ieee xplore download23123

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 59, NO. 5, MAY 2012 2277

Measurements and Simulations of DTC VoltageSource Converter and Induction Motor Losses

Lassi Aarniovuori, Lasse I. E. Laurila, Markku Niemelä, and Juha J. Pyrhönen, Member, IEEE

Abstract—Energy efficient pulse-width modulation invertersare widely used to control electrical machines accurately forprocess needs. The pulse-width modulation, however, has alsoadverse effects and produces additional losses in the motor. Theselosses increase the motor temperature and result in derating of themachine power in converter use. A reliable and reasonably accu-rate loss model of an induction motor drive system is important forthe performance prediction of a variable-speed drive. A two-levelfrequency converter main circuit model is coupled to a finite-element method motor model. The drive model is controlled byclosed-loop direct torque control. The frequency converter lossesare calculated analytically, and the finite-element method motormodel provides an analysis of the motor losses. The simulationresults are compared with measurement results.

Index Terms—AC motors, induction motors, magnetic losses,numerical simulation, power semiconductor switches pulse-widthmodulation converters, variable-speed drives (VSDs).

NOMENCLATURE

Esw,fw−diode Switching loss energy of fw-diode.Esw,IGBT Switching loss energy of IGBT.ESRC Equivalent series resistance of capacitor.ESRL Equivalent series resistance of input choke.f Frequency.fN Nominal frequency.fsw Switching frequency.idiode Instantaneous value of diode current.IC Capacitor current RMS value.Ifund Fundamental wave current amplitude.iL Instantaneous value of line current.im Instantaneous value of motor current.Im Motor current RMS value.IN Nominal current.IRated Rated current.IRMS RMS Current.n Rotational speed.nN Nominal speed.Nsw,change Number of switch state changes.Pad Motor additional losses.PC Capacitor bank losses.Pchoke Power loss in frequency converter input choke.

Manuscript received December 3, 2010; revised February 19, 2011,April 26, 2011, and June 5, 2011; accepted June 13, 2011. Date of publicationJune 30, 2011; date of current version February 3, 2012.

The authors are with the Department of Electrical Engineering, Lappeen-ranta University of Technology (LUT), 53850 Lappeenranta, Finland (e-mail:[email protected]; [email protected]; [email protected]; [email protected]).

Digital Object Identifier 10.1109/TIE.2011.2161061

PDC−link Intermediate circuit losses.Pdischarge Discharge resistor losses.Pdiode,cond Forward diode conduction losses.Pdiode,on Instantaneous conduction losses of diode.Pfw−diode,sw Fw-diode switching losses.PExtra Frequency converter extra losses.PIGBT,cond IGBT conduction losses.PIGBT,sw IGBT conduction losses.Pin Frequency converter input power.Pmech Mechanical power.PN Nominal power.Pout Frequency converter output power.RCE0 IGBT on-state resistance.Rdischarge Discharge resistor resistance.RF On state resistance of diode.RF0 On state resistance of fw-diode.UC Capacitor RMS voltage.UCE0 IGBT threshold voltage.UDC DC-link voltage.UF Forward voltage drop of diode.UF0 Fw-diode threshold voltage.Ufund Fundamental wave voltage amplitude.UN Nominal voltage.t Time.T Torque.TN Nominal torque.

I. INTRODUCTION

ADDITIONAL losses caused by PWM methods have beenstudied widely in the literature, e.g., in [1]–[6]. Several

methods have been proposed to calculate the iron losses undernonsinusoidal excitation [7]–[10]. In [11], a model is proposedfor accurately estimating the iron losses in rotating electricalmachines. The impact that PWM harmonics, amplitude modu-lation index, and switching frequency have on induction motoriron losses is investigated in [12] with a special test motor withplastic rotor cage.

IEC is developing a new standard: IEC 60034-2-3: Rotatingelectrical machines—Part 2–3: Specific test methods for de-termining losses and efficiency of converter-fed ac machines.Converter-fed motors will get their energy efficiency classes.However, there is no generally accepted method to evaluateadditional losses caused by PWM in electrical machines. Theac motor power derating caused by PWM losses varies from 0%to 20% [13]. In [14], it is stated that the additional losses in themotor caused by the frequency converter can increase the totalmotor losses up to 15%–20% compared with the grid operation.

0278-0046/$26.00 © 2011 IEEE

2278 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 59, NO. 5, MAY 2012

TABLE ITEFC MOTOR PARAMETERS

The aim of this paper is to find drive loss estimation methodsthat could be used without extensive confirming measurements.It is important to assure that the motor temperature at fullload does not exceed the thermal limits of the insulation andthus have a negative impact on the motor lifetime [15]. Thetemperature rise of the machine is the main dimensioningboundary condition for the machine.

Direct torque control (DTC) is one type PWM control strat-egy which can be considered as an alternative for vector controltechnique. DTC was proposed for ac drives by Depenbrock andTakahashi in the 1980s [16], [17]. The DTC has advantagesof high torque response, simple design, and robustness againstparameter variations. The variable switching frequency andhigh torque ripple are drawbacks of the classical DTC. DTChas been a topic of numerous scientific works over the pasttwo decades; the switching frequency of the DTC is analyzed in[18]–[20]. Numerous improvements in the classical DTC havebeen proposed for instance in [21]–[26].

The total losses of frequency converters are not studiedwidely. In [27], a unified loss model of a converter inductionmachine system is presented that includes steady state as wellas dynamic behavior of both machine and converter. The studiesare mainly focused on IGBT bridge losses [28] and the effectof modulation method in inverter losses [29], [30].

This paper provides results on how much the induction motorlosses and temperature rise of the motor increase when a PWMsupply is used. A 37-kW industrial totally enclosed fan-cooled(TEFC) class 130 (B) temperature rise induction motor is usedin the tests. The motor parameters with delta connection aregiven in Table I.

A commercial DTC frequency converter is used as a pulse-width modulation (PWM) supply. The sinusoidal voltages areproduced by a synchronous generator in the 25 Hz and 40 Hzpoints; while in the 50 Hz point, normal utility grid voltage isused.

The temperature rise tests were carried out with differentaverage switching frequencies of the inverter and with differentrotational speeds of the motor. It should be noted that the cool-ing conditions with different rotational speed are quite differentfor a TEFC-motor. A coupled field-circuit system simulatorwith a closed-loop control system was used to separate andanalyze the drive system losses with the sinusoidal and PWMsupply.

This paper is organized as follows. Section II concentrateson temperature rise tests, Section III presents the frequencyconverter loss models used in this study. In Section IV, thesimulation method is presented and different loss componentsare analyzed. The comparison of the simulated and measured

Fig. 1. Measurement setup. The continuous input and output power of37-kW voltage source converter supplying 37-kW TEFC induction motor wasmeasured with power analyzers. Rotational speed and torque were measuredwith a torque transducer. The induction motor temperature was measured withPt-100 sensors.

efficiencies is carried out in Section V. Section VI concludesthe paper.

II. EXPERIMENTAL RESULTS

The measurement setup is given in Fig. 1. The frequencyconverter was set to the frequency control mode with no slipcompensation. Thus, the slip of the induction motor depends onthe motor load and losses. The input voltage of the frequencyconverter was accurately set to 400 V RMS value with atransformer. Because of the nature of the DTC, the switchingfrequencies used in this paper represent average values of 1-stime intervals. Four different average switching frequencies(1, 2, 3, and 3.75 kHz) were used to find out the impact of theswitching frequency on the motor losses. As a load, anotherDTC-controlled induction machine was used. The nominalpoint (n = 100%, T = 100%) of the motor always representsan overload for the motor in a normal frequency converter use.This is a result of the additional harmonic components pro-duced by the PWM and, particularly, converter field-weakeningoperation in the 50-Hz operating point if six-step modulationis not used. The nominal load of the 37-kW induction machinein a 50-Hz sinusoidal 400-V supply is 239 Nm. The load valuewas set to 220 Nm resulting in 92% of the nominal load. Withthis load, the RMS value of the stator current of the invertersupplied motor is equal to the nominal RMS current in the50-Hz operating point.

To obtain the thermal equilibrium, the motor was kept run-ning for 9 h at a constant load in each test. The temperaturerise curves in Figs. 4, 7, and 10 are the average values ofthree factory installed Pt-100 sensors located in the stator wind-ings. The laboratory temperature was recorded and subtractedfrom the results. During the temperature tests, the laboratorytemperature varied between 25 ◦C and 28 ◦C. The continuousinput and output power of the 37-kW voltage source convertersupplying the 37-kW TEFC induction motor were measuredby two calibrated Yokogawa PZ4000 power analyzers on bothsides of the frequency converter. In the following tables, theinput power to frequency converter is defined as Pin, and

AARNIOVUORI et al.: MEASUREMENTS AND SIMULATIONS OF DTC VOLTAGE SOURCE CONVERTER 2279

TABLE IIMEASURED RESULTS IN THE 25-Hz OPERATING POINT

TABLE IIIMEASURED RESULTS IN THE 40-Hz OPERATING POINT

the output power from the converter to motor as Pout. Therotational speed, torque, and the shaft power (Pmech in tables)were recorded by a 500-Nm Magtrol torque transducer. Boththe THD and TD values provided in Tables II and III arecalculated from the measured currents. The THD50 values arecalculated from 50 lowest current harmonics. The TD20000

values include all harmonic, interharmonic, and subharmoniccomponents from 1 Hz to 20 kHz with 1-Hz resolution. In bothvalues, the fundamental wave RMS value is used as a scalingfactor. In the electric power measurement, the averages of 30 ×10 s samples with a 10 µs sample time were used to minimizethe errors.

A. 25-Hz Operating Point

In the 25-Hz operating point, a synchronous generator wasused to produce the nearly sinusoidal supply to the motor. Theresults of the discrete Fourier analysis of the voltage show thatthe most significant harmonic components in the voltage arein the 2nd, 3rd, 5th, and 7th order as shown in Fig. 2. Fig. 3presents the measured DTC voltage spectra in 25-Hz operatingpoint with different average switching frequencies.

Even though the voltage waveform is not purely sinusoidal,the harmonic content of the voltage is minimal compared withthe voltage produced with PWM. The THD50 value of thevoltage is 0.89%. The motor temperature rises during the last60 test minutes are shown in Fig. 4. The difference between

Fig. 2. Waveforms of the 25-Hz three-phase voltages produced with a syn-chronous generator and their harmonic amplitudes given in percents of thefundamental wave amplitude.

Fig. 3. Measured DTC voltage spectra in 25-Hz operating point. The averageswitching frequencies are (a) 1 kHz, (b) 2 kHz, (c) 3 kHz, and (d) 3.75 kHz.

Fig. 4. Temperature rises at the end of the temperature test at 25-Hz operatingpoint. The motor has reached its thermal equilibrium.

the sinusoidal and frequency converter supply in the motortemperature rise and motor losses is shown in Fig. 5 as afunction of the switching frequency.

In Fig. 4, the temperature rise behavior at the end of thetest with the sinusoidal supply is due to slight unintentionalincrease in the generator voltage. In Figs. 4 and 5, we can seethat the switching frequency of the inverter has a significantinfluence on the motor temperature. The decreasing rates ofthe losses and temperature rises are almost similar. In the25-Hz point, the temperature rise of the machine is roughly onedegree Celsius higher against 25 watts of losses. Table II showsthat in the sinusoidal supply, the motor current is larger than


Fig. 5. Temperature rises and changes in losses compared with the 25-Hzsinusoidal supply as a function of switching frequency.


in the frequency converter supply, because the DTC converteruses slightly more voltage than the sinusoidal supply.

DTC includes flux controller which keeps the flux linkageconstant regardless of the load. The lower rotational speed inthe sinusoidal supply is a consequence of the slightly lowerfrequency (24.84 Hz) than what was desired. If the rotationalspeed is scaled to 25-Hz voltage, the rotational speed willbe 734 rpm. The motor losses increase by 4%–14% in thefrequency converter supply compared with the sinusoidal sup-ply. The drive efficiencies are given in Section V, where thesimulation results are compared with the measured ones.

B. 40-Hz Operating Point

Similar as in the 25-Hz operating point, in the 40-Hz op-erating point, a synchronous generator was used to producethe sinusoidal supply to the motor. The THD50 value of thegrid synchronous generator voltage is 0.65% measured at themotor terminals. The voltage spectra of the DTC in the 40 Hzoperating point are shown in Fig. 6.

The motor temperature rises during the last 60 test minutesin the 40-Hz operating point are shown in Fig. 7, and the dif-ference between the sinusoidal and frequency converter supply

Fig. 7. Temperature rises at the end of the temperature test at 40-Hz operatingpoint. The motor has reached its thermal equilibrium. More efficient fan coolingresults in lower temperatures than at 25-Hz supply.

Fig. 8. Temperature rises and changes in losses compared with the 40-Hzsinusoidal supply as a function of switching frequency.

in motor temperature rise and motor losses is shown in Fig. 8as a function of the switching frequency. When comparing theresults to the 25-Hz operating point, it should kept in mind thatcooling of the motor in the 40-Hz operating point is much betterthan in the 25-Hz point. Results in Figs. 7 and 8 do not behaveas assumed.

The motor temperature rise with 3-kHz switching frequencyis smaller than with 3.75-kHz switching frequency, althoughthe direct loss measurement gives smaller losses at higherswitching frequency.

In the 40-Hz point, the temperature rise of the machine isroughly one degree Celsius higher against 37 watts of losses.

The numerical results in Table III show that the slip of themotor remains constant with all switching frequencies. In the40-Hz operating point, the motor losses increase by 6%–14%in the frequency converter supply compared with the sinusoidalsupply. When the motor is driven with a frequency converter,the slip is about 1 min−1 greater than when using sinusoidalsupply. Similar as at the 25-Hz operating point, the DTC is



TABLE IVMEASURED RESULTS IN THE 50-Hz OPERATING POINT

driving the motor with a higher voltage than what was used withsinusoidal supply. This leads to a slightly smaller RMS current.

C. 50-Hz Operating Point

In the 50-Hz operating point, the same load and switchingfrequencies were used as in the 25-Hz and 40-Hz point. Thegrid voltage THD50 is 1.24%. Fig. 9 shows the voltage spec-tra for different average switching frequencies. In frequencyconverter use, the 50-Hz point is either in the field weakeningor overmodulation range because of the voltage losses in thefrequency converter input rectifier and filters. In addition to thevoltage losses, the converter selects a suitable voltage reserveto be able to control quick load changes. The mechanicalpower remains almost constant in all measurements when thefrequency converter supply is used. The inverter output powerdecreases as the switching frequency is increased, Table IV.

As shown in Fig. 10, in the 50-Hz point, the temperature riseof the machine increases roughly by one degree Celsius against12 watts of losses. The temperature rises in Fig. 11 at 2, 3,and 3.75 kHz switching frequencies are almost constants andso are the measured losses. The increase in the motor lossesis 18–20% in the frequency converter supply compared withthe sinusoidal supply. The THD values in Tables II–IV showthat in the 50-Hz operation point, the currents in the frequencyconverter supply have a higher harmonic content than in 25-Hzand 40-Hz operating points.

Fig. 10. Temperature rise differences and loss changes compared with the50-Hz sinusoidal supply as a function of average switching frequency in theDTC supply.

Fig. 11. Temperature rises at the end of the 50-Hz temperature rise test.

In frequency converter supply, this is an outcome that resultsfrom the field-weakening operation. The DTC controller leavesa voltage reserve; therefore, the studied converter does not usethe full modulation index and hence the field weakening startsat 45 Hz. Another reason of the higher harmonic content is thatthe inverter can, naturally, with the same switching frequencyuse only half of the number of switchings per one fundamentalwave when compared with the 25-Hz point, therefore, resultingin a more coarse voltage waveform.

In Fig. 12, the grid voltage and the motor current during thetemperature test with the 3-kHz average switching frequencyare shown. The motor phase current amplitude follows thechanges in the grid voltage. In fact, the similar temperature risesin Fig. 10 with different switching frequencies can be explainedby different operating points of the motor. As given in Table IV,the fundamental wave voltage amplitude is slightly higher,when the switching frequency is low and the fundamentalvoltage is decreasing from the switching frequency of 1 kHz to


Fig. 12. Grid phase voltage and motor current during a temperature rise test.The voltage and current curves are from the temperature rise test carried outwith a 3-kHz average switching frequency.

Fig. 13. Frequency converter loss model has five main components.

3 kHz. The higher fundamental wave voltage leads to a lowercurrent, and the dominating resistive losses become smaller.This distorts the impact of the switching frequency on the motorlosses.

III. FREQUENCY CONVERTER LOSS MODELS

The losses in the frequency converter are divided into fivegroups: input inductor, diode rectifier, intermediate circuit,IGBT module, and extra losses. Losses in auxiliary devicessuch as fans are included, Fig. 13.

A. Input Inductor Losses

The input inductors of the frequency converter dissipatepower in the core and in the windings. The core losses canbe divided into hysteresis losses and eddy current losses [31].Although the exact calculation of these losses is complicated,they can be estimated using data sheet parameters availablefrom magnetic component suppliers or they can be defined bymeasurements. The input inductor losses and their frequencydependency can be modeled with the lumped parameters model,Cauer, or Foster equivalent models [32]. In the analysis, onlythe inductance at nominal point, the dc resistance, and totallosses at one load point are known. Therefore, lumped param-eters cannot be used, and the losses have to be modeled with

a single series resistance-inductance model. The total losses ofthe input inductor are

Pchoke(t) = 3i2L(t)ESRL (1)

where iL(t) is the instantaneous inductor line current, andESRL is the equivalent series resistance. ESRL is calculatedby using the measured current and losses of the input inductor.

B. Diode Bridge Losses

The power dissipation of a diode in forward conductionand reverse blocking state can be modeled as a function offorward and reverse leakage currents and voltages. Losses inthe blocking state are negligible. Diode switching losses canbe considerable but for a line frequency diode bridge rectifier,the switching losses are marginal (fall within error marginal)and only the conduction losses are considered. The parametersneeded for calculating the diode bridge losses are the forwardvoltage drop UF and the on-state resistance RF. The instanta-neous conduction losses of the diode are [28]

Pdiode,on(t) = UFidiode(t) + RFi2diode(t). (2)

The diode bridge losses are six times the single diode losses.

C. Intermediate Circuit Losses

The intermediate circuit of the modeled device does notcontain dc link reactors. Therefore, the losses consist of onlythe losses in the capacitor bank and discharge resistors. Thetotal energy loss in a capacitor bank is a function of dielectriclosses attributed to the polarizing mechanisms of the electricfield on the molecular structure of the dielectric, and ohmicloss from electrodes and termination metals. Dissipative lossesof the capacitor can, again, be represented by equivalent seriesresistance whose values or a curve of frequency dependency canbe found in manufacturers’ datasheets. The number of the seriesand parallel capacitors in the bank has to be taken into accountto calculate the capacitor bank losses correctly. The dc linkcurrent consists of a dc component IDC, harmonics produced bythe diode rectifier bridge, and the switching harmonics by theIGBT-inverter bridge. The dc voltage produced by a three-phasefull-bridge rectifier carries large amounts of n times sixth-orderharmonics. The average capacitor losses PC can be written

PC =∑

n

I2C(6n)ESRC(6n) (3)

where IC(n) is the RMS value of nth order of the capacitorcurrent and ESRC(n) is the equivalent series resistance ofthe capacitor for a particular frequency. ESR values for fivedifferent harmonic frequencies have been used in the lossmodel. At 50-Hz line frequency, these harmonic frequenciesare 300, 600, 900, 1200, and 3000 Hz. The lower frequencyESR values describes the losses at harmonic currents producedby the rectifier bridge, and the highest frequency ESR value


describes the losses in the switching frequency range. Theresistive losses in the parallel discharge resistor are

Pdischarge =U2

C

Rdischarge(4)

where UC is the capacitor RMS voltage. The intermediatecircuit conduction losses are neglected in this case. Thus, thetotal intermediate circuit losses are

PDC−link = PC + Pdischarge. (5)

D. IGBT Module Losses

The IGBT module losses comprise the conducting andswitching losses of a particular device. The same loss modelsare used for both semiconductor devices—the IGBT and itsantiparallel free-wheeling diode. For the IGBTs and diodes, theinstantaneous conducting losses are [33]

PIGBT,cond = UCE0im(t) + RCE0i2m(t) (6)

and

Pdiode,cond = UF0im(t) + RF0i2m(t) (7)

where im is motor phase current, UCE0 is the IGBT’s thresholdvoltage, RCE0 is the IGBT’s on-state resistance, and UF0 andRF0 are the corresponding values of the diode. For the switch-ing losses of the IGBTs and diodes, the same linear loss modelis used [33]. The average switching losses for a specific periodof time are

PIGBT,sw =UDC

Urated

Im

IratedEsw,IGBTNsw,change (8)

and

Pfw−diode,sw =UDC

Urated

Im

IratedEsw,fw−diodeNsw,change (9)

where Nsw,change is the number of the switch changes duringthe specific time period. Esw is the switching loss energy of aparticular device given for the reference commutation voltageand current. UDC and Im are the actual commutation voltageand current.

E. Auxiliary Devices Losses

The auxiliary devices losses are comprised of the inverterself-usage, for instance microcontroller, internal power supply,display, keyboard, bus-communication, digital and analog in-puts and outputs, and the blower and control system powerconsumption. In this case, these losses are about constant. Thefan speed is not load dependent.

IV. SIMULATIONS

In [34], the coupling of 2-D FEM equations with externalcircuit equations is exhaustively addressed. [34] also reviews,analyzes, and classifies the different coupling methods. Also,

Fig. 14. Principle of the simulation software.

TABLE VFREQUENCY CONVERTER LOSS SIMULATION PARAMETERS

the main features and problems of different techniques are sum-marized. In this paper, the coupling method referred to as “cur-rent output approach” in [35] is applied. The circuit simulator isused to calculate the frequency converter losses, and the FEMmotor model calculates motor losses; additional information ofcoupling method can be found in [36] and [37] The FEM modelis based on a 2-D finite-element method program and a circuitequation of the windings [38]. The copper loss calculation isexplained in [38] and the iron loss calculation in [39]. Theprinciple of simulation software is shown in Fig. 14 and thefrequency converter simulation parameters are given in Table V.The first-order mesh of the 37-kW induction motor used in theFE analysis is shown in Fig. 15. The FEM motor model iscalculated with 25 µs time steps. The FEM motor losses are anaverage value of a 1-s simulation time containing 40 000 steps.The motor loss results could be improved using second-elementmesh or using smaller time steps in the calculation, but in thiscontext, when the total drive system is simulated, the first-ordermesh gives adequate results. The circuit simulator is runningwith 1 µs time steps, and the voltage fed to the FEM modelis the average value of 25 steps. The simulation software doesnot include any cooling or thermal model motor or frequencyconverter.


Fig. 15. The first-order mesh of the 37-kW induction motor used in the FEanalysis. The mesh contains 1448 elements and 917 nodes.

TABLE VIMOTOR LOSS SIMULATION RESULTS IN THE 25-Hz OPERATING POINT

TABLE VIIFREQUENCY CONVERTER LOSS SIMULATION RESULTS

IN THE 25-Hz OPERATING POINT

A. Simulation Results at 25 Hz

The simulation software does not take friction and windageor all additional (stray) losses into account. These losses areadded to the simulated losses. The results are given in Table VI.The additional losses are proportional to the square of the loadcurrent and to the power of 1.5 of the frequency [40]

Pad ≈ I2f1.5. (10)

Friction losses of the motor are directly proportional to thespeed, and windage losses are proportional to the third powerof speed. Because the ratio of the friction and windage lossesis not known, the total friction and windage losses are assumedhere to be proportional to the square of the speed. These lossesgiven in Tables IV and VI are scaled loss values obtained

TABLE VIIIMOTOR LOSS SIMULATION RESULTS IN THE 40-Hz OPERATING POINT

TABLE IXFREQUENCY CONVERTER LOSS SIMULATION RESULTS


TABLE XMOTOR LOSS SIMULATION RESULTS IN THE 50-Hz OPERATING POINT

by efficiency measurements (IEC 60034-2-1) in the nominalpoint.

In Table VI, the simulated and measured motor loss val-ues are close to each other. The rotor losses decrease as theswitching frequency is increased, as assumed. The simulationresults show that the extra iron losses are the most significantadditional losses that the PWM produces. The frequency con-verter simulation results are given in Table VII. It is obviousthat the input choke, diode bridge, and dc link losses remainconstants when the switching frequency is changed. The simu-lator underestimates the effect of the switching frequency on theIGBT module losses. In the measurements, rising the switchingfrequency from 1 to 3.75 kHz gives 220 watts of more ironlosses, whereas the simulated losses increase only by 120 watts.

B. Simulation Results at 40 Hz

The simulation results are given in Tables VIII and IX.In Table VIII, the simulated and measured motor loss values

with PWM converter supply are close to each other. In the


TABLE XIFREQUENCY CONVERTER LOSS SIMULATION RESULTS


Fig. 16. Measured and simulated efficiencies in the 25-Hz point.

simulated losses, the most significant loss change can be seenin rotor copper losses; they are decreasing when the switchingfrequency is increased, as assumed. Total iron losses decreasefrom 791 W to 775 W when the average switching frequency isincreased from 1 to 3.75 kHz. Also, minor loss changes in thestator copper losses can be seen.

The simulated frequency converter losses are slightly smallerthan the measured ones. The measured frequency converterlosses are increasing linearly as the switching frequency isincreased. Similarly, as in the 25-Hz point, the simulator un-derestimates the effect of switching frequency on frequencyconverter losses.

C. Simulation Results at 50 Hz

The 50-Hz simulated losses behave more logically than themeasured ones because of the different operating conditions


of the motor as explained above. The differences betweenmeasured and simulated losses are very small. The motor lossesin the 50-Hz points behave similarly as in the 25-Hz and 40-Hzpoints, even though the changes in the losses are smaller thanin the 25- or 40-Hz points. Corresponding results can be seen inthe measured temperature rises and losses. The impacts of theswitching frequency and the PWM-caused losses are smallerin the 50-Hz points than in the 25-Hz points. The simulatedfrequency-converter-caused motor losses are given in Table Xand the simulated frequency converter losses in Table XI. Themeasured losses at the 3.75-kHz switching frequency are lessthan at 3 kHz; one reason for this is the slightly smaller currentat 3.75 kHz and the other one that the switching frequencyof the frequency converter has been limited for thermal pro-tection of the IGBTs. Otherwise, the simulated and measuredfrequency converter losses behave similarly as in the 25-Hz and40-Hz points.

V. COMPARISON OF THE RESULTS

The simulated motor losses with sinusoidal supply are over200 watts less in the 25-Hz and 40-Hz operating points thanthe measured ones. The difference can be explained with puresinusoidal voltage that was used in the simulations and har-monic components that exist in the generator produced voltagethat was used in the measurements. There exists no otherexplanation than measurement error in the shaft power, why thesimulated motor losses at 50-Hz operating point with sinusoidalsupply are greater than the measured ones.



The simulated and measured frequency converter, motor,and drive efficiencies are shown in Figs. 16–18. In the25-Hz point, the measured motor efficiency is 91.0% with thesinusoidal supply, and the efficiency varies from 90% to 90.8%in the frequency converter use. In the 40-Hz point, the measuredmotor efficiency is 93.6% with sinusoidal supply and variesfrom 91.9% to 92.4% in the frequency converter use. In the50-Hz point, the measured motor efficiency is 93.6% and variesfrom 92.5% to 92.9% in the frequency converter use.

In the 50-Hz point, the measurement shows that the drivesystem efficiency is at its best value at the low switchingfrequency. In the 40-Hz operating point, the highest switchingfrequency maximizes the drive efficiency. In the 25-Hz point,the 2-kHz switching frequency maximizes the efficiency of thedrive system. Even though the simulated losses do not exactlymatch the measured ones, the efficiencies of the drive system indifferent operational point can be simulated with a reasonableaccuracy.

VI. DISCUSSION AND CONCLUSION

Frequency converter losses are difficult to measure accu-rately because of the distorted input and output currents andoutput voltages of the converter. Extra care should be takenat 50-Hz operation with a frequency converter. An accurateefficiency result of the variable-speed drives requires a veryconstant load and line voltage level as well as good measuringequipment. The differences between the simulated and mea-sured efficiencies are so small that they cannot be separatedfrom the error originating from measuring setup.

The additional losses are a function of rotational speed andswitching frequency. In this case, the motor efficiency drop was0.2%–1.7% units when PWM supply was used compared withthe sinusoidal supply depending on the switching frequency andoperating point of the motor.

Furthermore, the simple, linear loss models used for fre-quency converter loss calculation are a source of error andmay not give accurate results in every operating point of theconverter. For the scope of this paper, to obtain the drive systemlosses without any measurements, the loss models are wellsuitable.

The coupled circuit simulator used in this study gives encour-aging results in the VSD simulation. The ability to estimatedrive system efficiencies with a reasonable accuracy withoutconfirming measurements is a powerful resource in scientificand practical work.

REFERENCES

[1] E. N. Hidebrand and H. Roehrdanz, “Losses in three-phase inductionmachines fed by PWM converter,” IEEE Trans. Energy Convers., vol. 16,no. 3, pp. 228–233, Sep. 2001.

[2] A. Boglietti, A. Cavagino, and A. M. Knight, “Factors affecting lossesin induction motors with non-sinusoidal supply,” in Conf. Rec. Ind. Appl.Conf., New Orleans, LA, 2007, vol. 1, pp. 1193–1199.

[3] J.-J. Lee, Y.-K. Kim, H. Nam, K.-H. Ha, J.-P. Hong, and D.-H. Hwang,“Loss distribution of three-phase induction machines fed by pulsewidth-modulated inverter,” IEEE Trans. Magn., vol. 40, no. 2, pp. 762–765,Mar. 2004.

[4] Y. Wu, R. A. McMahon, Y. Zhan, and A. M. Knight, “Impact of PWMschemes on induction motor losses,” in Conf. Rec IEEE 41st Ind. Appl.Conf., Oct. 2006, vol. 2, pp. 813–818.

[5] A. Ruderman, “Electrical machine PWM loss evaluation basics,” in Proc.Energy Efficiency Motor Driven Syst., Heidelberg, Germany, Sep. 2005,pp. 58–68.

[6] Y. Zhan, A. M. Knight, Y. Wu, and R. A. McMahon, “Investigation andcomparison of inverter-fed induction machine loss,” in Proc. IEEE Ind.Appl. Soc. Annu. Meeting, Oct. 2008, pp. 1–6.

[7] Z. Gmyrek, A. Boglietti, and A. Cavagnino, “Estimation of iron losses ininduction motors: Calculation method, results, and analysis,” IEEE Trans.Ind. Electron., vol. 57, no. 1, pp. 161–171, Jan. 2010.

[8] A. Boglietti, A. Cavagnino, and M. Lazzari, “Fast method for the ironloss prediction in inverter-fed induction motors,” IEEE Trans. Ind. Appl.,vol. 46, no. 2, pp. 806–811, Mar./Apr. 2010.

[9] A. Boglietti, A. Cavagnino, D. M. Ionel, M. Popescu, D. A. Staton,and S. Vaschetto, “A general model to predict the iron losses in PWMinverter-fed induction motors,” IEEE Trans. Ind. Appl., vol. 46, no. 5,pp. 1882–1890, Sep./Oct. 2010.

[10] L. Ruifang, C. C. Mi, and D. W. Gao, “Modeling of iron losses of electricalmachines and transformers fed by PWM inverters,” IEEE Trans. Magn.,vol. 44, no. 8, pp. 2021–2028, Aug. 2008.

[11] K. Yamazaki and N. Fukushima, “Iron loss model for rotating machinesusing direct eddy current analysis in electrical steel sheets,” IEEE Trans.Energy Convers., vol. 25, no. 3, pp. 633–641, Sep. 2010.

[12] A. Boglietti, A. Cavagnino, and A. M. Knight, “Isolating the impact ofPWM modulation on motor iron loss,” in Conf. Rec. IEEE IAS Annu.Meeting, Edmonton, Canada, Oct. 2008, pp. 1–7.

[13] T. Haring, “Design of motors for inverter operation,” in Energy Ef-ficiency Improvements in Electronic Motors and Drives, P. Bertoldi,A. T. de Almeida, and H. Falkner, Eds. Berlin, Germany: Springer-Verlag, 2000.

[14] Rotating Electrical Machines—Part 31: Guide for the Selection and Ap-plication of Energy-Efficient Motors Including Variable Speed Drives,IEC/TS 60034-31 Ed 1, 2009.

[15] A. Mihalcea, B. Szadados, and J. Hoolboom, “Determining total lossesand temperature rise in induction motors using equivalent loading meth-ods,” IEEE Trans. Energy Convers., vol. 16, no. 3, pp. 214–219,Sep. 2001.

[16] I. Takahashi and T. Noguchi, “A new quick response and high efficiencycontrol strategy of an induction motor,” IEEE Trans. Ind. Appl., vol. IA-22, no. 5, pp. 820–827, Sep. 1986.


[17] M. Depenbrock, “Direct self-control (DSC) of inverter-fed inductionmachine,” IEEE Trans. Power Electron., vol. 3, no. 4, pp. 420–429,Oct. 1998.

[18] D. Casadei, G. Serra, and A. Tani, “Analytical investigation of torque andflux ripple in DTC schemes for induction motors,” in Proc. 23rd Int. Conf.Ind. Electron., Control Instrumen., New Orleans, LA, Nov. 1997, vol. 2,pp. 552–556.

[19] J.-K. Kang and S. K. Sul, “Analysis and prediction of inverter switchingfrequency in direct torque control of induction machine based on hystere-sis bands and machine parameters,” IEEE Trans. Ind. Electron., vol. 48,no. 3, pp. 545–553, Jun. 2001.

[20] V. Ambrozic, M. Bertoluzzo, G. S. Buja, and R. Menis, “An assessmentof the inverter switching characteristics in DTC induction motor drives,”IEEE Trans. Power Electron., vol. 20, no. 2, pp. 457–465, Mar. 2005.

[21] N. R. N. Idris, C. L. Toh, and M. E. Elbuluk, “A new torque and fluxcontroller for direct torque control of induction machines,” IEEE Trans.Ind. Appl., vol. 42, no. 6, pp. 1358–1366, Nov./Dec. 2006.

[22] N. R. N. Idris and A. H. M. Yatim, “Direct torque control of inductionmachines with constant switching frequency and reduced torque ripple,”IEEE Trans. Ind. Electron., vol. 51, no. 4, pp. 758–767, Aug. 2004.

[23] M. Hajian, J. Soltani, G. A. Markadeh, and S. Hosseinnia, “Adaptivenonlinear direct torque control of sensorless IM drives with efficiencyoptimization,” IEEE Trans. Ind. Electron., vol. 57, no. 3, pp. 975–985,Mar. 2010.

[24] C. Lascu, I. Boldea, and F. Blaabjerg, “Direct torque control of sensor-less induction motor drives: A sliding-mode approach,” IEEE Trans. Ind.Appl., vol. 40, no. 2, pp. 582–590, Mar./Apr. 2004.

[25] K.-K. Shyu, J. K. Lin, V.-T. Pham, M.-J. Yang, and T.-W. Wang, “Globalminimum torque ripple design for direct torque control of induction mo-tor drives,” IEEE Trans. Ind. Electron., vol. 57, no. 9, pp. 3148–3156,Sep. 2010.

[26] J.-K. Kang and S.-K. Sul, “New direct torque control of induction mo-tor for minimum torque ripple and constant switching frequency,” IEEETrans. Ind. Appl., vol. 35, no. 5, pp. 1076–1082, Sep./Oct. 1999.

[27] G. C. D. Sousa, B. K. Bose, J. Cleland, R. J. Spiegel, and P. J. Chappell,“Loss modeling of converter induction machine system for variable speeddrive,” in Proc. IEEE IECON, Nov. 1992, vol. 1, pp. 114–120.

[28] J. W. Kolar, H. Ertl, and F. C. Zach, “Influence of the modulation methodon the conduction and switching losses of a PWM converter system,”IEEE Trans. Ind. Appl., vol. 27, no. 6, pp. 1063–1075, Nov./Dec. 1991.

[29] Y. Wu, M. A. Shafi, A. M. Knight, and R. A. McMahon, “Comparisonof the effects of continuous and discontinuous PWM schemes on powerlosses of voltage-sourced inverters for induction motor drives,” IEEETrans. Power Electron., vol. 26, no. 1, pp. 182–191, Jan. 2011.

[30] Y. Wu, S. Shao, and R. A. McMahon, “Power loss study of inverter-fedmachine drives using discontinuous pulse width modulation,” in Proc.IEEE Ind. Conf Sustainable Energy Technol., Now. 2008, pp. 1172–1177.

[31] R. Erickson, Fundamentals of Power Electronics. London, U.K.:Chapman & Hall, 1997.

[32] F. de León and A. Semlyen, “Time domain modeling of eddy currenteffects for transformer transients,” IEEE Trans. Power Del., vol. 8, no. 1,pp. 271–280, Jan. 1993.

[33] M. H. Bierhoff and F. W. Fuchs, “Semiconductor losses in voltage sourceand current source IGBT converters based on analytical derivation,” inProc. IEEE 35th PESC, Aachen, Germany, Jun. 2000, vol. 4, pp. 2836–2842.

[34] B. Asghari, V. Dinavahi, M. Rioual, J. A. Martinez, and R. Iravani,“Interfacing techniques for electromagnetic field and circuit simulationprograms IEEE Task Force on Interfacing Techniques for SimulationTools,” IEEE Trans. Power Del., vol. 24, no. 2, pp. 939–950, Apr. 2009.

[35] S. Kanerva, “Simulation of electrical machines, circuits and control sys-tems using finite element method and system simulator,” Ph.D. disserta-tion, Helsinki Univ. Technol., Espoo, Finland, 2005.

[36] S. Kanerva, S. Seman, and A. Arkkio, “Simulation of electric drivesystems with coupled finite element analysis and system simulator,” inProc. 10th Eur. Conf. Power Electron. Appl. EPE, Toulouse, France,Sep. 2–4, 2003.

[37] S. Kanerva, C. Stulz, B. Gerhard, H. Burzanowska, J. Järvinen, andS. Seman, “Coupled FEM and system simulator in the simulation ofasynchronous machine drive with direct torque control,” in Proc. ICEM,Cracow, Poland, Sep. 5–8, 2004.

[38] A. Arkkio, “Analysis of induction motors based on the numerical solutionof the magnetic field and circuit equations,” Ph.D. Dissertation, HelsinkiUniv. Technol., Otaniemi, Finland, 1987.

[39] A. Belachen and A. Arkkio, “Comprehensive dynamic loss model ofelectrical steel applied to fe simulation of electrical machines,” IEEETrans. Magn., vol. 44, no. 6, pp. 886–889, Jun. 2008.

[40] J. Pyrhönen, T. Jokinen, and V. Hrabovcová, Design of Rotating Ma-chines. Hoboken, NJ: Wiley, 2008.

Lassi Aarniovuori was born in 1979, Finland, inJyväskylä. He received the M.Sc. and D.Sc. degreesin electrical engineering from Lappeenranta Univer-sity of Technology (LUT), Lappeenranta, Finland, in2005 and 2010.

He is currently a Researcher for the Department ofElectrical Engineering at LUT. His current researchinterests are in the field of electric drives, particu-larly simulation of electric drives, efficiency mea-surements, and calorimetric measurement systems.

Lasse I. E. Laurila was born in Helsinki, Finland,in 1971. He received the M.Sc., Lic.Tech. and D.Sc.degrees from Lappeenranta University of Technol-ogy (LUT), Lappeenranta, Finland, in 1996, 2000and 2004, respectively.

His research interests include power electronicconverters, variable-speed drives, and their control.He is currently interested in electrical energy recov-ery in mobile work machines. He is currently anAssistant Professor at the Laboratory of ElectricalDrives Technology in LUT.

Markku Niemelä was born in Mäntyharju, Finland,in 1968. He received the B.Sc. degree in electricalengineering from Helsinki Institute of Technology,in 1990. He received the M.Sc. and D.Sc. (Tech-nology) degrees from Lappeenranta University ofTechnology (LUT), Lappeenranta, Finland, in 1995and 1999, respectively.

He is currently a Senior Researcher with the Care-lian Drives and Motor Centre in LUT. His currentinterests include motion control, control of line con-verters, and energy efficiency of electric drives.

Juha J. Pyrhönen (M’06) was born in Kuusankoski,Finland, in 1957. He received the D.Sc. degreefrom Lappeenranta University of Technology (LUT),Lappeenranta, Finland, in 1991.

He has been a Professor of Electrical Machinesand Drives since 1997. He is Head of the Departmentof Electrical Engineering in the Institute of LUTEnergy. He is engaged in research and developmentof electric motors and drives.

ieee xplore download23123

Documents