts of the magnet with current loading higher than 380kA/m (where the knee point is at 180˚C) will be demagnetized at these three levels of the fault.
63
Chapter 5. Thermal effects
(a)
(b)
(c)
Fig 5.14 Current loading during inter-turn short circuit fault a) six slots b) three slots c) six turns in one slot
64
Chapter 5. Thermal effects
5.2 Thermal check
Since the fault transient is very short, there is no need to do a very complex thermal analysis. Use a simple model to calculate the temperature rise of the copper is sufficient for this project. Due to the short time, the heat transfer is also not considered here. The temperature rise is caused by the copper loss, as a result, only the faults that will result in high current are discussed here. They are: phase short circuits, inverter faults and inter-turn short circuit.
The temperature rise during the fault
∫
[5] (5.1)
Where is the fault duration time, is the phase current, they can be found in JMAG output data of each fault.
[2] is the thermal coefficient of copper.
The circuit phase resistance
(5.2)
The copper mass of one phase is:
(5.3)
()
5.2.1 Phase short circuit
5.2.1.1 Ground fault
The generated heat is separated into two parts. One part is from the fault happens till 0.0378s (seven electrical periods) when the motor already works in the steady state after fault; the other part is the steady state after fault. The time of the second part depends on how long the fault will last.
Analysing the current data from JMAG, the temperature rise from fault happens to 0.0378s of three phases can be obtained. =0.4218˚C, =0.3860˚C, = 0.0211˚C.
65
Chapter 5. Thermal effects
Since the winding insulation material was considered to be H class, which means that it can withstand the temperature up to 180˚C. In city driving condition, the winding temperature is around 80˚C. [2] Assume that the temperature of 180˚C is reached. Then the maximum fault time after steady state can be calculated.
(5.4)
, , .
As a result, if the fault lasts longer than t=0.0378+1.0150= 1.0528s, the insulation will break down.
5.2.1.2 Results of phase short circuits
In the same way as 5.2.1.1, the results of two- and three- phase short circuit can be found. Combined with the result of ground fault, table 5.4 is shown.
Temperature rise from fault happens until 0.0378s [˚C] Phase A Phase B Phase C 0.4218 0.3860 0.0211 0.6889 0.6889 3.398 0.9451 0.8338 0.8480
Allowed
maximum fault time [s] 1.0528 0.6449 0.5013
Ground fault
2-phase short circuit 3-phase short circuit
Table 5.4 Thermal check results of phase short circuit faults
5.2.2 Results of inverter faults and inter-turn short circuit
Temperature rise from fault happens until 0.0378s [˚C] Phase A Phase B Phase C 0.2035 0.1381 0.1103 59.65 18.32 18.04
Allowed
maximum fault time [s] 1.8256 0.0401
1 switch open circuit 1 switch short circuit
Table 5.5 Thermal check results of inverter faults
Temperature rise from fault happens until 0.0378s [˚C] Phase A Phase B Phase C
0.4017 0.3529 6 slots inter-turn short 0.4101
circuit
0.1999 0.1887 3 slots inter-turn short 0.2142
circuit
0.1314 0.1493 6 turns in one slot inter-0.1583
turn short circuit
Allowed maximum fault time [s] 0.9588 1.7978 2.8162
Table 5.6 Thermal check results of inter-turn short circuit fault
66
6. Conclusions and further work 6 Conclusions and further work
6.1 Conclusions
At the beginning of this thesis, a failure modes and effects analysis (FMEA) was carried out to identify the important faults. Several faults were then selected: Field-weakening, short circuit, open circuit, high impedance, magnets fall off and demagnetization (ranked from high to low by risk priority number). One more fault, uneven magnetization, was added later to the fault lists. However, since uneven magnetization is simulated in the same way and has the same effect as demagnetization fault when it happens in some of the magnets, uneven magnetization fault was not analysed separately anymore.
End-winding is considered here and modelled as an extra inductor and resistor series connected together. This end winding model is obtained by theoretical calculation and added into the JMAG simulation circuit.
Four new JMAG models with end-windings were then developed to simulate different types of faults. Three new models are developed first in chapter 3: current source model with one pole, voltage source model with one pole and the full model with 26 poles. These three models were then simulated in normal condition to verify that they have the same outputs. In chapter 4.8, the fourth model with PWM voltage source was developed for the inverter faults. But due to the high amplitude of the harmonics, the output torque of this model under normal condition is lower than other three models.
In chapter 4.1, it was then verified in JMAG that when the motor speed is lower than the base speed, the motor has the same output torque at different speeds, which means that normally if field-weakening does not happen (speed lower than base speed), only one simulation at one typical speed needs to be done. This speed is chosen to be the motor rated speed 850rpm. In chapter 4.9, the motor base speed is found to be 1162rpm by JMAG.
In chapter 4 different types of faults are simulated, followed by a demagnetization check and a simple thermal check in chapter 5. It can be found that the most dangerous fault is the “one switch of the inverter short circuit” fault. In this fault the torque pulsates in the range of around ±1000Nm, which is around 6 times the average torque in normal condition. In addition, the demagnetization check shows that the magnets will be completely destroyed at the temperature higher than 60˚C during this fault.
Phase short circuit faults are the second dangerous fault. During the transients when these faults happen, there will be very high torque pulsation (up to a peak-to-peak value of around 700Nm). When the three phase short circuit fault happens, the magnets will be destroyed when the temperature is above 60˚C.
67
6. Conclusions and further work Apart from the two faults above, the fault “one inverter switch open circuit” and “inverter shuts down (during high speed)” will also create notable negative torque, which means that the motor is braking. Although the open circuit faults also create negative torque, but the negative torque during open circuit faults is only the negative part of the cogging torque, which can be neglected.
At maximum speed in field weakening range, when two-phase, three-phase open circuit happens or “inverter shuts down” fault happens during low speed, the average torque will reduce to 0.
Based on the summary above, there are some important things to be noticed. All the dangerous faults mentioned here should be avoided by using some protection measures, especially the inverter faults. The PWM voltage source contains large harmonics thus a filter or better modulation strategy should be implemented. The motor should be well protected from high temperature since at the temperature above 150˚C, demagnetization could happen even in normal condition.
6.2 Recommendations for further work
In this thesis, multiple faults are not considered, while different types of faults could happen at the same time. For example, when the motor is working at high temperature, thermal demagnetization and inter-turn short circuit could happen at the same time. Multiple faults simulations should be done to make the results closer to reality. And all the faults happening at different speeds in field-weakening range should be simulated.
Since some parameters including the load torque, viscous damping constant, tire mass and road friction are not clear yet, it is hard to estimate the speed change due to the torque change. However, the speed is also an important data when analysing the faults. Thus it is necessary to build up a precise model to calculate the speed change.
In high impedance, inter-turn short circuit and demagnetization faults, the output torque at some fault severity levels are found out. In order to estimate the output torque at other severity levels, a mathematical model should be built based on the obtained torque at certain different severity levels.
68
References
References
[1] M. Jonasson and O. Wallmark, “Control of electric vehicles with autonomous corner modules: implementation aspects and fault handling”, Int. J. Vehicle Systems modelling and Testing, Vol. 3, No. 3, 2008
[2] Sergio Romero Perez, “Analysis of a light permanent magnet in-wheel motor for an electric vehicle with autonomous corner modules”, M.Sc. thesis, KTH, 2011
[3] (2008) Failure mode and effects analysis – Wikipedia. [Online]. Available: http://en.wikipedia.org/wiki/Failure_mode_and_effects_analysis
[4] Otto, Kevin; Wood, Kristin (2001). Product Design - Techniques in Reverse Engineering and New Product Development. Prentice Hall. ISBN 0-13-021271-7
[5] Chandur Sadarangani, Electrical machines. Design and Analysis of Induction and Permanent Magnet motors. Stockholm, Sweden: Royal Institute of Technology, 2006
[6] Oskar Wallmark and Christer Danielsson, “Modulation of power electronic converters”, course material, 2010
[7] B. Vaseghi, N.Takorabet, F. Meibody-Tabar, “Analytical Circuit-based model of PMSM under Stator Inter-turn Short-circuit fault Validated by Time-stepping Finite Element Analysis”, XIX International Conference on Electrical machines, 2010
[8] Luís Romeral, Julio César Urresty, Jordi-Roger Riba Ruiz, Antonio Garcia Espinosa, “Modeling of surface-mounted permanent magnet synchronous motors with stator winding interturn faults”, IEEE Transactions on Industrial Electronics, VOL. 58, NO.5, 2011
[9] Data sheet rare-earth permanent magnets VACODYM-VACOMAX, Vacuumschmelze, edition 2003.
http://www.vacuumschmelze.de/index.php?id=52&L=2
[10] J. Soulard, \"Design of permanent magnet synchronous machines\course material, 2010
[11] Ned Mohan, Tore M. Undeland, William P. Robbins, Power Electronics: Converters, Applications, and Design, WIE ISBN: 0-471-42908-2
[12] D. Grahame Holmes, Thomas A. Lipo, Pulse Width Modulation for Power Converters, ISBN: 0-471-20814-0
[13] W. L. Soong, T.J.E. Miller, \"Field-weakening performance of brushless synchronous AC motor drives\IEE Proc-Electr. Power Appl., Vol. 141, No. 6, November 1994
69
References
[14] JMAG group, JSOL Corporation, JMAG manual
[15] Electrical Machines and Drives, Textbook, KTH, 2010
[16] Moment of inertia, [Online]. Available:
http://www.physics.uoguelph.ca/tutorials/torque/Q.torque.inertia.html
[17] JMAG group, JSOL Corporation, “Motor design”, Online support, Available: http://www.jmag-international.com/solutions/motordesign2.html
70
List of Symbols and Abbreviations List of Symbols and Abbreviations
List of Abbreviations
AC DC Emf FEM FMEA PWM RPN
Alternating current Direct current
Electromotive force Finite element method
Failure modes and effects analysis Pulse width modulation Risk priority number
List of Symbols
subscript n indicates normalised parameters
subscript b indicates base (or nominal/rated) parameters ˆ indicates the peak value of a sinusoidal parameter copper area of one conductor ̂peak air gap flux density residual flux density air gap diameter copper diameter induced voltage J moment of inertia total inductance of end winding in one phase N number of turns per slot the number of equivalent turns per pole pair P copper loss end- winding inner radius Rh_1 insulation resistance in JMAG end- winding outer radius total end winding resistance of one phase extra resistance during high impedance fault phase resistance of the active part of the winding R_insulation insulation resistance during inter-turn fault T torque motor output torque load torque temperature rise phase voltage ̂ the maximum value of peak fundamental phase voltage ̂ peak fundamental phase voltage at 850rpm viscous damping coefficient
[m2] [T] [T] [mm] [m] [V] [kg ] [mH] [W] [mm] [ ] [mm] [ ] [ ] [ ] [ ] [Nm] [Nm] [Nm] [˚C] [V] [V] [V] 71
List of Symbols and Abbreviations f i the thermal coefficient of copper connection factor frequency fundamental winding factor stator current in αβ frame phase current an empirical value in the range 0.15- 0.2 winding active length end- winding length of one turn at one side copper mass the number of conductors per slot number of poles number of slots per pole and phase time fault duration time maximum fault time after steady state pitch length (in number of slots) air gap length specific permeance coefficient pole pitch mechanical angular velocity slot pitch air gap radius copper resistivity the percentage of the broken strands stator flux in αβ frame magnet flux d-axis flux generated by winding currents q-axis flux generated by winding currents resulting flux of Ψm and Ψd [ ] [Hz] [A] [m] [mm] [kg] [s] [s] [mm] [H/m] [mm] [Hz] [mm] [mm] [ ] m p q t t_fault ω ς Ψm Ψd Ψq Ψ
72
App. A JMAG Designer App. A JMAG Designer
JMAG Designer is a FEM software package. It is chosen in this project to analyze the designed motor. A short description of the functions of JMAG is shown as follows: [17] Check basic characteristics: Induced voltage, torque, inductance Analyze complex phenomena: Thermal demagnetization, vibration
Magnetic circuit Torque View magnetic saturation, flux leakage, and Calculate torque in various motors with acting point. load and no-load, taking into account the skew effects. Inductance Calculate motor inductance accurately. Utilize specialized tools for calculating d/q-axis inductance of permanent magnet motors. Losses Analyze eddy current and hysteresis losses separately for the iron losses, and eddy current losses in magnets. Temperature Calculate temperature increase and hot points due to losses. Structural analysis Investigate the effects of centrifugal force in IPM motors such as stress and displacement of the magnet. For more information please check:
http://www.jmag-international.com/support/en/index.ht
73
App. B Demagnetization image App. B Demagnetization image
Images (a), (c) and (e) are the images with the demagnetized part of magnet circled. Images (b), (d) and (f) are the images under normal condition, the magnet is not demagnetized.
(a) (b)
(c) (d)
(e) (f)
74
因篇幅问题不能全部显示，请点此查看更多更全内容