IEEE 1999 International Conference on Power Electronics and DriveSystems, PEDS99, J uly 1999, Hong Kong. A NEW TORQUE CONTROL METHOD FORINDUCTION MOTOR DRIVES VINAYAK N.SHETUDAYKUMAR R. Y.KUSHA SHETTY K Dept. of Electrical Engg. IlTBombayI I TBombayIlT Bombay Email: uday@ne.iitb.emet.in Energy Systems EngineeringDept. of Electrical Engg Prof. V.P. SUNDARSINGH .Dept. of Electrical Engg. I l T Bombay Mumbai-400 076 I NDI AEmail: vps@ee.iitb.ernet.in Fax: 091-22-5783480 Abstract-Conventional vectorcontrolmethodinvolves CO- ordinatetransformationsandhencerequiresrealtime computationswhichinvolvestimeconsumingcomplex calculations. On the other hand,it isshown in thispaper that in drive systems involving periodicload change cycles the fast torque response can be achieved by giving a phase lead to the input voltage whenever load changes. In addition, a fixed amount of short term voltage boost can also given for furtherimprovingthetorqueresponse.Thisprincipleis simulated on an Inductionmachineand the resultsshow a markedimprovementintheresponsetime.Experiments were conducted based on this principle and the results show that this method improves the transientresponse. Thus, by precalculatingthephaseanglecorrectionrequired,this methodcould beeffectively usedwheretheloadchanges undergo cyclic fixed changes. I. INTRODUCTION Torque control in AC motors is achieved by controlling the motor currents,and the phase angle. In other words, current vectorhastobe controlled [l ] andisusedto obtainhighdynamicperformancesimilartotheone obtainedwithseparatelyexcitedDCmachine.For achievingthis, undertransientconditions,thetorque whichisaproductofthefluxproducingandtorque producing current components and which are inspace quadrature must be controlled. Different vector control strategies are evolved usingdifferent formulations. The vectorcontrolschemes arerotorfluxoriented control, Mutualfluxorientedcontrolandstatorfluxoriented control. Among all the approaches, the rotor flux oriented control is the most widely used method[2].Position and magnitude of the stator current changes whenever either torquecomponentorfluxcomponentchanges. Implementation methods are classified depending on how the unit vectors cos(w,t) and sin(w,t) are generated. These decide the position of the reference frame, which in turn decides the space angle of the stator current. Since Induction machine is a singly fed machine, the torqueproducingandfluxproducingcurrentsare simultaneously present in thestator winding.Hence, to obtain thequadrature components ofthe stator current, informations about the modulus andspace angle ofthe rotor flux are required. These can be determined either by direct method or indirect method. In the former method, fluxisdirectly measured using sensors like Halleffect sensor, search coil or tapped windings of the machine. In the latter method, modulus and space angle are obtained bymonitoringstatorcurrentsandrotorspeed.The implementation usingdirectmethodisnotpreferred becauseofmechanicalconsideration.Theintrusive sensors are not preferred by mechanical designers. On the other hand the indirect method requires that actual speed is to be measured. Hence the preferred method today is the indirect method where flux estimation rather than the measurementoffluxisused.Inthis method,fluxis estimatedusingterminalquantitieslikevoltagesand 0-7803-5769-8/99/$10.000 1999IEEE 56 1 currents byusingobservers.Anotherareawhichhas drawn interestistomakethecontrolindependentof parameter variations,[3] particularly ofrotor quantities. Inalltheabove method,themainrequirement ofthe control systemis that as soon as the mechanical torque is applied, the machine should develop the required torque with minimumoscillations and the magnetizing current of the machine must bekept at the rated value to produce the ratedfluxinthemachinemagnetic circuit.Hence,to obtain fast changes in torque, it is required that the torque producingcompohentchangestothefinalvalue immediately after the torque changes. Achange in the mains voltage is not feasible because the flux also depends onfresuency andany change inthe voltage has tobe accompanied with proportional change in frequency such thatVlfismaintainedconstant.Thus, alltheabove approachescallforcomplexandtimeconsuming calculations.Tosi mpl e thisamodifiedapproachis described here. In an Induction machine, theair gap flux voltageis aligned with quadrature axi s, and the direct axis current is responsible for the air gap flux. Now, if the input voltage vector position is changed then the air gap voltage can not change, since for sudden change of voltage flux linkage is zero. This will force the current in the rotor branch of the machinetochange,whichinturn willchangethe magnitudeofthetorqueproducedsincetorqueis proportional to the rotor current. If this phase correction is optimum, then, the required torque will be generated instantaneously. This calls for an open loop control and pre- calculation of parameters fromthe equivalent circuit. Thus, various steady state quantities such as slip, torque, rotor current etc, are calculated for a particular machine andtheydecide thephasecorrection tobe given.For calculating the parameters, the machine is represented in a two-axis modelusing equations given byTian-Hua et al[4].Asanexample,atypicalmachineparameter referred to the stator measured on amachine are given below for a 2 HP machine. I$=7.1422; LS=37mH;R=7.6222; L=37mH; L,,,=214mH; Voltage=240/415V M.I. (J)=0.008 kg-m2, 4-pole, 50 Hz, F.L.speed=1440 rpmBasedontheseparameters,simulationusinga computer was carried out assuming that at the instant of application oftorque, themachine input conditions are modified namely phase and input voltage (around 10Y0). Whenever thespeed changes are more,such as for fast response the gradient method demands large variation in theinputvoltage,whichmaynotbe feasibleinthe practical case. And also when the voltage is increased to much higher value thanthe rated value for that frequency, the magnetic circuit will saturate resulting in poor torque response. Itisevident fromthesteady state equivalent diagram shown in Fig.1, that the change in the phase of the input will not change the steady state operating point. Butsince,duringthetransientconditions,equivalent circuit of Fig.1 does not hold good (with the presence of slip termin rotor equivalent circuit) and hence a two-axis model as stated in Equation (1) is usedtodetermine the results. These are given by r iLc s Jwhere cpis the angle between the Q-axi sand'phasea' voltagephasor.Similarly,otherstatorparameterslike statorcurrentandfluxcanbe transformed tothis reference frame. The coding of the machine model and other excitation conditionaredoneusingCprogramming.Loading conditions and other related modifications are done using if-elseconditions.Tuningofthephaseistriedusing switch statement. Using this model, the machine response 562 is simulated and the results are shown in fig.2 & 3. Figure 2 shows the speed variation with and without the phase angle correction. If the machine is loaded to full load fromno load condition at t=0.54 sec, with a phase correction of 7.6 degree and in addition if after F0.5435 secs the phase angle correction is increased to 8.8 degree along with an increase of10% in input voltage the response is shown in fig. 2@).It is seen that curve'A' which corresponds to a speedchangewithoutthecorrectionshowsdamped oscillatory behaviour. But, it can be seen that the steady state valuesare thesame in boththecases. Thus,the applied phase angle lead improves the transient response of the machine. Figure3 shows the response for torque changes withandwithoutcorrection applied underthe above loading condition. The speed and torque response foralowcompensation angleof3degree andahigh compensation angle of14 degree was also studied. Figure 4 and 5 show the speed and torque changes with very low compensation angle and high compensation angle. When the compensation angle is 3 degree the speed response is sluggishandapproachnaturalspeedresponseasthe phaseanglecorrectionisreduced.Whenthe compensation angle is14 degree thespeedresponseis veryfast butthe overshoot is more.Thus theoptimumphase angle can bechosen once the allowable overshoot is fixed. The torque response is found to be sluggish for the case with a correction of 3 degree and faster for the other case.Hence,withproper phaseangle changeoptimumresponsecanbe obtainedand itgreatlyimproves the transient response. Figure 6 shows the change in the input current vector when the load is applied. IL EXPERIMENTAL SETUP AND RESULTS To venfy the validity of the simulation results ofthe proposedmethod,adrivesystemisimplemented with transistors as switching devices usingsinusoidalPWM method to obtain nearsinusoidal Inverter output. PWM Inverter is supplied in the front end by a diode rectifier, which directly converts line AC to DC. It is followed by a large capacitor filter. The frequency and amplitude of the fundamental component in the outputcan be varied by varying the frequency and amplitude of themodulating waverespectively. The base frequency chosen is 40Hz. By keeping V/f ratio constant, same torque-current sensitivity can beobtained. The modulating waveamplitude is taken as 0.95. Number of pulses per half cycle is chosen to be 21. PWM pulse pattern at 40 Hz is generated bysoftware andthesamevaluesarestoredinanEPROMfor reproduction. The complete block diagramof the control set-up is shown in fig.7. Whenever there is torque change, there will be suddenchangeinDClinkcurrent. This changeiscapacitivelycoupledtoacomparator, which generates apulse, which in turn starts ADCto digitise the input. The input to ADC decides the phase shift given to the output voltages. When conversion is complete, the end-of -conversion (EOC) is generated. This EOC signal isusedtostrobe the newaddresslocation to counters. Then onwards the base drive sequence will start fromthis newlocation. This resembles to the phase-shift given to the machine voltages. Figure 8 shows the torque response whenevertheloadisappliedwithoutanycorrection. Figure9showsthevariationinthetorqueresponse whenever the load is applied with a correction. As seen fromthe above waveforms the torque response is better for the case in which correction has been applied. This shows that the transient response has improved because ofthe control circuit. 111. CONCLUSIONS A method is proposed to get fast speed response for an Inductionmachine.Simulationoft h i s methodwas caniedout and it is seen that the phase angle correction method yields satisfactory results as far as the transient conditions are considered. By carrying out experiments, it is also proved that the response can be improved bythe method proposed here. Hence, it can beconcluded that the 563 correctionmethodsworkswellwithpredetermined parameters. I. 2. 3. 4. 5.REFERENCES: Bose B.IC, "Power electronics and AC drives" Englewood clif&, NJF'rinticeHall 1986. Lansen P.L. and Lorenz RD."A Physically insightful approach to the designandaccuracy assessmentoffluxobservers forfield oriented Induction machinedrives" IEEE Transactions on Industry Applications, Vol. 30, No.1, Feb'94, pp.101-109. Tunpimoht IC, Peng F.Z.and Fuko T.,"Robust Vector Control of Inductionmotorwithoutusingstatorandrotorcircuittime constant" IEEE trans. onIAS,Vo1.30,No.5,Sept.'94,pp.1241- 1246. Tim-Hua,Jen-RemFu,and T.ALipo,"A Strategy forImproving Reliability of Field-Oriented Controlled Induction Motor Drives" IEEETrans. onIAVo1.29, No.& Sept93,pp.910-917. Yang G. and Chin T. H..,"Adaptivespeed identification schemefor a vector controller speedsensor less Inverter Induction motor drive" IEEE Trans. on I AS, Vol. 29, No.4, hgust'93. 564 Fig. 1Stcady state equivalent diagramof IM. 1 .oo z0.98 C U a .- 0.96 0.94 0.92~~-~~~~~0.300.400.500.M)0.700.80 timeinseconds Fig. 2Speed change with and without phase ande corrcction. A Fig. 4Speed change with under and over phase angle correction 2.003 -0.40 1 -0.80 0.300.400.500.600.700.8C timeinsec I 1 -0.50 Fig. 3Torque change with and without phasc angle correction. 565 : if c -5.130 5 Fig. 8 Torque response without correction Fig. 9 Torque response withcorrection Dc hq,l Fig. 7 Block diagramof experimental set-up 566