Constanta Maritime Universitys Annals

Year XIII, Vol.18

Improvements of the Direct Torque Controlled Induction Motor Drives

1

Lecturer dr.eng. Sanda Victorinne Paturca, Lecturer dr.eng Valeriu Bostan, Lecturer dr.eng Leonard Melcescu
1

University Politehnica of Bucharest

ABSTRACT This paper presents two methods for reducing the torque ripple of the conventional Direct Torque Control (DTC) of induction motor drives. The methods were implemented on a Digital Signal Controller, their effectiveness was evaluated for a set of motor operating points and the experimental results are comparatively presented. Keywords: Direct torque control, torque ripple, pulse width modulation, space vector modulation

1.

INTRODUCTION

Direct Torque Control (DTC) is a high performance control strategy for induction motor drives fed by Voltage Source Inverters (VSI) [1]. The main advantages of DTC are the simple control scheme, a very good torque dynamic response, absence of the inner current loops and rotor position/speed measurement. These advantages make DTC an attractive option for applications that require a very fast torque response, torque reference input (no speed control, but torque control only) and improved reliability in harsh environments (dust, vibrations conditions), due to the absence of the rotor position transducer. However, the main drawback of the conventional DTC is the high torque ripple generated in steady state operation. In conventional DTC, the voltage vector selection is based on the torque and flux errors, but small and large errors are not differentiated by the hysteresis controllers. The voltage vectors are applied for the entire sample period, even for small errors, resulting large torque overshoots in steady-state regime. One approach to reduce the ripple is to increase the number of voltage vectors applied in a sampling period, using some sorts of pulse-width modulation (PWM). Among the possible choices, a simpler one is to use two voltage vectors: a nonzero one, applied for a fraction of the sampling period, and the null vector for the rest. The duty ratio must be calculated each sample period, and by varying it between its extreme values, it is possible to apply more voltage levels to the motor, according to the desired torque variation. In [2], an analytical online algorithm calculates the optimum duty ratio each sampling period, by using a torque ripple minimization condition, which is based on ripple equations. However, this algorithm requires high computational effort and additional motor parameters to be known. In [3] the duty ratio value is provided by a new fuzzy logic module, whose inputs are the stator flux position, the electromagnetic torque and an input defining the motor operating point, given by the speed and the torque values. This algorithm involves expert knowledge and needs the rotor speed. To overcome this problem, the authors proposed two modified DTC schemes, which are based on the idea of

increasing the number of voltage vectors applied in a sample period. This makes it possible to obtain more voltage levels at the inverter output, in accordance to the desired torque and flux variations. The proposed schemes were experimentally tested for a set of motor operating points. Their effectiveness was evaluated by calculating the root mean square (RMS) deviation of the instant torque values with respect to the load torque, in steady state regime. The paper presents the DTC principle, the proposed schemes, graphical and numerical experimental results of the tests that were conducted to evaluate the effectiveness of the proposed control schemes, and the conclusions. 2. DIRECT TORQUE CONTROL PRINCIPLE

The basic model of the classical DTC induction motor scheme is shown in figure 1. It consists of torque and stator flux estimators, torque and flux hysteresis comparators, a switching table and a VSI. The basic idea of DTC is to choose the optimum inverter voltage vector in order to control both stator flux and electromagnetic torque of machine simultaneously [1].
Two level Hysteresis

s* +
Flux reference

Controller
1
1

Switching Table

S A , S B , SC

T*
Torque reference

1
0
1

u s (0,1,...,7 )

Voltage Source Inverter

U DC

Three level Hysteresis Controller

S T

Sk (1,2,...,6 )

Flux Sector detection

Torque and Flux calculator

ia ib

Figure 1 Block diagram of the conventional DTC The stator flux space vector, , is calculated in the stationary reference frame ( ) using the stator
s

Constanta Maritime Universitys Annals

voltage equations [5]. The circular trajectory of stator flux is divided into six symmetrical sectors referred to inverter voltage vectors. The components of the stator flux are used to determine the sector in which the flux vector is located. The calculated magnitude of stator flux and electromagnetic torque are compared with their reference values in their corresponding hysteresis comparators. Finally, the outputs of the comparators and the number of sector at which the stator flux space vector is located are fed to a switching table to select an appropriate inverter voltage vector [4]. 2. THE PROPOSED DTC METHODS FOR TORQUE RIPPLE REDUCTION In conventional DTC, a single stator voltage vector of the inverter standard topology is selected during every control sampling period, and it is maintained constant for the whole period. By this switching technique, based on hysteresis, large and small torque and flux errors are not differentiated, which causes an extra torque ripple in motor steady state operation. The both proposed methods are based on the idea of increasing the number of voltage vectors applied in a sample period. Their principles are differentiated by the voltage vector calculation criteria and mode of applying this vector to the inverter. 2.1 DTC with the variable amplitude of the inverter basic space phasors This first method consists in the modulation of the nonzero voltage vector duration over a sampling period, according to the torque and stator flux errors, using symmetric PWM. The nonzero voltage vectors are selected using a switching table, like in classical DTC. The null vectors are automatically inserted by using PWM, so they are no more needed in the switching table. Consequently, in the proposed control scheme, shown in figure 2, simple comparators, with no hysteresis, are used. A duty ratio calculator and a PWM block have been added to the classical scheme.

Year XIII, Vol.18

strategy. The PWM period is considered equal to the control sample period. To apply to the inverter the selected voltage vector (denoting the corresponding inverter switching states S i , i = A, B, C ), with a given duty ratio, it was adopted the following method: for S i = 0 , the pulse duty ratio of the corresponding inverter leg takes zero value, otherwise it takes the calculated value. The calculation procedure of the duty ratio is presented below. In order to preserve the very good dynamic response of the classical DTC, the voltage duty ratio is modified only when the actual torque value is located in a proximity zone, around the reference value, otherwise it is kept at 100%. This zone, whose width is denoted by Z T , is similar to the torque hysteresis band of the classical DTC torque comparator. The duty ratio, as a function of the torque and flux error, is calculated as in (1), so that to be a fractional value.

T + F ZT ZF

1 T 1 ZT

(1)

where: T and F are the torque error and respectively the flux error; Z T and Z F are the corresponding base values used for normalization. The second term in (1) was introduced in order to avoid the flux decrease at low frequencies, due to the stator ohmic drops, which can occur when the torque error becomes too small. This term takes into account the flux error, which is normalized in the same manner as the torque error. The duty ratio is limited to a minimum value, min , in order to consider the maximum switching frequency of the inverter. Considering the voltage duty ratio, the line voltage at the inverter output is given by (2). U l = U DC (2) where: U l is the line voltage at the inverter output, U DC is the DC link voltage.

At current sample time, the ( , ) components of the stator voltage are calculated using (3) and (4).

S SC 2 U DC S A B 3 2 S SC 2 , u s = U DC B 3 3 u s =

(3) (4)

where is the duty ratio calculated at previous sample time. 2.1.2 Figure 2 Block diagram of the first proposed method Every sample time, one of the six nonzero voltage vectors are selected from the switching table, and it is applied to the inverter using symmetric PWM switching DTC using Space Vector Modulation

In this second method, a reference stator voltage space vector is calculated, in terms of magnitude and phase, using the instant values of torque and stator flux errors, and the flux position.

Constanta Maritime Universitys Annals

A normalized value of the reference voltage magnitude was calculated using equation (5). The right side of the equation is the same as that utilized for the duty ratio calculation in the previous scheme.

Year XIII, Vol.18

u 3 (010)

F
u 2 (110)

m=

T T 1 + F 1 ZT ZF Z T

U ref

(5)

U ref

For torque and flux errors situated in these zones, m is calculated using the expression (5), otherwise m equals unity [9]. As regarding the voltage vector phase, its calculation is based on the conventional DTC principle. As presented in Section 2, the main idea of DTC is to choose the optimum voltage vector in order to achieve a simultaneous and decoupled control of the stator flux and electromagnetic torque. Every control sample time, one of the inverter voltage vectors is selected according to the torque and flux errors and the sector in which the actual flux vector is situated. The phase difference, , between the selected voltage vector and the middle of the flux sector, is k / 3 , where k {2, 1, 1, 2} . It can be proved that for a given voltage vector and a flux sector, the torque slope depends on the flux vector phase angle. This determines a irregular torque ripple in steady state operation, especially when the sector changes. In order to obtain a uniform torque ripple, the above mentioned dependency can be eliminated by considering a voltage vector which is phase shifted with relatively to the flux vector phase angle ( ), instead of the middle of the sector like in the conventional DTC. For simplicity, there were preserved the phase difference quantities. The phase of the reference voltage vector is calculated by adding to the flux phase angle the phase difference , whose value is simply derived from the sign of the torque and flux errors, like in equation (6).

u 4 (011)

u1 (100)

u 5 (001)

u 6 (101)

Figure 3 The voltage space phasor angle for two cases of torque and flux commands The principle of the presented method is illustrated in figure 3, for two cases of torque and flux commands. In figure 4 it is presented the block diagram of the proposed DTC-SVM scheme [5].

= sign ( T ) = +

(3 sign ( F ) ) 6

Figure 4 Block diagram of the second proposed method 3. EXPERIMENTAL RESULTS

(6) (7) The proposed methods, along with the conventional DTC, were implemented on a Digital Signal Controller, and tested for a series of motor operating points, given by torque and speed. The results obtained for an induction motor drive by using the conventional DTC and DTC with the proposed methods of torque ripple reduction are comparatively presented in this section. In figure 5 it is presented the experimental set-up.

The reference voltage vector, with the calculated magnitude, m , and phase , is applied to the inverter using Space Vector Modulation. In the linear region, the voltage vector at the inverter output is:

U ref = m

3 u i e j 2

(13)

where: u i is the modulus of the inverter basic space vectors.

m=

U ref U max

is the normalized value of the

reference voltage vector magnitude;

U ref is the magnitude of the reference voltage.

Figure 5 Picture of the experimental set-up

Constanta Maritime Universitys Annals

The experimental scheme used for the tests is shown in figure 6.

Year XIII, Vol.18

The motor parameters are: Pn = 370 W; n n = 1390 rpm; Tn = 5,15 Nm

I n = 2,1 A ( Y ); R s = 10 ;

Figure 6 The scheme of the experimental set-up The components of the experimental set-up are: 3-Phase IGBT Power Module, 750W, Technosoft ACPM 750 v3.4; Controller board MSK28335, with TMS320F28335 Digital Signal Controller, floating point, 150 MHz; Induction motor Sieber, 370 W, with incremental encoder; Electromagnetic brake, with magnetic powder, DeLorenzo DL1019P, capable of maintaining constant torque regardless the operating speed; Software development tool Technosoft DMC28x Developer Pro, used for application development, reference input and data logging; PC with RS232 communication.

The control sampling period was set to 0.1 ms for all three control schemes, and in both proposed control schemes the PWM period was set to 10 kHz. The tests were conducted so that to emphasize: - the torque time response, as compared to the conventional DTC. - the effectiveness of the proposed methods of torque ripple reduction, at different speeds and loads. All the subsequent tests graphical results are expressed in relative units. In order to correctly evaluate the torque transient response provided by the proposed methods, there was performed a prime test in which the torque reference was directly imposed, with no outer speed loop. The reference was prior programmed using the reference generator built into the DMC tool in the following sequence: at t = 0 s, Tref = 2Tn, followed by Tref = Tn at t = 0.2 s and Tref = 0 at t = 0.4 s. The results, comparatively presented in figure 7, for conventional DTC and the two proposed DTC methods, show that the fast torque response of the conventional DTC was preserved, with an insignificant delay. Moreover, the steady state error of the torque, which is inherent in conventional DTC [9], is eliminated by the proposed schemes, due to the torque reference correction [8].

Figure 7 Comparative results from the first experimental test, emphasizing the torque response in the case of conventional DTC (a), and DTC with the first and second proposed methods (b) and (c) respectively. The quantities represented, up to down, are: two stator phase currents, estimated stator flux module and the flux component on phase alpha, torque reference and estimated torque, measured speed.

Constanta Maritime Universitys Annals

Year XIII, Vol.18

Figure 8 Comparative results from the second experimental test, emphasizing the torque ripple reduction, as compared to conventional DTC (a), achieved by the first and second proposed methods (b) and (c) respectively. The quantities represented, up to down, are: two stator phase currents, estimated stator flux module and the flux component on phase alpha, torque reference and estimated torque, measured speed. The second test was conducted to evaluate the effectiveness of the proposed methods of torque ripple reduction , at different values of speed and load torque. For this purpose, an external speed loop was added to the control scheme, so that the torque behavior can be clearly observed at constant speed. Note that the speed transient response is not relevant in this case, since the test was focused on the steady state behavior. The test consists in running the motor at low and high speeds, for different loads, in the cases of conventional DTC and respectively the two proposed methods of torque ripple reduction. The test revealed that the torque ripple is significantly reduced by using the proposed methods, for both low and rated speed operation. Sample results of the performed tests are comparatively presented in figure 8, for the case of load equal to the rated torque, and two speed levels of 25% and 100% of the rated speed. The quantities represented are: two stator phase currents, estimated stator flux module and the flux component on phase alpha, torque reference and estimated torque, measured speed. All quantities are expressed in relative units, with respect to the corresponding rated values. For a quantitative evaluation of the proposed methods effectiveness, in table 1 it is presented the root mean square (RMS) deviation of the developed torque, with respect to the load torque. The presented values are calculated for a set of operating points, defined by speed and load torque, in steady state regime. For better readability, the speed and torque are expressed in percent of the corresponding rated values, and the torque standard deviation is expressed in percent of the rated torque. Table 1. Root mean square (RMS) deviation of the instant torque values from the load torque RMS of the torque deviation in percent of TN, M speed [%] 25 100 5. torque Conventional [%] DTC 25 100 25 100 34 23 33.5 17.3
( pc)

[%] DTC using SVM 5.3 5.7 6.4 5.8

DTC with variable duty ratio 8.3 8 9.9 8.5

CONCLUSIONS

The paper presented two methods of torque ripple reduction, for the DTC. In order to test the proposed methods, and to compare the resulted motor behavior, all the methods were implemented on a Digital Signal Controller and tested on a 3-phased voltage inverter drive. The first proposed method has the advantage of preserving the structural simplicity of the conventional DTC, while adding very little computation effort. In the second method, the torque ripple is reduced in a significant larger amount, but in the expense of an increased complexity, which is inherently added by the SVM technique. However, SVM offers a better utilization of the inverter switching capability. The experimental results proved the effectiveness of both proposed methods. Moreover, the corresponding control algorithms are simple and allow easy tuning.

Constanta Maritime Universitys Annals

6. ACKNOWLEDGMENTS

Year XIII, Vol.18

[5] S. V. Paturca, M. Covrig, C. Cepisca, G. Seritan, Proposed Schemes for Improving the Steady State Behaviour of Direct Torque Controlled Induction Motor, Proceedings of the 7th WSEAS International Conference on POWER SYSTEMS (PE'07), Beijing, China, 2007. [6] S.V. Paurc, M.Covrig, Direct Torque Control with Variable Voltage Duty Ratio for Inverter Fed Induction Motor Drives, Proceedings of the 10th WSEAS International Conference on Automatic Control, Modelling & Simulation (ACMOS08), Turkey, 2008. [7] S.V. Paurc Comenzi moderne n sisteme de acionri, Bucuresti, Ed. Matrix Rom, 2011. [8] S.V. Paurc, Contribuii la comanda numeric cu controlere industriale, a acionrilor electrice, Ph.D. thesis, University Politehnica of Bucharest, 2010. [9] L. Romeral, A. Arias, E. Aldabas and M. G. Jayne, Novel Direct Torque Control (DTC) Scheme With Fuzzy Adaptive Torque-Ripple Reduction, IEEE Trans. Ind. Applicat., vol. 50, 2003. [10] S.V. Paturca, A. Sarca, M. Covrig. A Simple Method Torque Ripple reduction for Direct Torque Control of PWM Inverter-Fed Induction Machines Drives, 8th International Conference on Applied Electricity, ICATE 2006.

The work has been co-funded by the Sectorial Operational Programme Human Resources Development 2007-2013 of the Romanian Ministry of Labor, Family and Social Protection through the Financial Agreement POSDRU/89/1.5/S/62557. 7. REFERENCES

[1] I. Takahashi, T. Noguchi, A new quick-response and high efficiency control strategy of an induction motor, IEEE Trans. Ind. Applicat., IA, 22, 1986. [2] J. Kang, S. Sul, New direct torque control of induction motor for minimum torque ripple and constant switching frequency, IEEE Trans. Ind. Applicat., vol. 35, 1999. [3] L. Romeral, A. Arias, E. Aldabas and M. G. Jayne, Novel Direct Torque Control Scheme With Fuzzy Adaptive Torque-Ripple Reduction, IEEE Trans. Ind. Applicat., vol. 50, June 2003. [4] P. Vas, Sensorless Vector and Direct Torque Control. New York: Oxford University Press, 1998.