Professional Documents
Culture Documents
Doctor of Philosophy
in the Faculty of Engineering
by
Rajib Datta
Acknowledgements
I am grateful to Prof. V.T.Ranganathan for his guidance and help during the course of my research work. Many of the ideas presented in the thesis are results of numerous sessions of lively discussions with him. His instructive suggestions and encouragement are deeply acknowledged. I thank Prof. V.Ramanarayanan for all that I have learnt from him, including power electronics. I gratefully acknowledge his efforts in providing an excellent laboratory infrastructure. I sincerely thank Prof. Indraneel Sen for his advice and careful review of the thesis. Acknowledgements are due to Mr. T.Chouridas and other members of the Electrical Engineering Workshop for their help in fabricating the experimental setup. I thank Mrs. Silvi Jose for maintaining the Power Electronics Store and helping in purchase of components. I also appreciate the assistance and cooperation provided by Mr. D.M.Channe Gowda and his colleagues in the Electrical Engineering Office. Prof. Jamadagni gladly extended the various facilities available at the Centre for Electronic Design and Technology, whenever required. His useful suggestions during the design of the DSP hardware are duly acknowledged. I sincerely thank Prof. Giri Venkataraman of University of Wisconsin, Madison, for his keen interest in my work and, for ensuring a steady flow of many critical hardware components to our lab. The DSP-based hardware activities in the Power Electronics Group have been made possible due to the excellent cooperation of Mr.Sanjeev Das Mahapatro of Texas Instruments, India. Thanks are due to the members of the Power Electronics Lab, ER&DC, Trivandrum, for their help in providing valuable information and data regarding a practical wind energy conversion system. I owe a great deal to my lab mates G.Narayanan, Debiprasad Panda and Parthasarathi Sen Sharma for their help, support and constant motivation. Ebenezer Vidyasagar has been equally involved with me in the design of the TMS320F240 board. Enlivening discussions with Souvik and Gautam has helped my understanding of various technical issues. It has been a pleasure to work with Rajapandian, Giridharan, Venkatesh, Mahesh, Biju, Ramananurthy and other members of the Power Electronics Group. The synergetic atmosphere in the lab has truly been an inspiring and educative experience for me.
iii
iv
Abstract
This thesis deals with modeling, simulation and implementation of rotor side control strategies for a grid-connected wound rotor induction machine. In the system under consideration, the stator is directly connected to the constant frequency three phase grid and, the rotor is supplied by two back-to-back three phase voltage source inverters with a common dc link. Such a configuration is attractive in large power applications with limited speed range of operation. The rotor currents are controlled at any desired phase, frequency and magnitude to control the active and reactive powers of the machine independently. A stator flux oriented model of the doubly-fed wound rotor induction machine is presented. The front-end converter is similarly modeled in the stator voltage reference frame. The current controllers are designed in the field coordinates. Simulation waveforms exhibit excellent transient response of the current loops; the dynamics of the direct and quadrature axes are also observed to be decoupled. Two position sensorless algorithms for grid-connected wound rotor induction machine are proposed. In the first method the field oriented current controllers are retained; the rotor position is estimated by using simple transformations between the stator and rotor coordinates. The second algorithm directly controls the active and reactive powers by instantaneous control of the rotor flux. It uses a novel strategy to update the sector information of the rotor flux. The salient features for both these methods are starting on-the-fly, stable operation at zero rotor frequency and minimal dependence on machine parameters. An experimental setup consisting of IGBT inverters and a TMS320F240 DSP based digital controller is developed in the laboratory to implement the control algorithms. Relevant experimental waveforms are presented; they are observed to be in good agreement with the simulation results. Application of rotor side control of doubly-fed induction machine to wind energy conversion systems is studied. The scheme is compared against the existing systems using cage rotor induction machine. Peak power point tracking algorithm in the conventional torque control mode is first implemented. A dc motor driven by a commercial thyristor drive is used to simulate the turbine characteristics. Subsequently, a method is proposed to track the peak power point locus which operates in the speed control mode. Unlike the previous case, here precise information about the turbine characteristics is not required. This strategy is also implemented and verified experimentally.
vi
Contents
Acknowledgements Abstract List of Symbols 1 Introduction 1.1 1.2 1.3 1.4 1.5 General Basic Concept of Rotor Side Control Background Rotor Side Control with Back-to-back Inverter Configuration Rotor Side Control: Modes of Operation 1.5.1 Mode 1: Subsynchronous Motoring 1.5.2 Mode 2: Supersynchronous Motoring 1.5.3 Mode 3: Subsynchronous Generation 1.5.4 Mode 4: Supersynchronous Generation 1.6 1.7 Present Status Scope of the Thesis
2 Modeling and Simulation 2.1 2.2 2.3 Introduction Machine Model in Field Coordinates Field Oriented Control 2.3.1 Rotor Equation in Field Coordinates 2.3.2 Design of Rotor Current Controller in Field Coordinates 2.4 2.5 2.6 2.7 2.8 2.9 Simulation Results - Rotor Side Control Front end Converter System Description Principle of Operation and Control Modeling of the Power Circuit Front end Converter Controller Design
Contents 2.9.1 Design of the Current Controller 2.9.2 Design of the Voltage Controller 2.10 Simulation Results - Front end Converter 2.11 Conclusion 3 Hardware Organization and Experimental Results for Conventional Field Oriented Rotor Side Control and Front end Converter Control 3.1 3.2 Introduction Organization of the Power Circuit 3.2.1 IGBT Converter 3.3 DSP Based Control Hardware 3.3.1 TMS320F240 - A Brief Overview 3.3.2 TMS320F240 Based Digital Control Platform 3.4 Software Organization 3.4.1 Task Scheduling 3.4.2 Program Flow 3.4.3 Description of Tasks 3.4.5 Generation of Unit Vectors Synchronized to the Supply Voltage 3.4.6 Generation of Unit Vectors from Incremental Position Encoder Pulses 3.4.7 Scaling and Signal Monitoring through DAC 3.5 Experimental Results 3.5.1 Rotor Side Control 3.5.2 Front end Converter Control 3.6 Conclusion 63 63 65 65 67 68 68 72 72 73 74 77 78 80 80 81 87 90 91 91 92 94 96 97 97 viii 50 52 53 61
4 Rotor Side Field Oriented Control without Position Sensors 4.1 4.2 4.3 Introduction Review of Existing Schemes Proposed Algorithm for Position Sensorless Control 4.3.1 Computation of i ms 4.3.2 Starting 4.3.3 Speed Estimation
Contents 4.4 4.5 4.6 Simulation Implementation and Experimental Results Conclusion 99 106 111 113 113 114 116 117 119 119 120 120 121 121 122 124 127 128 133 138
5 Direct Power Control - Concept and Implementation 5.1 5.2 5.3 Introduction Concept of Direct Power Control Voltage Vectors and their Effects 5.3.1 Effect of Active Vectors on Active Power 5.3.2 Effect of Active Vectors on Reactive Power 5.3.3 Effect of Zero Vector on Active Power 5.3.4 Effect of Zero Vector on Reactive Power 5.4 Control Algorithm 5.4.1 Measurement of Stator Active and Reactive Power 5.4.2 Defining References and Errors 5.4.3 Switching Vector Selection 5.5 5.6 5.7 5.8 5.9 Sector Identification of Rotor Flux Starting Simulation Results Implementation and Experimental Results Conclusion
6 Using Doubly-fed Wound Rotor Induction Machine for Wind Power Generation - Design Considerations and Control Strategies 6.1 Introduction 6.1.1 Wind Turbines 6.1.2 Isolated and Grid-connected WECS 6.1.3 Choice of Wind Electric Generators 6.2 Conventional Fixed Speed System 6.2.1 Wind Turbine Characteristics 6.2.2 Conventional Fixed Speed System 6.3 Variable Speed System using Cage Rotor Induction Machine ix 139 139 139 140 141 143 143 144 147
Contents 6.3.1 Design Example 6.3.2 Operating Region and Control 6.4 Variable Speed System using Wound Rotor Induction Machine 6.4.1 Design Example 6.4.2 Operating Region and Control 6.5 Simulation of WECS 6.5.1 Fixed Speed System 6.5.2 Variable Speed System using Cage Rotor Induction Machine 6.5.3 Variable Speed System using Wound Rotor Induction Machine 6.6 Detailed Simulation of Variable Speed System WECS using Wound Rotor Induction Machine with Rotor Side Current Control 6.7 Practical Implementation of Variable Speed System using Wound Rotor Induction Machine in Torque Control Mode 6.7.1 Simulation of the turbine characteristics 6.7.2 Experimental Results 6.8 Peak Power Tracking in Speed Control Mode 6.8.1 Peak Power Tracking Algorithm 6.8.2 Selection of Sampling Frequency 6.8.3 Selection of Kt 6.8.4 Experimental Results 6.9 Conclusion 168 172 172 175 177 179 180 181 183 185 185 185 187 189 195 201 201 164 147 150 152 152 154 156 157 160 164
7 Conclusion 7.1 7.2 7.3 General Summary of the Present Work Scope for Further Research
References A B Machine Model in Stationary Coordinates Details of Major Power Circuit Components B.1 Wound Rotor Induction Machine
Contents B.2 IGBT Power Converters B.3 Front end Converter B.4 Position Encoder and Mounting Arrangement B.5 DC Motor and Drive C MATLAB Data Files C.1 Machine and Controller Parameters used for Simulation and Implementation of Rotor side Control C.2 Power Circuit and Controller Parameters used for Simulation and Implementation of Front end Converter Control D E Relevant Data of Vestas V-27 Wind Turbine Calculation of DC Bus Capacitance 208 211 213 205 202 202 203 204 205
xi
List of Symbols
i s1 , i s2 , i s3 i r1 , i r2 , i r3 i fe1 , i fe2 , i fe3 i g1 , i g2 , i g3 i c , i dc , i l u s1 , u s2 , u s3 u r1 , u r2 , u r3 u ac1 , u ac2 , u ac3 u fe1 , u fe2 , u fe3 u dc i sa , i sb i ra , i rb i ra , i rb i fea , i feb i msa , i msb Instantaneous values of stator phase currents Instantaneous values of rotor phase currents Instantaneous values of front end converter phase currents Instantaneous values of grid phase currents Instantaneous values of capacitor current, dc bus current and load current respectively for the front end converter Instantaneous values of stator phase voltages Instantaneous values of rotor phase voltages Instantaneous values of transformer secondary phase voltages for the front end converter Instantaneous values of the front end converter phase voltages Instantaneous value of the dc bus voltage Instantaneous values of the a-axis and b-axis stator currents respectively Instantaneous values of the a-axis and b-axis rotor currents respectively Instantaneous values of the a-axis and b-axis rotor currents respectively Instantaneous values of the a-axis and b-axis front end converter currents respectively Instantaneous values of the a-axis and b-axis stator flux magnetizing currents respectively u sa , u sb u ra , u rb u aca , u acb Instantaneous values of the a-axis and b-axis stator voltages respectively Instantaneous values of the a-axis and b-axis rotor voltages respectively Instantaneous values of the a-axis and b-axis transformer secondary voltages for the front end converter respectively u fea , u feb y sa , y sb Instantaneous values of the a-axis and b-axis front end converter voltages respectively Instantaneous values of the a-axis and b-axis stator flux respectively xiii
List of Symbols y ra , y rb i sd , i sq i rd , i rq i fed , i feq & i& rd , i rq & i& fed , i feq Instantaneous values of the a-axis and b-axis rotor flux respectively Instantaneous values of the d-axis and q-axis stator currents respectively Instantaneous values of the d-axis and q-axis rotor currents respectively Instantaneous values of the d-axis and q-axis front end converter currents respectively Instantaneous values of the d-axis and q-axis rotor current references respectively Instantaneous values of the d-axis and q-axis front end converter current references respectively u sd , u sq u rd , u rq u acd , u acq u fed , u feq & u& rd , u rq & u& fed , u feq Instantaneous values of the d-axis and q-axis stator voltages respectively Instantaneous values of the d-axis and q-axis rotor voltages respectively Instantaneous values of the d-axis and q-axis transformer secondary voltages for the front end converter respectively Instantaneous values of the d-axis and q-axis front end converter voltages respectively Instantaneous values of the d-axis and q-axis rotor voltage references respectively Instantaneous values of the d-axis and q-axis front end converter voltage references respectively ir sir is i& r i ms i ms us ur yr Space phasor of rotor currents in rotor reference frame Space phasor of rotor currents in stator reference frame Space phasor of stator currents in stator reference frame Reference space phasor of rotor currents in rotor reference frame Space phasor of stator flux magnetizing current in stator reference frame Instantaneous magnitude of the stator flux magnetizing current space phasor Space phasor of stator voltages in stator reference frame Space phasor of rotor voltages in rotor reference frame Space phasor of rotor flux xiv
List of Symbols ys ym Rs, Rr Ls, Lr Lo rs, rr r Ts, Tr md ml Kc J L fe X fe R fe T fe T ir T ife T vfe K pir K pife K pvfe u tri Gr G fe K ir K ife l e Space phasor of stator flux Space phasor of air gap flux Resistances of the stator and rotor phase windings respectively Self-inductances of the stator and rotor phase windings respectively Magnetizing inductance Leakage factors of the stator and rotor phase windings respectively Total leakage factor Stator and rotor time constants respectively Machine torque Load torque Torque constant Moment of inertia Ac side inductance per phase for the front end converter Ac side coil reactance per phase for the front end converter Ac side coil resistance per phase for the front end converter Time constant of the ac side coil in the front end converter Time constant of the current loops in rotor side control Time constant of the current loops in the front end converter control Time constant of the voltage loop in the front end converter control Proportional gain for PI controller used in rotor side current control Proportional gain for PI controller used in front end converter current control Proportional gain for PI controller used in front end converter voltage control Peak of triangular carrier waveform for sine-triangle modulation Gain of the rotor side converter Gain of the front end converter Gain of the current sensors used for rotor side control Gain of the current sensors used for front end converter control Angle between the stationary axis and the stator flux space phasor Angle between the stationary axis and the rotor axis xv
List of Symbols h q1 q2 dp zs z ms ze z est z P Ps Pr Pt P t arg et Cp A q v R zt k K opt Angle between the stationary axis and the stator voltage space phasor Angle between the stationary axis and the rotor current space phasor Angle between the rotor axis and the rotor current space phasor Angle between the stator flux and rotor flux space phasors Angular velocity of the stator voltage Angular velocity of the stator flux Angular rotor velocity in electrical rads.s-1 Estimated angular velocity rotor velocity in electrical rads.s-1 Angular rotor velocity in mechanical rads.s-1 Total power Stator power Rotor power Wind turbine power Target power in peak power tracking Power coefficient of the wind turbine Swept area of the wind turbine blades Air-density Wind velocity Radius of the wind turbine blades Angular velocity of the wind turbine blades Tip-speed ratio of the wind turbine Proportional gain used in peak-power tracking algorithm in torque control mode Kt Proportional gain used in peak-power tracking algorithm in speed control mode
xvi
Chapter 1
INTRODUCTION
1.1 General
Efficient control of electric power, both at the generation and utilization ends, has been an important contributing factor for industrial growth in the twentieth century. Bulk of this power is generated and utilized through electromechanical energy conversion. Variable speed operation of electrical machines enables this conversion of power in a controlled manner. With the availability of power semiconductor devices the efficiency of conversion is high and, if desired, fast dynamic response can also be achieved. Even though, dc machines can be easily controlled and are inherently suitable for high dynamic performance, several disadvantages associated with the mechanical commutator have restricted their usage in the recent past. On the other hand, squirrel cage induction machines have become increasingly popular due to their rugged construction and maintenance-free operation. Using field oriented control techniques, the flux and torque of an induction machine can be controlled in a decoupled manner and hence, fast dynamic performance, similar to that possible with dc machines, can also be achieved. While cage rotor induction machines are mainly used for medium power drive applications, the wound rotor or slip-ring induction machines are commonly used in large power drives having limited range of operating speeds. The increased cost of a slip-ring machine is justified by the reduced size of power electronic converter in the rotor circuit. So far, such machines were used as slip power recovery drives with pump or fan type of mechanical loads. However, with the emergence of variable
Chapter 1 Introduction speed constant frequency (VSCF) generation applications such as wind power generation, there is an increased attention towards wound rotor induction machines controlled from the rotor side.
P s
P r
P m
Fig.1.1 Power flow in cage rotor induction motor
Fig.1.2 Speed control of wound rotor induction motor with external rotor resistance
With the availability of thyristors this concept was utilized to introduce a dynamically varying resistance in the rotor circuit [1] as shown in Fig.1.3. It is shown that the torque produced by the machine is approximately proportional to the dc link current. Therefore, a speed controlled drive can be designed whose inner loop controls the dc link current by adjusting the duty ratio of the switch. High starting torque is available at low starting current. Also improved power factor is possible over a 2
Chapter 1 Introduction wide range of speed. However, the method is inefficient because of the power lost in the external resistors and, is only used in intermittent speed control applications where the efficiency penalty is not of great concern. If the slip power is absorbed by an appropriate electrical source instead of being wasted in the resistive elements, the same objective can be achieved. The rotor power, in this case, is regenerated back in electrical form. It is possible to control the amount of power absorbed by the source and hence the shaft speed can be varied. If the source has both sourcing and sinking capabilities, power can be absorbed from or injected into the rotor circuit. The slip can therefore, be positive or negative enabling subsynchronous and supersynchronous operation.
Slip Power
Fig.1.3 Wound rotor induction machine control with dynamic rotor resistance
1.3 Background
Historically the controllable electrical source in the rotor circuit was another auxiliary machine. The slip power was recovered back either in mechanical form or in electrical form. The former was proposed by Kramer and the latter by Scherbius in the same year (1906). These schemes can be viewed in simplified forms as in Fig.1.4(a) and Fig.1.4(b). In Kramer drive the torque contribution of the dc motor reduces the mechanical load taken by the induction motor. On the other hand, electrical recovery by Scherbius scheme uses another induction generator which feeds back the slip power to the grid at power frequency. In both cases slip is controlled by controlling the field of the dc machine.
Chapter 1 Introduction Input Power Wound Rotor Induction Motor Diode Bridge DC Motor
Diode Bridge
DC Motor
Slip Power (b) Scherbius system Fig.1.4 Slip power recovery schemes using auxiliary machines With the advent of controllable power devices like SCRs, it was possible to dispense with the additional machines. The variable frequency slip power could be recovered by introducing a phase-controlled converter at the grid interface. This was proposed by several researchers in the 1960s. Erlicki [2] proposed a scheme in 1965 where the rotor circuit was fed by an inverter operating at a frequency greater than the grid frequency. The inverter is the power source of the drive and part of the slip power is fed back from the stator to the grid. A phase-controlled rectifier is provided between the inverter and the network, which permits continuous voltage control of the dc inverter 4
Chapter 1 Introduction supply. However, the rating of the inverter and the rectifier in the rotor circuit has to be more than the mechanical output obtained from the drive. The more conventional scheme where the rotor power is rectified and fed back to the grid by a phase-controlled inverter (Fig.1.5) was subsequently proposed by Lavi and Polge [3], and Shepherd and Stanway [4]. This method of control became popularly known as the Static Scherbius system. A large dc link choke is used to interface the diode bridge output with the grid-side inverter. This ensures that the rotor current is continuous and proper control over the speed can be exercised by varying the inverter firing angle. However, the inverter consumes reactive power because of phase control and the overall system power factor is poor. The reactive power demand of the inverter also depends on the slip range, being ideally zero when the rotor runs at the synchronous speed. To improve the system power factor a transformer with proper turns ratio is connected between the inverter and the grid. The drive is started through external resistors in the rotor circuit which are subsequently cut-off when the designed slip range is reached.
Transformer
Slip Power
Diode Bridge
Fig.1.5 Static slip power recover scheme With the diode bridge and inverter arrangement, a commutatorless Kramer drive for large capacity induction machines was proposed by Wakabayashi et.al.[5] in 1976. In this scheme the inverter in the rotor circuit drives a synchronous motor whose shaft is coupled to the main motor shaft; hence the slip power adds to the mechanical output. The inverter is load commutated; hence control at low speeds is not possible because of insufficient back emf. 5
Chapter 1 Introduction In all these schemes the rotor power can flow in one direction only; so the machine can operate either at subsynchronous or at supersynchronous speeds. However, instead of the dual converter system, use of a cycloconverter in the rotor circuit permits power flow in both directions. This was proposed by Long and Schmitz [6], Weiss [7], Chattopadhyay [8] and Mayer [9]. The cycloconverter permits a reversible power flow naturally and speed control is possible for
subsynchronous as well as supersynchronous operation by controlling the injected rotor voltage. Long and Schmitz described cycloconverter control of a doubly-fed induction motor giving speed torque characteristics similar to that of a dc series motor. Weiss reported the application and performance of an ac drive using a cycloconverter and a doubly-fed wound rotor motor for pump and compressor applications. In [8], a simple rotor position-detector is used to switch the thyristor configuration in a sequential manner to generate an output voltage having a predominant slip-frequency component. The speed-torque characteristics obtained are similar to that of a dc shunt motor, and the drive is reported to be inherently stable. A 15000 hp cycloconverter-fed wound rotor induction motor drive with high dynamic response of stator active and reactive powers is presented in [9]. An orthogonal control scheme is employed to determine the rotor voltages. However, in the absence of proper current control loops, the stator power flow can be smoothly controlled only above 35% slip. However, the use of cycloconverters in industrial drives has been restricted because of the large number of thyristors used in the power circuit, the complexity involved in the firing and commutation circuits and the complex interaction with the grid. Wind energy recovery using Static Scherbius induction generator was proposed by Smith et.al. [10]. In their scheme a current source inverter is used on the rotor side and a fully-controlled rectifier on the line side. A novel signal generator concept, which is locked in phase to the rotor emf controls the secondary power to provide operation over a wide range of subsynchronous and supersynchronous speeds. The VA rating of the current source is determined as a function of the gear ratio and the operating range. However, the need for a large dc link choke and commutation capacitors were the major disadvantages. Subsequently, Holmes et.al. [11] proposed a cycloconverter-excited divided-winding doubly-fed machine as a wind power converter. The system was conceptually similar to the previous scheme, except that, the current source inverter was replaced by a cycloconverter in the rotor circuit.
Chapter 1 Introduction
Transformer
Series Inductors Slip Power Rotor side Converter Front end Converter
Fig.1.6 Rotor side control scheme with back-to-back PWM converters with capacitive dc link The converter used at the grid interface is termed as the line-side converter or the front end converter (FEC). Unlike the rotor side converter, this operates at the grid frequency. Flow of active and reactive powers is controlled by adjusting the phase and amplitude of the inverter terminal voltage with respect to the grid voltage. Active power can flow either to the grid or to the rotor circuit 7
Chapter 1 Introduction depending on the mode of operation. By controlling the flow of active power, the dc bus voltage is regulated within a small band. Control of reactive power enables unity power factor operation at the grid interface. In fact, the FEC can be operated at a leading power factor, if it is so desired. Since, the inverter operates at a high frequency, usually between 1 kHz to 5 kHz, the harmonics in input current are largely reduced. It should be noted that, since the slip range is limited, the dc bus voltage is lesser in this case when compared to stator side control. A transformer is therefore necessary to match the voltage levels between the grid and the dc side of the FEC. This arrangement presents enormous flexibility in terms of control of active and reactive powers in variable speed applications. In the following section, the concept of controlling the power flow in the machine by injecting currents in the rotor circuit is explained by deriving suitable phasor diagrams and power flow diagrams.
MODE 1 0 MODE 3
MODE 2
Speed
MODE 4
s Fig.1.7 Operating region of the doubly fed induction machine with rotor side control The operating region of the system in the torque-speed plane is shown in Fig.1.7. As stated earlier, the rotor side control strategy is advantageous within a limited slip range. Hence the operating region is spread out on both sides of the synchronous speed z s implying both subsynchronous and supersynchronous modes of operation. Moreover, the machine can operate in the motoring and generating modes irrespective of the speed. Thus four distinct modes of operation can be achieved through rotor side control corresponding to the four quadrants in the torque-speed plane. In Fig.1.7 mode 1 refers to positive torque and subsynchronous speed; this is termed as subsynchronous motoring (i.e. normal motoring) operation. Mode 2 corresponds to positive torque 8
Chapter 1 Introduction and supersynchronous speed; this mode is called supersynchronous motoring. Similarly mode 3 corresponds to subsynchronous generation and mode 4 corresponds to supersynchronous generation. The following sections describe how these different modes of operation can be achieved through rotor side control.
is us
s Lo
si
Fig.1.8 Approximate equivalent circuit with rotor side control A simplified equivalent circuit of the doubly-fed wound rotor induction machine controlled from the rotor side is shown in Fig.1.8. It is assumed that the rotor currents can be injected at any desired phase, frequency and magnitude. Therefore, the rotor circuit can be represented by a controllable current source. The equivalent circuit is drawn in the stator reference frame; hence the rotor current is represented as s i r . The steady-state phasor diagram and power flow diagram for the subsynchronous motoring mode of operation are shown in Fig.1.9. Neglecting the stator resistance, it may be assumed that the stator flux y s remains constant in magnitude and frequency since the stator is connected to the power grid. y s has two components; the stator leakage component and the magnetizing component. The former is due to the stator current alone, while the latter is due to both the stator and the rotor currents. An equivalent current i ms can be defined in the stator reference frame, which is responsible for the stator flux. This is termed as the stator flux magnetizing current [12]. The direction of y s (which is in phase with i ms ) is defined as the d-axis and, the direction of the stator voltage, which is at quadrature to y s , is termed as the q-axis. It is possible to resolve i s and s i r along and perpendicular to i ms . (The components of the currents along the d-axis are represented with subscript 'd', and those along the q-axis with subscript 'q'.) The mathematical relations between the currents in this stator flux reference frame is derived in Chapter 2 where the field oriented model is presented. Here, a qualitative approach is taken to understand the effect of current injection in the rotor circuit. 9
is s Lo s es
is s Lo s es is
isd
ims s d-axis
m
isd
ird
ims s d-axis
m
irq
ir (b)
isd
ird ims
d-axis
P s
P r
irq B (c)
ir
P m
(d)
Fig.1.9 Phasor diagram and power flow diagram during subsynchronous motoring Since y s is constant, it implies that i ms is also constant and equals the sum of i sd and i rd . With current control being exercised in the rotor circuit, an injection of positive i rd will naturally result in a lesser value of i sd being drawn from the stator terminals. The stator power factor is thereby improved. This feature is clearly depicted in Fig.1.9(a) and Fig.1.9(b). Fig.1.9(a) shows i ms being fully supplied from the stator side, as in the case of a cage rotor induction machine, whereas in Fig.1.9(b) it is partially supplied from the rotor side and partly from the stator side. It may be noted here that i rd will never be made negative. This would mean that the stator has to supply the 10
Chapter 1 Introduction magnetizing energy of the machine, as well as the reactive energy demand of the rotor circuit, bringing down the stator power factor to a very low value. Along the q-axis, the magnitude of the active component of stator current i sq is directly proportional to i rq , but opposite in sign. In fact, the induction machine can be looked upon as a current transformer as far as the active power flow in the stator and rotor circuits are concerned. Hence, to produce a motoring torque (i.e. positive torque), i rq has to be negative. This is evident from Fig.1.9; a negative i rq induces a positive i sq , implying flow of active power into the stator circuit. Below the synchronous speed the rotor falls behind the air-gap flux and the rotor induced emf e r lags the mutual flux y m by 900 as shown in Fig.1.9(a) and Fig.1.9(b). The locus of i s and i r for constant active power flow is shown in Fig.1.9(c). As the tip of the rotor current phasor is moved from B to A, the stator current phasor locus moves in the opposite direction from B to A. From this phasor diagram it may be appreciated that some amount of reactive power can as well be delivered to the source from the stator side, when the reactive power supplied from the rotor side is more than the machine requirement. This is, however, possible when the active load demand is low and there is adequate current margin in the rotor coils. In order to utilize the copper in the stator and rotor circuits effectively, it is advisable to divide the reactive power demand between the two ports. Under the condition of subsynchronous motoring the stator voltage phasor u s leads the air-gapd voltage e s (= e r ) under all conditions of load which indicates power flowing into the stator. Also the rotor current i r makes an angle less than 900 with e r , the rotor induced emf, implying that active power is being extracted from the rotor circuit. This rotor power, or the slip power, is recovered from the rotor circuit and fed back to the mains, thereby increasing system efficiency. The mechanical power output is roughly the difference between the stator and rotor powers. This is illustrated in Fig.1.9(d).
Chapter 1 Introduction It may be noted that in this mode of operation, if the stator input power is 1 p.u. and the motor is running at a slip of s p.u., the mechanical output that can be obtained is (1+s) p.u. which is more than the apparent power rating of the machine. q-axis us isq is s Lo s es = er is
isd
ird
ims s d-axis
m
P s
P r
irq
ir (a)
P m
(b)
Fig.1.10 Phasor diagram and power flow diagram during supersynchronous motoring
isd
d-axis
P s
P r
ird ims s
isq er
is
P m
(b)
(a)
Fig.1.11 Phasor diagram and power flow diagram during subsynchronous generation
12
Chapter 1 Introduction If a positive i rq is injected into the rotor circuit, i sq changes direction and becomes negative. Therefore, the active power flow into the stator becomes negative indicating that the machine is generating. This can also be appreciated from the fact that the stator terminal voltage vector u s now lags the stator induced emf. The phase angle between i r and e r exceeds 900, implying that power is fed into the rotor circuit. The power flow and phasor diagrams are given in Fig.1.11.
P s
isd ird ims m s d-axis
P r
isq
P m
is (a) (b)
Fig.1.12 Phasor diagram and power flow diagram during supersynchronous generation
13
Chapter 1 Introduction
14
Chapter 1 Introduction estimation of doubly-fed wound rotor machines [17, 18, 19]. The fact that, both the stator and rotor currents are directly measurable quantities in this case, provides interesting options for accurate position estimation, even at zero rotor frequency. In [17] the desired angle of the rotor current in the rotor reference frame, is compared with the angle of the rotor current vector in the stator reference frame, to generate the rotor frequency. Torque angle estimation is proposed in [18]. The algorithm uses rotor voltage integration; hence estimation at or near synchronous speed is difficult. In [19] transformations between the rotor reference frame and the stator flux reference frame are employed for estimating the rotor position. However, the algorithms proposed in the literature so far do not address to all the requirements of a VSCF generating system. In order to control the active and reactive power flow in the machine without position sensors, an alternative approach may be considered where, instead of the rotor current, the rotor flux is directly controlled. Direct self control (DSC) of induction motor has been proposed [20] where the stator flux is controlled to track a hexagonal trajectory. The switching scheme is such as to control the torque within a defined band. Subsequently direct torque control (DTC) schemes [13, 21, 22, 23] have been reported; the primary difference from the earlier method being a circular trajectory of the stator flux. So far, the application of direct torque control has been primarily addressed to cage rotor induction motors and permanent magnet synchronous motors [13]. However, it is possible to extend the concept of DTC to directly control the stator active and reactive powers in case of a wound rotor induction machine. The implementation of such control algorithms demands fast real-time computations. Apart from the transformations and current control loops associated with field oriented control, most of the sensorless algorithms require the machine model to be computed parallelly within the controller. Microprocessor-based digital controllers have been effectively used to implement field oriented control techniques for ac machines [24]. However, in order to improve the dynamic performance, the need for faster computation was felt. The recent availability of high-speed digital signal processors (DSP) has enabled easy implementation of computationally intensive algorithms in real time with high sampling frequencies [25]. The current trend is to incorporate application-specific peripheral hardware along with the processor in the same silicon package. PWM generation engines are also provided internally. This simplifies the design and minimizes chip count. Therefore, the hardware cost and the software overhead are reduced to a great extent.
15
Chapter 1 Introduction
16
Chapter 1 Introduction Finally, relevant experimental results to demonstrate the steady-state and dynamic performance are presented. The actual experimental responses are found to be consistent with the simulation results. Chapter 4 deals with position sensorless vector control of the wound rotor machine. The requirements of a position sensorless algorithm when used in VSCF operation are clearly brought out. A novel control strategy based on transformations between the stator and rotor coordinates is proposed. The algorithm is simple to implement, can be started on the fly (an important consideration for wind power generation) and, can run stably at zero rotor frequency (i.e. at synchronous speed). It is also independent of any critical system parameter and does not involve any dynamic angle controller. The performance of the algorithm under different conditions, e.g. during starting, transient in active power and parameter variation is studied through exhaustive simulation. It is observed that the scheme exhibits excellent dynamic and steady-state performance. Details of implementation along with relevant experimental results are provided. The experimental and simulation results are found to be in good agreement. In Chapter 5, a method for direct decoupled control of active and reactive power is presented. The algorithm extends the switching logic of direct torque control (DTC) (normally used in cage rotor induction machine) to rotor side control of doubly-fed wound rotor induction machine. By selecting the appropriate vectors in the rotor circuit the stator active and reactive powers are controlled within narrow hysteresis bands. But unlike DTC, the exact position of the rotor flux is not calculated. It is observed that the information of the sector in which it resides is sufficient for switching the correct inverter state. A novel method for updating the sector information, based on the direction of change of reactive power in the stator circuit is proposed. The direct power control algorithm uses only stator quantities for active and reactive power measurement and is inherently position sensorless. It is computationally simple and does not incorporate any machine parameter. The algorithm can start on the fly and operates stably at synchronous speed. Simulation and experimental results to validate the concept are presented. Chapter 6 deals with application of rotor side control of grid connected slip ring induction machine to wind power generation. A brief review of wind-turbines and their characteristics is first presented. The proposed VSCF system is compared with the existing systems using cage rotor induction machines (both fixed speed and variable speed). Characteristics of a practical turbine are considered to design the major electrical components used in the three systems. It is observed that, in spite of a higher machine cost, the proposed overall system employing rotor side control is 17
Chapter 1 Introduction economically competitive. The performances of these systems are compared through extensive simulation. The proposed scheme is seen to be superior in terms of energy output. The motivation of variable speed operation is to maximize the generator energy output by tracking the peak power point locus of the turbine. This is first demonstrated with the more conventional torque control mode of operation. Since, the control law in this case depends on the mechanical characteristics of the turbine and air-density, it is felt that an alternative approach of speed control can be used to make the peak power point tracking method more robust and parameter-independent. An algorithm is proposed which searches the zero slope on the power-speed characteristics of the turbine. The generator is run in the speed control mode with the speed reference being generated as a function of the change in active power. Both these algorithms are implemented on the laboratory experimental setup and relevant experimental results are furnished. Chapter 7 concludes the thesis with suggestions for further work.
18
Chapter 2
2.1 Introduction
The rotor side control scheme for doubly-fed wound rotor induction machine, introduced in Chapter 1, shows that independent control of active and reactive powers can be accomplished through current injection in the rotor circuit in a desired manner. In order to design such a current controller, field oriented control is employed. Stator flux orientation, as proposed by Leonhard [12], is a natural choice for decoupling the dynamics of the active and reactive current loops. Contrary to the cage rotor induction machine, there is an uncontrolled voltage source connected to the stator of a doubly-fed wound rotor induction machine. This acts as a disturbance variable in the plant model. The current controller, therefore, needs to be designed in such a way that the effect of the disturbance terms are nullified [26]. The control of the front end converter has attracted major attention of researchers over the last decade. Several current control techniques have been proposed for high power factor PWM rectifiers used in single phase and three phase utilities [27-31]. A hyteresis current controller has been discussed in [27], whereas, fixed frequency switching controllers are proposed in [28, 30, 31]. The intensive whistling noise in the ac side reactor due to fixed frequency switching can be reduced by using a multiple-frequency carrier [29]. In [32], a direct power control strategy is reported which selects the optimum switching state of the converter by measuring the instantaneous active and reactive powers. Even though most of these control methods promise satisfactory performance, the selection of a particular strategy is based on the ease of implementation. It is observed that if the front end converter is modeled in the synchronous reference frame with stator voltage orientation, the structure of the active and reactive current loops closely resembles the corresponding current loops
19
Chapter 2 Modeling and Simulation for rotor side control. Similar software modules can, therefore, be used for simulating and implementing the machine side and the front end converter control. The objective of this chapter is to formulate a mathematical model of the doubly-fed grid-connected wound rotor induction machine and the front end converter. A design methodology is evolved for developing the current controllers. Simulation results are presented to confirm the design and modeling. The implementation of field oriented control and experimental results are given in the next chapter.
s Lo
us
s = i ms L o
Lo
si
20
Chapter 2 Modeling and Simulation The magnetizing current vector, which is qualitatively explained in Chapter 1, is defined as i ms = (1 + r s ) i s + i r e je = (1 + r s ) i s + s i r (2.2)
i ms is the equivalent current vector in the stator reference frame responsible for producing the stator flux y s as depicted in the equivalent circuit of Fig.2.1. Hence it may be called the stator flux magnetizing current. In terms of i ms , Eq.(2.1) can be written as di R s i s + L o dt ms = u s (2.3)
Since the quantity over which direct control can be exercised is i r (and not i s ), Eq.(2.3) needs to be expressed in terms of the rotor currents. Substituting for i s in Eq.(2.3) using Eq.(2.2), we get i ms i r e je Rs 1 + r s or, where T s = + L o d i ms = u s dt (2.4)
The above equation Eq.(2.4) is defined in the stator reference frame. It can now be expressed in terms of a coordinate system fixed to the stator flux y s or equivalently to the current i ms . In order to do this i ms is first expressed in polar form with respect to stator coordinates as i ms = i ms e jl (2.5)
where i ms is the instantaneous magnitude of the current space phasor i ms and, l is its instantaneous position with respect to the stationary axis. The various phase relationships are shown in Fig.2.2. Now Eq.(2.4) can be written as 1+r T s d i ms e jl + i ms e jl = R s u s + i r e je dt s or, Ts dl jl di ms jl 1+r e + T s i ms j e + i ms e jl = R s u s + i r e je dt dt s (2.6)
21
Chapter 2 Modeling and Simulation ir irq ms Field Axis ird e Rotor Axis
Stator Axis
Fig.2.2 Angular relations of current vectors for doubly fed induction machine The above equation can be transformed into field coordinates by multiplying both sides with the operator e jl Ts dl 1+r di ms + j T s i ms + i ms = R s u s e jl + i r e j(e l ) dt dt s (2.7)
In the field coordinates, the stator voltage and rotor current space phasors can be represented as u s e jl = u sd + j u sq , and i r e j(e l) = i rd + j i rq (2.8)
Substituting this in Eq.(2.7) and separating the real and imaginary parts yields the following equations. Ts di ms 1+r + i ms = R s u sd + i rd dt s (2.9) (2.10)
dl 1 + rs = z ms = 1 R s u sq + i rq dt T s i ms
Eq.(2.9) and Eq.(2.10) represent the dynamics of the field vector magnitude and angle respectively. The stator voltage and rotor current vectors are the two inputs to the system, of which the former is not controllable (hence a disturbance variable) and the latter is the control variable. The electromagnetic torque developed can also be expressed in terms of the field current vector. In the stationary coordinates the torque equation is given by P je & md = 2 3 2 L o Im i s i r e
(2.11)
22
Chapter 2 Modeling and Simulation Substituting for i s using Eq.(2.2), we get i ms s i r P 2 & s ir m d = 3 2 L o Im 1 + r s P Lo s & = 2 3 2 1 + r s Im i ms i r P Lo jl = 2 3 2 1 + r s Im i ms e P Lo = 2 3 2 1 + r s i ms i rq i rd + j i rq e jl &
(2.12)
The complete set of equations that describe the machine dynamics in the field coordinates can be therefore, written as Ts 1+r di ms + i ms = R s u sd + i rd dt s (2.13 a) (2.13 b) (2.13 c) (2.13 d)
dl 1 + rs = z ms = 1 R s u sq + i rq dt T s i ms P Lo J dz = 2 3 2 1 + r s i ms i rq m l dt de = P z = z e 2 dt
The simulation block diagram of the doubly-fed wound rotor induction machine modeled in the field coordinates is given in Fig.2.3. The inputs to the system are the stator voltage and the rotor currents. It is assumed that there is a controlled current source in the rotor circuit which is capable of injecting currents at appropriate phase, frequency and magnitude. The rotor currents are first transformed to the stator reference frame using the operator e je . Subsequently the stator voltages and rotor currents (in the stator reference frame) are transformed to the field coordinates by multiplying with e jl . The angle is derived by solving the q-axis equation Eq.(2.13b). The magnitude of i ms is computed using the d-axis equation Eq.(2.13a). It may be noted that the alternating quantities in the stationary coordinates, when transformed to the field coordinates appear as dc quantities.
23
us1
Rs e -j
us usd s +1 usq s +1
Rs
us2
us3
3/2
us
ir1
+ + + e -j()
irq
+ * *
ms
24 L0 -P2 2 3 s +1
ir2
ir3
3/2
irb
Ts
ms e +
1 J
P 2
ml
Fig.2.3 Block diagram of the doubly-fed wound rotor induction machine model in the field coordinates
Controller
Fig.2.4 Schematic block diagram of the system arrangement for doubly-fed SRIM
is in the rotor reference frame and needs to be transformed to the field coordinates for field-oriented control. Moreover, the equation has to be expressed only in terms of i r , the variable to be controlled, and i ms , the field variable. 25
Chapter 2 Modeling and Simulation From Eq.(2.2) i ms i r e je is = 1 + r s Substituting for i s using Eq.(2.15), Eq.(2.14) may be expressed as i ms e je i r d d Rrir + Lr ir + Lo 1 + rs dt dt Now, and L2 Lo 0 Lr ( ) = Ls Lr = 1 r Lr 1 + rs i ms = i ms e jl . Therefore, Eq.(2.16) can be written as R r i r + r L r d i r + (1 r )L r d i ms e j(l e) dt dt = ur (2.17) = ur (2.16) (2.15)
Writing in terms of the direct and quadrature axis components of rotor current and voltage vectors R r i rd + ji rq e j(l e ) + rL r d dt d i ( ) + (1 r ) L r dt ms e j l e = u rd + ju rq e j(l e ) or, d i + ji ( ) R r i rd + ji rq e j(l e ) + rL r dt rq e j l e rd + rL r j di d(l e ) i rd + ji rq e j(l e ) + (1 r ) L r ms e j(l e ) dt dt d(l e ) j(l e ) e dt (2.18) i rd + ji rq e j(l e )
+ (1 r ) L r i ms j =
u rd + ju rq e j(l e )
In order to transform this equation to the field-oriented reference frame, both sides have to be multiplied by e j(l e ) . Thereby the following complex equation can be derived. d i + ji R r i rd + ji rq + rL r dt rq + j (z ms z e )rL r i rd + ji rq rd
26
Finally, the real and the imaginary parts are separated to get the d-axis and q-axis equations respectively as given below. rT r di rd u di + i rd = Rrd + (z ms z e ) rT r i rq (1 r )T r ms dt dt r (2.21)
di rq u rq rT r dt + i rq = R (z ms z e ) rT r i rd (z ms z e ) (1 r )T r i ms r (2.22) These equations represent the dynamics of the rotor currents in the field coordinate system. It is observed that due to the presence of the rotational emf terms, there is some amount of cross-coupling between the d and q axes. However, the current-loop dynamics in the two axes can be made independent of each other by compensating for these cross-coupling terms. It is interesting to note the connotations of the terms underlined in these two equations. Multiplying these terms with R r gives the corresponding voltages, which can be interpreted as follows. (a) (z ms z e ) r T r i rq % R r = (z ms z e ) r L r i rq : This is the rotational emf induced in the d axis due to the q axis rotor current. Since the relative speed of the rotor with respect to the field axis is (z ms z e ), the frequency term involved in this equation corresponds to the slip frequency. (b) (1 r ) T r d i ms % R r = (1 r ) L r d i ms : This denotes the transformer induced voltage in dt dt the d-axis due to the field current . Obviously this term will not appear in the q-axis. (c) (z ms z e ) r T r i rd % R r = (z ms z e ) r L r i rd : This term gives the rotationally induced emf in the q-axis due to the d-axis current, similar to the first term. (d) (z ms z ) (1 r )T r i ms % R r = (z ms z e ) (1 r ) L r i ms : This denotes the speed emf induced in the q-axis due to the field. The frequency term involved in this equation is again the slip frequency. 27
Chapter 2 Modeling and Simulation Due to the presence of these terms, there exists some coupling between the two axes. However, as the slip range is limited, the contribution of the terms (a) and (c) is relatively small compared to the speed emf term (d). The transformer emf term (b) also does not exist after the flux has built up, provided the stator voltage is constant in magnitude and frequency. While designing the rotor current controller it is possible to compensate for these terms and make the loop dynamics in the two axes independent of each other. This is discussed in the following section.
T ir is the desired current-loop time constant K ir is the current sensor gain. The task is now to find out the d-axis component of the instantaneous inverter terminal
voltage required to produce the current dynamics given by Eq.(2.23). Substituting for di rd from Eq.(2.23) in Eq.(2.21) gives dt rL ( ) di ms ( ) u rd = T Kr i & rd K ir i rd + R r i rd + 1 r L r dt z ms z e r L r i rq ir ir (2.24) The inverter can be modeled as a gain block G r . For sine-triangle modulation the inverter gain depends on the dc bus voltage u dc and the peak of the triangle u tri . u dc G r = 2$u tri
(2.25)
In order to make the inverter gain constant, the peak of the carrier triangular waveform u tri is made proportional to u dc .
28
Chapter 2 Modeling and Simulation Therefore, the reference for the d-axis component of rotor voltage is given by u rd u& rd = G r (1r ) L r di ms (z ms z e ) r L r Rr rL K i + i + i rq = T K rG i & ir rd rd rd Gr Gr Gr dt ir ir r (2.26) Assuming same current loop dynamics for the q-axis the reference for the q-axis component of the rotor voltage can be expressed as u rq u& rq = G r rL R = T K rG i & K ir i rq + Gr i rq rq ir ir r r + (z ms z e ) (1r ) L r (z ms z e ) r L r i ms + i rd Gr Gr (2.27)
If the impressed rotor voltages are in accordance with Eq.(2.26) and Eq.(2.27), the field and quadrature axes rotor currents can be controlled independently. The rotor current controller, therefore, does not contribute significantly to the dynamics of the system. It only ensures that the rotor currents track the reference signals produced by the outer loops. It may be appreciated that the current loop dynamics can be made much faster than the rotor time constant. However, a practical limitation to the bandwidth of the current controller is imposed by the switching frequency. Since the controller is designed in the field coordinates, all the quantities are dc and implementation of the controller becomes simpler. The simulation block diagram of the rotor current controller is shown in four parts. Fig.2.5 shows the computation of the flux vector and transformation of the rotor currents to the field coordinates. Assuming that at t = 0, the rotor and stator axes are aligned, the rotor position e can be directly obtained in the simulation by integrating the shaft speed. The rotor currents are first transformed to the stationary coordinates with this angle information e. The angle l, which the field axis makes with the stator coordinates, as well as the magnitude of i ms , can be calculated from the stator and rotor currents as follows. In the stator coordinates, i ms e jl = (1 + r s )(i sa + ji sb ) + (i ra + ji rb )(cos e + j sin e)
29
Chapter 2 Modeling and Simulation = [(1 + r s ) i sa + i ra cos e i rb sin e] + j [(1 + r s ) i sb + i ra sin e + i rb cos e] Therefore, i ms = [{(1 + r s )i sa + i ra cos e i rb sin e} 2 + {(1 + r s )i sb + i ra sin e + i rb cos e} 2 ] 1/2 {(1+r s )i sb + i ra sin e + i rb cos e} l = arctan {(1+r )i + i cos e i sin e} s sa ra rb (2.28) (2.29) (2.27)
i s
Eq.(2.28)
ims
Eq.(2.29)
d/dt
3/2
is ira
d ims dt
d/dt
ms e
3/2
irb
e j
e -j
ird irq
Fig.2.5 Flux computation and rotor current transformation blocks The d-axis controller block diagram along with the plant is given in Fig.2.6(a). All the parameters necessary for constructing the feedforward terms have already been computed in the & previous stage. After generation of the reference voltages u & rd and u rq , they are again transformed back to the rotor reference frame by multiplying with the inverse transformation operator e j (le ) . However, for the sake of simplification, transformation of u & rd from the field coordinates to the rotor reference frame in the controller, and the corresponding forward transformation in the machine model are not shown in the diagram. The plant model in the d-axis is shown with shaded blocks. It is evident from this diagram that the controller just adds or subtracts the disturbance inputs to the machine model reducing the plant to merely an integrator. Hence, the closed loop system behaves like a first-order lag circuit whose time-constant can be modified by changing the proportional gain of the system, as in Fig.2.6(b) and Fig.2.6(c). The q-axis current controller can be similarly modeled; the compensating terms will only be different in this case. In the controller block diagram the current sensor gain Kir is assumed to be unity for simplifying the diagrams. 30
Chapter 2 Modeling and Simulation The implementation of this controller is discussed in Chapter 4. The details for the practical implementation vary slightly from the theoretical design of the controller, even though the overall concept remains the same. (1r )L r di ms Gr dt i& rd + i rd rL r G r T ir + + + + R r i rd Gr + Gr + (z ms z e )rL r i rq + + + R r i rd i 1 1 rd s rL r
(z ms z e ) rL r i rq Gr (a)
(1 r )L r di ms dt
i& rd
rL r T ir (b)
1 1 rL r s
i rd
i& rd
1 1 + sT ir (c)
i rd
Fig.2.6 Block diagram of d-axis proportional rotor current controller (b) Proportional Integral Controller In the proportional controller, the steady-state error in the rotor currents will depend on the accuracy of computation of the feedforward terms. Any error in the compensation terms would result in slight modification of the dynamic response. In the practical implementation, it is extremely difficult to perfectly nullify the disturbance terms owing to measurement and computational errors. Therefore, a proportional integral controller needs to be incorporated. This is particularly important when the machine is run in the torque-control mode without any outer speed loop. The flux computation and transformations shown in Fig.2.5 remain identical. The plant is modeled as a first-order lag system with the two rotational emfs as disturbance inputs as shown in Fig.2.7(a). These two terms are canceled through feedforward compensation as before. In the controller, the PI 31
Chapter 2 Modeling and Simulation time-constant is made equal to rT r , so that the dynamics of the system is decided by the proportional gain, as shown in Fig.2.7(b) and Fig.2.7(c). The choice of proportional gain follows from the equation K pir = desired effective time constant of the current loop. (1r )L r di ms Gr dt i& rd + i rd K pir (1 + rT r s ) rT r Gr + + + Gr (z ms z e )rL r i rq + + (1 r )L r di ms dt i rd rT r T ir R r , where T ir is the
1 R r (1 + rT r s )
(z ms z e ) rL r i rq Gr (a) (1 + rT r s ) rT r
i& rd
K pir
1 R r (1 + rT r s )
i rd
(b) i& rd
1 1 + T ir s (c)
i rd
32
2 Vsb
3 Vsc
3ph to 2ph Stator Side 7 W 3ph to 2ph Rotor Side Rotor to Stator Electrical Subsystem P/2 Pole Pair * We Sin E, Cos E 3 4 Is3 Ir1 Current Inspection Block 6 Ir3 2 Is2 5 Ir2 1 Is1 Mechanical Torque Converter Subsystem 7 Ml
4 Vra
5 Vrb
6 Vrc
As an example, the modeling of the doubly-fed SRIM may be considered. The machine model consists of the transformation blocks, electrical subsystem, torque converter block and, the mechanical subsystem. This is shown in Fig.2.8. The transformation blocks include 3 phase to 2 phase transformation, rotor coordinate to stator coordinate transformation and, the corresponding inverse transformations (used in the current inspection block). The electrical subsystem is modeled in the stationary coordinate system. The state space equations (Eq.(A32)) to compute the stator and rotor currents in the stator reference frame are derived in Appendix A. The torque equation, given by Eq.(A33), is executed in the torque converter block. Finally, the mechanical subsystem computes the machine speed from the mechanical dynamics, as given by Eq.(A22d). The position of the rotor with respect to the stator is obtained through integration of the shaft speed in the sin e, cos e block. These modules with appropriate interconnections are grouped together to form the SRIM block in the system simulation model of Fig.2.9. In a similar manner, the other functional modules of Fig.2.9 are developed.
33
5 Ird* 8 w*
4 DC Bus
Inverter
Sensors
A speed controlled drive using grid-connected doubly-fed SRIM is simulated. Stator flux orientation, as discussed in the earlier sections, is employed. The d-axis and q-axis current controllers are designed in the field reference frame. For the speed loop, a PI controller is employed which & generates the q-axis / active current reference i & rq . The d-axis reference i rd is set in open loop. The machine parameters, sensor gains and, controller gains are given in Appendix B and Appendix C. In order to emulate the implementation of the controller in the DSP-based hardware platform, the controller module is modeled in per unit. The base values are selected appropriately as given in the MATLAB data file of Appendix C. The results are presented in per unit terms for the sake of uniformity. The speed response of the drive under no load is given in Fig.2.10(a). The motor is started DOL with the rotor shorted. At t=0.25s, the rotor side control is released with a speed reference z & =0.75 p.u. At t=1.25s, the speed reference is given a step change from 0.75 p.u. to 1.25 p.u. The corresponding motor torque and i rq are given in Fig.2.10(b) and Fig.2.10(c) respectively. The speed controller time constant is set to 100 ms. At t=1.75s, i & rd is given a step change to 0.75 p.u. This results in transfer of the reactive power from the stator to the rotor side. However, a change in i rd does not affect i rq , as can be seen from these plots. 34
Speed (p.u.)
0 .5 0 0
0 .5
1 secs
1 .5
1 .5
0 .5
-0.5
-1
-1.5
0 .5
1 secs
1 .5
35
0 .5
irq (p.u.)
-0.5
-1
-1.5
0 .5
1 secs
1 .5
1 .5
0 .5
ird (p.u.)
-0.5
-1
-1.5
0 .5
1 secs
1 .5
36
0 .5
irq (pu)
-0.5
-1 0 .7
0 .72
0 .74 secs
0 .76
0 .78
0 .8
0 .5
ird (pu)
-0.5
-1 0 .7
0 .72
0 .74 secs
0 .76
0 .78
0 .8
37
Fig.2.11(c) Corresponding simulated response of i s along with u s for the grid-connected SRIM
Fig.2.11(d) Corresponding simulated response of s i r along with u s for the grid-connected SRIM
38
0 .5
ird (pu)
-0.5
-1
1 .02
1 .04 secs
1 .06
1 .08
1 .1
0 .5
irq (pu)
-0.5
-1
1 .02
1 .04 secs
1 .06
1 .08
1 .1
39
Fig.2.12(c) Corresponding simulated response of i s along with u s for the grid-connected SRIM
Fig.2.12(d) Corresponding simulated response of s i r along with u s for the grid-connected SRIM
40
speed (p.u.)
1 0 .8 0 .6 0 .4 0 .2 0
0 .8
1 .2
1 .4 secs
1 .6
1 .8
Fig.2.13(a) Simulated response of speed through the synchronous speed for the grid-connected SRIM
0 .6 0 .4 0 .2
ir (p.u.)
41
0 .6 0 .4 0 .2
Ps (p.u.)
Fig.2.13(c) Corresponding simulated response of stator power for the grid-connected SRIM
0 .4 0 .3 0 .2 0 .1
Pr (p.u.)
0 .8
1 .2
1 .4 secs
1 .6
1 .8
Fig.2.13(d) Corresponding simulated response of rotor power for the grid-connected SRIM
42
The dynamic response of the current loops are given in Fig.2.11 and Fig.2.12. The q-axis current loop time constant is designed for 1ms and, the d-axis loop time constant is designed for 4ms. (The d-axis reference is not required to be varied dynamically; so the dynamic response of i rd need not be very fast.) The actual stator currents and, the rotor currents in the stator reference frame are also shown. In Fig.2.11 i rd is zero, so the stator supplies the reactive power and the rotor power factor is unity. In Fig.2.12, the reactive power is transferred from the stator to the rotor side, resulting in an improvement in the stator power factor. The transition through synchronous speed is shown in the plots of Fig.2.13(a) through Fig.2.13(d). With an initial speed of 0.6 p.u., the simulation is run with i & rq = 0.5 p.u. and a load torque of -0.88 p.u. (i.e. driving torque). Therefore, the stator power remains constant, while the rotor power goes from positive to negative. The transition of the rotor phase current through synchronous speed is also shown. Using the aforestated method of rotor current control the four modes of operation as described in chapter 1 can be achieved. During subsynchronous motoring and supersynchronous generation, the direction of rotor power flow is from the rotor circuit to the grid. Therefore, an ordinary diode bridge rectifier cannot be employed as the line side converter. A current controlled IGBT based PWM converter is used instead. In the following sections, the method of control for this front end converter and relevant simulation results are presented.
Chapter 2 Modeling and Simulation It is understood that by employing stator voltage orientation, the active and reactive currents at the input of the front end converter can be controlled in the synchronous reference frame. The control essentially has the same structure as the rotor side controller. The objectives of control and, modeling of the power and control circuit are described in the following sections.
u s1 u s2 u s3
Inverter udc + i _ l
PWM Signals
Feedback Signals
Digital Controller
The schematic block diagram of the front end converter is shown in Fig.2.14. The transformer in the input side is used to match the voltage levels between the dc bus and the ac side. The rotor side converter operates within a limited frequency range. Hence, the dc bus voltage requirement is less when compared to the stator side control schemes of cage rotor induction machine. The saving in the converter rating in the rotor side is achieved due to reduction in voltage rating of the power devices. With a reduced dc bus voltage it, however, becomes necessary to use a transformer at the input of the front end converter. Since the rotor side need not be isolated from the ac grid, an autotransformer can be used instead. This reduces the cost and weight of the equipment considerably. The PWM switching converter is connected to the secondary side of the transformer through series chokes. These inductors act as buffers between the two voltage sources. The choice of the values for these inductors depends on the switching frequency, allowable harmonics in the input current waveform and the reactive power requirement. 44
Chapter 2 Modeling and Simulation The objectives for the control of the converter are, i) Voltage regulation of the dc bus, ii) bi-directional power flow, iii) operation at any desired power factor, and iv) low current harmonics.
i fe
uac i fe x fe
uac
ufe
ufe i fe uac
(d) Reverse power flow at upf
ufe i fe x fe i fe uac
(e) Reverse power flow at leading power factor
i fe x fe
Fig.2.15 Equivalent circuit and phasor diagrams for the front end converter The primary objective of control is dc bus voltage regulation. A change in the dc bus voltage can be attributed to an imbalance between the active powers between the ac and dc sides. (The effect of reactive power on the dc bus is to produce ripples in the voltage even though the average value remains the same.) Hence, the voltage error in the dc side is an indication of the active power demand in the ac side. If the demand is positive, active power drawn from the grid needs to be increased; if
45
Chapter 2 Modeling and Simulation the demand is negative, it has to be fed back to the grid. Since the converter has bidirectional switches, current flow can be in either direction and it is possible to source or sink active power in the ac side. Fig.2.15(a) shows the single phase equivalent circuit of the ac side of the front end converter. If the load current in the dc side for a given dc bus voltage is known, the current drawn from the ac side at any desired power factor can be calculated applying power balance between the ac and dc sides. Consequently, subtracting the reactive drop from the source voltage, the magnitude and phase of the inverter terminal voltage with respect to the source can be computed. The inverter, therefore, acts as a fixed frequency source with controllable phase and magnitude. Fig.2.15(b) and Fig.2.15(d) represent steady-state phasor diagrams at unity power factor operation when power is flowing from ac to dc side and vice versa. The corresponding phasor diagrams for leading power factor operation are illustrated in Fig.2.15(c) and Fig.2.15(e). It is observed that the magnitude of the inverter terminal voltage increases in this case. The amount of reactive power that can be injected into the grid depends on the available dc bus voltage and the value of per unit inductance in the ac side. The terminal voltage of the inverter will also contain switching harmonics apart from the fundamental. As far as the harmonics are concerned the ac source acts as a short-circuit and the effective impedance to the harmonic current is nX fe . For high frequency switching (more than 1kHz), the harmonic impedance is quite high resulting in very low distortion of the ac side current waveform. With the above scheme of control it is possible to achieve the desired objectives as stated earlier. The salient features of the control strategy can be summed up as the following. 4 Employs an outer voltage control loop to regulate the dc bus voltage and an inner current control loop to control the ac side inductor current. 4 Outer voltage loop decides the value of inductor current to meet the active power balance between the two dc and ac sides. 4 The current loop tracks this reference by adjusting the inverter terminal voltage so that proper phase relationship between the supply voltage and the inductor current is maintained for a given power factor operation. 4 Employs current-controlled sinusoidal PWM for current tracking. 4 The current controller is designed in synchronously rotating reference frame with orientation being done with respect to the supply voltage space phasor.
46
Chapter 2 Modeling and Simulation Fig.2.16 shows the schematic block diagram of the control structure of the front end converter. The voltage controller is a proportional-integral controller with feed-forward of the load & or the reference for the active component of current. The reference for the current and generates i feq & , is set in open loop; e.g. it is set to zero for unity power factor reactive component of current i fed operation. The current controller operates in the synchronous reference frame and generates u & fed and u& feq for the inverter terminal voltage. These references are first transformed back to the stationary reference frame, and then from two phase to three phase quantities. Finally, the three phase references are compared with a triangular carrier to generate the PWM signals for the inverter switches. The modeling of the power circuit and design of the controllers are discussed in detail in the following sections. sin * udc + Voltage Contrl * ifeq + _ D-Axis Contrl Tranf. & PWM Genr. cos
udc
_ +
ifeq
Q-Axis _ Contrl
* ifed
ifed
Tranf.
udc
sin cos
Fig.2.16 Schematic block diagram of the control structure of the front-end converter
47
Chapter 2 Modeling and Simulation di u ac1 = u fe1 + L fe dt fe1 di u ac2 = u fe2 + L fe dt fe2 di u ac3 = u fe3 + L fe dt fe3 The above equations can also be represented in terms of space phasors as di u ac = u fe + L fe dt fe where, di u aca = u fea + L fe dt fea di u acb = u feb + L fe dt feb d Stator voltage space phasor q (2.33) (2.34) (2.35) (2.30) (2.31) (2.32)
Fig.2.17 Stationary and synchronous reference frame The stationary reference frame equations are transformed to synchronously rotating reference frame; the orientation being done with respect to the supply voltage space phasor as indicated earlier. The relative orientation of the stationary and synchronous reference frames is shown in Fig.2.17. To maintain compatibility with the d-axis and q-axis definitions used in rotor side control, the supply voltage phasor axis is taken as the q-axis in this case. The unit vectors cos h and sin h can be directly obtained from u aca and u acb respectively. In terms of the d-axis and q-axis variables Eq.(2.34) and Eq.(2.35) can be rewritten as follows. u acq e jh = ( u feq + j u fed ) e jh + L fe d dt or, u acq e jh = ( u feq + j u fed ) e jh + L fe d dt + j dh L fe i feq + j i fed e jh dt i feq + j i fed e jh i feq + j i fed e jh (2.36)
48
Chapter 2 Modeling and Simulation Eq.(2.36) describes the system dynamics in the stationary coordinates in terms of the synchronous reference frame variables. Since the orientation is done with respect to the grid voltage, u acd is zero. Transforming this to the rotating reference frame by multiplying both sides with e jh and separating the real and imaginary parts the following d-axis and q-axis equations can be obtained. di u fed + L fe dt fed + z s L fe i feq = 0 di u feq + L fe dt feq z s L fe i fed = u acq (2.37) (2.38)
Due to the transformation, rotational emf terms appear in the voltage equations in the synchronous reference frame, giving rise to cross-coupling between the two axes. These rotational emf terms are required to be compensated by appropriate feedforward signals in order to independently control the active and reactive components of current.
i dc i feq
2 . uacq 3 udc
ic i l udc
C
Fig.2.18 DC Bus Model The dynamics of the dc bus voltage can be modeled by considering the balance between the active power flow between the ac and the dc sides. If the series inductor is lossless, 2 u 3 acq $ i feq = u dc i dc (2.39)
where i dc is the dc bus current as indicated in Fig.2.18. This current can be written in terms of the capacitor charging current i c and, the load current i l as du i dc = i c + i l = C dc + i l dt Substituting i dc from Eq.(2.40) in Eq.(2.39) the following equation can be derived. C u acq du dc = 2 3 u dc $ i feq i l dt (2.41) (2.40)
Since the dc bus voltage is regulated within a narrow band and, the ac side voltage is u acq nominally constant, the factor u dc may be considered as a constant ratio to transform the 49
Chapter 2 Modeling and Simulation synchronous reference frame current to the dc link current. The model is schematically shown in Fig.2.18.
T ife is the desired current-loop time constant K ife is the current sensor gain. Substituting di feq from Eq.(2.42) in Eq.(2.38) gives the q-axis component of the instantaneous dt
inverter terminal voltage required to produce the desired current dynamics. L fe u feq = u acq + z s L fe i fed T K i& feq K ife i feq ife ife u dc The inverter can be modeled as a constant gain block G fe = 2$u tri (2.43) as explained in Section
2.3.2(a). The reference for the q-axis component of rotor voltage is, therefore, given by & = u feq u feq G fe u acq z L i L = G + s Gfe fed T K fe G i & feq K ife i feq fe fe ife ife fe
(2.44)
The plant along with the controller for the q-axis is shown in Fig.2.19. The current sensor gain K ife is taken as unity to simplify the diagram. Assuming same current loop dynamics for the d-axis, the reference for the d-axis component of the rotor voltage can be expressed as the following. & = u fed u fed G fe
50
Chapter 2 Modeling and Simulation = z s L fe i feq L T K fe G i & fed K ife i fed G fe ife ife fe (2.45)
If the impressed inverter voltages are in accordance with Eq.(2.44) and Eq.(2.45), the active and reactive components of the ac side current can be controlled independently. u acq z s L fe i fed G fe i& feq + i feq L fe T ife G fe + + G fe + + + u acq (a) i& feq L fe T ife 1 1 L fe s i feq
z s L fe i fed G fe
1 1 L fe s
i feq
(b)
& i feq
1 1 + T ife s (c)
i feq
Fig.2.19 Block diagram of q-axis proportional front end current controller So far it has been assumed that the resistive drop for the inductor is negligible, which in practice is a valid assumption. However, if a proportional controller is used, the inclusion of the resistive drop as compensating terms in (2.44) and (2.45) minimizes the steady-state error. b) Proportional Integral Controller The design of the proportional-integral controller is similar to that discussed in Section 2.3.2(b). The resistive drop is taken into consideration and the plant is represented by a first-order lag along with the cross-coupling and input voltage terms. The q-axis plant and controller are given in
51
Chapter 2 Modeling and Simulation Fig.2.20. The PI time-constant T pife is made equal to T fe (= L fe /R fe ) and the proportional gain K pife T is selected as T fe R fe , where T ife is the effective current-loop time constant. ife u acq G fe i& feq + _ i feq K pife (1 + T fe s ) T fe s G fe _ + + G fe + z s L fe i fed _ + + _ u acq (a) i& feq (1 + T fe s ) T fe s 1 R fe (1 + T fe s ) i feq
z s L fe i fed G fe
K pife _
1 R fe (1 + T fe s )
i feq
(b)
& i feq
1 1 + T ife s (c)
i feq
52
Chapter 2 Modeling and Simulation the dynamics of the current loop can be neglected while designing the voltage controller. The forward path transfer function of the system becomes K pvfe 1+sT vfe s2 CT vfe (2.46)
From the bode plot of the system, shown in Fig.2.21(b), it can be inferred that 1/T vfe has to be selected slightly lower than the desired bandwidth. The gain K pvfe is then adjusted to make the phase margin close to 90o. il u& dc + _ u dc + (1+T s ) + K pvfe T vfe vfe s 3 u dc 2 u acq i& feq i feq 2 u acq 3 u fe + il _ 1 sC u dc
dB Frequency (logscale) 1/T vfe 00 90 0 180 0 Fig.2.21(b) Magnitude and Phase plot of the voltage loop Frequency (logscale)
53
Chapter 2 Modeling and Simulation Finally, the dynamics of the capacitor voltage, given by Eq.(2.41), is modeled to get the dc bus voltage. The modeling of the entire system along with the feedbacks and controllers is given in Fig.2.23. The individual functional modules can be identified clearly from this diagram.
5 U1* 6 U2* 7 U3* 2 Vs1 3 Vs2 3 Ph Inverter AC Side Demodulator 2 3 Is1 4 Is2 1 Is3 Iload DC Side 1 Vc
4 Vs3
Fig.2.22 SIMULINK model of the three phase front end converter power circuit
3ph Source Uref Uref Voltage Controller Current Controller PWM Generator Power Circuit Load
Sensors
Fig.2.23 SIMULINK system model of the three phase front end converter The simulation results for the three phase front end converter under various transient conditions are given in Fig.2.24 through Fig.2.28. The parameters for the power circuit and the controller used in this simulation are taken to be the same as in the experimental setup. These are listed in a MATLAB file in Appendix C. The controller module is modeled in per unit in order to emulate the actual implementation of the controller. 54
Chapter 2 Modeling and Simulation The current loop dynamics is first investigated. The output of the voltage loop is disconnected and, step changes in the current references are given. The initial dc bus voltage is set to the rated value of 300V. The value of the capacitor is made much higher than the actual value of 4000 F to keep the bus voltage constant during the forced transients in the active and reactive current loops. The true response of the current loops can, therefore, be determined. It may be noted here that such decoupling of the voltage and current loops is not possible during the implementation. In Fig.2.24(a), responses of i feq and i fed for step change in i & feq from 0 to 0.5 p.u. with i& fed = 0, are shown. The designed current loop time constant is 2ms. The corresponding response of the ac side current is given in Fig.2.24(b) along with the supply voltage. Since i & fed = 0, the input power factor is observed to be unity. The reversal of active current from 0.5 p.u. to -0.5 p.u. under unity power factor operation is shown in Fig.2.25. Current responses for step change in i & fed from 0 to & = 0.5 p.u. are plotted in Fig.2.26. The input current waveform shows that the front 0.25 p.u. with i feq end operates at a leading power factor, thereby supplying reactive power to the source. There is, of course, a limit to the reactive current that can be injected. This depends on the dc bus voltage and the line side inductance value. Since the reactive voltage drop adds to the line voltage in the same phase, the maximum value of i fed can be written by the following equation. i fed,max = 2. 3 2.
u dc m us 22 max
zL
(2.47)
The decoupling of the dynamics of the d-axis and q-axis current loops is clearly evident form these plots. The initial charging of the dc bus voltage from 178V (i.e. 2.125) to 300V and, transient due to application of positive load of 0.75 p.u. is shown in Fig.2.27. The response of the dc bus voltage when the load is reversed from 0.35 p.u. to -0.35 p.u. is plotted in Fig.2.28. The voltage loop time constant is designed to be 100 ms.
55
Fig.2.24(b) u ac and response of i fe & Fig.2.24 Simulation results for step change in i & feq from 0 to 0.5 p.u. with i fed = 0.
56
Fig.2.25(b) u ac and response ofi fe & Fig.2.25 Simulation results for step change in i & feq from 0.5 p.u. to -0.5 p.u. with i fed = 0.
57
Fig.2.26(b) u s and response ofi fe & Fig.2.26 Simulation results for step change in i & fed from 0 to 0.25 p.u. with i feq = 0.5 p.u.
58
1 .2 1
udc (p.u.)
0 .8 0 .6 0 .4 0 .2 0
0 .5
1 secs
1 .5
2 .5
Fig.2.27(a) Response of u dc
1 .5
0 .5
ifeq (p.u.)
-0.5
-1
-1.5
0 .5
1 secs
1 .5
2 .5
Fig.2.27 Simulation results of response of dc bus voltage when a positive load of 0.75 p.u. is suddenly applied on the dc side
59
1 .2 1
udc (p.u.)
0 .8 0 .6 0 .4 0 .2 0
0 .5
1 .5 secs
2 .5
Fig.2.28(a) Response of u dc
1 .5
0 .5
ifeq (p.u.)
-0.5
-1
-1.5
0 .5
1 .5 secs
2 .5
Fig.2.28 Simulation results of response of dc bus voltage when the dc side load is reversed from 0.35 p.u. to -0.35 p.u.
60
2.11 Conclusion
A stator flux oriented model has been derived for the wound rotor induction machine. Current controllers designed in the field reference frame comprise proportional or proportional-integral controllers with subsequent addition or subtraction of the compensating terms. The design method is simple as it directly follows from the rotor voltage equations. Simulation results show that the dynamics of the active and reactive current loops are decoupled as required. The front end converter is modeled in the stator voltage reference frame. The structure of the current loops is similar to that for the rotor side control. The front end converter has been exhaustively simulated for forward and reverse power flow conditions. It is shown that leading power factor operation is possible upto a certain limit. The voltage controller exhibits excellent transient response during sudden impact of load on the dc bus.
61
Chapter 3
HARDWARE ORGANIZATION AND EXPERIMENTAL RESULTS FOR CONVENTIONAL FIELD ORIENTED ROTOR SIDE CONTROL AND FRONT END CONVERTER CONTROL
3.1 Introduction
A major emphasis of the present work is to develop a generalized hardware platform for high-performance ac drives. The system organization for rotor side control of doubly-fed wound rotor induction machine presents a versatile case where both the machine side and line side converters are necessary. In order to demonstrate the application of such a system to wind power generation, the wind turbine characteristics also need to be simulated with a dc drive. In this chapter a detailed description of the experimental setup is provided. The DSP-based software implementation of conventional field-oriented rotor side control with position sensors and, control of the front end converter are discussed. Finally, typical experimental results are furnished. The transient responses of the control loops are compared with those obtained through simulation to verify the system modeling and controller design.
63
ig1
ig3
us1 uac2
+ -
Encoder
us2 uac3
S1' S2' S3' S1 S2 S3
us3
ife3
64
TMS320F240 DSP based Digital Controller
DC machine Controller
DC Drive
65
Chapter 3 Hardware Organization and Experimental Results PFR 818 Status Signal HP3101 Sht ckt protection logic Collector
Gate Pulse
+15V SMPS
+15V
Fig.3.2 Schematic block diagram of IGBT gate drive circuit (b) Current Sensor Card Hall effect low profile current transformer Telcon HTP50 is used for current sensing [37]. Two current sensors are mounted on a single card. The sensor output is scaled by a linear amplifier to !10V. Extensive measurements are performed to evaluate the performance of the sensors. The sensor circuit offers very low non-linearity (maximum of 0.2%) and, high bandwidth (100 kHz). (c) Voltage Sensor Card A precision ac/dc voltage transducer is fabricated with a high CMR isolation amplifier HCPL-7800 [36, 38]. The isolated output of the transducer is scaled to!10V by using a differential amplifier stage. The measured non-linearity of the card is within 0.5% and, -3dB bandwidth is 20 kHz. (d) Protection and Delay Card This card accepts the gating pulses for the top devices in each leg of the converter from the digital controller. It then generates the complementary gating signals for the bottom devices and, introduces the dead-time between the two. The different sensor outputs and gate drive status signals are also routed to this board. Comparator logic is used to generate relevant protection signals like overcurrent, overvoltage, undervoltage etc. All these signals are then AND-ed using wired logic. The final enable signal is used for blanking out the base drive pulses under any fault condition.
66
Chapter 3 Hardware Organization and Experimental Results (e) Indicator Card The various protection signals from the protection and delay card are routed to this board. Any fault condition is indicated by switching on a particular LED in the front panel of the cabinet. The feedback signals for the control of the front end converter consist of the ac side currents, ac side voltages, the dc side load current and, the dc bus voltage. Apart from the ac side voltage transducers, all the other sensors are integrated within the front end converter cabinet. In the case of machine side converter, the rotor current sensors and the dc bus voltage transducer are integrated within the rotor side converter cabinet. However, the stator side voltages and currents also need to be measured. The ac side voltage for the front end converter differs from the stator voltage by the transformer turns ratio. In practice, a suitable tapping from the input transformer is used for measuring the ac side voltage. The stator voltage is then derived by scaling this signal appropriately within the software. The rotor side converter and the front end converter operate from 300V dc bus. This allows a u maximum rotor induced voltage of dc m max i.e. 106V for m max = 0.85. The maximum allowable 22 slip is therefore, 0.375 p.u. To ensure that the rotor circuit is open when the slip exceeds this limit, a contactor is introduced between the converter and the rotor terminals. This contactor opens the rotor circuit if the slip exceeds this limit or, the dc bus voltage exceeds 325V.
67
Chapter 3 Hardware Organization and Experimental Results architecture. In this section, a brief overview of the 'F240 DSP and the digital control board is first presented, so as to make the implementation details more clear later.
68
Digital I/O
FEC Analog Signals Scaling and R-C PAL 4 MHz Filter Scaled Analog Signals Power-on reset Quad DAC 32K SRAM Buffer Shifting Circuit
TMS320F240
69 XDS510
Chapter 3 Hardware Organization and Experimental Results (a) Analog Signal Conditioning Board The internal ADCs of the 'F240 are unipolar and operate between 0 to 5V. Since the sensor outputs (from the power circuit) are in the range of !10V, these signals need to be properly scaled before feeding them to the ADCs. This scaling is done in the analog signal-conditioning board. First, the !10V signals are scaled down to !2.5V; then they are shifted by adding a reference voltage of +2.5V. Finally, the signals are clamped between 0 and 5V with appropriate zeners for protection of the ADCs. A total of 10 analog channels can be handled by the signal conditioning board. (b) Processor Board The sensor outputs, appropriately scaled, are routed to the ADC inputs of the DSP in the processor board. First order R-C filters are provided for the purpose of noise elimination and anti-aliasing. For signal monitoring and debugging, a quad-channel 12-bit digital-to-analog converter (DAC) DAC4815 [41] is used. DAC4815 is bilpolar and operates between !10V. The variables of interest are properly scaled and output through the DAC. The PWM outputs and other digital I/Os from the 'F240 are buffered using 74ALS245. The PWM signals are routed to the inverter interface board. The QEP signals, after similar buffereing, are terminated on a connector, which interfaces to the external circuit associated with the incremental position encoder. The board has a total of 64K words of external, on-board memory. The memory is partitioned in the following manner: 32K words of EEPROM as external program memory and 32 K words of SRAM as external data memory. Since this digital hardware platform is targeted towards system development, provision of onboard RAM makes it convenient to modify the software and directly download on to the external memory. If the internal flash EEPROM is used the memory needs to be erased and burnt every time the software is modified; this is extremely inconvenient and time-consuming. However, the external memory needs to be fast so that the number of wait-states introduced can be reduced. In the design, 15 ns CY7C199 SRAM from Cypress [42] is used; hence additional wait-states need not be introduced. The use of high speed external memory, of course, necessitates the printed circuit board (PCB) to be mutilayered. (The present board is 4-layered; the two internal layers being the power plane and the ground.)
70
Chapter 3 Hardware Organization and Experimental Results A 4 MHz reference crystal is used externally, which in conjunction with the on-chip oscillator circuit generates the input clock to the internal PLL module. The PLL can be programmed to generate the required CPU clock. A power-on reset circuit is also included to drive the DSP into a pre-defined state after the power supply is switched on. Apart from the 5V supply required for the processor and buffers, the onboard DAC operates from !15V. A power supply unit of appropriate rating is designed for the processor board, analog signal conditioning board and, the incremental encoder circuit. The processor communicates with the PC through an emulator (XDS510). The emulator card is seated on the PC mother board and connects to the DSP hardware via the JTAG port. (c) Power Converter Interface The interface board provides a single-point connection between each converter unit and the digital control platform. The power circuit and the DSP hardware communicate through an FRC cable, which carries the PWM signals generated by the DSP and the analog feedback signals from the converter. There is also provision for transmitting the PWM signals through optical fiber links; however, in the present work this is not used. (d) Position encoder interface For determining the rotor position with respect to the stator, an incremental position encoder is mounted on the machine shaft. An encoder having 2500 pulses/rev from Stegmann is used. The encoder generates two trains of pulses through its two quadrature pulse channels and an index pulse through the third channel. It is mounted in such a way that, when the rotor a-phase coil axis coincides with the stator a-phase coil axis, the index pulse is generated. The test procedure for determining the coil axes and, accordingly, orienting the encoder is given in Appendix B. The quadrature encoder pulses and the index pulse have to be routed to the encoder interface connector in the processor board. However, it is observed that due to long length of the cable, the signals are prone to pick-up noises. Hence, at the encoder end, these signals are fed to 75172 current drivers [43]. The corresponding receivers 75173 [43] are physically located close to the processor board. These receivers convert the current signals into TTL voltage levels which can be directly interfaced to QEP channels of the 'F240.
71
The execution of the different tasks are driven by a timer interrupt. A general purpose timer of the 'F240 (Timer1) is initialized to generate an interrupt at every 56.8s (2048 X 27.7ns) (The 'F240 PLL is set to produce a CPU clock of 36 MHz, corresponding to a time period of 27.7 ns.) Each individual task gets executed in this time slot of 56.8s. To complete all the tasks, six time slots are required, corresponding to a time period of 340.8s. The FEC current control is executed in every alternate time slot. Therefore, the sampling rate of the FEC current loop is 56.8 X 2s i.e. 113.6s. The rotor side current control and FEC voltage control are executed once in 340.8s. The manner in which the tasks are actually organized in the different time slots are shown in Fig.3.4. The task scheduler keeps track of the present task that is 72
Chapter 3 Hardware Organization and Experimental Results being executed and switches to the subsequent one at the next Timer1 interrupt. This is implemented by using a software counter.
73
74
Chapter 3 Hardware Organization and Experimental Results Subroutine GET_Uref: & , u & ( Synchronous to stationary reference frame transformation using Compute u fe a feb sin h, cos h) & , u & , u & ( 2 phase to 3 phase transformation) Compute u fe 1 fe2 fe3 Subroutine UPDATE_PWM_FE: Update compare registers (of Full Compare Units) for PWM signal generation for FEC. (b) TASK2 Subroutine SOC_Is: Start of Conversion for ADC channels connected to the i s1 , i s2 . Subroutine GET_POS: Read internal timer associated with QEP circuit (Timer2). Scale timer value to get e. Get sin e, cos e. Compute rotor speed. Subroutine READ_Is: Read ADC FIFO to get i s1 , i s2 , adjust for channel offset and, and scale data. Subroutine SOC_Ir: Start of Conversion for ADC channels connected to the i r1 , i r2 . Subroutine READ_Ir: Read ADC FIFO to get i r1 , i r2 , adjust for channel offset and, and scale data. Subroutine GET_Ims: Compute i s3 . Compute i sa , i sb (3 phase to 2 phase transformation) Compute i r3 . Compute i ra , i rb (3 phase to 2 phase transformation) Compute i ra , i rb (Rotor to stator reference frame transformation using sin e, cos e) Compute i msa , i msb Compute i 2 ms Compute i ms (using stored square root table) 75
Chapter 3 Hardware Organization and Experimental Results Compute sin l, cos l Subroutine FLD_ORIENT: Compute i rd , i rq (Stationary to stator flux reference frame transformation using sin l, cos l) (c) TASK4 Subroutine SOC_UIdc: Start of Conversion for ADC channels connected to the u dc , i l . Subroutine GET_FF_TERMS: Compute feed-forward terms for rotor side control. Subroutine ROT_CURR_CONTRL: & Execute rotor side current control loop (using PI control) and generate u & rd , u rq . Subroutine GET_Uref_R: & Compute u & ra , u rb ( Stator flux to stationary reference frame transformation using sin l, cos l) & Compute u & ra , u rb ( Stationary reference frame to rotor reference frame transformation using sin e, cos e) & & Compute u & r1 , u r2 , u r3 ( 2 phase to 3 phase transformation) Subroutine UPDATE_PWM_RE: Update compare registers (of Simple Compare Units) for PWM signal generation for rotor side control. Subroutine READ_UIdc: Read ADC FIFO to get u dc , i l , adjust for channel offset and, and scale data. Subroutine VOLT_LOOP: Execute FEC voltage control loop (using PI control) and generate i & feq . The subroutines of related functionalities are included in the same file so that they can share the common variables locally. However, there are a few variables which need to be accessed by a large number of routines. These are defined globally in one place and referenced in the other files. All these files are assembled and linked together to form the executable file (.out file), which can be
76
Chapter 3 Hardware Organization and Experimental Results directly downloaded into the internal and external RAMs in the processor board. The memory allocation and mapping are defined in a command file (.cmd file) and are done through the linker. The implementation of the routines is a matter of coding detail and is not presented in the thesis. However, the generation of the unit vectors for the FEC and, for the rotor shaft position is slightly involved. The algorithm for these routines are briefly explained in the following subsections.
360
<
+
Count Value No. of Samples
us1
Positive ZCD
Ts / TCLK
sin [k] cos [k]
+
[k-1]
[k]
Look-up Table
Fig.3.5 Algorithm for generation of unit vectors synchronized to the grid In order to generate the unit vectors sin h, cos h synchronized to the grid voltage, the 'F240 general purpose timer (Timer3) is made use of. The CPU clock, suitably prescaled, acts as the clock to the timer and, it is programmed in continuous up-count mode. At every positive zero crossing of u ac1 , the timer value is read off and the timer is reset to zero. This count is, therefore, proportional to the time period of the supply voltage for the previous cycle and, is used to compute the increment in h i.e. Dh at every sampling instant over the present cycle. With this estimated value of h the unit vectors are then read off from a sine lookup table in the memory. This is schematically shown in Fig.3.5. To take an example, it may be assumed that the grid frequency was exactly 50 Hz in the previous cycle. Therefore, number of samples taken for the present cycle is 177 with a sampling period of 57s. This means that at the 177th sample, the pointer points to the extreme end of the lookup table. If the grid frequency slightly reduces, the duration of the present cycle will be more than 20 ms. Therefore, in the subsequent sampling interval the value of h calculated would be more than 3600 and the pointer
77
Chapter 3 Hardware Organization and Experimental Results would try to access a location beyond the table. This is not allowed. The value of h is checked at every sampling interval and, when it exceeds 3600, cos h is forced to zero and sin h is forced to -1. If on the other hand, the frequency increases over the present cycle, the zero crossing detection will occur before the pointer goes to the end of the table. In that case, there will be a small discontinuity in the unit vectors at positive zero crossing. Since the variation of the grid frequency is extremely slow, the number of samples gained or lost in every cycle is observed to be not more than 1 or 2. Hence, the error in the unit vectors near the positive zero crossing is negligible and, proper interlocking of the unit vectors with the supply voltage is ensured. It may be noted that if the unit vectors are derived directly from the phase voltages, the presence of harmonics results in their distortion in turn leading to distortion of the line current waveforms. The line voltage u ac1 and cos h are shown in Fig.3.6.
78
Chapter 3 Hardware Organization and Experimental Results the index pulse is generated, then the timer count value is proportional to the rotor shaft position at any instant of time. However, this is implemented in a slightly different manner.
Sampling interupt
Fig.3.7 Resetting logic of the timer in QEP circuit The index pulse from the encoder is connected to the capture input of the QEP module. The capture unit is also associated with Timer2. When a signal undergoes a desired transition at the capture input, the count value of Timer2 gets stored in a FIFO register (CAP3FIFO). This is illustrated in Fig.3.7. As shown in this diagram, CAPFIFO3 now contains the value T1. In the software, this event is signaled by setting a corresponding interrupt flag (EVIFRC), though the actual interrupt is disabled. At the subsequent sampling interrupt, the setting of EVIFRC is detected and the Timer2 counter is reset to the value (T2 - T1). The capture module is also reset simultaneously. This ensures that when the position information is read, the Timer2 count is always proportional to e. Subsequently, the timer value is appropriately scaled and, the unit vectors sin e, cos e are read off from the sine lookup table in the memory. In Fig.3.8, the rotor position e and the unit vectors are given.
79
80
(c) Channel 1- u s, channel 2- s i r , channel 3-i & rq Fig.3.9. Experimental results showing the transient response of the q-axis current control loop. A & step in i & rq is given from 0 to 0.5 p.u. and i rd is maintained at 0.75 p.u.
82
(c) Channel 1- u s, channel 2- s i r , channel 3-i & rd Fig.3.10. Experimental results showing the transient response of the d-axis current control loop. A & step in i & rd is given from 0 to 0.75 p.u. and i rq is maintained at zero.
83
& Fig.3.11(a) Simulated response of i rd and i rq for step change of i & rq from 0 to 0.5 p.u. with i rd held constant at 0.75 p.u.
& Fig.3.11(b) Simulated response of i rd and i rq for step change of i & rd from 0 to 0.75 p.u. with i rq held constant at zero.
84
85
& Fig.3.13(a) Experimental results showing steady-state waveforms for i & rq = 0.75, i rd = 0.5 at 1275 rpm (subsynchronous operation)
& Fig.3.13(b) Experimental results showing steady-state waveforms for i & rq = 0.75, i rd = 0.5 at 1435 rpm (synchronous operation)
86
(3.1)
u dc
Rl
u diode
u dc = 300V R l = 66W
u diode = 360V R se = 9W
The steady-state ac side voltage and current waveforms for forward power flow is given in Fig.3.15(a). The same plots for regenerative mode of operation are given in Fig.3.15(b). In both these cases, i & fed is kept at zero to demonstrate unity power factor operation. Leading power factor & kept at 0.25 p.u. is shown in Fig.3.15(c). The current waveforms are observed to operation with i fed be smooth without any perceptible ripple. This is because of the high value of the series inductors (0.66 p.u.) used in the experiment. The transient response of the dc bus voltage due to application of positive and negative loads is shown in Fig.3.16(a) and Fig.3.16(b) respectively. The designed voltage loop response of 100 ms is reflected in the observed waveforms. In Fig.3.16(c), the input side current waveform is shown due to reversal of dc load. Since, i fed is zero, and i feq goes from positive to negative through zero, the ac side current also has to go down to zero before changing its phase.
87
(a) Forward power flow at unity power factor (S1 on and S2 off)
(b) Reverse power flow at unity power factor (S1 on and S2 on)
(c) Forward power flow at leading power factor (S1 on, S2 off and i & fed = 0.25 p.u.) Fig.3.15 Experimental results showing steady-state waveforms u ac1 , i fe1
88
Fig.3.16 Experimental results showing dc bus voltage and currents during transient loading
89
Fig.3.17 Simulated transient response of u dc and i feq for sudden application of positive load
The simulation plots of dc bus voltage and active current transients due to similar positive loading are given in Fig.3.17 respectively. The experimental and simulation results show close resemblance establishing the validity of the system modeling and design of the controllers.
3.6 Conclusion
A power hardware platform for implementing the rotor side control strategies has been built. The hardware is designed in a modular fashion and has been standardized in the laboratory for general motor control applications. A TMS320F240 DSP based digital control board is also designed and developed. This platform is powerful enough to execute all the control loops associated with rotor side control and front end converter control. Conventional field oriented control using shaft position sensor and front end converter control is first implemented. Experimental results presented in this chapter show decoupled response for the active and reactive current loops. They are also in close agreement with the simulated waveforms. Several control algorithms are developed subsequently and are presented in the following chapters. All these algorithms are implemented using the same hardware setup.
90
Chapter 4
4.1 Introduction
For field oriented control of ac machines the position of the rotor with respect to the stator is necessary. This information is usually derived by mounting a suitable position encoder on the shaft of the machine. The performance of a vector controlled drive depends on the accuracy of the position information and hence on the accuracy and resolution of the position encoder. The use of a position encoder (incremental/ absolute/ resolver) introduces additional interfacing hardware between the instrument and the controller. These factors while adding to the cost simultaneously reduce the reliability of the drive. In doubly-fed induction machines, it is moreover necessary to mount the encoder in a specific orientation with respect to the stator. The preferred orientation would be such that an index pulse is generated when the rotor and stator 'a' phase coil axes coincide. This would ensure that the rotor position e is directly derived by counting the quadrature encoder pulses (discussed in Chapter 3). Hence, apart from the higher component cost, the cost of having precise mounting arrangement also needs to be considered. Quite naturally a major challenge to researchers in this area has been to eliminate the use of this encoder and, yet obtain similar dynamic performance.
91
Chapter 4 Rotor Side Control Without Position Sensors Position sensorless control of ac machines have attracted a lot of attention in recent times [13, 33, 34]. However, the major focus of activity has been restricted to cage rotor induction machines and permanent magnet synchronous machines due to their higher usage in industrial vector controlled drives. Doubly-fed wound rotor induction machines being mostly used in conventional slip-power recovery schemes, fast dynamic response is not so far required. Currently, the requirement for VSCF operation in applications like wind power generation has led to the use of field oriented control of such machines to independently control the active and reactive powers (discussed in earlier chapters). Hence, the rotor position information needs to be acquired. In wind power generation, there is a large physical separation between the generator (which is coupled to the turbine shaft through gears) and the power electronic equipment (which is at ground level). It is, therefore, desirable that there is minimum interface between the two and, for higher reliability, a control scheme without shaft position sensors. There are two major challenges in designing a position sensorless scheme for a doubly-fed wound rotor induction machine. The foremost requirement is that the algorithm should work stably at or near synchronous speed. The synchronous speed operation corresponds to zero rotor frequency; hence this is analogous to zero speed operation in case of cage rotor induction machine. The second criterion is that the algorithm should be able to start on the fly. It is understood that the rotor side control strategy operates over a restricted speed range. The rotor circuit is closed and the control is initiated when the speed rises above a minimum threshold. Hence, the position estimation algorithm should start while the rotor is already in motion, without the knowledge of any initial condition. The literature available on sensorless control of doubly-fed wound rotor induction machines is rather sparse and algorithms proposed do not address the aforementioned requirements simultaneously. A position sensorless algorithm, capable of starting on the fly and stable operation at or near synchronous speed has been developed and presented in this chapter.
92
Chapter 4 Rotor Side Control Without Position Sensors only variables that one can access are the stator currents and voltages whereas, in case of doubly-fed wound rotor induction machine, the rotor currents can also be measured directly. Thus precise information about another state variable is available. However, most of the earlier publications tend to overlook this fact and, relatively complicated schemes based on angle controllers have been proposed. In [17], the desired angle of the rotor current in the rotor reference frame is computed from the active and reactive powers in the stator circuit. The actual angle of the rotor current vector in the stator reference frame is simultaneously computed. These two variables are fed to an angle controller which generates the rotor frequency. The actual rotor currents are subsequently used for current control. The details of the angle controller and the rotor current control method are not presented in [17]; hence it becomes difficult to access the dynamic behavior of such an angle control method. The method proposed in [18], on the other hand, uses the rotor voltages and currents to design a torque angle controller. The torque of the induction machine can be expressed as a cross-product of the rotor flux and rotor current vectors. From the measured rotor currents and voltages, the rotor flux is first computed by integrating the PWM rotor voltage. Subsequently, the angle between the rotor flux vector and rotor current vector ( d ) is estimated. The reference angle d & is set by the torque demand decided by an outer speed control loop. The error between the reference and estimated torque angles d & d drives a voltage controlled oscillator (VCO) to generate the slip frequency. The VCO output is simultaneously integrated to generate an angle h s ; this is used for transformation of the rotor currents to the synchronous reference frame. Finally, the rotor current controller is designed in the synchronous reference frame. The major drawback of this scheme is the method employed for computation of rotor flux. The integration of rotor voltage at or near synchronous speed, is analogous to the integration of the stator voltage at or near zero speed in case of a cage rotor induction machine. Hence, similar problems of integrator saturation resulting in incorrect estimation of the rotor flux is inevitable. Use of this algorithm has to be restricted upto a certain minimum slip and operation through synchronous speed is not possible. The scheme proposed in [19] is by far the most comprehensive one available in the literature. The system developed (ROTODRIVE) is a commercial product aimed at the variable speed high power drive market (>300 kW). The objective is mainly to provide a wide speed range with a reduced size of the converter. The system is started with the stator circuit shorted and the rotor being fed from
93
Chapter 4 Rotor Side Control Without Position Sensors a PWM converter (Mode I). After the speed has reached 0.5 p.u., the stator is opened (Mode II) and, subsequently connected to the grid (Mode III). After this, rotor side field oriented control is employed upto a maximum speed of 1.5 p.u. With the stator flux remaining constant it is possible to maintain rated torque capability upto this maximum speed. The sensorless method proposed for Mode III uses coordinate transformations for estimating the rotor position. The stator voltage vector is taken as the synchronous reference frame. The stator currents are measured and the rotor current vector in the synchronous reference frame is estimated using the machine parameters. The rotor current vector is directly measured in the rotor coordinates. From this information, the angle between the stator and rotor axes is determined. This algorithm provides stable operation at or near synchronous speed and, can be started on the fly. However, the accuracy of the estimation process depends on machine parameters like L s , R s and the supply frequency. The proposed algorithm is also based on axis transformations. However, it is more direct and does not involve the synchronous reference frame. The dependence on machine parameters is also largely reduced.
Stator flux
0
Rotor axis
Stator axis
Fig.4.1 Location of different vectors in stationary coordinates The proposed sensorless algorithm can be explained with the help of Fig.4.1. Here, the rotor current vector is shown along with the stator and rotor axes. Seen from the stator coordinate system, i r makes an angle q1. The same vector makes an angle q2 with the rotor axis. The problem, therefore, is to compute q1and q2, so that e = (q1 q2) can be determined. With the knowledge of the stator
94
Chapter 4 Rotor Side Control Without Position Sensors flux and the stator currents, the rotor current in the stator reference frame i.e. s i r can be computed. In the rotor reference frame i r can be directly measured. From this information, the angle between the two reference frames can be computed by using simple trigonometric relations. In Fig.4.1, the stator voltage vector u s is also shown. Assuming that the stator resistance drop is negligible, the stator flux axis i.e. the d-axis is at quadrature to it. Hence, the stator flux magnetizing current vector i ms makes an angle (h - 900 ) with the stator axis. It is also assumed that the magnitude of i ms vector (denoted as i ms ) is already known. (The estimation of i ms is presented later). Therefore, the a, b components can be written as follows. i msa = i ms $ sin h i msb = i ms $ cos h (4.1) (4.2)
Using this value of i ms and the measured value of i s , the rotor currents can be computed in the stationary coordinates as i ra = i msa (1 + r s )i sa i rb = i msb (1 + r s )i sb |s i r | = 2 i2 ra + i rb 1/2 (4.5) (4.3) (4.4)
The unit vectors for s i r are given by cos q1 = i ra / | s i r | sin q1 = i rb / | s i r | (4.6) (4.7)
The rotor currents are directly measured in the rotor circuit and, the unit vectors for i r can be derived as cos q2 = i ra / |i r | sin q2 = i rb / |i r | (4.8) (4.9)
Equations (4.6), (4.7) and equations (4.8), (4.9) represent the unit vectors in the two reference frames; the former rotating at synchronous speed and, the latter at slip frequency. The unit vectors pertaining to the rotor position e = (q1 q2 ) can be now easily computed. 95
Chapter 4 Rotor Side Control Without Position Sensors sin e = sin(q1 q2 ) = sin q1 $ cos q2 sin q2 $ cos q1 cos e = cos(q1 q2 ) = cos q1 $ cos q2 + sin q1 $ sin q2 (4.10) (4.11)
It may be noted here that the unit vectors sin e and cos e corresponding to the rotor position suffice for executing the vector control algorithm. The actual angle need not be computed through inverse functions, as in [19].
4.3.1 Computation of i ms
The accuracy of this computation depends on the value of i ms , since the other quantities are directly measured. The stator flux can be calculated directly by stator voltage integration so that the variations in the grid voltage and frequency are taken into account. However, in order to estimate i ms , the magnetizing inductance L o is required. If there is a substantial boost in the grid voltage or, a dip in the grid frequency, L o is most likely to saturate. This will lead to an incorrect estimation of i ms . Also, the presence of distortion components in the grid voltage normally gives rise to integration problems. The objective is, therefore, to make the estimation process minimally dependent on any machine parameter and if possible, avoid integration of the stator voltage. Any change in the magnitude of the stator flux being much slower than the sampling frequency (2.9 kHz), i ms can be correctly estimated by adopting the following method of recomputation. First, i ms for the present sampling interval is computed by transforming the present rotor current sample to the stator coordinates using the unit vectors computed in the previous interval. This is formulated as follows. i ra [k ] = i ra [k ] $ cos e[k 1 ] i rb $ sin e[k 1 ] i rb [k ] = i rb [k ] $ cos e[k 1 ] + i ra $ sin e[k 1 ] i msa [k ] = (1 + r s ) $ i sa [k ] + i ra [k ] i msb [k ] = (1 + r s ) $ i sb [k ] + i rb [k ] i ms =
2 + i 2 i ms a msb
1/2 (4.16)
The superscript ' indicates intermediate variables used in the computation. i ms as calculated from Eq.(4.16) is passed through a low-pass first-order filter with a time-constant of 1 ms. This ensures 96
Chapter 4 Rotor Side Control Without Position Sensors that even if there is any small error in the previous sample of sin e and cos e, it is not directly reflected in the present i ms estimate. The estimation of i ms is the first step in the position estimation algorithm. With this value of i ms , the algorithm proceeds from Eq.(4.1) till Eq.(4.11).
4.3.2 Starting
It is understood that the algorithm has to start with a known value of e in order to compute i ms by using Eq.(4.12) through Eq.(4.16). Since the rotor side control needs to be started on the fly, it is not possible to assign an initial value of e. Instead, the algorithm starts with an initial value of i ms , which is the same as its nominal value given by u s /(z s L o ). The position of i ms is computed from the stator voltage phasor as before. After a few sampling intervals, the algorithm switches over to the recomputation method. The estimation process thus becomes independent of variations in the stator voltage and frequency. The only machine parameter on which the algorithm depends is the stator leakage factor r s . The leakage is only a small percentage of the stator inductance and is not subjected to any saturation. Even a significant error in the value of r s does not introduce any appreciable error in the estimation of sin e and cos e. This is verified through extensive simulation. The instantaneous nature of computation also ensures jitter-free estimation during transients in the active and reactive power.
The usual method of differentiation of rotor position would require e to be computed from the unit vectors using inverse trigonometric functions. This is avoided in this method of speed estimation. Moreover, the unit vectors are smoothly varying continuous functions unlike e (which is discontinuous at e = 2o) and, are easier to differentiate without checking for discontinuity. However, the differential terms contribute to some noise which is eliminated by employing a first-order low-pass filter. The position and speed estimation block diagrams are shown in Fig.4.2. 97
sin 2
i rb
ir | 1/ | __
us sin
Compute
=
sin 1 1 __ irs | | __ cos 1 1 2 cos
us cos i ms i ms i ms i ms - (1+ s ) i s
Compute angle of
i ms
i ms
i ra
i rb
Compute magnitude
is
of
is
-1
i ms
98
sin d/dt
X
est
cos - d/dt
4.4 Simulation
The simulation of the system is carried out in MATLAB-SIMULINK platform to study the starting and, the effect of parameter variation on the proposed algorithm. The simulation block diagram is the same as given in Chapter 2; the controller now incorporates a position estimation block as shown in Fig.4.3. The same machine and controller parameters are selected as used in the laboratory experimental setup [Appendix C]. The plots of estimated and actual sin e during starting are given in Fig.4.4 through Fig.4.5. The rotor side control is released with an active current reference of i & rq = 0.5p.u. and, with different initial speeds. In these runs, the speed is held constant by increasing the system inertia. It is observed that the estimated position catches up with the actual position almost instantaneously irrespective of the initial speed. The steady-state relations between sin q1 and sin q2 for z = 0.75 p.u. and z = 1.0 p.u. are shown in Fig.4.6(a) and Fig.4.6(b) respectively. At synchronous speed sin q2 is perfectly dc, implying that the rotor currents in the rotor reference are also dc. This illustrates the stable steady-state operation at synchronous speed.
1 Us1f 2 Us2f 3 Ir1f 4 Ir2f 5 Is1f 6 Is2f Stator to Field
9 Usqf
8 Usdf
3ph to 2ph
Flux Estimator
1 Irdf 2 Irqf
0 .5
sinE_est
-0.5
-1
-1.5 0 .8
0 .82
0 .84 secs
0 .86
0 .88
0 .9
0 .5
sinE
-0.5
-1
-1.5 0 .8
0 .82
0 .84 secs
0 .86
0 .88
0 .9
Fig.4.4(b) Actual sin e & Fig.4.4 Estimated and actual sin e at starting with i & rq = 0.5 p.u.and i rd = 0.75 p.u. The initial speed is set to 1 p.u. and, the inertia is made high so that the shaft speed does not change during this transient.
100
0 .5
sinE_est
-0.5
-1
-1.5 0 .4
0 .42
0 .44 secs
0 .46
0 .48
0 .5
0 .5
sinE
-0.5
-1
-1.5 0 .4
0 .42
0 .44 secs
0 .46
0 .48
0 .5
Fig.4.5(b) Measured sin e through encoder & Fig.4.5 Estimated and actual sin e at starting with i & rq = 0.5 p.u.and i rd = 0.75 p.u. The initial speed is set to 1.25 p.u. and, the inertia is made high so that the shaft speed does not change during this transient. 101
Fig.4.6(b) sin q1, sin q2 at synchronous speed Fig.4.6 Steady-state waveforms of sin q1, sin q2 at subsynchronous (0.75 p.u.) and synchronous & speed (1 p.u.) for i & rq = 0.5 p.u.and i rd = 0.75 p.u.
102
0 .5
sinE, sinE_est
-0.5
-1
-1.5 0 .25
0 .255
0 .26 secs
0 .265
0 .27
0 .5
sinE, sinE_est
-0.5
-1
-1.5 0 .31
0 .315
0 .32 secs
0 .325
0 .33
Fig.4.7(b) Estimated (solid lines) and actual sin e (dotted lines) Fig.4.7 Estimated and actual sin e when (a) value of r s used in computation equals 1.5 times the actual and, (b) value of r s used in computation equals 0.5 times the actual value
103
1 .2
1 0 .8 0 .6 0 .4 0 .2 0 0 .2
0 .25
0 .3
0 .35
0 .4 secs
0 .45
0 .5
0 .55
1 .2
1 0 .8 0 .6 0 .4 0 .2 0 0 .2
0 .25
0 .3
0 .35
0 .4 secs
0 .45
0 .5
0 .55
Fig.4.8 Estimated speed before and after the low-pass filter when the estimation algorithm is started with an actual rotor speed of 1 p.u.
104
0 .75
ims (p.u.)
0 .7
0 .65
0 .6 0 .2
0 .25
0 .3
0 .35
0 .4 secs
0 .45
0 .5
0 .55
0 .75
ims_est (p.u.)
0 .7
0 .65
0 .6 0 .2
0 .25
0 .3
0 .35
0 .4 secs
0 .45
0 .5
0 .55
Fig.4.9(b) Estimated i ms during starting & Fig.4.9 Actual and estimated i ms during starting with i & rq = 0.5 p.u. and i rd = 0.75 p.u.
105
Chapter 4 Rotor Side Control Without Position Sensors Fig.4.7(a) and Fig.4.7(b) shows the effect of variation in r s in the estimation process. In the first case, r s used in the estimation algorithm is 1.5 times the actual value and, in the second case, it is 0.5 times. The plots show almost negligible errors in the computation of the unit vectors, even at starting. The estimation of speed is shown in Fig.4.8. The estimated speed signals before and after the low-pass filter are given in Fig.4.8(a) and Fig.4.8(b). The filter time constant is set to 25 ms. In Fig.4.9(a) and Fig.4.9(b), the actual and estimated waveforms of i ms are plotted during starting with an active current reference of i & rq = 0.5 p.u. and initial speed of 1 p.u. The increase in i ms magnitude is attributed to the stator resistive drop. (It may be appreciated that for a small laboratory scale machine the stator resistance drop is relatively high; however, in a practical case, rotor side control is only employed for wound rotor machines of ratings greater than 100 kW where the stator resistance drop can be neglected.)
106
Fig.4.10 Experimental waveforms showing sin q1, sin q2 and sin e for z = 1190 rpm
Fig.4.11(a) Experimental waveforms showing estimated and actual sin e at starting for z = 1460 rpm
Fig.4.11(b) Experimental waveforms showing estimated and actual sin e at starting for z = 1600 rpm
107
Fig.4.12 Experimental waveforms showing estimated and actual sin e for step in i & rq from 0 to 0.5 p.u
Fig.4.13(a) Experimental waveforms showing estimated and actual z at starting before filtering
Fig.4.13(a) Experimental waveforms showing estimated and actual z at starting after filtering
108
& Fig.4.14(a) Experimental results showing step in i & rq from 0 to 0.5 p.u. with i rq = 0.75 p.u.
& Fig.4.14(b) Experimental results showing step in i & rd 0 to 0.75 p.u. and i rq = 0
109
Fig.4.16(a) Experimental results showing steady-state u s1 , i s1 , i r1 with i & rd = 0.75 p.u. and i& rq = 0.5 p.u. at 1620 rpm (supersynchronous operation)
Fig.4.16(b) Experimental results showing steady-state u s1 , i s1 , i r1 with i & rd = 0.75 p.u. and i& rq = 0.5 p.u. at 1428 rpm (synchronous operation)
110
Chapter 4 Rotor Side Control Without Position Sensors The estimated speed signals (from sin e and cos e) before and after filtering are shown in Fig.4.13(a) and Fig.4.13(b). During the initial period there is a large error in the estimated speed due to the filter time-constant. If this incorrect value of estimated speed is used to determine the slip-dependent cross-coupling terms in the rotor current control, it will give rise to undesired transients and, in turn, erroneous estimation. In practice, the estimated position would not be able to catch up with the actual position. Hence, during starting the computed slip is forced to zero for about 100 ms; after this the estimated speed signal is used to calculate the slip. The transient response of the q-axis and d-axis rotor current loops are shown in Fig.4.14(a) and Fig.4.14(b) respectively. The responses are observed to be identical to those shown in Chapter 3, thereby establishing the fact that, with the position estimation algorithm the same dynamic performance can be achieved. In Fig.4.14(b), the response is taken during switching on of the rotor converter; hence a small transient is observed in the rotor currents (before the estimated position catches up with the actual one). Fig.4.15 shows the estimated i ms before and after the low-pass filter. In Fig.4.16(a) and Fig.4.16(b), the steady-state stator and rotor currents in their own reference frames are shown along with the stator voltage waveform for supersynchronous and synchronous modes of operation. The reactive power is supplied from the rotor side; hence the stator power factor is unity. In Fig.4.16(b), the rotor currents are dc showing that the estimation algorithm operates stably at zero rotor frequency. The transition through synchronous speed is also observed to be smooth, as is illustrated in Fig.4.17.
4.6 Conclusion
Position sensorless control of wound rotor induction machine is a desirable feature of VSCF generation systems like wind power generation. The proposed position sensorless algorithm meets the requirements for such applications. The algorithm can be started on the fly without the knowledge of the initial rotor position. Operation at synchronous speed, corresponding to zero rotor frequency, is stable; also it can ride through synchronous speed smoothly. The proposed method of computation of stator flux magnetizing current makes the estimation process independent of critical machine parameters. The simulation and experimental results show that the dynamic performance of the system compares with that using position sensors.
111
Chapter 5
5.1 Introduction
In field oriented control technique, the transient response of the active and reactive powers is dependent on the degree of decoupling between the direct and the quadrature axes. This, in turn, depends upon the accuracy of computation of the stator flux magnetizing current and accuracy of rotor position information. As proper alignment of the position encoder is difficult in doubly-fed machine, sensorless methods as discussed in chapter 4 are employed. These methods which make use of field oriented control require mathematical computations involving coordinate transformation and parameter estimation. An alternative approach may be considered where, instead of the rotor current, the rotor flux is directly controlled to control the active and reactive power flow in the machine. Direct self control (DSC) of induction motor has been proposed [20] where the stator flux is controlled to track a hexagonal trajectory. The switching scheme is such as to control the torque within a defined band. Direct torque control (DTC) schemes have also been proposed [21-23]; the primary difference from the earlier method being a circular trajectory of the stator flux. Two hysteresis controllers, namely a torque controller and a flux controller, are used to determine the switching states for the inverter. The method of control is computationally simple and, does not require the rotor position information. However, the problem associated with low frequency sensorless operation exists.
113
Chapter 5 Direct Power Control So far, the application of direct torque control has been primarily restricted to cage rotor induction motors and, permanent magnet synchronous motors [13]. In this chapter, an algorithm is proposed which extends the switching concepts of DTC to rotor side control of doubly-fed wound rotor induction machine. Here the directly-controlled quantities are the stator active and reactive powers; hence, the algorithm is referred to as direct power control in this text. The sector in which the rotor flux is presently residing is identified and the switching vectors are selected to control its trajectory in a desired manner with respect to the stator flux. The sector information is updated in a novel way based on the direction of change of the reactive power due to the application of a switching vector. This method is inherently position sensorless and does not use any machine parameter in the computation. The concept of direct power control is first introduced. The details of the control strategy are subsequently presented with relevant simulation and experimental results.
isq
s
is
is us
s
ir ur
isd ird ims
s m r
L0
d-axis
q-axis us isq is
q-axis us isq
isd ims B s
p m
d-axis
ims = ird
p
s m
d-axis
r
A irq (a)
C irq (b)
Fig.5.3 Phasor diagrams showing variations in rotor flux with change in active and reactive powers
The effect of injection of these rotor currents on the air-gap and rotor fluxes can be derived by subtracting and adding the respective leakage fluxes. The variation of the rotor flux with variations in the active and reactive power demand is shown in Figs.5.3(a) and Fig.5.3(b). In Fig.5.3(a) i rd = 0, i.e. the reactive power is fed completely from the stator side. Under this condition if i rq is varied from 0 to full load, the locus of y r varies along A-B which indicates a predominant change in angle d p between y s and y r , whereas the magnitude of y r does not change appreciably. In other words, a change in the angle d p would definitely result in a change in the active power handled by the stator in a predictable fashion. For example, in Fig.5.3(a) which indicates motoring mode of operation, the active power can be increased by decelerating the rotor flux with respect to the stator flux. Conversely it can be reduced by accelerating the rotor flux. In Fig.5.3(b) the stator active power demand is maintained constant so that i rq is constant and i rd is varied from 0 to the rated value of i ms . Here the locus of y r varies along C-D, resulting in a predominant change in magnitude of y r , whereas the variation of d p is small. Therefore, the reactive power drawn from the grid by the stator can be reduced by increasing the magnitude of the rotor flux and vice-versa. It may be noted that the phasor diagrams as indicated in Figs.5.3(a) and 5.3(b) remain the same irrespective of the reference frame; the frequency of the phasors merely changes from one reference frame to the other. It can be concluded from the above discussion that; 115
Chapter 5 Direct Power Control i) The stator active power can be controlled by controlling the angular position of the rotor flux vector. ii) The stator reactive power can be controlled by controlling the magnitude of the rotor flux vector. These two basic derivations are used to determine the instantaneous switching state of the rotor side converter to control the active and reactive power as discussed in the following section.
S1 (1 0 0)
S2 (1 1 0)
S3 (0 1 0)
S4 (0 1 1)
S5 (0 0 1)
S6 (1 0 1)
S0 (0 0 0)
S7 (1 1 1)
116
Chapter 5 Direct Power Control Fig.5.4 shows the 8 possible switching states of a three phase VSI of which six are active states (S1, S2,....S6) and two are zero states (S0, S7). Assuming that the orientation of the three phase rotor winding in space at any instant of time is as given in Fig.5.5(a), the six active switching states would correspond to the voltage space vectors U1, U2 ....U6 [Fig.5.5(b)] at that instant. In order to make an appropriate selection of the voltage vector the space phasor plane is first subdivided into six 600 sectors I,II..VI. The instantaneous magnitude and angular position of the rotor flux space phasor can now be controlled by selecting a particular voltage vector depending on its present location. The effect of the different vectors as reflected on the stator side active and reactive powers, when the rotor flux is positioned in Sector 1 is illustrated in the following subsections.
U3 S3 (010) Phase b Sector 3 U2 S2 (110) Sector 2
S4 (011) Phase a
Sector 4 U4
Sector 1 U1 S1 (100)
Sector 5 Phase c S5 (001) U5 Fig.5.5(a) Orientation of the rotor winding in space with respect to which the voltage space phasors are drawn
Sector 6 S6 (101) U6
subsyn
subsyn
s
U4
p
r
U4
p
U1
U1
s
supersyn U5 (b) U6
supersyn U5 (a) U6
Fig.5.6 Flux vectors in (a) motoring mode and (b) generating mode
In the motoring mode of operation in Sector I, application of voltage vectors U2 and U3 accelerates y r in the positive direction. This reduces the angular separation between the two fluxes resulting in a reduction of active power drawn by the stator. At subsynchronous speeds, U2 and U3 cause y r to move in the same anti clockwise direction as y s ; hence the effect on P s depends on the difference between the angular velocities of the two fluxes. The factors effecting the angular velocities of the fluxes y s and y r are the slip speed and the dc bus voltage respectively. In the rotor reference frame, y s rotates at slip speed and the rate of change of y r depends on the dc bus voltage and the applied inverter state. So, for a given bus voltage, higher the slip lesser is the relative angular velocity between the two flux vectors, thereby effecting a slower change in P s and vice-versa. At supersynchronous speeds the relative velocity is additive and change in P s is faster. In the generating mode of operation, application of vectors U2 and U3 result in an increase in angular separation between the two and thereby an increase in the active power generated by the stator. (P s being negative for generation, U2 and U3 still results in a reduction of positive active power). The relative speeds of the vectors in subsynchronous and supersynchronous generation are same as in motoring operation; hence the same conclusions can be drawn. Similarly it can be seen
118
Chapter 5 Direct Power Control that the effect of U5 and U6 on the active power would be exactly opposite to that of U2 and U3 in both the motoring and generating modes. Power drawn by the stator being taken as positive and power generated being taken as negative, it may be concluded that, if the rotor flux is in the kth sector, application of vectors U(k+1) and U(k+2) would result in a reduction in the stator active power and application of vectors U(k-1) and U(k-2) would result in an increase in the stator active power.
119
Chapter 5 Direct Power Control same conclusion holds true for the generating modes as well. The rate of change of P s depends on the slip speed alone as y r remains stationary in the rotor reference frame.
Subsynchronous
Supersynchrnous
Chapter 5 Direct Power Control powers (P s and P r ). From this set value and the present speed, the reference torque m & d can be computed. The reference for the stator power can, therefore, be calculated as & P& s = md $ zs Q& s is set according to the desired power factor at the stator terminals. (5.1)
where
i sa = 3 2 i s1 i sb = 3 2 i s1 + 2i s2
121
Chapter 5 Direct Power Control && P s P err = P s if (P err > 0 ) & P && s = P s + P band else & P && s = P s P band In a similar manner the error and reference for the reactive power can be written as Q err = Q && s Qs if (Q err > 0 ) & Q && s = Q s + Q band else & Q && s = Q s Q band P** s P* s Ps A C && Fig.5.7 P s , P & s , Ps B P band (5.9) (5.8)
122
Chapter 5 Direct Power Control Table 5.2(b) Selection of active switching states when (P err > 0 ) Sector 1 Qerr > 0 Qerr <= 0 S5 S6 Sector 2 S6 S1 Sector 3 S1 S2 Sector 4 S2 S3 Sector 5 S3 S4 Sector 6 S4 S5
If the rotor side converter is switched in accordance to these tables it is possible to control the active and reactive powers in the stator side within the desired error bands. But the use of active vectors alone would result in non-optimal switching of the converter and also a higher switching frequency. The effect of the zero vectors on P s and Q s has been summarized in Table 1. Since the zero vectors affect both these parameters the usual logic for zero vector selection to enhance/reduce the torque as used in direct torque control cannot be applied here. The algorithm for incorporating the zero vector logic is as follows. if (P & s m 0) { if (z [ z s ) { if (Q err m 0 && P err m 0) Sn = Sz else Sn = Sa } ;supersynchronous motoring Sn = Sz else Sn = Sa } } else {if (z [ z s ) { if (Q err < 0 && P err m 0) Sn = Sz else Sn = Sa } ;supersynchronous generation Sn = Sz else Sn = Sa } } else {if (Q err m 0 && P err < 0) ;subsynchronous generation else { if (Q err < 0 && P err < 0) ;subsynchronous motoring
Here, S n represents the switch state to be selected, S z represents a zero state and S a an active state.
123
Chapter 5 Direct Power Control It has already been mentioned that the effect of zero vector is primarily on the active power; the effect on reactive power is minimal. Also, it is observed that the effect on P s is opposite in the subsynchronous and supersynchronous modes of operation. This criterion is used in detecting the transition from subsynchronous to supersynchronous operation and vice-versa. It can be illustrated with an example. It is assumed that the machine is operating in subsynchronous generation mode. Therefore, the use of a zero vector increases P s and consequently P err should decrease. The amount of reduction in P err depends on the slip speed (for a constant dc bus voltage). When the slip becomes negative, P err will start increasing (instead of reducing) with the application of a zero vector. This direction of change of P err is detected and it is inferred that the mode of operation has now changed to supersynchronous generation. The zero vector logic is then modified accordingly. The choice between S0 and S7 is done depending on the minimum inverter switching. For example while switching to a zero vector from S1, S0 is selected. On the other hand if the transition to the zero vector is from S2, S7 is selected. Both these transitions then would result in switching of only one arm of the inverter. It may be concluded that these switching strategies would result in close tracking of P & s and Q& s within the prescribed error bands using near-optimum switching of the rotor side converter.
124
Chapter 5 Direct Power Control Q s would also be opposite. These reversals in the direction of change of Q s , when a particular vector is applied, can be detected and a decision of sector change may be taken on this basis. Similarly, if the flux vector is rotating in the clockwise direction (supersynchronous operation) the effect of states S2 and S5 on Q s would change in direction when y r changes over from Sector 1 to Sector 6. Thus in any particular direction of rotation there are two vectors which can provide the information for sector change. Since the rotor flux vector cannot jump through sectors the change will always be by one sector, either preceding or succeeding. In this method, even though the exact position of the flux is unknown, the sector information can be updated just by observing the changes in Q s due to the applied vectors. It may be noted that the effect of the vectors on P s would not provide a conclusive inference about the change in sector. The expected direction of change in Q s due to the application of any switching vector in the different sectors can be summed up in the following table.
Table 5.3 Expected direction of change in Q s . S0 Sector 1 Sector 2 Sector 3 Sector 4 Sector 5 Sector 6 0 0 0 0 0 0 S1 + + + S2 + + + S3 + + + S4 + + + S5 + + + S6 + + + S7 0 0 0 0 0 0
Note: + indicates increment in Q s , - indicates decrement in Q s , 0 indicates no change (however, application of the zero vectors will result in some small changes, but zero vectors are not taken into consideration to infer sector changes).
It may, however, be noted that in a particular sector not all vectors will be applied. For example, in sector k, vectors U(k) and U(k+3) will never be applied. These vectors would have predominant effect on the reactive power, but their effect on the active power would depend on the actual position of the rotor flux vector in the sector. In most applications there is hardly any requirement for fast transient changes in reactive power; so it is not necessary to apply the strongest vector to effect any change in Q s . In the switching logic, therefore, only those vectors are selected 125
Chapter 5 Direct Power Control which have uniform effects on P s and Q s in terms of their direction of change irrespective of the position of the rotor flux in a particular sector. For any given vector applied in a particular sector the expected direction of change in Q s can be read off from Table 5.3. The actual direction of change can be computed from the present value of Q s and its previous value. If they are in contradiction then a decision on change of sector is taken. Whether the sector change has to be effected in the clockwise or anti clockwise direction depends on the applied vector and the observed change in Q s . This information is stored in another lookup table as furnished below.
Table 5.4 Direction of change of sector S0 Sector 1 Sector 2 Sector 3 Sector 4 Sector 5 Sector 6 0 0 0 0 0 0 S1 0 +1 -1 0 +1 -1 S2 -1 0 +1 -1 0 +1 S3 +1 -1 0 +1 -1 0 S4 0 +1 -1 0 +1 -1 S5 -1 0 +1 -1 0 +1 S6 +1 -1 0 +1 -1 0 S7 0 0 0 0 0 0
Note: 0 indicates no change, + 1 indicates the sector has to be updated to its next value in the anti clockwise direction, -1 indicates the sector has to be updated to its previous value.
To illustrate the algorithm with an example it may be assumed that the rotor flux vector is presently residing in Sector 1 and rotating in the anti clockwise direction (corresponding to subsynchronous speed operation). As long as the flux vector is within the boundary of Sector 1, the direction of change in Q s will be as expected and the computed direction will match with that stored in Table 5.3. The quantitative change in Q s due to the effect of the vectors will obviously depend on the position of y r in the sector but the direction of change should be in accordance with this table. In this example, since y r is rotating in the anti clockwise direction the most widely used active states in Sector 1 will be S2 and S3. When the flux vector has crossed over to Sector 2, S2 will have a more pronounced effect on Q s in the same direction as in Sector 1, but the effect of S3 will reverse its direction. The application of S3 will cause a decrement in Q s whereas in Sector 1 it is expected to increase. Hence, the computed direction of change will be opposite to that stored in Table 5.3. When 126
Chapter 5 Direct Power Control this is detected the corresponding entry in Table 5.4 is looked at. For Sector 1 and switch state S3 the entry in Table 4 indicates a positive change in sector. Hence it is updated to Sector 2. Similarly it can be verified that if the flux vectors are rotating in the clockwise direction (corresponding to supersynchronous speed) the most commonly used active states will be S6 and S5. When y r crosses over to Sector 6, S6 will have a predominant effect on Q s in the same direction as in Sector 1, but S5 will cause Q s to reduce instead of increasing it. This direction of change in Q s is detected from Table 5.3, and the corresponding entry in Table 5.4 indicates a change in sector in the negative direction. Hence the sector information is updated from Sector 1 to Sector 6. For reliable detection of the direction of change of Q s a minimum switching period of 6 sampling periods (336 s) of a particular switching state is maintained. This also puts a maximum switching limit of 4.5 kHz for the rotor side converter. This method of sector identification is independent of any machine parameter but relies on directly measurable fixed frequency quantities. It is also independent of the rotor frequency and can work stably at or near synchronous speed.
5.6 Starting
Before the rotor side converter is switched on, the entire reactive power is drawn from the stator side. Initially Q & s is set to the computed value of Q s after passing it through a low-pass filter with a time constant of 100 ms. Thereby it is ensured that at the instant of switching the rotor converter, Q err is within Q band and the sector estimation algorithm can be used for correcting to the appropriate sector. Since, the minimum switching period on the basis of which a definite decision about sector change is made is about 336 s, the algorithm locks onto the correct sector within 1 ms ( j 336x3) even if the actual sector is opposite to the computed sector at switch-on. The system can be thus, started on-the-fly without any appreciable transient in rotor or stator currents. Q & s is then slowly ramped down to zero (or any other reference) so that the sector updating logic can function properly. It may be noted here that the sector correction logic will give improper inferences for a sudden step change in Q & s . However, a transient demand of reactive power is not a practical requirement for the present system, and a gradual change in Q s is acceptable.
127
Ps, Qs
Sector Update
128
P (p.u.)
0 .35
0 .36
0 .37 secs
0 .38
0 .39
0 .4
(a) Response of P s
0 .8 0 .6 0 .4 0 .2
Q (p.u.)
0 .35
0 .36
0 .37 secs
0 .38
0 .39
0 .4
(b) Response of Q s Fig.5.9 Simulation results showing transient responses of P s and Q s due to step change in P & s from 0 to 0.5 p.u. with Q & s = 0 at z = 0.9 p.u.
129
P (p.u.)
0 .26
0 .27 secs
0 .28
0 .29
0 .3
(a) Response of P s
0 .8 0 .6 0 .4 0 .2
Q (p.u.)
0 .26
0 .27 secs
0 .28
0 .29
0 .3
(b) Response of Q s Fig.5.10 Simulation results showing transient responses of P s and Q s due to step change in P & s from 0 to -0.5 p.u. with Q & s = 0 at z = 0.9 p.u.
130
0 .5
us, is (p.u.)
-0.5
-1
-1.5 0 .34
0 .35
0 .36
0 .37 secs
0 .38
0 .39
0 .4
0 .5
us, is (p.u.)
-0.5
-1
-1.5 0 .25
0 .26
0 .27 secs
0 .28
0 .29
0 .3
(b) Response of i s for generation Fig.5.11 Simulation results showing transient responses of i s along with u s due to step change in P & s (a) from 0 to 0.5 p.u. (b) from 0 to -0.5 p.u., with Q & s = 0 at z = 0.9 p.u.
131
0 .5
-0.5
-1
-1.5 0 .5
0 .55
0 .6
0 .65 secs
0 .7
0 .75
0 .8
Fig.5.12 Simulation waveform showing identification of sector with rotor flux component y ra
1 .5
0 .5
ir (p.u), Sector
-0.5
-1
-1.5 0 .5
0 .6
0 .7
0 .8
0 .9 secs
1 .1
1 .2
Fig.5.13 Simulation waveform showing i r and sector information during transition through synchronous speed
132
Chapter 5 Direct Power Control The rotor flux along with the sector information is given in Fig.5.12. The sector information is shown in the form of steps; there are 6 steps corresponding to the six sectors. The rotor current during transition through synchronous speed is plotted in Fig.5.13. As the rotor passes through the synchronous speed, the slope of the steps change from positive to negative, thereby indicating that the rotor flux changes its direction of rotation.
133
Fig.5.14(a) Experimental waveforms showing transient response of P s and Q s due to step change in & P& s from 0 to -0.5 p.u. and Q s = 0
Fig.5.14(b) Experimental waveforms showing transient response of i s due to step change in P & s from 0 to -0.5 p.u. and Q & s =0
134
& Fig.5.15(a) Experimental results showing steady-state waveforms u s , i s , i r for P & s = -0.5 p.u., Q s =0 at 1300 rpm (subsynchronous operation)
& Fig.5.15(b) Experimental results showing steady-state waveforms u s , i s , i r for P & s = -0.5 p.u., Q s =0 at 1430 rpm (synchronous operation)
& Fig.5.16 Experimental results showing steady-state y ra , y rb for P & s = -0.25 p.u., Q s =0
135
Fig.5.17(b) Experimental results showing P s , Q s and sector information during steady-state & operation with P & s = -0.5 p.u.,Q s =0 at 1300 rpm
Fig.5.17(c) Experimental results showing i r and sector information during transition through synchronous speed
136
Chapter 5 Direct Power Control Transient in active power for a step change in P & s from 0 to -0.5 p.u. is shown in Fig.5.14(a). & Q& s is maintained at 0. As P s is changed P err goes out of the prescribed band of 0.05 p.u.. This results in the selection of only the active vectors thereby effecting the fastest possible change in P & s. The slope of change of P is decided by the rate of change of rotor current which in turn depends on & the dc link voltage. It may be noted from these waveforms, that, the transient responses in P & s and Q s & are perfectly decoupled. The steady-state ripple in P & s and Q s due to switching between the positive and negative error bands can also be observed. Fig.5.14(b) illustrates the effect of the active power transient as reflected in the stator current waveform. The steady-state current waveform in Fig.5.14(c) clearly shows unity power factor operation. The steady-state stator voltage and current waveforms for subsynchronous and synchronous operations are given in Fig.5.15(a) and Fig.5.15(b) respectively. The synchronous speed operation is observed to be perfectly stable. In Fig.5.16, the steady-state rotor flux waveforms y ra and y rb are presented. One of the important requirements of the wind power generators is that the machines have to be "cut-in" when the turbine speed crosses a given limit. The method of "on-the-fly" starting has been discussed in section 5.6. The relevant waveforms of Q s and the computed sector are given in Fig.5.17(a). (The sector information is scaled and output through DAC such that the analog output voltage is 1V multiplied by the sector number.) Before the rotor converter is switched on, the sector information as can be seen from the plot is erroneous. However, the computed sector locks onto the actual sector instantaneously as the rotor circuit is excited. Q s is gradually ramped down to zero. Fig.5.17(b) shows the steady-state waveforms of P s and Q s along with the sector information for P & s = -0.5 p.u. and Q & s =0. Fig.5.17(c) gives the rotor current waveform along with the sector information for transition through synchronous speed. During subsynchronous speed operation, the flux vectors rotate in the anti clockwise direction in the rotor reference frame; hence the sector number increases from 1 to 6 and resets back to 1. This is represented by the ascending staircase waveform. As the rotor moves over to supersynchronous speed the flux vectors start rotating in the clockwise direction. Therefore, the sector number changes in the reverse order as seen by the descending staircase. The changeover from subsynchronous to supersynchronous speed is observed to be smooth without any transients in P s and Q s .
137
5.9 Conclusion
A method of direct power control for doubly-fed, slip ring induction machine is presented. The stator active and reactive powers are controlled within hysteresis bands by adopting a switching algorithm on the rotor side. It is proposed that instead of estimating the exact position of the rotor flux, the information of the sector in which it resides is sufficient for switching the correct inverter state. A novel method for sector identification based on the direction of change of reactive power is proposed. The control algorithm uses only stator quantities for active and reactive power measurement and is inherently position sensorless. It is computationally simple and does not incorporate any machine parameter. Relevant simulation and experimental results to validate the concept are presented. The direct power control method can be an attractive proposition for slip-ring induction generators in wind-energy application.
138
Chapter 6
DOUBLY-FED WOUND ROTOR INDUCTION MACHINE FOR WIND POWER GENERATION - DESIGN CONSIDERATIONS AND CONTROL STRATEGIES
6.1 Introduction
Harnessing wind power by means of windmills can be traced back to about four thousand years from now, when they were used for milling and grinding grains and, for pumping water. Even today there are over one million windmills in operation around the world used for traditional applications. However, there has been a renewed interest in wind energy in recent years as it is a potential source for electricity generation with minimal environmental impact [44, 45]. With the advancement of aerodynamic designs, wind turbines which can capture hundreds of kilowatts of power, are readily available. When such wind energy conversion systems (WECS) are integrated to the grid, they produce a substantial amount of power, which can supplement the base power generated by thermal, nuclear or hydro power plants.
139
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation direction. This is accomplished by yaw control (slow rotation by gear arrangement) of the nacelle (assemblage comprising wind turbine, gears, generator, bearings, control gear etc. mounted in a housing). These turbines are commercially available with ratings upto 1650 kW [50] and are universally employed in the generation of electricity. VAWTs have an axis of rotation perpendicular to the wind direction; so they can harness wind from any direction without the need to reposition the rotor when the wind direction changes. The Darrieus VAWT, which is the earliest VAWT, has a huge egg-beater like structure with curved blades attached at the top and bottom of the same vertical shaft. This shape is structurally suitable for withstanding relatively high centrifugal forces. It is however, difficult to manufacture, transport and install. This led to the proposition of straight-bladed VAWTs, the H-type VAWT and V-type VAWT. Due to higher manufacturing costs, VAWTs have not become economically competitive with HAWTs. In the present work, the system design and control has been proposed with a standard a three-bladed HAWT. However, the same methods can be applied to VAWTs also.
140
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation though the initial investment is high, the advantages are higher efficiency and smaller size of the diesel generator. Hybrid energy systems using wind power have not become popular because of high cost, control complexity and requirement of large battery banks. Rather than serving a localized load WECS can be integrated to large power grids. The energy generated from wind is readily absorbed without affecting the supply quality. (The amount of intermittently generated power which a grid can absorb depends on the grid condition; typically 10% power penetration is within permissible limits [49].) Wind turbines used for grid connected systems are of larger size, normally rated above 100 kW (typical ratings being 225 kW, 600 kW, 660 kW, 1650 kW [50]). In locations of continuous favorable wind conditions, several such WECS are connected to the grid forming a wind-farm. Very large scale WECS (>1 MW) have been installed in a few places (e.g. a 3 MW unit was installed in 1982 in Maglap, Sweden) on experimental basis. However, such large units are very expensive and uncommon.
141
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation power is determined based on a cubic function of speed. The required frequency is then computed depending on the machine parameters. The annual energy production of the system was estimated using Raleigh annual wind distribution. It is reported that for a 5m radius turbine, the annual energy production was 49.6 MWH, compared to 37.2 MWH for a corresponding fixed speed system. Vector controlled squirrel cage induction generators for VSCF wind power systems are also commercially available [52]. Instantaneous control over the machine torque can be exercised leading to smoother variations in generator power and speed. The front end converter is simultaneously controlled for unity power factor operation under all wind conditions. Direct torque control (DTC) algorithm can also be employed for decoupled control over the generator flux and torque. Recently a 225 kW prototype using DTC on the machine side and similar switching logic for the FEC has been successfully implemented and tested [53]. In spite of the disadvantages associated with slip-rings, the wound rotor induction machine has been a potential candidate as wind electric generator. By suitable integrated approach towards design of a WECS, use of a slip-ring induction generator is found to be economically competitive. Control of both grid-connected and isolated variable speed wind turbines with doubly fed induction generator has been implemented by Pena et.al. [16], [54]. Conventional vector control using a position sensor has been employed from the rotor side in both the cases for independent control of active and reactive power. For the isolated WECS, control of an auxiliary load in parallel with the main load, allows the system to track the optimal wind turbine speed for maximum energy capture. In case of the grid-connected system the generator is run in speed-control mode with the help of a torque observer for optimum operating point tracking. Implementation of a torque observer is, however, difficult. Beyond the rated operating point, the algorithm tries to reduce the shaft speed to limit the generator power. This requires sufficient torque capability of the generator to overcome the instantaneous turbine torque and, may not be a feasible solution in practice. In this chapter a comprehensive study on variable speed grid-connected WECS using wound rotor induction machine as the wind electric generator is presented. The motivation for variable speed control is explained and the proposed scheme is compared against the existing fixed speed and variable speed systems using cage rotor machines. The turbine characteristics are generated by a dc drive in the laboratory setup. Peak power point tracking algorithm in the conventional torque control mode is first implemented. A dc motor driven by a commercial thyristor drive is used to simulate the turbine characteristics. Subsequently, a novel technique for tracking the peak power points using 142
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation speed controlled operation is proposed. The technique searches the zero slope point on the power-speed characteristics of the turbine. The peak power tracking is made independent of turbine characteristics, air density etc. in this method. This strategy is also implemented and verified experimentally. Note: In this chapter, the sign convention adopted for active power is different from that of the earlier chapters. Since only generator operation is being considered, active power is taken as positive if it flows out of the induction machine terminals, both at the stator as well as the rotor. The power developed by the turbine is always taken to be positive.
where q is the air density, A is the swept area (cross-sectional area) of the turbine and v is the wind velocity. Cp is called the power coefficient and is dependent on the ratio between the linear velocity of the blade tip (R $ z t ) and the wind velocity (v). This ratio, known as the tip-speed ratio, is defined as k = z t $R v (6.2)
where R is the radius of the turbine. An idealized Cp Vs. k curve, taken from [56], is shown in Fig.6.1. It is observed that the power coefficient is maximum for a particular tip-speed ratio. This implies that for any wind velocity there is a particular rotor rpm for which maximum power transfer takes place. The prime motivation for variable speed control of WECS is to track this rotor speed with changing wind velocity so that Cp is always maintained at its maximum value. Using the Cp-k curve of Fig.6.1, the power-speed
143
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation characteristics are plotted for a commercially available turbine (Vestas V27) by using a Mathcad program. In the following sections these power curves, as shown in Fig.6.2 and Fig.6.3, are used for comparison between the different control schemes and designs. 0.5 0.4 Cp 0.2 0 .
10
15
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation From these characteristics it is observed that at a wind-speed of around 14 m/s the rated power of the machine (225 kW) is reached at 43 rpm (corresponding to approximately 1008 rpm of the rotor shaft). This is indicated by the point B. It can also be seen that the maximum power that the turbine can actually generate at a wind-speed of 14 m/s is about 460 kW, provided the turbine shaft speed is allowed to vary upto 85 rpm and a generator of adequate capacity is used. The fixed speed system in this case is designed to operate at 1000 rpm, whereas operation at 1500 rpm would result in substantially higher generation. The reason is as follows. It may be observed from the turbine characteristics that at 1000 rpm the change in turbine power for a large change in wind velocity is not significant. Even for a wind velocity of 20 m/s the power generated can only be 250 kW. This ensures that the generator is not overloaded to a great extent even if there is a sudden gust of wind. Pitch control comes into operation once the rated power is reached; but this hydraulically operated mechanism being sluggish, transient overshoot of the generator power cannot be prevented. So operating at 1000 rpm ensures that the generator is not overloaded under sudden high wind conditions. However, at 1500 rpm the rise in power with wind velocity is much sharper as observed from Fig.6.3. Fixed speed systems are therefore, designed for lower shaft speeds where the turbine power curves (for different wind velocities) are close to each other. Thus a natural protection against overslip and overload is provided for the generator, but utilization of the turbine capability is poor. A cage rotor induction generator when connected to the grid draws the magnetizing current from the line thereby reducing the stator power factor. Under low wind conditions, when the active power generation is low, the machine mainly draws reactive power from the grid and the stator power factor is extremely poor. The lagging reactive power is compensated by connecting capacitor banks across the line. Depending on the active power generation, these capacitors are either cut-in or cut-out to regulate the average power factor of the generator between 0.95 and 1. But the random switching of the capacitor banks gives rise to undesirable transients in the line currents and voltages. In a grid, where hundreds of such machines are installed, these capacitive switchings can cause severe overvoltage problem. From Fig.6.3 it may be noted that if the machine is always operated at 1000 rpm, then the power generated for low wind velocities (<5 m/s) will be extremely small. In order to boost the generated power under such circumstances another winding of reduced power capacity (50 kW) is added to the motor with 4 pole pairs, with a synchronous speed of 750 rpm. The controller switches 145
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation over from one winding to the other depending upon the wind condition. This feature makes the machine nonstandard and expensive. Design compromises are also associated with this addition of a separate winding. It is seen that the no-load current for the main generator is about 235 A (about 0.59 p.u.) [Appendix D], which is unusually high for a machine of 225 kW rating.
500
v10
100
80
Fig.6.2 Power curves of the wind turbine against turbine shaft rpm (v1=5m/s, v2=6m/s .. v10= 14m/s)
500 v10
400
v9
v8
v7
1000
2000
2500
Fig.6.3 Power curves of the wind turbine against generator shaft rpm with gear-ratio of 1:23.4 146
147
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation (a) Generator and Gearbox Selection From Fig.6.2, it may be noted that the maximum power that can be delivered by the turbine at the wind velocity of 12 m/s is about 290 kW corresponding to the shaft speed of 75 rpm. On this basis, a 300 kW, 415V squirrel cage induction machine is selected. The synchronous speed of the machine is kept at 1000 rpm by using a 6-pole machine. Assuming that the rated power is reached at the rated frequency, the gear ratio works out to be 1000:75 i.e. 13.3:1. The cost of the gearbox can be brought down by using a lower gear-ratio. This is possible by selecting a machine with higher pole pairs. However, with increase in the number of pole pairs the machine frame size increases. The magnetizing current requirement also increases significantly. (b) Converter Rating For variable speed control, the back-to-back PWM converter configuration as shown in Fig.6.4 is used on the stator side. The stator side converter supplies the required reactive power and also handles the full active power generated by the machine. The line side converter transfers the generated active power to the grid at unity power factor and regulates the dc bus voltage. The size of the converters, therefore, is dictated by the generator rating. With a provision for overloading, the ratings of the converters can be taken as 375 kVA. For this power rating, IGBT modules are ideally suited, so that a high switching frequency of about 5 kHz can be employed to limit the current ripple in both the converters to less than 10%. The device ratings can be computed as follows. i s,rms(max) = 375000/(3 $ 415 ) = 522A i s,peak = 738A Allowing 10% peak-to-peak switching ripple, the peak current rating for the device may be taken as 5% more than i s,peak . i s,peak(max) = 1.05 $ 738 = 775A Using a maximum modulation index of 0.9 with sine-triangle modulation, the dc bus voltage that is required is given by u dc = 2 $ (415 $ 2/3 ) $ 0.9 = 753V The dc bus can be designed for 750 V.
148
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation Two paralleled 1200V, 400A IGBT modules can be selected. The same devices may be used for both the converters, even though the line side converter will be operating at upf and need not carry any reactive current. (c) DC Bus Capacitor Assuming no-load current to be 15% of the rated current (i.e.62.6A) and allowing 0.5% dc bus voltage ripple due to reactive loading, the dc bus capacitance can be computed [Appendix E] as C = 2 $ i nl / RPU $ u dc $ 24 $ f rated = 2 $ 62.55/(0.005 $ 750 $ 24 $ 50) = 19658 lF i ripple,rms = 2/3 $ i nl = 2/3 $ 62.55 = 51A The capacitors are divided in three banks corresponding to three phases. (Each phase consists of two legs, one for the front end converter and the other for the machine side converter). The voltage rating of 750V cannot be achieved with a single electrolytic capacitor in each parallel branch. 2, 3300F, 450V capacitors may be used in series for each branch. 12 such parallel units (a total of 24 capacitors) need to be connected to meet the required capacitance value. Hence, each bank comprises 4 such units. The effective capacitor value becomes C = 12 $ 3300/2 = 19800lF. The rms ripple current in each branch is 4.25A, which is within allowable limits. (d) Input Transformer The line side converter is interfaced to the power grid through a transformer. Potential wind sites are usually remote and the transmission grid is available at a higher voltage. (For example, in south Tamil Nadu, India, the wind farms are connected to 6.6 KV power grid). The transformer can be rated for 375 KVA with a turns ratio of 6.6KV:415V. (e) Line Side Inductance The line side series reactance decides the current ripple for the front end converter. Hence, a very low value cannot be used. With 5 kHz switching frequency and 0.25 p.u. choke the switching ripple is within the design limits of 10% (checked through simulation). Using this value, L fe = 0.25 $ (415/3 )/ i s,rms $ 2 $ o $ 50 149
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation = 0.25 $ (415/3 )/(417 $ 2 $ o $ 50 ) = 457.5lH Therefore, 450H series inductance per phase can be selected.
v9 300 v8
Ptarget
200 v6 100 v4 v3 v1 0 0 200 400 600 800 1000 1200 1400 v2 v5 v7
12 m/s
Fig.6.5 Operating region of WECS with cage rotor induction machine in the P z plane
5000
4000
Mdtarget
v8 2000 v7 v6 1000 v4 v3 v1 0 0 200 400 600 800 1000 1200 1400 v2 v5 12 m/s
Fig.6.6 Operating region of WECS with cage rotor induction machine in the m z plane
151
Wind turbine 3 Phase Transformer 3 Phase Front end Converter Gear box 3 Phase Inverter Wound Rotor Induction Machine Fig.6.7 Variable speed grid-connected WECS using doubly-fed wound rotor induction machine
152
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation (b) Converter Rating The converter rating in this case depends on the range of operating speed. Assuming 0.5 p.u. slip on either side of the synchronous speed, the converter rating can be half the power rating of the stator. Allowing the same amount of overloading as earlier, the converter can be rated for 375*0.5 i.e. 187.5 KVA. It may be noted that the equivalent current ratings for the stator and rotor are same for the selected turns ratio. So, the current rating of the devices, as selected in the previous case, also applies here. However, the maximum voltage that needs to be applied to the rotor terminals is half the stator voltage. Therefore, a dc bus of 375V (750/2) is sufficient in this case for the same maximum allowable modulation index of 0.9. For the rotor side converter and front end converter, 2 paralleled 600V, 400A IGBT modules can be selected. If desired, the reactive power can be distributed equally between the rotor and the stator sides, and the front end converter can be operated at leading power factor to compensate for the lagging VAR drawn by the stator. This results in equal current loading for the rotor side and front end converters. However, since the reactive current is only about 10-15% of the rated current, it does not make a significant difference even if the rotor side converter supplies the full reactive power and the front end converter is operated at unity power factor. (c) DC Bus Capacitor Since the reactive current requirement is almost the same for a cage rotor machine and a wound rotor machine of 300 kW rating [67], the same capacitor value needs to be used. However, the voltage rating is reduced by half. Using single 450V capacitors for each parallel branch may not provide sufficient voltage margin when the bus is charged to 375V. A more realistic design would be to use 2, 250V capacitors in series in each branch. The configuration of the capacitor banks remains the same i.e. a total of 24, 250V, 3300 F capacitors are required for the entire bank. The rms ripple current rating in each branch remains at 4.25A as in the previous case. (d) Input Transformer The transformer connecting the system to a 6.6KV grid should have two secondaries; one winding connecting the stator being rated at 415V and the second winding, connecting the front end converter being rated at 415/2 i.e.208V. Without this voltage reduction on the rotor side, it is not possible to operate the dc bus at 375V. Consequently the voltage ratings for the devices and the 153
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation capacitor bank cannot be optimized. The rating of the transformer also has to be boosted up by 50% because of the extra power being generated from the rotor side during supersynchronous operation. Therefore, a 375*1.5 i.e. 560 KVA transformer is to be used with two windings having turns ratio of 6.6KV:415V/208V. (e) Line Side Inductance The same per unit reactance of 0.25 can be used in this case. However, since base impedance is reduced to half (due to reduction in input voltage), 225H per phase inductance is sufficient for the present scheme.
154
Pgen
v9
300 v8
Pstator
200 v6 100 v4 v3 v1 0 v2 6 m/s v5 v7 12 m/s
Protor
-100 0 200 400 600 800 1000 1200 1400
Fig.6.8 Operating region of WECS with wound rotor induction machine in P z plane
5000
4000
Mdtargtet
3000
v10 14 m/s v9
v8 2000 v6 1000 v3 v1 0 0 200 400 600 800 1000 1200 1400 v2 6 m/s v5 v4 v7 12 m/s
Fig.6.9 Operating region of WECS with wound rotor induction machine in m z plane (ii) The stator flux is constant over the entire operating region. Therefore, the torque can be maintained at its rated value above the synchronous speed. This results in higher power above the synchronous speed (i.e. at high wind velocities) when compared to a cage rotor induction generator of the same frame size. Thus the machine utilization is substantially improved. 155
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation (iii)A lower dc bus voltage is required. This reduces the voltage rating of the capacitor bank and significant saving in the cost of the capacitor. (iv)The line side inductance value is also reduced. Table 6.1 Summary of the design results for the three WECSs. Constant Speed System with Variable Speed System Variable Speed System with Cage Rotor Induction Machine with Cage Rotor Slip Ring Induction Induction Machine Machine Turbine Generator V27 Two winding cage rotor induction machine Main:415V, 225 kW, 6 pole Auxiliary:415V, 50 kW, 8 pole 23.4:1 None V27 Cage rotor induction machine 415V, 300 kW, 50Hz, 6-pole 13.3:1 375 KVA, 1200V,400AX2 IGBT modules 750V 19800 F (2, 450V, 3300F in series in each branch X 12 parallel branches) 450 H 375 kVA 6.6KV:415V 300 kW 0 - 1500 rpm V27 Wound rotor induction machine 415V, 300 kW, 50Hz, 6-pole 13.3:1 187.5 KVA, 600V,400AX2 IGBT modules 375V 19800 F (2, 250V, 3300F in series in each branch X 12 parallel branches) 225 H 560 kVA 6.6KV:415V/208V 450 kW 500 - 1500 rpm
FEC Inductance Transformer Maximum Power Capture Speed Range (generator shaft)
156
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation particular section is to determine the energy capture for the three systems under consideration. A wind function is defined and, the turbine power and target power for the generator are determined. Current controller dynamics are ignored for the time being and it is assumed that the actual generated power tracks the reference instantaneously. A more detailed simulation including the rotor side current controllers and a realistic wind profile is presented later where the rotor current control scheme is also included. Wind is a randomly fluctuating variable. Therefore, it is difficult to model and evaluate the performance of a WECS theoretically without implementing it and subjecting it to actual environmental conditions. Some theoretical predictions are possible with the statistical data of wind variations at a particular location [60]. However, these are more appropriate for optimal planning of WECS in terms of cost, overall energy output per unit land area etc. [61], rather than evaluating the relative performances of the different generating schemes. For effective performance evaluation the system has to be operated over the entire range of wind variations so that all the design limits are reached. This is simulated by defining a wind function as Vw = 10 - 2.cos(2.pi/20)t - 5.cos(2.pi/600)t (6.3)
Fig.6.10(a) shows the wind profile; it is observed that the wind varies between maxima and mimima with a periodicity of 10 secs and these peaks and troughs are modulated over a longer period with a periodicity of 10 mins. The minimum touches the cut-in speed of 3 m/s, whereas the global maximum reaches 25 m/s. All the three systems are subjected to this wind function and the shaft speed, generator power and generated energy are plotted.
157
18 16 14
12 10 8 6 4 2
100
200
500
600
40
30
20
10
100
200
500
600
158
250
200
150
100
50
100
200
300 secs
400
500
600
30 25 20 15 10 5 0
100
200
300 secs
400
500
600
159
It is seen that the target power varies as the cube of the rotor speed, other parameters being dependent on the turbine characteristics and assuming air density to be constant. Hence the torque corresponding to the peak-power locus varies as the square of the rotor rpm. m d,t arg et = K opt $ z 2 t (6.6)
The generator torque is always set in accordance to this desired locus. The speed of the shaft is free to vary and therefore, it settles at an operating point where the generated torque equals the turbine torque. Since the intersecting points between the two curves correspond to the maximum power points, it is ensured that the generator always extracts the maximum possible power from the turbine irrespective of the wind speed. This can be explained with reference to Fig.6.2. Let the system be operating at the point S, corresponding to a wind velocity of v4 (=8 m/s) and generating about 90kW. Under this condition if the wind velocity increases to v7 (=11 m/s), immediately the turbine power rises to 200 kW as indicated by point P. Since the speed of the system cannot change instantaneously, the generator still continues to deliver 90 kW. The driving power being more, the 160
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation system accelerates, and with increasing speed the generated power also rises until it reaches the stable operating point T. When the wind speed falls the operating point for the turbine shifts to Q, but the generator power does not immediately change. This gradually decelerates the system back to S. In this process a part of the energy from the wind is stored by the inertia of the system during increasing wind velocities; which is released during deceleration in a controlled manner. From Fig.6.5 and Fig.6.6 it can be seen that, beyond the rated speed the generator torque is varied as m d,t arg et = P g,rated zt (6.7)
The system is simulated with the same turbine data [Appendix D]. It is observed that due to constant variation in the wind velocity the system is always in the transient state searching for the optimum operating point. The large inertia of the rotating blades tend to reduce the fluctuations in torque and power to a substantial extent. The relevant plots are given in Fig.6.11(a) through Fig.6.11(d). The system is started with an initial speed of 10 rpm. (Before this point the turbine torque is almost zero and the system fails to accelerate in the simulation model. In practice the pitch angle is controlled to start the system.) The generator power is limited to 300 kW and the generator shaft speed is limited to 1500 rpm. The corresponding turbine shaft speed is 113 rpm. It is assumed that the pitch control comes into operation beyond these limits and restricts the operating region. However, the dynamics related to the pitch control mechanism are neglected for simplification. It is seen that the energy output for the defined wind function (Eq.6.3) over 10 minutes is 28.5 kWH, an increment of 32.5% with respect to the previous case. It is interesting to note that this control mechanism eliminates the need for measurement of wind velocity to track the peak-power locus. Moreover, since the torque of the machine is directly controlled the generator is never allowed to exceed its maximum torque capability; so the problem of overloading or pulling out does not arise and the pitch control mechanism can operate more effectively in this mode of operation.
161
18 16 14
12 10 8 6 4 2
100
200
500
600
100
80
60
40
20
100
200
500
600
Fig.6.11(b) Turbine shaft speed for VSCF system using cage rotor induction machine
162
100
200
300 secs
400
500
600
Fig.6.11(c) Generator power for VSCF system using cage rotor induction machine
40 35
30 25 20 15 10 5 0
100
200
300 secs
400
500
600
Fig.6.11(d) Generated energy for VSCF system using cage rotor induction machine
163
6.6 Detailed Simulation of Variable Speed WECS using Wound Rotor Induction Machine with Rotor Side Current Control
In this section simulation of the doubly-fed grid connected wound rotor induction generator operating under with a realistic wind profile is presented. Generation of the wind profile using a random function generator is shown in Fig.6.13. The function generator output is passed through a first order filter to smoothen out sharp fluctuations. The output of the function generator varies from 0 to 2. This is scaled by the filter gain and added to an average value of 12. The variation in wind velocity is thus in the range of 0 to 18 m/s.
164
18 16 14
12 10 8 6 4 2
100
200
500
600
100
80
60
40
20
100
200
300 secs
400
500
600
Fig.6.12(b) Turbine shaft speed for VSCF system using wound rotor induction machine
165
100
200
300 secs
400
500
600
Fig.6.12(c) Stator power of wound rotor induction machine for VSCF system
200 150 100
100
200
300 secs
400
500
600
Fig.6.12(d) Rotor power of wound rotor induction machine for VSCF system
166
500
400
300
200
100
100
200
300 secs
400
500
600
Fig.6.12(e) Total generated power for VSCF system using wound rotor induction machine
40 35
30 25 20 15 10 5 0
100
200
300 secs
400
500
600
Fig.6.12(f) Generated energy for VSCF system using wound rotor induction machine
167
1 Wind Velocity
Fig.6.13 SIMULINK block diagram of the random wind profile generator The torque reference is generated using the same torque control law as described in section 6.5. The q-axis current reference i & rq is derived from this torque reference assuming the stator flux is constant at the nominal value. The d-axis reference is kept at 0, so that, the stator supplies the total reactive power required by the machine. The same rotor current controller is employed with rotor position feedback. The simulation results are presented in Fig.6.14(a) through Fig.6.14(f). The speed variations of the generator shaft (Fig.6.14(b)) tend to follow the pattern of the wind profile (Fig.6.14(a)) to track the peak power. The generator torque (Fig.6.14(c)) and the stator generated power (Fig.6.14(d)) differ by a scaling factor, which is the synchronous frequency. The generator torque is saturated at its rated value so that the stator power is limited to its nominal rating of 300 kW. The rotor power varies between -40 kW (in the subsynchronous range) and 110 kW (in the supersynchronous range) as seen from Fig.6.15(e). This is in agreement with the operating point locus shown in Fig.6.8 The generator output power is the sum of the stator and the rotor powers and it is observed that a substantial amount of power is generated from the rotor side even after the stator power saturates to its rated value.
6.7 Practical Implementation of Variable Speed System using Wound Rotor Induction Machine in Torque Control Mode
The peak power tracking algorithm is implemented on the experimental setup. The turbine characteristics are simulated by a dc motor driven by a commercial drive. The system is run for different wind velocities and the corresponding steady-state operating points are found to be close to the peak power points in the P z curves. Transients due to step changes in wind velocity are also recorded and presented in this section.
168
25
20
15
10
100
200
300 secs
400
500
600
1600 1400
100
200
300 secs
400
500
600
3000
2500
2000
1500
1000
500
100
200
300 secs
400
500
600
100
200
300 secs
400
500
600
150
100
50
-50
-100
-150
100
200
300 secs
400
500
600
300 250 200 150 100 50 0 0 100 200 300 secs 400 500 600
Since under rated condition of input voltage and frequency, y s is unity, 172
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation & i& rq (pu) = m d (pu) $ (1 + r s ) = z(pu) 2 $ (1 + r s ) (6.9)
With such scaling, the software implementation of the peak power tracking algorithm in torque control mode becomes very simple. The rotor side reactive current reference i & rd is set to zero for the present experiment. The software modules for the simulation of the turbine characteristics and generation of the rotor current reference are executed in the 6th slot of the task schedule, as discussed in Chapter 3. Therefore, the torque references are updated every 341 s. The system was run for each of the wind velocities and the operating points were recorded (Fig.6.15). It is seen that the results are in close proximity to the peak power points in the P z characteristics. The small errors can be attributed to (i) the resolution of the turbine characteristics as stored in the memory and (ii) the losses incurred in the system.
1.4
1.2
v9
v8
0.8
v7
0.6
v6
0.4
0.2
173
Fig.6.16(a) Experimental result for transients in z and i & rq due to change in wind velocity between v6 and v8 while operating in torque control mode
Fig.6.16(b) Experimental result for transients in z and i & rq due to change in wind velocity between v7 and v9 while operating in torque control mode The transients in speed and active current reference i & rq for step changes in wind velocity from v8 to v6 and vice versa are given in Fig.6.16(a). Similar responses are recorded for transitions between v7 and v9 in Fig.6.16(b). The settling time between two operating points is decided by the system inertia and the mechanical losses. Since, the actual system inertia cannot be scaled in the laboratory setup, the responses do not match the practical dynamic behavior of a WECS. Nevertheless, it can be concluded that with varying wind conditions, the torque control law works and the reference torque is dynamically updated to follow the optimum operating point locus. It is also 174
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation observed from Fig.6.16(b) that beyond 1 p.u. speed (corresponding to 5V in the oscilloscope plot), the system runs with constant torque with i & rq limited to its rated value. Note: For these experiments, the base power is taken as 3 kW (instead of 4.5 kW as considered in the earlier chapters). With the base speed still at 1500 rpm, the base torque is 3/4.5 i.e. 0.67 times the previous case. The scalings for the inner current control loops are not modified; therefore, i & rq is limited to (1 + r s ).0.67 i.e. 0.738 p.u. In Fig.6.16(b) it can be seen that i & rq saturates at about 3.7 V (corresponding to 0.738 p.u.).
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation the algorithm presented here, the generator is operated in the speed control mode with the speed reference being dynamically modified in accordance with the magnitude and direction of change of active power. The peak power points in the P z curve correspond to dP/dz = 0. This fact is made use of in the optimum point search algorithm.
500
400
Pgen
300
200
100
200
400
600
800
1000
1200
1400
15
10
0 0
100
200
300 secs
400
500
600
Fig.6.18 Generated energy for q = 0.5*1.225 kg/m3 with correct value of K opt (continuous line) and earlier value of K opt (dotted line) 176
Turbine Power
P3 P2
P4 v2
P1
P6 P5 v1
Fig.6.19 Shift of operating points in the proposed peak power tracking algorithm The proposed algorithm is explained with the help of Fig.6.19, where the P z curves corresponding to two wind velocities are shown. Let the present wind velocity be v1. The generator is run in the speed control mode with a speed reference of z1 (which corresponds to the optimum operating point P1 for v1). The generator output power and speed are sampled at regular intervals of time. If the wind velocity is steady at v1 the difference between successive samples of active power P i.e. DP will be very small and no action is taken. Now let there be a step jump in wind velocity from v1 to v2. Since the speed is constant, this would result in a change of operating point from P1 to P2. Therefore, DP would be large and positive. Corresponding to this change in DP a positive change in speed reference is commanded. The change in speed reference Dz & is made proportional to DP. This shifts the operating point from P2 to P3 resulting in a smaller positive change in DP. Since this change in DP is due to a positive change in Dz & it implies that the peak power point is further to the right hand side of the curve. Thus a further positive change in Dz & is commanded in proportion to DP. In this process when DP becomes very small (within some defined band) no further change in speed command is given and the system keeps operating at P4. Now if the wind velocity again changes from v2 to v1, the operating point shifts to P5 resulting in a large negative change in DP.
177
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation Thus a negative change in speed reference in proportion to DP is applied. However, this results in a positive change in DP as the operating point shifts to P6. Since the positive change is due to a negative change in speed command the peak power point is to the left of P6. Therefore, the speed reference is further reduced. The algorithm continues until DP is within the pre-defined band and the operating point again slides back close to P1. The algorithm is implemented in the following manner. The active power is sampled at a particular rate and the incremental change is computed as DP(k) = P(k) P(k 1) The magnitude of Dz & (k) is given by | Dz & (k) | = | DP(k) % K t | (6.8)
(6.9)
where K t is the proportional constant and needs to be selected judiciously. This is discussed later. However, the sign of Dz & (k) has to be properly assigned. If Dz & (k 1) is zero i.e the speed reference was not changed in the previous sample then the sign of DP(k) alone decides the sign of Dz & (k). IfDz & (k 1) is non-zero, the product of the signs of Dz & (k 1) and DP(k) determines the sign of Dz & (k). This can be formulated as follows. if ( Dz & (k 1) == 0 ) S = Sign( DP(k) ) else S = Sign( DP(k) )* Sign( Dz & (k 1) ) Dz & (k) = S . | DP(k) * K t | The reference speed is sampled at the same frequency as the active power. If the magnitude of DP(k) is within some small defined band then the reference speed is not changed, otherwise it is changed by Dz & (k). if( |DP(k)| <= P band ) z & (k) = z & (k 1) else z & (k) = z & (k 1) + Dz & (k) With this reference, the machine is operated in the speed control mode. The controller block diagram is given in Fig.6.20. A PI controller is employed for the speed loop. The output of the speed controller is saturated at the rated torque of the machine. Therefore, beyond the rated operating point, the operation is similar to that in the torque control mode. This is explained with the help of Fig.6.19. 178
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation Let P4 denote the rated operating point. Now, if the wind velocity increases to v3, the speed controller will try to increase to generator torque so that the operating point shifts to P7. However, the rated torque is already applied at P4; hence the prime-mover accelerates the system until a stable operating point at P8 is reached. Hence, the control naturally ensures that the operating region remains the same as indicated earlier in Fig.6.8.
+ _
m* d 1/Kc i* rd
S1 S2 S3
179
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation speed loop. It may be assumed that the speed loop settles down within 4 times the designed loop time constant. Therefore, the sampling period is taken as 4 times the speed loop time constant of the system. The inertia of the system being high the speed loop time constant cannot be made very small; this would require a machine of very high torque capability. It is noted that a fast speed controller also gives rise to large transients in the machine torque, which is reflected in the generated power. This is a disadvantage of employing speed control in generating systems. However, the dynamics of the speed controller can be made slower so that the fluctuations in the machine torque can be reduced. Also in a wind-farm where many such systems are connected to the grid there will be some averaging effect in the overall power generated and, the power fluctuations of the individual machines would not be directly reflected on the grid.
6.8.3 Selection of Kt
350
300
250
200
150
100
50
K t determines the change in speed reference for a given change in P. Therefore, it depends on the slope of the P z characteristics. To choose a value of K t , an approximate idea of the turbine characteristics is needed. The P z characteristics in the region of operation of the peak power tracking algorithm are considered. This is shown in the V27 power curves of Fig.6.21. The wind 180
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation velocities over this region vary between 6 m/s and 12 m/s. The approximate changes in Dz for successive changes in wind velocities and hence DP are also shown. It is obvious that the Dz/DP is more for lower wind velocities and vice-versa. If K t is set to the maximum value of Dz/DP in the operating range then, for changes in wind velocities during high wind conditions, the increment in speed reference would be more than desired. This would result in overshooting of the optimum operating point. The system would oscillate about the peak power point before it settles down. Therefore, the maximum value of K t is limited by the lowest value of Dz/DP. A large value of K t will also result in a large transient in generator torque which is not desirable. Hence the value of K t selected is substantially lower than the limit imposed by the minimum value of Dz/DP. From Fig.6.21 it can be seen that the P z curves are flat-topped near the peak power points. Therefore, the change in DP for an increment in speed would be very small in this region. The P band may be set at 5% of the nominal power rating of the generator. So the final operating point may not move exactly to the peak power point, but may settle down close to it.
181
Chapter 6 Wound Rotor Induction Machine for Wind Power Generation presented for changes in wind velocity between v7 and v9. The saturation of the torque beyond the rated speed is clearly observed from the plot of i & rq .
1.4
1.2
v9
v8
0.8
v7
0.6
v6
0.4
0.2
Fig.6.23(a) Experimental result for transients in z and i & rq due to change in wind velocity between v6 and v8 while operating in speed control mode with proposed algorithm
182
Fig.6.23(b) Experimental result for transients in z and i & rq due to change in wind velocity between v7 and v9 while operating in speed control mode with proposed algorithm
6.9 Conclusion
A comprehensive study on the design and performance of WECS using wound rotor induction machine with rotor side control has been presented. A comparison with the existing schemes shows, that for a machine of similar rating, energy capture can be enhanced by using a wound rotor machine. In this case, the rated torque is maintained even at supersynchronous speeds whereas, in a system using cage rotor machine, field weakening has to be employed beyond synchronous speed, leading to reduction of torque. It is therefore, possible to operate the proposed system upto higher wind velocities. The voltage rating of the power devices and dc bus capacitor is substantially reduced. The size of the line side inductor is also decreased. Hence, the use of wound rotor induction machine promises to be economically feasible and attractive for wind power generation. Peak power tracking using conventional torque control mode is implemented and experimentally verified. An algorithm for searching the optimum operating point in speed control mode is proposed. This technique makes peak power tracking independent of the turbine characteristics and the air density. Experimental verification shows that the performance of the control algorithm compares well with the conventional torque control method.
183
Chapter 7
CONCLUSION
7.1 General
Rotor side control of grid-connected wound rotor induction machine has shown considerable promise for use in variable speed constant frequency applications with limited speed range. The ratings of the power converters to be used are reduced and the machine utilization is improved. Use of field oriented control technique provides effective control over the active and reactive powers handled by the machine. The recent availability of high-performance DSPs with integrated peripheral units for motor control applications, allows easy implementation of sophisticated control strategies. The scheme has the potential of being a reliable and cost-effective solution to wind power generation as well as industrial drives.
185
Chapter 7 Conclusion The doubly-fed wound rotor induction machine is modeled in the stator flux oriented reference frame and the front end converter in the stator voltage oriented reference frame. Direct and quadrature axes current controllers, consisting of PI loops with appropriate compensating terms to decouple the dynamics of the two axes, have been designed for both the converters. The design procedure is direct and follows from the voltage equations derived in the field coordinates. Simulation results show excellent dynamic response for the current loops. Steady-state unity power factor operation at the stator terminals and at the grid interface of the front end converter is demonstrated. The front end converter is shown to be capable of operating at a leading power factor upto a certain limit. The control algorithms are implemented on an experimental setup in the laboratory. The IGBT based power converters are designed and fabricated in a modular way. They are subjected to rigorous testing including short-circuiting of the dc bus. These power converters have become standard modules for other drive applications in the laboratory. A generalized digital control platform is also built using a TMS320F240 DSP. The hardware has sufficient number of digital and analog inputs/ outputs to interface with two converters and the computing power necessary to execute all the control loops associated with the rotor side and front end converters at a fast sampling rate (8.85 kHz for the fastest loop). The software is designed for multitasking with a task scheduler coordinating the execution of the various modules. Firstly, the conventional rotor side field oriented control scheme is implemented with position sensors. Transient responses for the active and reactive components of rotor current are in close agreement with the corresponding simulation results. Unity power factor is achieved both at the front end and stator terminals. The voltage control loop exhibits good regulation and fast dynamic response. Even for step reversal of load on the dc side, the deviation of the bus voltage from the set value is within 10%. A position sensorless algorithm for field oriented control is proposed. The algorithm makes use of simple transformations to estimate the rotor current in the stator reference frame. With the rotor current being directly measured in the rotor reference frame, the angle between the two axes is easily determined. A method of correctly estimating the stator flux without integration of the stator voltage is employed. The proposed method can be started on-the-fly, i.e., when the rotor is already in motion. It also operates stably at synchronous speed which corresponds to zero rotor frequency. These features are verified through extensive simulation and subsequently demonstrated through laboratory experiments. 186
Chapter 7 Conclusion An algorithm for directly controlling the active and reactive powers in the stator circuit is proposed. In this method, the instantaneous magnitude and angular velocity of the rotor flux vector with respect to the stator flux is controlled by selecting an appropriate switching state of the rotor side inverter. Instead of integrating the rotor PWM voltage to obtain the rotor flux (as practised in conventional DTC techniques), a novel strategy to update the sector information is used. The algorithm does not make use of any machine parameter but relies on the measured active and reactive powers in the stator for controlling the rotor flux. Features like on-the-fly start and stable synchronous speed operation are also obtained. The direct power control algorithm is simulated extensively and verified experimentally. Excellent dynamic response for change in active power is demonstrated. Application of rotor side control of doubly-fed induction machine to wind energy conversion systems is studied. The scheme is compared with the existing fixed speed and variable speed systems using cage rotor induction machine. Power circuit component cost decreases considerably with the proposed scheme. A lower power rating of the converters also increases the reliability of the system. The rated torque is maintained even beyond the synchronous speed and hence, energy capture is improved at higher wind velocities. This additional power is generated through the rotor circuit resulting in better utilization of the machine rating compared to systems with cage rotor machines. An algorithm for the tracking the peak power points on the turbine characteristics in the conventional torque control mode is first implemented. A dc motor driven by a commercial thyristor drive is used to simulate the turbine characteristics. The machine is run in current control mode with appropriate reference signals generated from the DSP. Experimentally obtained steady-state operating points match closely with the optimum operating points on the turbine characteristics. Subsequently, a novel method is proposed to track the peak power point locus which operates in the speed control mode. Unlike in the previous case, precise information about the turbine characteristics is not required. By intelligently varying the speed, the algorithm searches the zero slope locations on the turbine power-speed curves. This method of tracking the peak power has also been experimentally verified.
187
Chapter 7 Conclusion The complete potential of these systems are yet to be fully exploited and there exists scope for further research. In case of a grid connected system, the rated torque of the machine can be maintained upto twice the synchronous speed, provided a 1 p.u. converter is used in the rotor circuit. At the highest speed, the machine, therefore, delivers 2 p.u. of power from the same frame size, 1 p.u. being drawn from the stator and 1 p.u. from the rotor. However, with the stator connected to a constant frequency source, the speed cannot be further increased. Moreover, it is not possible to obtain speed reversal with the present arrangement. If, instead of connecting the stator to the grid, it is also fed from a voltage source inverter, the speed range of the system can be further extended and speed reversal can also be achieved. Such a scheme has been recently proposed by Kawabata [70] et. al., where, a double-inverter-fed vector-controlled drive is implemented with a wound rotor induction machine using position sensors. This scheme will be attractive in high speed, high torque applications. The position sensorless algorithms proposed in the thesis, can be directly applied to this system. It is possible to achieve stable zero speed operation, which is still a problem in sensorless schemes for cage rotor induction machines. Direct torque control algorithms can be developed for such doubly-controlled machines. The flexibility to control both the stator and rotor flux can improve the dynamic performance of the drive. It simultaneously opens many possibilities for selecting the switching states of the inverters. There is a growing awareness that power electronic systems need be modularized, so that systems of larger ratings can be developed by integrating low power modules. High frequency IGBT inverters are now commercially available upto ratings of 250 kW. A doubly-controlled induction machine may be designed with split-phase rotor and stator windings, each being fed from a standard inverter module. The number of split phase windings used will depend on the required power rating of the drive. Such an approach towards development of very high power drives can prove advantageous compared to the current trend for using multilevel converters.
188
References
[1] P.C.Sen and K.H.J.Ma, Rotor Chopper Control for Induction Motor Drive: TRC Strategy, IEEE Trans. Ind. Appl., Vol. 11, No. 1, pp. 43-49, Jan/Feb 1975 [2] M.S.Erlicki, Inverter Rotor Drive of an Induction Machine, IEEE Trans. Pow. App. and Sys, Vol. 84, No. 11, pp. 1011-1016, Nov 1965 [3] A.Lavi and R.J.Polge, Induction Motor Speed Control with Static Inverter in the Rotor,, IEEE Trans. Pow. App. and Sys., Vol. 85, No. 1, pp. 76-84, Jan 1966 [4] W.Shepherd and J.Stanway, Slip Power Recovery in an Induction Motor by the Use of a Thyristor Inverter, IEEE Trans. Ind. and Gen. Appl., Vol. 5, No.1, Jan/Feb 1969 [5] T.Wakabayashi, T.Hori, K.Shimzu, and T.Yoshioka, Commutatorless Kramer Control System for Large Capacity Induction Motors for Driving Water Service Pumps, Conf. Rec. 1976 IEEE/IAS Anual Meet, pp. 822-828 [6] W.E.Long and N.L.Schmitz, Cycloconverter control of the doubly fed induction motor, IEEE Trans. Ind. and Gen. Appl., Vol. 7, No. 1, pp. 95-100, Jan/Feb 1971 [7] H.W.Weiss, Adjustable Speed AC Drive Systems for Pump and Compressor Applications, IEEE Trans. Ind. Appl., Vol. 10, No. 1, pp. 162-167, Jan/Feb 1974 [8] A.Chattopadhyay, An Adjustable-speed Induction Motor Drive with a Cycloconverter-Type Thyristor-Commutator in the Rotor, IEEE Trans. Ind. Appl., Vol. 14, No. 2, pp. 116-122, March/April 1978 [9] C.B.Mayer, High Response Control of Stator Watts and Vars for Large Wound Rotor Induction Motor Adjustable Speed Drives, Conf. Rec. 1979 IEEE/IAS Anual Meet., pp. 817-823 [10] G.A.Smith and K.A.Nigim, Wind-energy recovery by a static Scherbius induction generator, IEE Proc., Vol. 128, Pt. C, No. 6, pp. 317-324, Nov 1981 [11] P.G.Holmes and N.A.Elsonbaty, Cycloconverter-excited divided-winding doubly-fed machine as a wind-power converter, IEE Proc., Vol. 131, Pt. B, No. 2, pp. 61-69, March 1984 [12] W.Leonhard, Control of Electrical Drives, Springer-Verlag, 1985 [13] P.Vas, Sensorless Vector and Direct Torque Control, Oxford University Press, 1998.
189
References [14] M.Yamamoto and O.Motoyoshi, Active and Reactive Power Control for Doubly-Fed Wound Rotor Induction Generator, IEEE Trans. Power Electronics, Vol. 6, No. 4, pp. 624-629, Oct 1991 [15] Y.Tang and L.Xu, A Flexible Active and Reactive Power Control Strategy for a Variable Speed Constant Frequency Generating System,, Conf. Rec. 1993 IEEE/IAS Anual Meet., pp. 568-573 [16] R.Pena, J.C.Clare, G.M.Asher, Doubly fed induction generator using back-to-back PWM converters and its application to variable-speed wind-energy generation, IEE Proc., Vol. 143, Pt.B, No. 3, pp. 231-241, May 1996 [17] E.Bogalecka, Power Control of a Double Fed Induction Generator without Speed or Position Sensor, Conf. Rec. EPE Brighton 1993, 8, (377), Chap. 50, Pt. 8, pp. 224-228. [18] L.Xu and W.Cheng, Torque and Reactive Power Control of a Doubly fed Induction Machine by Position Sensorless Scheme, IEEE Trans. Ind. Appl., Vol. 31, No. 3, pp. 636-642, May/June 1995 [19] L.Morel, H.Godfroid, A.Mirzaian, J.M.Kauffmann, Double-fed induction machine: converter optimisation and field oriented control without position sensor, IEE Proc., Vol. 145, Pt. B, No.4, pp. 360-368, July 1998 [20] M.Depenbrock, Direct Self Control (DSC) of Inverter-fed Induction Machine, IEEE Trans. Power Electronics, Vol. 3, No. 4, pp. 420-429, Oct 1988 [21] I.Takahashi and T.Noguchi, A New Quick Response and High Efficiency Control Strategy of an Induction Motor, IEEE Trans. Ind. Appl., Vol. 22, No. 5, pp. 820-827, Sept/Oct 1986 [22] I.Takahashi and Y.Ohmori, High Performance Direct Torque Control of an Induction Motor, IEEE Trans. Ind. Appl., Vol. 25, No. 2, pp. 257-264, March/April 1989 [23] J.N.Nash, Direct Torque Control of Induction Motor without an Encoder, IEEE Trans. Ind. Appl., Vol. 33, No. 2, pp. 333-341, March/April 1997 [24] B.K.Bose (Ed.), Microcomputer control of power electronics and drives, IEEE Press, 1987 [25] Irfan Ahmed (Ed.), Digital Control Applications with the TMS320 Family - Selected Application Notes, Texas Instruments, 1991 [26] A.M.Khambadkone, R.Datta and V.T.Ranganathan, Modeling and Rotor Current Control of Doubly-fed Induction Machine with Complex Signal Flow Graphs, Conf. Rec. 1998 PEDES, Vol. 2, pp. 972-977
190
References [27] O.Stihi and B.T.Ooi, A single-phase controlled-current PWM rectifier, IEEE Trans. Power Electronics, Vol. 3, No. 4, pp. 453-459, Oct 1988 [28] R.Wu, S.B.Dewan, and G.R.Slemon, A PWM ac-to-dc converter with fixed switching frequency, IEEE Trans. Ind. Appl., Vol. 26, No. 5, pp. 880-885, Sept/Oct 1990 [29] J.Holtz and L.Springob, Reduced harmonics PWM controlled line-side converter for electric drives, IEEE Trans. Ind. Appl., Vol. 29, No. 4, pp. 814-819, July/Aug 1993 [30] K.Thiyagarajah, V.T.Ranganathan and B.S.R.Iyengar, A High Switching Frequency IGBT Rectifier/Inverter System for AC Motor Drives Operating from Single Phase Supply, IEEE Trans. Power Electronics, Vol. 6, No. 4, pp. 576-584, Oct 1991 [31] T.Ohnuki, O.Miyashita, T.Haneyoshi, and E.Ohtsuji, High Power Factor PWM Rectifiers with an Analog Pulsewidth Prediction Controller, IEEE Trans. Power Electronics, Vol. 11, No. 3, pp. 460-465, May 1996 [32] T.Noguchi, H.Tomiki, S.Kondo, and I.Takahashi, Direct Power Control of PWM Converter Without Power-Source Voltage Sensors, IEEE Trans. Ind. Appl., Vol. 34, No. 3, pp. 473-479, May/June 1998 [33] B.K.Bose (Ed.), Power Electronics and Variable Frequency Drives, IEEE Press, 1997 [34] K.Rajashekara, A.Kawamura and K.Matsuse (Ed.), Sensorless Control of AC Motor Drives, IEEE Press, 1996 [35] Semikron, Power Electronics Data Manual, 1997/1998 [36] Hewlett Packard, Isolation and Control Components Designers Catalog, 1996 [37] Telcon, Low Profile Hall Effect Current Transformers Data Sheet, 1994 [38] G.Narayanan, R.Datta and V.Ramanarayanan, A High Performance Isolated AC/DC Voltage/Current Transducer, Conf. Rec. 1996 International Conference on Instrumentation ICI, Bangalore, India, pp. 343-347 [39] Texas Instruments, TMS32024x DSP Controllers - CPU, System, and Instruction Set, 1997 [40] Texas Instruments, TMS32024x DSP Controllers - Peripheral Library and Specific Devices, 1997 [41] Burr Brown, Mixed Signal Products IC Data Book, 1996 [42] Cypress, Semiconductor Data Book CD-ROM, 1999 [43] Texas Instruments, The Interface Circuits Data Book for Design Engineers, 1981
191
References [44] Suresh R, Urja Bharati - Special Report on Wind Energy, Ministry of Non-coventional Energy Sources, India, Vol.6, April 1995 [45] D.F.Warne, Generation of electricity from the wind, Proc. IEE, Vol. 124, No. 11R, pp. 963-985, Nov 1977 [46] C.V.Nayar, Small scale wind electricity generation - design criteria, TIDE (Tata Energy Research Institute Information Digest on Energy), Vol. 3, No. 2, pp. 111-123, June 1993 [47] C.V.Nayar, Wind Power - The near term commercial renewable energy source, Australian Science, Summer Issue, Vol.16, No.4, pp. 25-26, 1995 [48] C.V.Nayar, S.J.Philips, W.L.James, T.L.Proyor and D.Remmer, Novel Wind/Diesel/Battery Hybrid Energy System, International Journal of Solar Power, Vol. 51, No. 1, pp. 65-78, June 1993 [49] Getting Connected - Integrating Wind Power with Electric Utility Systems, Report of American Wind Energy Association, 1997 [50] Vestas, Danish Wind Technology V27-225 kW, 50 Hz Windturbine Product Brochure, 1994 [51] A.Miller, E.Muljadi and D.S.Zinger, A variable speed wind turbine power control, IEEE Trans. Energy Conversion, Vol.12, No.2, pp. 181-187, June 1997 [52] Kenetech Wind Power Technology - A New Generation of Power, Product Brochure of KVS-33 Wind Turbine, Published by Wind Energy Division, Abban Loyd Chiles Offshore Limited [53] A.K.Unnikrishnan, R.Sudeepkumar, P.S.Chempakavally, G.Poddar, A.Joseph and S.Bindu, A Variable Speed Power Controller for Wind Turbines, Proc. Workshop on Wind Power Generation and Power Quality Issues, Thiruvananthupuram, 1999 [54] R.Pena, J.C.Clare, G.M.Asher, A doubly fed induction generator using back-to-back PWM converters supplying an isolated load from a variable speed wind turbine, IEE Proc., Vol. 143, Pt. B, No. 5, pp. 380-387, Sept 1996 [55] P.W.Carlin and L.J.Fingersh, Some Analyses of Energy Production from the NWTC Variable Speed Test Bed, Conf. Rec. 1999 AIAA Wind Energy Symposium, pp. 1-10 [56] L.J.Fingersh, P.W.Carlin, Results from the NREL Variable-Speed Test Bed, Conf. Rec. 1998 AIAA Wind Energy Symposium, pp. 233-237 [57] S.Heier, Grid Integration of Wind Energy Conversion Systems, John Wiley & Sons, 1998 [58] Wind Energy Information Guide, Published by US Department of Energy, 1996 [59] NGEF Ltd., Standard Induction Motors Catalogue, 1995 192
References [60] D.S.Zinger and E.Muljadi, Annualized Energy Improvement Using Variable Speeds, IEEE Trans. Ind. Appl., Vol. 33, No. 6, pp. 1444-1447, Nov/Dec 1997 [61] S.Roy, Optimal Planning of Wind Energy Conversion Systems Over an Energy Scenario, IEEE Trans. Energy Conversion, Vol.12, No.3, pp. 248-254, Sept 1997 [62] M.G.Simoes, B.K.Bose and R.J.Spiegel, Design and Performance Evaluation of a Fuzzy-Logic-Based Variable-Speed Wind Generation System, IEEE Trans. Ind. Appl., Vol. 33, No. 4, pp. 956-965, July/Aug 1997 [63] Y. Kawabata, E. Ejiogu, and T. Kawabata, Vector-Controlled Double-Inverter-Fed Wound-Rotor Induction Motor Suitable for High-Power Drives, IEEE Trans. Power Electronics, Vol. 35, No. 5, pp. 1058-1066, Sept/Oct 1999
193
Appendix A
irb
is
ira
The doubly-fed wound rotor induction machine has symmetrical three phase coil systems both on the stator and the rotor [Fig.A1(a)], which can be represented by two equivalent two phase coil systems [Fig.A1(b)]. The rotor axis makes an angle (t) with respect to the stator axis. The current space phasors defined for the stator and the rotor currents in their own reference frames, namely i s (t) and i r (t) respectively can be written as i s (t) = i sa (t) + j i sb (t) i r (t) = i ra (t) + j i rb (t) where i sa = 3 2 i s1 i sb = and 3 2 i s2 i s 3 (A1) (A2) (A3) (A4) (A5) 195
i ra = 3 2 i r1
It is assumed that both the stator and rotor windings have isolated neutrals so that i s1 (t) + i s2 (t) + i s3 (t) = 0 i r1 (t) + i r2 (t) + i r3 (t) = 0 The flux linkages of the stator coils along the and axes are given by y sa = L s i sa + L o i ra cos e L o i rb sin e y sb = L s i sb + L o i ra sin e + L o i rb cos e (A9) (A10) (A7) (A8)
where L s is the self inductance of the stator coils and L o is the maximum value of the mutual inductance between the stator and rotor coils. Equations (A9) and (A10) can be combined to get the stator flux linkage space phasor as y s = y sa + j y sb = L s (i sa + j i sb ) + L o (i ra + j i rb ) (cos e + j sin e) = L s i s + L o i r e je Similarly the rotor flux linkage space phasor can be written as y r = y ra + j y rb = L r i r + L o i s e j e (A12) (A11)
where L r is the self inductance of the rotor coil. (It is assumed for the sake of simplicity that the rotor coils have the same number of turns as the stator coils.) The voltage equations for the stator and rotor coils can be written as dy u sa = R s i sa + dt sa dy u sb = R s i sb + dt sb dy u ra = R r i ra + dt ra 196 (A13) (A14) (A15)
where R s is the stator resistance and R r is the rotor resistance. Combining Eq.(A13) with Eq.(A14) and Eq.(A15) with Eq.(A16), the complex voltage space phasors can be derived. dy u s = R s i s + dt s dy u r = R r i r + dt r (A17) (A18)
Substituting y s and y r from Eq.(A11) and Eq.(A12) into Eq.(A17) and Eq.(A18) respectively gives d L i + L i e je u s = R s i s + dt s s o r d i + L d i e je = R s i s + L s dt s o dt r d L i + L i e j e u r = R r i r + dt r r o s d i + L d i e j e = R r i r + L r dt r o dt s (A20) (A19)
Equations (A19) and (A20) represent the electrical dynamics of the stator and rotor circuits in their respective coordinate systems. The instantaneous electromagnetic torque developed by the machine is given by P L Im i i e je & md = 2 s r 3 2 o i r e je & (A21)
represents the complex conjugate of the rotor current space phasor in the stator coordinate
system. Therefore the complete set of equations that describe the behaviour of the machine is as follows. d i + L d i e je = u R s i s + L s dt s o dt r s d i + L d i e j e = u R r i r + L r dt r o dt s r z + B z = 2 P L Im i i e je & Jd s r 3 2 o dt de = P z = z e 2 dt ml (A22a) (A22b) (A22c) (A22d)
197
Appendix A Machine Model in Stationary Coordinates The rotor current and rotor voltage space phasors in the stator reference frame can be written as follows. s i r = i r e je s u r = u r e je (A23) (A24)
With the new representation of the rotor current vector in the stationary coordinates, the stator and rotor voltage equations i.e. Eq.(A19) and Eq.(A20) may be rewritten as d i + L d si R s i s + L s dt s o dt r and, or, = us (A25)
d s i e j e + L d i e j e = u R r s i r e je + L r dt r o dt s r d s i e je L s i j de e je R r s i r e je + L r dt r r r dt d i e je L i j de e je = u + L o dt s o s dt r (A26)
Even though Eq.(A26) is written in terms of the stator and rotor current space phasors in the stationary coordinates, it is still in the rotor reference frame. To transform it to the stator reference frame both sides have to be multiplied by e je . Thus, d si j z L si + L d i j z L i = su R r s i r + L r dt r e r r o dt s e o s r (A27)
Now, Eq.(A25) and Eq.(A27) depict the electrical dynamics of the machine in the stationary coordinates. The next step is to represent this machine model in the standard state-variable form suitable for simulation using available software platforms. Substitution of d ( s i r ) from Eq.(A27) in Eq.(A25) and, d i s from Eq.(A25) in Eq.(A27) dt dt and subsequent simplification results in the following. rT s di s us Ts s = R i s j z e (1 r)T s i s j z e (1+r ir dt s s) T 1 si 1 su + T s (1+ r r ) ( r s R s 1+r r ) r (A28)
198
r s and r r are the defined as the leakage factors for the stator and the rotor respectively, and r is the total leakage factor. These complex equations can be split into real and imaginary parts using the following definitions. i s = i sa + ji sb s i r = i ra + ji rb u s = u sa + ju sb s u r = u ra + ju rb (A31a) (A31b) (A31c) (A31d)
After substitution and separation of the real and imaginary parts, we get the electrical circuit equations in state-space form as given below. = AX + BU X where X = i sa i sb i ra i rb r1 T T (A32a)
A =
z e (1+r ) r
(A32b)
199
Appendix A Machine Model in Stationary Coordinates 1 rL s B = 0 1 r(1+r s )L r 0 0 1 rL s 0 1 r(1+r s )L r T 1 r(1+r r )L s 0 1 rL r 0 0 1 r(1+r r )L s 0 1 rL r (A32c)
U = u sa u sb u ra u rb
(A32d)
The electromagnetic torque developed can be derived from Eq.(A21) as follows. P md = 2 3 2 L o Im i sa + j i sb P = 2 3 2 L o i sb i ra i sa i r b i ra + j i rb &
(A33)
(The model of a squirrel cage induction machine becomes a special case of Eq.(A32) when u ra = 0 and u rb = 0.)
200
Appendix B
Various Base and Per Unit Values are Base voltage = 415 V = 239.6 V 3 Base current = 6.26 A Base impedance = 239.6 6.26 W = 38.34 W Base power = 3 $ 6.26 $ 239.6 W = 4500 W Base angular frequency = 2 $ o $ 50 = 314.16 rad/s Base torque = 4500 = 28.65 Nm. ) (2 4 $314.16 Electrical Parameters of the Machine Nominal Parameters Stator Resistance (R s ) Rotor Resistance (R r ) Stator Inductance (L s ) Rotor Inductance (L r ) Magnetizing Inductance (L o ) Stator Leakage Factor (r s ) Rotor Leakage Factor (r r ) Total Leakage Factor (r ) Values in SI Units 1.557 W 2.62 W 195 mH 195 mH 177 mH 0.1017 0.1017 0.1761 Values in p.u. 0.0406 0.0683 1.5978 1.5978 1.4503 0.1017 0.1017 0.1761
201
202
Appendix B Power Circuit Components (In the experiments 17 mH is connected in opposition to 100 H tapping; however, the actual measured value of inductance with LCR meter comes to 17.9 mH.) Various base and per unit values for the front end converter are AC side base voltage = 125 V = 72.169 V 3 AC side base current = 8.5 A Base impedance = 72.169 8.5 W = 8.49 W DC side base voltage = 300 V DC side base current = 6.133 A Base power = 3 $ 72.169 $ 8.5 W = 1840 W $17.9%10 3 = 0.6624 Per unit ac side inductance = 2$o$508.49
203
Drive: 4 quadrant AUTOCON drive from Autodata Three phase fully controlled anti parallel bridge for four quadrant operation. Input 125V, 3 phase, 50 Hz input Current rating 45 A.
204
Appendix C
Appendix C MATLAB Data Files P=4 Kc = -2/3*1/(1+SigmaS)*P/2 B = 0.0161 J = 0.1 Ts = Ls/Rs Tr = Lr/Rr Tm = J/B
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Sensor Gains (machine end) %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Ki = 1/3.55 Kv = 1/32.7 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Inverter (machine end) %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Vdc = 300 % Maximum modulation index MImax = 0.85 % Maximum allowable slip Smax = ((Vdc/(2*1.414))*MImax)/240 % Gain-factor machine side Gr = MImax/Smax
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Sine Pulse Width Modulator %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FreqTri = 4464 T = 1/FreqTri %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Machine side Controller Parameters %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Rotor current controller time constant Tir_d =0.004 %Desired d_axis time constant Tir_q =0.001 %Desired q_axis time constant
206
Appendix C MATLAB Data Files Kpir_d = (Sigma*Tr/Tir_d)*Rrpu Kpir_q = (Sigma*Tr/Tir_q)*Rrpu % Speed controller time constant Tw = 100e-3 % Rotor current limits (25% margin) IrdMax = 1.25 IrqMax = 1.25 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Base Quantities for machine side control %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Usbase = 1.414*240 Isbase = 1.414*6.26 Irbase = 1.414*6.26 Zbase = Usbase/Isbase Wbase = 157.08*2 Tbase = 20e-3 Rspu = Rs/Zbase Rrpu = Rr/Zbase Xspu = 2*pi*50*Ls/Zbase Xrpu = 2*pi*50*Lr/Zbase Xopu = 2*pi*50*Lo/Zbase Imsrated = 1/(Xspu)*Isbase*1.5 Mdbase = -Kc*Lo*Imsrated*(Irbase*1.5) Tipu = Ti/Tbase %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % END FILE % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %Proportinal gain - d_axis %Proportinal gain - q_axis
207
C.2 Power Circuit and Controller Parameters used for Simulation and Implementation of Front end Converter Control
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % Data File for Simulation of Front-end Converter % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % 3ph Source %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Vph = 125*(1.414/1.732) f = 50 w = 2*pi*f r = 2*pi/3 %peak of ac side pahse voltage
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Power Circuit %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Lfe = 17.9e-3 Rfe = 0.64 C = 4000e-6 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Sensor Gains (front end) %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Kife = 1/3.55 Kvfe = 1/32.7 Kidc = 1/3.55 Kvdc = 1/32.7 %ac side current sensor gain %ac side voltage sensor gain %dc side current sensor gain %dc side voltage sensor gain
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Base Quantities for front end converter control %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Uacbase = 125/1.732 Isbase = 8.5*1.414 Zbase = Uacbase*1.414/Isbase Vdcbase = 300 208
Appendix C MATLAB Data Files Idcbase = 6.133 Rpu = R/Zbase Tfe = Lfe/Rfe Tl = C*Vdcbase/Idcbase %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Front-end Controller Parameters %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %Voltage Controller Tvfe = 110e-3 Kpvfe = 1.516 %Current Controller Tpife = Tfe Tife = 0.002 Kpife = (Tfe/Tife)*Rpu
%PI time constant %Desired time constant of the current loop %PI proportional gain
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Inverter (machine end) %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Gfe = Uacbase*1.414/(Vdcbase/2) %Inverter Gain - trinagle peak is 1 p.u.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Sine Pulse Width Modulator %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% FreqTri = 4464 T = 1/FreqTri %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % END FILE % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
209
Appendix D
Swept Area: 573 m2 Rotational speed, generator 1: 43 rpm Rotational speed, generator 1: 33 rpm Number of blades: Cut-in speed: Rated wind speed (225 kW): Cut-off wind speed: Survival wind speed: 3 3.5 m/s 14 m/s 25 m/s 56 m/s
Gearbox Nominal power: Ratio: Generator1 225 kW, 400 V, 396 A, 50 Hz, 1008 rpm, 163 kVAR Generator2 50 kW, 400 V, 101 A, 50 Hz, 760 rpm, 48 kVAR 433 kW 1:23.4
211
Appendix E
Fig.E.1 Ripple current in the dc link capacitor The reactive component of the fundamental load current, when reflected on the dc link, appears as shown in Fig.E.1. In order to derive a simplified expression for the ripple in the dc bus voltage, this current waveform can be approximated as a saw-tooth waveform (shown as dotted line in the Fig.E.1). From t = 0 to t = T s /12, the capacitor current can then be expressed as ic = Ip t T s /12 (E.1)
where, I p is the peak of the current waveform and T s is the time period of the fundamental cycle. The voltage ripple Du dc can therefore, be calculated as I p .T s Du dc = 24C (E.2)
If the allowable voltage ripple is RPU.u dc (where RPU is the p.u. ripple in the dc voltage), then the dc bus capacitance required can be derived by using Eq.(E.2). The value of C should be estimated for the maximum reactive power which is drawn by the load at the rated frequency.
213
Appendix E Calculation of DC Bus Capacitance Ip 1 C = RPU.u dc 24.f rated 2.I nl 1 C = RPU.u dc 24.f rated The rms value of the ripple current can be derived as i c,rms = = 1 Ts Ip 3 T 0
s
(E.3)
For an induction machine, I p is approximately the peak of the no-load current. Therefore, (E.4)
I p 2 1/2 T s t dt (E.5)
214