A Variable Speed Drive

Published Date: 02 Nov 2017

Introduction

1.1 Variable Speed Drive:

Rotational industrial loads require operation at any one of a wide range of operating speeds. Such loads are generally termed as variable speed drives or adjustable speed drives. The variable speed drive systems are also an integral part of automation. They help to optimize the process and to reduce investment costs, energy consumption, and energy cost. The system efficiency can be increased by the introduction of variable speed drive operation in place of constant speed operation [1, 2].

There are three basic types of variable speed drive systems: electrical drives, hydraulic drives and finally mechanical drives. In this thesis, only electrical drives are focused. Drives employing electric motors are known as electrical drives. Block diagram of an electric variable speed drive system is shown in Fig 1.1. It consists of three basic components: the electric motor, the power electronic converter and the control system. The electric motor is connected directly or indirectly (through gears) to the load. The power electronic converter controls the power flow from power supply to the motor by appropriate control of power semiconductor switches. With recent advances in power semiconductor devices and converter topologies, electric variable speed drives are witnessing a revolution in a wide variety of applications such as machine tools and robotics drives, fans, pumps, compressors, paper mill, steel industries, automation, traction applications, ship propulsion and cement mills.

Power Supply

Power Electronic Converter

Gear System

Drive Control Unit

Load

Supervisory Control System

Electric Motor

Fig.1.1. Block diagram of an electric variable speed drive system

1.2 Classification of Variable Speed Drives:

According to the type of electric motor, the electric variable speed drives can be classified into two categories.

DC motor drives

AC motor drives

1.2.1 DC Motor Drives:

From the characteristic analysis, the separately excited DC machines were the right choice for applications in variable speed drives, where good dynamic response and steady state performance are required. The speed control of separately excited DC motor is very straightforward, mainly because of the commutator within the motor. The commutator and brush allow the developed torque of the motor to be proportional to the armature current if the field current is held constant. The DC machines also have the simple and excellent dynamic performance over a wide range of operating conditions due to inherent decoupling between field flux and armature current. Applications are steel industries, robotic drives, printers, machine tools, textile and paper industries, etc. On the other side, dc machines are heavy; require frequent maintenance, with low value of torque-to-weight ratio, in addition to having severe commutation problems. Moreover, the mechanical commutator limits the maximum applicable voltage to about 1500V and the maximum power capacity to a few hundred kilowatts. The commutator also limits the maximum armature current and its rate of change.

1.2.2 AC Motor Drives:

AC motors exhibit highly coupled, nonlinear and multi variable structures as opposed to much simpler decoupled structures of separately excited DC motors. The AC motors have a number of advantages: light weight, inexpensive and have low maintenance compared with DC motors. They require control of frequency, voltage and current for variable speed applications. However, the advantages of AC drives outweigh the disadvantages. AC drives are replacing DC drives and are used in many domestic and industrial applications. The AC motor drives can be classified into two categories.

Induction motor drives

Synchronous motor drives

1.2.2.1 Induction Motor Drives:

The induction motors drives can be classified into two types, namely

Squirrel-cage induction motor drives

Slip-ring induction motor drives

Both the motors are electrically equivalent as long as attention is confined to the fundamental sine-waves of voltage, current, flux, etc except the former has rotor-winding terminals permanently shorted inside the motor. In case of slip-ring induction motor, the terminals of the rotor three-phase winding are externally available to the user.

1.2.2.1.1 Squirrel-cage Induction Motor Drives:

The Nikola Tesla exhibited a crude type of three-phase induction motor at the Frankfort exhibition of 1891. An improved construction, with a distributed stator winding and a cage rotor, was built by Dolivo Dobrowolsky in conjunction with the Maschinenfabrik Oerlikon and described in 1893. This motor is by far the most widely used motor in the industry. Traditionally, it has been used in constant and variable speed drive applications that do not cater for fast dynamic processes. Because of recent development of several new control technologies, such as vector control, sensorless control and direct torque control (DTC), the situation is changing rapidly. Squirrel-cage induction motors are much cheaper and more rugged than the dc motor. They require little maintenance. They can be designed as totally enclosed motors to operate in dirty and explosive environments. All these features make them attractive for use in industrial drives. The some of speed control methods are listed below, which are widely used.

Scalar control

Vector control or Field Oriented Control (FOC)

Sensorless control

Direct Torque Control (DTC)

1.2.2.1.2 Slip-ring Induction Motor Drives:

The slip-ring induction motors with three rotor slip rings have been used in adjustable speed drives for many years. In early slip-ring induction motor drives, adjustable speed is achieved by dissipating the energy in external resistances, connected to the slip-ring terminals of the rotor. Modern slip-ring induction motor drives use an inverter to recover the power from the rotor circuit, feeding it back to the supply system. So, the speed control methods employed for slip-ring induction motor drives are

Rotor resistance control

Slip power recovery schemes (Static Kramer drive & Static Scherbius drive)

Generally, slip-ring induction motors are used for high power applications where a small speed range is required.

1.2.2.2 Synchronous Motor Drives:

The speed of synchronous motors with constant rotor excitation is determined by the stator supply frequency and the number of poles. So, a variable frequency static inverter can extend its operation as a variable speed drive. The main applications are gearless rolling mills, mine hoists, traction, etc.

In this thesis, main attention is given to squirrel-cage induction motor drives only.

1.3 Control Strategies for Squirrel-cage Induction Motor

Induction motors are known as workhorses of industry. This is because they are most widely used motors due to their lower cost, rugged construction and high power to volume/weight ratio. When operated directly from the ac line voltage, induction motor operates nearly at constant speed. However by means of power electronic converters, it is possible to change the speed of an induction motors. Even though the induction motors are desirable, their speed control is not as straight forward as that of a dc motor. Therefore, it was natural for the researchers to think of ways, which would take the induction motor control closer to that of a dc motor. The various speed control methods, which are used to control the speed of induction motors discussed in this section.

1.3.1 Volts/Hz Control of Induction Motor:

The volts/Hz control of induction motor is by far the most frequently held method of speed control because of its straight forwardness, and these types of motors are frequently used in industry [2, 3]. In this control, for adjustable speed applications, supply frequency is varied. However, it is required to maintain voltage to be proportional to variable frequency so that the flux remains constant, neglecting the stator resistance drop. Hence, in this method, is held constant. In steady state operation, the machine air gap flux is approximately related to. As the frequency nearly approaches zero, near zero speed, the magnitude of the stator voltage proportional to varied frequency also tends to zero and this low voltage is absorbed by the stator resistance. Therefore, at low speed by injecting the boost voltage the stator resistance drop is compensated, so that rated air gap flux and hence the full load torque is available up to zero speed. At steady state operation, if load torque is increased, the slip increases within stability limit and a balance will be maintained between the developed torque and the load torque.

Problems with Volts/Hz Control

If the supply voltage to the inverter fluctuates, the air gap flux will vary.

Also, increase in stator resistance with temperature results in variation in air gap flux. Hence, in constant control scheme torque sensitivity with slip frequency or stator currents are affected by the air gap flux drift. If correct ratio is not maintained, the flux may be weak or may saturate.

Torque pulsations are present at low speeds owing to presence of fifth, seventh and eleventh and higher harmonics.

Because of the presence of low frequency harmonics, the motor losses are increased at all speeds causing the derating of the motor.

These drawbacks can be overcome with the help of vector control technique where an induction motor is controlled on the same principles as a separately excited dc motor in which torque component and the flux component are decoupled.

1.3.2 Vector Control of Induction Motor:

In 1971, a new control strategy was proposed by F. Blaschke, in which the control of induction motor is similar to that of a separately excited dc motor, called Field Oriented Control (FOC) or vector control. As in the dc machines, torque control in ac machines is achieved by controlling the motor currents. However, in contrast to a dc machine, current phasor has to be controlled. Due to this scheme the terminology ‘vector control’ is pronounced. In the vector control, the induction motor is assessed as a synchronously rotating reference frame, in which the sinusoidal variables perform as dc quantities. To attain a good dynamic response the torque and the flux components are determined and regulated independently. Vector controlled techniques incorporating fast microprocessors and DSPs have made possible the application of induction motors for high performance applications where traditionally only dc drives were applied. However, it should be noted that:

In dc machines, the armature current and main flux distribution are fixed in space and where the torque can be established by independently controlling the excitation flux and armature current.

Where as in ac machine, it is much more difficult to realize this principle because these quantities are coupled and are stationary with respect to the stator and rotor. They also depend on modulus, frequency and phase angles of stator current.

There are essentially two general methods of vector control [2]. They are:

Direct vector control or Feedback method, developed by F. Blaschke

Indirect vector control or Feedforward method, developed by K. Hasse.

These two methods are different essentially by how the unit vector is generated for the control.

1.3.2.1 Direct Vector Control of Induction Motor:

The direct vector control depends on the generation of unit vector signals from the stator or rotor flux signals. The air-gap flux signals can be measured directly or estimated from the stator voltage or current signals. In these systems, rotor speed is not required for obtaining rotor field angle information. Here, the actual motor currents are converted to synchronously rotating frame currents using park’s transformation. The resulting dc quantities are compared with the reference d-axis and q-axis components. The outputs of the controllers are used to generate the pulse width modulated signals for switching the devices in the inverter bridge feeding the motor. The main disadvantages of this method are:

Direct method of rotor flux estimation depends on the machine parameter; the rotor resistance variation, especially, becomes dominant due to the temperature variation and skin effect.

The direct method of vector control can be applied typically above 10% of the base speed because of difficulty in accurate flux signal synthesis at low speeds.

Hence, due to these disadvantages, normally indirect method of vector control is preferred.

1.3.2.2 Indirect Vector Control of Induction Motor:

In this method, the unit vector signal that transforms the synchronously rotating stator voltages into stationary frame signals has been generated from the speed signal and slip signal. The drive can easily be operated from zero speed to constant power field-weakening region. It is the most popular vector control method in industry. The main disadvantages of this method are:

The machine parameter variation affects the slip gain, and correspondingly, both static and dynamic performances of the drive are affected.

The on-line tuning for parameter variation is more difficult.

The control methods discussed so far require a speed sensor for closed operation. The speed sensor has several disadvantages from standpoint of drive cost, reliability and noise immunity. The torque is controlled indirectly.

1.3.3 Sensorless Vector Control of Induction Motor:

Sensorless vector control of an induction motor drive practically implies vector control without any type of speed sensor [2, 4]. Here the terminology ‘sensorless’ describes just the speed and shaft sensors. It is possible to evaluate the speed signal from tacho-generator terminal voltages and currents. The speed estimation methods can generally be classified as follows:

Slip calculation

Direct synthesis from state equations

Model referencing adaptive system (MRAS)

Speed adaptive flux observer

Extended kalman filter (EKF)

Slot harmonics

Many of the sensorless techniques depend on the machine parameters, temperature, saturation levels, etc.

1.3.4 Direct Torque Control of Induction Motor:

In the mid 1980s the Direct Torque Control (DTC) principle was developed by Takahashi and Noguchi for low and medium power applications and Direct Self Control (DSC) principle was established by Depenbrock for high power applications. As the name suggests, DTC or DSC regulates the motor torque and flux directly. In the DTC approach, the reference torque and reference flux are compared to the estimated motor torque and the estimated stator flux respectively, both employing hysteresis controllers. The torque and flux hysteresis controller output logic signals are evaluated in an optimal switching logic table to generate the inverter switching device gate signals. The generation of inverter switching state is made to restrict the stator flux and electro-magnetic torque errors within the flux and torque hysteresis bands and to acquire the fastest torque response and highest efficiency at every switching instant. The DTC scheme is found to be very promising and valuable as compared to FOC. But, DTC has few drawbacks such as more steady state ripple in flux, torque and current and variable switching frequency operation of the inverter due to hysteresis bands.

1.4 Literature Review and State of the Art Assessment:

1.4.1 Introduction:

The one and a half century of progress in the electric machines field, about three quarters of a century of progress in the power electronics field, and about half a century of progress in the micro-electronics/macro-electronics and control fields are inherited in the state of the art pulsewidth modulated voltage source inverter (PWM-VSI) drives. Since they involve various disciplines of engineering and there has always been a strong demand for them in the market, PWM-VSI drives have continuously drawn the attention of many researchers all around the world. Among the various PWM-VSI drives, the induction motor drives with cage type machines have found wide range of applications and have become the workhorse of industry because of their simplicity and ruggedness. Induction motors can be used as variable speed drives by feeding them with Current Source Inverters or voltage source inverters. Recent advances in semiconductor technology have resulted in new generations of fast- acting power semiconductor switches like GTOs, MOSFETs, IGBTs, and recently IGCTs. The performance and features of these switches highly favor the VSI topology over the CSI one. This has been a significant factor for VSI fed induction motor drives becoming more popular compared to CSI fed induction motor drives. Pulse Width Modulation (PWM) techniques are required for switching the devices in a VSI properly to produce variable voltage, variable frequency, 3-phase AC required for the variable speed induction motor drive.

Following a brief review of the various control techniques for induction motor drives and state of the art DTC of induction motor drives will be described and the fundamental contributions to the area will be discussed in detail.

1.4.2 Control Techniques for Induction Motor Drives:

The various speed control techniques for three-phase squirrel cage induction motors are

Constant Volts per Hertz Control

Vector Control or Field Oriented Control (FOC)

Sensorless Vector control

Direct Torque Control (DTC)

1.4.2.1 Constant Volts per Hertz Control:

The block diagram of volts per hertz control of induction motor is given in Fig. 1.2. In this method, the inverter output voltage is varied proportionally to the reference frequency such that constant stator flux is maintained. In an induction motor drive, this operating mode results in shunt speed-torque characteristics (linear portion of the torque-speed curve), yielding low slip frequency and therefore high energy efficiency and good speed regulation. Therefore, the method gained wide acceptance in many industrial and residential induction motor drive speed regulation applications as given by B.K. Bose [2], R. Krishnan [3], D.A. Bradley et al [6] and B. Mokrytzki [7-8].

IM

Diode Bridge Rectifier

L

C

Sine Triangle PWM

G

+

+

Vo

V*S

ω*S

AC supply

Fig. 1.2 Open-loop volts per hertz speed control of induction motor

The performance of the control is not satisfactory, because the rate of change of voltage and frequency has to be low. A sudden acceleration or deceleration of the voltage and frequency can cause a transient change in the current, which can result in drastic problems. Moreover, they exhibit limited speed response, poor load torque disturbance characteristic, and inferior low speed characteristics. Some efforts were made to improve control performance, but none of these improvements could yield a torque controlled drive systems and this made DC motors a prominent choice for variable speed applications. In printing press applications, packaging applications, servo applications with very high resolution position control etc, where precise control is mandatory, the performance of drives is not satisfactory. In such type of applications, traditionally DC motor drives have been employed with a shaft encoder. Typical application areas of drives are pumps, ventilation systems, etc. which have passive torque-speed characteristics and no precise speed regulation requirement.

1.4.2.2 Vector Control:

In the volts per hertz control, the voltage and frequency are the basic control variables of the induction motor. In a voltage fed drive, both the torque and air gap flux are functions of voltage and frequency. This coupling effect is responsible for the sluggish response of the induction motor and moreover, the system is easily prone to the instability. However, the continuous progress in induction motor control theory, power electronics, and digital signal processors yielded the modern vector controlled induction motor drives [9-15] which can match the performance and reliability characteristics of dc drives and cost less. The invention of vector control, which is also known as decoupling, orthogonal, transvector or field oriented control (FOC) in the beginning of 1970s, and the demonstration that an induction motor can be controlled like a separately excited dc motor, brought a renaissance in the high-performance speed control of induction motor drives. Modern vector controlled squirrel cage induction motor drives meet the demanding performance criteria of most high performance speed control applications. In the vector control method, the speed controlling of the ac machine is similar to the speed controlling of the separately excited dc machine. This analogy is explained by B.K. Bose [2] as in Fig. 1.3.

(a)

IM

Vector Control

Inverter

(b)

Fig. 1.3 (a) Separately excited dc motor (b) vector controlled induction motor

In a dc machine, neglecting the armature demagnetization effect and field saturation, the torque is given by

(1.1)

where is the armature or torque component of current and is the field or flux component of current. In a dc machine, the control variables and can be considered as orthogonal or decoupled "vectors". In normal operation, the field current is set to maintain the rated field flux and torque is changed by changing the armature current. Since the the corresponding field flux or filed current is decoupled from the armature current, the torque sensitivity is unaltered and remains maximum in both steady state and transient operations. This kind of control can be extended to an induction motor also if the machine operation is done in a synchronously rotating reference frame where the sinusoidal variables appear as dc quantities. In Fig. 1.3 the induction motor with inverter and vector control is shown with two control inputs, and . The currents and are the direct axis component and quadrature axis component, respectively, of the stator current, where both are in a synchronously rotating reference frame. In vector control is analogous to the armature current and is analogous to the field current of a dc machine. Therefore, from the basic DC machine equations the torque can be expressed as

(1.2)

Thus the similarity between the production of the electromagnetic torque in a compensated dc machine and in symmetrical, smooth air gap induction machine has been established. However, it should be noted that:

In dc machines, the armature current and main flux distribution are fixed in space and where the torque can be established by independently controlling the excitation flux and armature current.

Where as in an induction machine, it is much more difficult to realize this principle because these quantities are coupled and are stationary with respect to the stator and rotor. They also depend on modulus, frequency and phase angles of stator current.

The search for simple control schemes, similar to those used for dc machines has led to the development of "vector controlled schemes". There are essentially two general methods of vector control. They are:

Direct vector control or Direct Field Oriented Control (DFOC)

Indirect vector control or Indirect Field Oriented Control (IFOC)

These methods are differentiated on how the unit vector signals are generated from stator, rotor or air-gap flux signals. The DFOC method was presented by F. Blaschke [9] and it employs flux sensors. The IFOC method was presented by K. Hasse [10] and it employs a shaft encoder to close the speed loop.

In FOC, the magnetizing flux and torque producing components of the stator currents are properly and independently distributed both during steady state and dynamic conditions. As explained by D.W. Novotny and T.A. Lipo [11] and Rik W. De Doncker and D.W. Novotny [12], by regulating each component independently with a high performance current controller, the drive torque can be controlled in the same precise manner as the DC machine. Since installing flux sensors in the stator or the air gap of a machine is difficult, and the operation is not reliable, the DFOC method is practically rarely employed in its original form. Employing flux observers, the DFOC method provides high performance torque control, in particular in the high speed region where the stator resistance voltage drop is small compared to the stator EMF and the stator flux observer is highly accurate. The stator flux oriented DFOC method is attractive for traction, spindle tool etc, applications which require operation in a wide field weakening region. However, near zero speed the stator flux observer estimator error becomes substantial due to the dominance of the stator resistive voltage component over the nearly zero EMF and the DFOC method looses performance. In a large number of applications requiring high performance in the low speed operating region the rotor flux oriented IFOC method is utilized.

With accurate parameter adaptation, the IFOC based induction machine drives can provide servo performance in a wide speed region. Since the torque regulation quality of an FOC induction motor drive is mainly dependent on the current controller accuracy and bandwidth, high performance motion control requires high performance current regulators. The hysteresis type current controllers which have superior dynamic performance have not gained acceptance in motor drives due to the difficulty in controlling their switching frequency and significant waveform distortion. Employing high switching frequency IGBT devices and high performance digital signal processors or microprocessors, high performance current controlled drives provide high torque/speed bandwidth, hence high motion quality. High performance FOC drives have been successfully employed in industrial and servo drive applications which are summarized by T. Kume and T. Iwakane [13]. The evolution of FOC drives from concept to industrial products and successful applications has been summarized by W. Leonard in [14-15] in detail.

1.4.2.3 Sensorless Vector Control:

The principle of sensorless vector control induction motor drive is vector control without any shaft encoder or speed sensor. In closed loop speed or position control in vector control drives an incremental shaft mounted speed sensor; usually an optical type is used. A speed encoder adds cost and poses reliability problems, and also this requires for a shaft extension and mounting arrangement. To reduce total hardware complexity, cost and to increase mechanical robustness, it is desirable to eliminate speed and position sensors in vector-controlled drives. Drives operating at higher speeds or in hostile environments the speed sensors cannot be mounted. In substitute of speed sensor, the information of the rotor speed is derived from measured stator currents and voltages at the motor terminals. Continuing research has concentrated on the elimination of the machine speed sensor at the shaft without diminishing the dynamic performance of drive control system. Speed estimation is an issue of particular interest with induction motor drives where the mechanical speed of the rotor is generally different from the speed of the revolving magnetic field. The advantage of sensorless induction motor drives are elimination of the sensor cable, reduced hardware complexity, reduced cost, reduced size of the drive machine, increased reliability, better noise immunity and less maintenance requirements.

The pioneering work in the shaft encoderless motor speed control area was reported by R. Jötten and G. Maeder in 1983 [16]. They employed the induction motor fundamental model to estimate the slip frequency and the back emf of the machine and provided a closed loop controller to regulate the slip such that superior dynamic performance could be obtained in a wide speed region, including the field weakening region. Although a large variety of shaft encoderless control methods have been reported from that time to the present date, only a few found practical applications, which are given by B.K. Bose [2], Peter Vas [4], Tsugutoshi Othani et al [17], C. Ilas et al [18] and J. Holtz [19-20].

1.4.2.4 State of the Art DTC of Induction Motor Drives:

In addition to vector control systems, instantaneous torque control yielding fast torque response can also be achieved by employing Direct Torque Control (DTC) [21, 23] or Direct Self Control (DSC) [22, 24]. As the name suggests, the DTC method regulates the motor torque and flux directly. In the mid 1980s the DTC principle was developed and discussed by Isao Takahashi and T.Noguchi [21] for low and medium power applications and DSC principle was established by M. Depenbrock [22] for high power applications. In this thesis, the attention will be mainly focused on the DTC scheme. In the DTC approach, the reference torque is compared to the estimated motor torque and the reference stator flux is compared to the estimated stator flux, both employing hysteresis controllers. The torque and flux hysteresis controller output logic signals are evaluated in an optimal switching logic table to generate the inverter switching device gate signals. The generation of inverter switching state is made to restrict the electromagnetic torque errors and the stator flux linkage within the respective torque and flux hysteresis bands and to obtain the highest efficiency and fastest torque response at every instant. The DTC scheme is found to be very promising and valuable as compared to FOC. Moreover, using the DTC it is possible to obtain a good dynamic control of the torque without any speed sensors or position sensors on the machine shaft. Thus, DTC can be considered as sensorless type control techniques.

In DTC, the stator flux can be calculated from the motor terminal voltages and stator resistance. Variations in the stator resistance result in significant errors in the stator flux, especially at low speeds. This problem can be overcome by using the slip relation from indirect rotor flux field orientation to locate the position of the rotor flux. The rotor flux position is then used to locate the position of the stator flux. Also, the motor speed and rotor resistance are used to calculate the position of the stator flux at low speeds. With this, the advantages of DTC scheme are maintained over the entire speed range as explained by Thomas G. Habetler et al [25]. Moreover, the robust start and improved operation in the zero speed region can be achieved easily by introducing the additional carrier signal to the torque comparator input as given by Kazmierkowski and Kasprowicz [26].

Thus, unlike FOC, DTC operates with closed torque and flux loops but without current controllers. In spite of its simplicity, DTC allows a good torque control in steady state and transient operating conditions to be obtained. Moreover, DTC has simple and robust control structure and is not sensitive to rotor parameters. A review of recently used DTC algorithms for VSI fed induction motor drives has been presented and discussed by Giuseppe S. Buja and Marian P. Kazmierkowski [27]. A detailed comparison between FOC and DTC, emphasizing advantages and disadvantages are provided by Domenico Casadei et al [28] and concluded that DTC might be preferred for high dynamic applications. Hence, the DTC scheme was introduced in commercial products by Asia Brown Boveri (ABB) and therefore created wide interest. This is very significant industrial contribution and it has been specified by ABB that DTC is the latest ac motor control method and it can be considered to be next generation motor control technologies. Therefore, DTC has gaining more industrial applications such as high performance applications, electric vehicle applications, etc as explained by Peter Vas [4], James N. Nash [29], Pekka Tiitinen and Surandra [30] and Jawad Fiaz et al [32].

Though DTC has high dynamic performance, it has few drawbacks that can be summarized as high current, torque and flux ripple, variable switching frequency due to hysteresis controllers and high noise level at low speeds, etc. The effect of torque and flux hysteresis band amplitudes on the performance of induction motor drive has been studied by D. Casadei et al [33] and Jun-Koo Kong et al [34-35]. The amplitude of the flux hysteresis band mainly affects the motor current distortion in terms of low order harmonics. Small flux hysteresis bands lead to sinusoidal current waveforms, while small torque hysteresis bands allow smoothed torque to be generated. On the other hand, small hysteresis bands usually determine high switching frequency thereby increasing the switching losses. Moreover, the switching frequency of the torque controller has a peak value at medium speed due to the effect of back emf, while that of the flux controller is proportional to operating speed [35]. The analytical determination of the relationships between the applied voltage vector and the corresponding torque and flux variations is given by D. Casadei et al [36], from which, it has been observed that the effects produced by a voltage vector are strongly dependent on both rotor speed and voltage vector direction relative to the rotor flux. The maximum torque variation is obtained by applying a voltage vector along the direction perpendicular to the rotor flux vector. Thus, the presence of torque and flux hysteresis bands in DTC causes ripples in stator current, stator flux and torque that results in more harmonics in the line current.

To increase the dynamic performance of DTC and to decrease the ripple in torque, various switching control strategies had been proposed in the literature. The effect of the applied voltage on the torque response is strongly dependent on rotor angular speed. To tackle the problem of stator flux drooping at low speeds, to reduce the harmonic contents in the stator current and to reduce the switching frequency, the method of "variable switching sectors" for DTC has been proposed by CG Mei et al [37]. To reduce the ripple in torque further, a series of switching control strategies have been presented by E. Galvan et al [38] and G. Escobar et al [39]. In conventional DTC (CDTC), which was proposed by Takahashi, the selected voltage vector is not always the best one since only the sector is considered where the flux linkage space vector lies without considering its accurate location. As the CDTC has a fewer number of selectable voltage vectors, it causes higher ripples in the flux and torque. To overcome this problem, a unified flux control (UFC) method for DTC has been developed by Joon Hyoung Ryu et al [40]. In UFC, a voltage space vector is calculated for a deadbeat action and a minimum-distance vector selection scheme replaces the switching vector look-up table to minimize the flux and torque ripples. Moreover, torque ripple can be reduced, by applying a suitable voltage vector from the switching table for the time interval required by the torque to reach the lower or upper limit of the band, where the required time interval is premeditated from a suitable modeling of the torque dynamics as explained by Vanja Ambrožič et al [41]. This method is also known as band-constrained technique in which, depending on the inverter voltage vector and the operating conditions, the time interval may extend over several sampling periods. Therefore, the inverter switching frequency settles automatically to the minimum value.

Nowadays, the intelligent controllers like fuzzy, neuro and neuro-fuzzy controllers play a major role in industrial applications. To improve the dynamic performance of torque and flux, stator resistance estimation, stator flux estimation, tuning procedure, etc, intelligent control algorithms given by Sayeed A. Mir et al [42], I.G. Bird and H. Zelaya De La Parra [43], Fatiha Zidani and Rachid [44], Luis Romeral et al [45] and Pawel Z. Grabowski et al [46] can be implemented to the DTC algorithm. As there are no sector borders, there is no current and torque distortion caused by the sector changes. During the low speed region also, the performance can be improved by using fuzzy logic or neuro-fuzzy controllers.

A substantial reduction in torque, flux and current ripples could be obtained using the discrete space vector modulation (DSVM) algorithm developed by D. Casadei et al [47] and Xin Wei et al [48]. DSVM uses prefixed time intervals within a cycle period that results more number of voltage vectors with respect to those used in conventional DTC. The increased number of voltage vectors allows the definition of more accurate switching tables in which the selection of voltage vectors is made according to the rotor speed, the flux error and torque error. In DSVM, one sampling time period is divided into ‘m’ equal time intervals. One of the VSI voltage vectors is applied in each time interval. The number of voltage vectors, which can be generated is directly related to ‘m’. However, a good compromise between the errors compensation and the complexity of the switching tables is achieved by choosing m = 3 [47-48]. Using DSVM algorithm with three equal time intervals, 36 synthesized non-zero voltage vectors are obtained. If the stator flux vector is assumed to be in first sector, then 19 voltage vectors can be used. Then, different voltage vectors are chosen for different speed ranges. A fuzzy logic controller can be designed to select synthesized voltage vectors in DSVM based DTC [48]. Thus, DSVM allows the performance of DTC scheme in terms of flux and torque ripple and current distortions to be improved without increasing the complexity of the power circuit and the inverter switching frequency.

To overcome the problem of variable switching frequency, and torque ripple, few controllers have been proposed in the literature [49-50]. This can be done in two ways. In first method, the optimal switching instant is calculated at each switching cycle to satisfy the ripple minimum condition based on the instantaneous torque slope equations. In second method the conventional three-level torque hysteresis comparator is replaced by a new controller, which consists of two triangular waveform generators, two comparators and a PI controller. To operate the DTC algorithm with constant switching frequency and to reduce the torque ripple few methods have been proposed in [51].

In recent years, to overcome the problem of ripples and varying switching frequency operation of a voltage modulation algorithm, which is known as Space Vector Pulsewidth Modulation (SVPWM) has been used in the literature [52-86]. Over the last decade, recently reported SVPWM algorithm [58-67] has become very popular. In this method, the reference is provided as a voltage space vector, which is sampled once in every subcycle and an average voltage vector equal to the sampled reference voltage vector is generated by time-averaging of the different voltage vectors produced by the inverter. The SVPWM is a superior PWM technique for three phase inverter drives compared to the traditional regularly sampled triangular comparison technique. Space vector approach has the advantages of lower current harmonics and a possible higher modulation index compared with the three phase sinusoidal modulation method and ease of digital implementation.

A novel scheme was reported by Thomas G. Habetler et al [52] that calculate the inverter switching pattern directly in order to control the torque and flux in a dead beat fashion over a constant switching period. This is accomplished by calculating the voltage space vector required to control the torque and flux on a cycle-by-cycle basis using the calculated flux and torque errors sampled from the previous cycle and estimated value of the back EMF in the machine.

To get constant switching frequency and to increase the inverter switching frequency for the same sampling frequency, the symmetrical regular sampled SVM technique was used by Yen-Shin Lai and Jian-Ho Chen [53] for inverter control of the DTC based drive. A new SFVC based DTC was reported by D. Casadei et al [54] along with a simple closed loop flux estimator to improve the drive performance in the very low speed region, including zero speed. Further, a simplified DTC algorithm based on SVPWM was reported by Lixin Tang et al [55-56], in which instead of the switching table and hysteresis controllers, a PI controller and reference flux vector calculator (RFVC) were used to determine reference stator flux linkage vector. The RFVC generates the reference flux vector according to the error in the torque, which is based on the current estimated flux linkage vector. Moreover, a special SVM pattern has been used to reduce the switching frequency of the inverter. Further, closed loop digital control for both flux and torque was implemented by Cristian Lascu et al [57] in a SVPWM based DTC to improve the transient performance and steady state ripple and to preserve the robustness. A sensorless hybrid DTC drive based on SVPWM for high volume low cost applications was reported by Cristian Lascu and Andrzej M. Trzynadlowski [57]. In this hybrid method, under the transient operating conditions, the drive is controlled by using the classical bang-bang DTC and in the steady state operation, using linear flux and torque controllers, the control system produces a reference voltage vector for the inverter.

Nowadays, the attention is paid to determine the switching losses of the inverter. The dependency of the switching losses of a bridge leg of a PWM converter system with a pulse rate was explored by Johann W. Kolar et al [64].

So far, a number of PWM techniques have been discussed for VSI fed induction motor drives. The techniques for the generation of PWM waveforms can be broadly divided into:

Offline PWM generation techniques

Online PWM generation techniques

Offline PWM techniques are those where the switching instants of the inverter are stored in the form of lookup tables, which are previously calculated and used. The online PWM techniques are more common where the fundamental cycle is divided into many subcycles in each of which the volt-second balance is maintained. The online PWM techniques can be further subdivided into two categories on the basis of approach, namely the triangle comparison (TC) approach and the space vector (SV) approach. In the TC approach, three-phase modulating waves are compared against a common triangular carrier to determine the switching instants of the three phases. The most common and popular modulating waves are sinusoidal waves. Any triplen frequency component can be added as zero sequence components to the 3-phase sinusoidal waves. The choice of these triplen frequency components is a degree of freedom in this approach. In the SV approach, the voltage reference is provided in terms of a revolving space vector. The magnitude and the frequency of the fundamental component are specified by the magnitude and frequency respectively of the reference vector. The reference vector is sampled once in every subcycle. The inverter is maintained in different states for appropriate durations such that a sampled reference vector is generated over the given subcycle will be equal to the average voltage vector. The inverter states used are the two zero voltage vectors and the two active voltage vectors, whose voltage vectors are the closest to the commanded voltage vector. The division of the zero voltage vector duration between the two zero states is a degree of freedom in the space vector approach. This division of zero vector time in a subcycle is equivalent to adding a common-mode component to the 3-phase average pole voltages. The same PWM waveform can be generated based on both the approaches as explained by G. Narayanan and V. T. Ranganathan [63].

The ripple in torque can also be decreased by using the multilevel inverters. An increase in the number of levels improves the torque quality reducing the ripple amplitude. Therefore, by using the multilevel inverters, the torque performance of direct torque control of induction motor in high power and medium power applications can be improved as explained by Kyo-Beum Lee et al [69], A. Damiano et al [70], Zhuohui Tan et al [71] and José Rodríguez [72]. But, in the multi level concept, though the torque performance is improved, the cost and complexity will be increased.

Though the look-up table based 3-level inverter fed DTC drives give good performance when compared with the look-up table based 2-level inverter fed DTC drives, it gives varying switching frequency operation of the inverter and gives more harmonic distortion. To obtain constant switching frequency operation and to achieve superior waveform quality, various PWM algorithms have been proposed in the literature. Nabae, et.al. Proposed a PWM algorithm for neutral point clamped (NPC) 3-level inverter in [73]. Nowadays, the multilevel inverter fed drives are becoming popular in many industrial and electrical vehicles applications especially for medium and high power applications [74-77]. A detailed survey on the multilevel inverters and various topologies of the multilevel inverters are discussed in [77]. The waveform quality can be increased by increasing the number of levels. But, as the number of level increases, the complexity involved in the PWM algorithm and power circuit also increases. To decrease the complexity involved in the PWM algorithms for a multilevel inverter, several simplified PWM algorithms have been proposed in the literature. A simplified SVPWM algorithm has been proposed for a three-level inverter by using the concept of SVPWM algorithm for a two-level inverter in [78-79]. In this algorithm, the switching times can be calculated similar to a two-level inverter. However, this algorithm requires angle and sector calculations, which increases the complexity of the PWM algorithm as the number of levels increases.

To decrease the complexity of the SVPWM algorithm, it is necessary to avoid the angle and sector calculations. A simplified approach for SVPWM algorithm is proposed in [80], which uses instantaneous phase voltages only for the calculation of gating times of the inverter. The same approach is extended to the various discontinuous PWM algorithms along with the SVPWM algorithm in [81]. However, these approaches are proposed for two-level inverters only. The same approach is extended to an n-level inverter in [82]. This algorithm also uses instantaneous phase voltages only. By using the concept of effective time, the algorithm is extended for multilevel inverters under both linear and over modulation regions.

Nowadays, many researchers have been focused their interest on open-end winding induction motor drives in medium and high power applications. The open-end winding induction motor drives offer many advantages when compared with the normal drives. The open-end winding induction motor drives fed by two inverters on either ends. By using the two 2-level inverters on both sides of the windings, the phase voltages can be obtained similar to the three-level inverter. To control these two 2-level inverters, various PWM approaches are presented in the literature [83-92].

Among the various PWM algorithms, decoupled PWM (DCPWM) and alternate inverter switching PWM (AISPWM) algorithms are popular approaches for open-end winding induction motor drives. In both the approaches, the two inverters will be operated with 180 degrees phase shift. Though the implementation of DCPWM algorithm is simple, it gives more harmonic distortion in line currents and voltages [91]. Hence, nowadays the research interests have been focused on AISPWM algorithm. In [83], a look-up table based AISPWM algorithm has been presented. However, this approach will generate large common mode voltage variations and large current and torque ripple. To overcome the drawbacks of AISPWM algorithm, which is presented in [83], various approaches have been proposed in the literature [86-92].

The acoustic noise radiated from the DTC based induction drive depends on the width of the hysteresis bands. Increment in flux hysteresis band increases the low switching frequency harmonics, which results in high noise emissions around these frequencies given by L.Xu et al [93]. T. Noguchi et al [94] has presented a novel simple method for a high frequency switching operation of a PWM inverter for Direct Torque Control of an induction motor to make it an acoustically silent drive. It has the drawback of variable switching frequency and moreover the designing of inverter output filter becomes difficult. Acoustic has been a major problem since the invention of the induction motor drive. Basically, it is caused by flux space harmonics are dependent upon certain factors, such as winding layout, fluctuation in air-gap permeance due to slots of stator and rotor eccentricity, local saturation etc.

The acoustic noise radiated by the induction machine increases when they are operated from non-sinusoidal power supplies, such as pulse width modulated inverters [95]. The acoustic noise radiated by the PWM controlled induction machine drives has been explained in detail by W.C. Lo et al [96] and S.L.Capitaeanu [97]. Electromagnetic noise due to PWM switching is generated at narrowband high frequencies, which cause communication obstacles and unpleasant high-frequency audible noise [98]. Physiological studies have shown that narrowband noise is more unpleasant than broadband noise. Several studies related to PWM techniques to reduce this audible switching noise have been reported.

Recently, new random pulse width modulation (RPWM) methods have been investigated increasingly for noise reduction [99-103], have been investigated extensively. These are used broad-band switching frequencies to spread the noise spectrum instead of using a specific fixed switching frequency.

RPWM methods operate with different carrier switching frequencies at each switching instant to spread the spectrum of the voltage, current and noise. RPWM strategies are attracting interest as excellent methods for noise reduction because of their simple algorithm. There are several ways to implement RPWM by modifying the conventional SVPWM technique. One approach uses the SVPWM technique extended to variable switching frequency operation, which effectively reduces the acoustic annoyance. Problems arise, however, in the control system, because a variable controller sampling frequency is needed if the modulator and the controller operate in synchronism.

Another method is based on placing a pulse either at the beginning or at the end of a switching interval. This is also called lead-lag modulation, and effectively gives random modulation [104]. It is difficult for this approach to be an effective method to decrease the harmonics and acoustic noise because of the limited freedom available for pulse placement at the leading or lagging edge. The third approach is to select the switching frequency randomly from among a few predetermined frequencies, limiting the problem of randomization in the control loop [105]. In general, it is difficult to select these different switching frequencies in a real application.

Recently a new RPWM called as random position space vector PWM was proposed by S.H Na [106]. In the proposed technique, each PWM pulse can be located at any place in each modulation interval as long as it does not corrupt the switching sequences for SVPWM. The duty ratio is calculated based on SVPWM and then each pulse is located at a randomly selected position. Due to the high degree of freedom in locating the pulse position, the power spectrum is made flatter and hence reduces the audible switching noise effectively. However, for employing these techniques, the high speed switching devices like IGBTs will be used to increase the carrier frequency of VSI-PWM inverters, thus leading to much better performance characteristics. In the extension of the reduction of the acoustic noise and to reduce the computational burden involved in the existing RPWM algorithms, a novel variable delay RPWM algorithm has been proposed in [107]. In this approach, the switching times have been calculated using the instantaneous phase voltages.

A different approach to increase the number of levels in the open end winding induction motor configuration with decoupled SVPWM algorithm is presented in [108], the methods presented in [109,110] are also employed the decoupled algorithm only for the generation of the 4- level SVPWM voltages.

Thus, though the DTC offers good dynamic performance, it has few drawbacks such as steady state ripple in torque, flux and current, varying switching frequency and sensitive to load torque disturbances. Hence alternatives must be explored to reduce the steady state ripple in torque, flux and current and memory size and to get constant switching frequency. Mainly, this research is focused on the various PWM algorithms to overcome the problems of steady state ripple, switching frequency variations and memory size. Moreover, various simplified PWM algorithms have been presented for open-end winding induction motor drives.