Influence of constructive parameters and power signals on sound quality and airborne noise radiated by inverter-fed induction motors

Measurement 73 (2015) 503–514

Contents lists available at ScienceDirect

Measurement journal homepage: www.elsevier.com/locate/measurement

Influence of constructive parameters and power signals on sound quality and airborne noise radiated by inverter-fed induction motors R. Lara a, R. Jiménez-Romero a, F. Pérez-Hidalgo b, M.D. Redel-Macías c,⇑ a Dep. Electrical Engineering, Ed Leonardo da Vinci, Campus de Rabanales, University of Cordoba, Campus de Excelencia Internacional Agroalimentario, ceiA3, 14071 Cordoba, Spain b Dep. Electrical Engineering, Ed Ingenierías, Campus Universitario de Teatinos, University of Málaga, 29071 Málaga, Spain c Dep. Rural Engineering, Ed Leonardo da Vinci, Campus de Rabanales, University of Cordoba, Campus de Excelencia Internacional Agroalimentario, ceiA3, 14071 Cordoba, Spain

a r t i c l e

i n f o

Article history: Received 27 March 2015 Received in revised form 21 May 2015 Accepted 29 May 2015 Available online 11 June 2015 Keywords: Acoustic Harmonic analysis Pulse Width Modulation (PWM) Signal analysis

a b s t r a c t This work presents the result of a study of the acoustic quality of noise emitted by a three-phase, induction motor fed with various modulation techniques (Pulse Width Modulation, PWM). It complements the classical study of noise emitted by rotating electrical machines, based on acoustic emission levels. For this, the psychoacoustic parameters of Specific and Total Loudness, as well as Total Roughness were used. These were applied to the noise emitted by a 4-speed, Dahlander-type motor to identify their contribution to the acoustic quality of distinct pole-number configurations as well as the different control-parameter values of each of the studied techniques. The modulation techniques employed have been analyzed in previous research in which was determined their contribution to the noise levels emitted by a machine of similar characteristics as that employed here. Results are checked experimentally and compared to the sound-pressure level; showing the noise emitted and the sound quality do not coincide for a given combination of parameters. Moreover, the choice of the best modulation strategy changes depending on if the aim to achieve is less harmonic distortion (HIPWM-FMTC), less heating of associated electronics (SLPWM) or a lower acoustic emission level (HIPWM-FMTC2). Finally, with the present study an attempt is made to compliment and corroborate the results obtained up until now, as well as initiate a new focus of analysis that can serve as the basis for future research of other operation forms and regimes, and also other kinds of rotating electrical machines. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Of the three components of noise that exist in electrical motor’s normal operation, the most relevant are aerodynamic and electromagnetic. Although the latter is the one that is most noticeable at low speeds of operation, it is

⇑ Corresponding author. Tel.: +34 957 212222; fax: +34 957 218550. E-mail address: [email protected] (M.D. Redel-Macías). http://dx.doi.org/10.1016/j.measurement.2015.05.049 0263-2241/Ó 2015 Elsevier Ltd. All rights reserved.

the aerodynamic component that gains prominence when the speed of the machine is increased. Many of the control techniques of electrical machines are based on what is known as Pulse Width Modulation or PWM techniques. These types of techniques pursue different outcomes by acting on the signal that feeds each type of machine, for example: improved performance, increased control, decreased consumption, or improved quality of the power signal itself. Some of these techniques have been recently developed and they have been tested

504

R. Lara et al. / Measurement 73 (2015) 503–514

Nomenclature f f

supply frequency, 50 Hz frequency of the fundamental of modulated signal fc carrier signal’s central frequency for HIPWM-FMTC technique fF frequency of the triangular carrier signal for HIPWM-FMTC2 technique MMF magneto motive force HIPWM-FMTC Harmonics Injection Pulse Width Modulation with Frequency Modulated and Triangular Carrier HIPWM-FMTC2 Harmonics Injection Pulse Width Modulation with Frequency Modulated and Triangular Carrier 2 k real number for slope PWM (SLPWM) technique

with the objective of reducing acoustic noise level [1–8]. On the other hand, certain constructive parameters of the machine such as pole number also condition the level of noise that it generates [9–13]. These do so; not only in terms of speed, but the distribution of windings in slots, input current waveform distortion, air gap permeance fluctuations, rotor eccentricity, and phase unbalance give rise to mechanical deformations, vibrations and changing electromagnetic acoustical noise [14]. In this sense, the results achieved in these researches should be applied to other similar motors whether these conditions are keeping, although acoustical noise could change depending on both the power and charge of the motor. Also there are determining factors at the time of selecting a specific technique or control parameter for the distribution of magnetic field harmonics in the machine’s air gap [15–20]. Commonly, the study of electrical motors’ acoustic behavior is focused on determining noise levels either in general terms or as a set of frequency bands within the audible range. In this way, it is understood that electrical machines, or better said specific operating modes of these machines that generate greater noise levels, are ‘‘worse’’ than others with a lower acoustic emission level, see Fig. 1. Traditionally, the sound pressure level A-weighted decibels (dBA) have been employed for this type of comparison; however, a more developed concept of acoustics has now emerged that seeks not only to minimize emitted sound but also to fundamentally improve it. It is based on a perception of sound that simulates the operation of the human ear. This concept is called psychoacoustics and it is used in a multitude of processes from appliances to vehicles in order to improve the quality of the sound that is emitted [13,21,22]. Additionally, comparison is done through jury testing where people evaluate emitted sound with developed metrics of sound quality based on sound parameters and perception and known as Loudness, Roughness or Sharpness, among others [23]. There are numerous studies focused exclusively on the noise level that certain electrical machines generate when they are fed by different modulation techniques

Kc kf L M mf R s1 s2 SLPWM SPL THD

a l

modulation constant for HIPWM-FMTC technique modulation constant for HIPWM-FMTC technique loudness (Sone) number of pulses per period modulating signal’s frequency roughness (Asper) number of stator slots number of rotor bars Slope Pulse Width Modulation sound pressure level (dBA) total harmonics distortion control parameter for the effective slope for HIPWM-FMTC2 technique slope of trapezoidal modulator wave for slope PWM (SLPWM) technique

[4,11,24,25]. Nonetheless, up until now there is no known research based on criteria different from the greater or lesser sound emission level of these machines, as with those related to the acoustic quality of emitted sound. In this work, a methodology of study is presented whose objective is to determine the best combination of a machine’s power-signal, control variables, which will translate into an optimal acoustic emission, not only in terms of noise level but above all in terms of sound quality. Fig. 1 shows an overview of the main elements and parameters studied in this work. 2. Materials and methods 2.1. Parameters of sound quality Parameters of sound quality refer to those whose values reflect the pleasantness or unpleasantness that a particular sound might achieve, and whose study constitutes what is known as ‘‘psychoacoustics’’. Methods have been developed that enable the evaluation of some properties of a sound through descriptors such as ‘‘Loudness’’, ‘‘Roughness’’, ‘‘Sharpness’’ or ‘‘Fluctuation Strength’’, among others. Only obtaining Loudness is standardized. With regard to the characterization of sound quality, there are no metrics defined for every sound [26,27]. For example, for an internal combustion engine, some authors employ only Loudness parameter [28], while others also use Roughness and Sharpness [29,30]. In the case of electrical machines, there are no previous studies on sound quality. Thus, for its characterization Loudness and in critical cases also Roughness are employed to observe if there is a just-noticeable difference between emitted sounds. These parameters are briefly defined below. 2.1.1. Loudness The parameter of Loudness is a subjective measure of the intensity with which sound is perceived by the human ear. This metric determines how strong one sound is in relation to another. Its unit is the ‘‘Sone’’.

R. Lara et al. / Measurement 73 (2015) 503–514

505

Fig. 1. Overview.

In contrast to the common spectrum which can be divided into octaves or thirds of an octave, frequency range is fractionated into the critical bands called ‘‘Barks’’ for the first 24 frequencies. Thus, for each of these bands (Barks) at the higher level of the parameter, the sensation is understood to be of greater sonority and in dealing with a noise, as in the case of a motor, will be even more annoying. The samples and trials carried out in this work, therefore, will be considered of worse quality with regard to sound perception or acoustic quality for those parameter combinations leading to a greater value of Specific Loudness for each Bark analyzed. 2.1.2. Roughness Roughness is a parameter that quantifies the degree of annoyance resulting due to rapid modulations. Its unit is the ‘‘Asper’’. An Asper is defined as the roughness produced by a 1000 Hz tone of 60 dB, modulated at 70 Hz with a 100% modulation index. The maximum value of Roughness for any sound is achieved by modulating it at 70 Hz [31]. As in the case of Loudness, the parameter of Roughness is often represented in terms of critical bands (Barks). 2.2. Materials used 2.2.1. Motor For the purpose of evaluating the relationship and dependency between speed and acoustic noise, it was

decided on the use of a three-phase, Dahlander type, induction motor that allowed for the study of a single machine running at different speeds without modifying constructive parameters. Thus, this eliminated the possible influences that these have on acoustic noise generation, as with the number of slots both of the stator and rotor or the thickness of the magnetic core [20]. Therefore, the results achieved should be applied to another electrical machines, which have both the same structural parameters and operating conditions, with different number of poles [14]. The nominal characteristics of the motor were 380 V, 1.38 A, 0.35 kW, cos u = 0.8 and 50 Hz, with a capacity to function at 2, 4, 6 and 12 poles. As well, the number of stator slots were s1 = 36 and the number of rotor bars is s2 = 30, with single-layer, stator winding. 2.2.2. Semi-anechoic chamber To avoid interference in the measurements, the test environment was reduced to a semi-anechoic chamber in accordance with the standardized test protocol for electrical machines, which ensures a reliable measurement of the acoustic values obtained. The room fulfils the ISO 1680 standard for measurement of airborne noise emitted by rotating electrical machines and the ISO 3745 standard (for classification of semi-anechoic rooms) in the frequency range of 100 Hz to 10 kHz. The set-up was placed inside a semi-anechoic room, where background levels remained below 35 dB (A) [32].

506

R. Lara et al. / Measurement 73 (2015) 503–514

2.2.3. Sound measurement equipment The equipment used for measurement of noise consisted of a Soundbook series instrument comprised of a portable meter of multiple channels. Additionally, both the measurement software package Samurai v1.7 by SINUS Messtechnik GmbH (Germany) and a B & K calibrator were used for calibration of microphones, which were of a free-field, pre-polarized G.R.A.S half-inch type. All measurements were recorded and post-processed using Matlab R2008a by MathWorks Inc. (USA) and the open-source software Audacity 1.3.14-beta (USA). A linear average was used to place equal emphasis on all spectra and time records. This type of average is useful for the analysis of stationary signals. 2.2.4. Power inverter The power inverter was composed of a non-controlled three-phase rectifier that provided, in combination with a filter, a quasi-ideal voltage source (IVS). From it, three branches were fed with two switches, each of them formed by Insulated Gate Bipolar Transistors (IGBTs), in whose midpoint were found the terminals of the phases that feed the motor. The specific model used was a Skiip 132 GDL120-412 CTV of the brand SemikronÒ, possessing a 20 kHz maximum switching frequency. 2.2.5. Network analyzer The network analyzer used for monitoring the electrical signal applied to the motor was the Power Visa by the firm DranetzÒ. It consists of a three-phase analyzer with four inputs for voltage from 1 to 600 V rms AC/DC and for currents with a range of 0.1–3000 A rms AC/DC, both with a capacity to record 256 samples per cycle. The harmonic spectrum of the signals analyzed, with regard as much to intensity and voltage as power, cover up to 63 harmonics. 2.2.6. Control equipment To generate control signals, the software Simulink by Matlab R2008a from MathWorks Inc. (USA) was used. It was run on an IntelÒ Pentium IV PC-type computer, in which at the same time was also installed a DSP-type card to communicate with the inverter via an interface for decoupling the control signals from those of power. This card was a dSPACEÒ 60 MHz DS1102. The digital output signals of said card were filtered and isolated from the power amplifier through an interface composed of high-speed optocouplers as well as operational amplifiers that allow transfer of logic levels to the inverter as accurately as possible to that of the original signals. Similar configurations have been used to remote monitoring of electrical parameters using PC [33]. 2.3. PWM techniques used The studies conducted so far, as stated in Section 1, only focus on the level of noise emitted: overall good or good for frequencies of interest. In this work, that study is extended using sound quality. Below, the techniques analyzed are briefly described, and their influence on the acoustic noise emission, specially the electromagnetic noise.

All the techniques use the same modulation order or number of pulse per cycle of the control signal (M). Thus, it is possible to avoid resonances machine frequency and MMF harmonics with high winding factor. The M values should meet a set of requirements [5]:  M should be a whole value to allow for phase constancy between the modulator and carrier signals.  It is preferable for M to be a multiple of three. This makes the carrier, which is the highest harmonic in the simple voltage, to disappear by subtraction in the line-to-line output voltage.  M should be odd in order to guarantee that even-numbered harmonics are eliminated. It is also recommended that M to be at least equal to 10, in order to make the sampling that the carrier subjects on the modulator as effective as possible, and allow for the gathering of sufficient information.  All the foregoing leads one to choose M = 15 as the modulation order, since 15 is the first value that meets all the above requirements. 2.3.1. The HIPWM-FMTC technique The HIPWM-FMTC technique (Harmonics Injection Pulse Width Modulation with Frequency Modulated and Triangular Carrier) attempts to obtain a PWM signal of low harmonic content. It reduces the rate of distortion of the inverter’s output signal, thus improving the quality of the signal that will feed the machine. Furthermore, induction machines have been designed to radiate a low noise when they are fed by a sinusoidal signal [7]. Since this technique achieves a low harmonic distortion (THD) and therefore, a high fundamental term similar to a sinusoidal signal. So, it is expected that the electromagnetic noise radiated by the induction machine will also be low [34]. All conditions for M values, mentioned above, are imposed by this technique. For a given M, the modulation imposed on the frequency of the carrier signal through a sinusoidal function of the modulator signal forces a greater number of commutations when the slope of the modulator signal is higher and vice versa. Thus, more information for the same value of M is transmitted. The rate of change of the frequency of the carrier signal depends on Kc, and, for a given value of M, it implies changing as well, following:

M¼

2f c  k 2

ð1Þ

where M is the number of pulses per period, fc is the carrier signal’s central frequency and k is a real number. The only two conditions are that they must be real positive numbers and that fc must be in the range from M to 2M [34]. The technique consists of using a PWM that has a sinusoidal modulation signal of frequency mf = 50 Hz, with an over-modulation of the fundamental term 15%, 27% of the third harmonic in-phase, and 2.9% of the ninth harmonic in anti-phase [35]. The carrier signal for its part consists of a triangular signal whose instantaneous frequency varies proportionally to the slope of a squared sinusoidal signal, and which is synchronized to the previous modulating signal’s frequency.

R. Lara et al. / Measurement 73 (2015) 503–514

The regulation parameter fc is the central frequency from which the frequency of the carrier signal varies. Additionally, there exists a parameter kf, which is a modulation constant that modifies the variation speed of said frequency. Some example patterns are shown in Fig. 2(a and b) for an average order or number of pulses per cycle of the control signal M = 15. 2.3.2. The SLPWM technique This strategy generates high-quality voltages with very few commutations per cycle. It is particularly appropriate for reducing commutation loss [22]. Although it presents some subharmonic characteristics in the 5f and 7f frequencies, the SLPWM technique (SLope Pulse Width Modulation) achieves a shift in its harmonic spectrum at a larger margin compared to the HIPWM-FMTC technique, whose limit is imposed by the maximum switching speed of the circuit breakers [36]. This strategy is based on a frequency modulation of the triangular carrier through a sinusoidal signal. It is used to decrease acoustic noise and to achieve a reduction in THD [31]. The main advantage of this strategy is that it is possible to modify the electrical spectrum while the number of pulses per period remains unchanged [37]. Then it is possible to control the time harmonics in order to avoid some harmonic sounds, reducing the level of electrical harmonics at the output of the inverter in comparison to other traditional strategies, providing a high fundamental term, which for a given value of the rms voltage and getting a smooth variation over the electrical spectrum using one parameter to modify the modulation order M, and other to control the output voltage level, avoiding mismatches between the space harmonics and the time harmonics of the inverter output.

507

The parameters l and k ensure the control of the output signal. The value of l fixes the maximum width for the switching pulses while the value of k determines the number of pulses during the transition. The combination of parameters defines the modulation order, and the proper selection of these achieves improving certain acoustic characteristics of the motor [31]. The modulation order M (number of commutation pulses per cycle) is fixed by the k parameter. Only odd M values are feasible: M = 3, 7, 11, 15 . . . are obtained when k has positive values and M = 5, 9, 13 . . . are obtained with negative values of k. The k range to obtain a given M value is (kM4, kM), where kM4 and kM are the upper and lower limits for a given M, and they are obtained by

kM ¼

1 2p cosðdM Þ

ð2Þ

The angle dM is the solution of the equation

dM ¼ tan ðdM Þ

ð3Þ

on the interval ððM þ 1Þp=2; ðM þ 2Þp=2Þ where M is number of pulses per period. For this study with the aim to compare the results in the same operating conditions, the same modulation order (M = 15) and k between 3.25 and 4.25 have been chosen for HIPWM-FMTC technique. Fig. 2(c and d) shows some patterns of this PWM strategy for l = 2 and different values of k parameter being M = 15. 2.3.3. HIPWM-FMTC2 technique The HIPWM-FMTC2 technique (Harmonics Injection Pulse Width Modulation with Frequency Modulated and Triangular Carrier 2) has as its main objective to minimize the noise emitted by an inverter-fed, induction motor. It

Fig. 2. Examples of PWM patterns for M = 15: (a) HIPWM-FMTC with fc = 18, (b) HIPWM-FMTC with fc = 24, (c) SLPWM with k = 3.24, (d) SLPWM with k = 4.24, (e).

508

R. Lara et al. / Measurement 73 (2015) 503–514

takes advantage of the two previously described techniques, though with some variation [3]. This technique uses a carrier signal’s frequency modulation based on the SLPWM technique, which while it degrades the signal obtained in terms of the rate of harmonic distortion, achieves further dispersal of the harmonics of the side-bands. Thus, it improves some of the acoustic characteristics of the noise generated by the machine [2]. On the other hand, in addition to selectively injecting harmonics through which is achieved less voltage distortion when exiting the inverter, the modulated wave’s most significant harmonics can be shifted to frequencies higher than those of the theoretical limit imposed by the HIPWM-FMTC technique. In this way, noise due to vibration of the stator and/or rotor can be reduced, or those frequencies avoided that permit the combination of machine-load to enter into resonance. The harmonic spectrum presents a progressive evolution when the control parameter a varies over a range, therefore it is possible to optimize the inverter behavior in order to achieve the THD and reduce the acoustic noise. In contrast to the SLPWM technique, in the HIPWM-FMTC2 technique a linear variation of the instantaneous frequency (frequency modulation) is proposed for a time that is adjustable. The practical limit for this strategy (as with the SLPWM technique) is imposed by the maximum switching speed that the inverter permits. As with the HIPWM-FMTC technique from which it derives, the expression that follows the carrier’s

instantaneous frequency and that determines the switching pulses is given by an expression that depends on this frequency’s maximum value and change constant. The effective slope angle a is the control parameter. Furthermore, the maximum, theoretical angle at which the pulse is cancelled is 90°, but because the objective is to increase switching during the interval of the carrier’s maximum slope, it is limited to 30°. Consequently, the results of Total Harmonic Distortion THD, like the Distortion Factor DF, are worse beginning at this value. As mentioned above, in order to compare the results achieved for the different techniques, the same modulation order (M = 15) has been chosen. The triangular carrier frequency is modulated by a triangular signal with frequency fF double of f, to maintain low the THD parameter [38]. Fig. 2(e and f) shows some examples of PWM pattern of the HIPWM-FMTC2 technique. 2.4. Test methodology Fig. 3 collects the integration and interconnection of the equipment and materials used during the data collection process. In summary, the methodology for carrying out measurements for each of the employed PWM modulation techniques and for any number of the tested machine’s pole pares consists of using SimulinkÒ software to generate each variable configuration’s control vector and then send it to a DSP card which manages communications with the inverter through the appropriate interface. For its part, the

Fig. 3. Experiment set-up.

509

R. Lara et al. / Measurement 73 (2015) 503–514

another kind of induction motors with the same number of poles, slots, distribution of windings in slots, rotor eccentricity, air gap permeance fluctuations operating at the same conditions and feeding with the same input current waveform. This should also be extrapolable to psychoacoustic results [14].

Table 1 Values of the control parameters used, according to the type of strategy. All correspond to a modulation order M = 15. HIPWM-FMTC 15 6 fc 6 30

SLPWM 3.24 6 k 6 4.24 (l = 2)

HIPWM-FMTC2 0° 6 a 6 35°

16.75 18 22 24 27 28 29

3.24 3.54 3.84 4.24

17 18 20 25 35 45

3. Results and discussion In order to study the noise emitted by the machine in terms of sound quality, the results obtained were presented after applying Loudness and Roughness parameters, as analysis filters to the wav files generated after each measurement. Within the possible margins, a number that was representative of the control parameters values for each of the techniques was selected with the objective of evaluating its performance in overall terms (Table 1). The chosen values are shown as are each control parameter’s possible intervals according to the PWM technique employed for the case of 15 pulses per cycle in the control signal (M = 15). In this same table, the values used for the psychoacoustic analysis that was performed are also included. During the tests and measurements, a large number of samples were taken within the possible intervals. Furthermore, it was opted for a sufficiently representative number of these values, ones that corresponded to combinations that offered just-noticeable-differences but which did not make the study exceedingly difficult.

motor which is found inside the ‘‘semi-anechoic’’ chamber is powered by the inverter. Moreover, for each configuration of pole pairs and for every value of the analyzed techniques’ control parameters, the corresponding measure, according to the protocol mentioned in Section 2.2.2, is carried out, and emitted noise is recorded for subsequent processing and analysis. During these measurements, the electrical signals from the network analyzer that feed the machine were simultaneously recorded [39]. The fundamental power frequency was 50 Hz in all trials. To avoid the phenomenon of magnetic saturation of the motor, all the measurements were carried out for voltage values lower than 80% of the machine’s nominal value. The noise emitted by the machine, when it was connected to the power supply, was measured and recorded for 30 s, for each of the pole number’s strategies and configurations. Due to the relationship between the constructive parameters of the induction machine and the acoustic noise emission under specific operating conditions, it is possible to claim that the results could be applied to

Fig. 4(a–d) shows the evolution of the Specific Loudness parameter (Sone/barks) for each value of pole number vs. frequency (barks).

4.5 4

10

3.5 3 15

2.5 2

20

1.5 1 18

22

24

27

28

29

5.5 5

5

4.5 4

10

3.5 3 15

2.5 2 1.5

20

1

0.5

16.75

18

22

4.5 4

10

3.5 3 15

2.5 2

20

1.5 1

f

c

0.5

27

28

29

0.5

6 5.5

Bandas de frecuencia (bark)

5

5

Specific Loudness (Sone/bark)

Bands of frequency (bark)

5.5

24

29

(d)

6

(b)

22

28

c

c

18

27

f

f

16.75

24

5

5

4.5 4

10

3.5 3 15

2.5 2 1.5

20

1 16.75

18

22

24

27

28

f

c

Fig. 4. Specific Loudness values for technique HIPWM-FMTC. (a) 2 Poles, (b) 4 Poles, (c) 6 Poles and (d) 12 Poles.

29

0.5

Specific Loudness (Sone/bark)

16.75

Bands of frequency (bark)

5

5

6

(c) Specific Loudness (Sone/bark)

5.5

Specific Loudness (Sone/bark)

6

(a) Bands of frequency (bark)

3.1. Analysis of the HIPWM-FMTC technique

R. Lara et al. / Measurement 73 (2015) 503–514

The influence of pole number on the Specific Loudness parameter’s value is clearly observed. When the pole number is low, the speed of the machine is greater (Fig. 4a). Moreover, the level of Specific Loudness undergoes an increase when frequency is decreased, coinciding with frequencies that correspond to the aerodynamic noise component which is predominant at high speeds. In this situation, the control parameter’s value has minimal influence on the machine’s level of Specific Loudness, given that it influences the electromagnetic noise component which, as previously stated, is masked at these speeds by the aerodynamic component. As the number of poles increases and, thus, speed decreases (Fig. 4b–d), the control parameter’s influence on the value of Specific Loudness is more apparent at the same time that the amount of said Specific Loudness also shows a reduced value due to the decreased presence of the aerodynamic component. As well, it can be observed how at low speeds, depending on the parameters selected, differences in the contribution of Specific Loudness are present even for the same amount of poles, see Fig. 4(c and d). Using a machine operating with 6 poles, Specific Loudness is considerably greater fc parameter values near the end of the feasible interval than for central values. Thus, with 6 poles the range of values to use for an fc parameter is understood to be between 24 and 27 in order to improve the quality of sound emitted by the motor. Beside, for these combinations of parameters poles and fc, it is possible to reduce by half the THD and avoiding at the same times the excitation of the stator resonance frequencies. It is due to the great amount of magnetic noise in the motor happens at frequencies with radial magnetic forces which are the same or very close to the natural mechanical frequencies of the stator system.

3.2. Analysis of the SLPWM technique For this situation, the values of Specific Loudness for each configuration of pole numbers are shown in Fig. 5(a–d). As with the HIPWM-FMTC strategy that was analyzed, Fig. 5(a and b) depicts how at high operating speeds the greatest levels of Specific Loudness appear at low frequencies due to the presence of high levels of noise having an aerodynamic origin. This is especially noticeable in the case of a machine operating with 2 poles, see Fig. 5a. In the case of 4 poles (Fig. 5b), significant levels of Specific Loudness also appear at high frequencies for some values of the control parameter k, especially for k = 3.84. In spite of this, the highest level of Specific Loudness is present at

4.5 4

10

3.5 3 15

2.5 2

20

1.5 1 3.54

3.84

4.24

5.5 5

5

4.5 4

10

3.5 3 15

2.5 2

20

1.5 1

0.5

3.24

3.54

k

4.24

6

6

(d)

5.5

4.5 4

10

3.5 3 15

2.5 2

20

1.5

Bands of frequency (bark)

5

5

Specific Loudness (Sone)

5.5

5

5

4.5 4

10

3.5 3 15

2.5 2

20

1.5

1 3.24

3.54

3.84

k

0.5

k

(b) Bands of frequency (bark)

3.84

4.24

0.5

1 3.24

3.54

3.84

k

Fig. 5. Specific Loudness values for technique SLPWM. (a) 2 Poles, (b) 4 Poles, (c) 6 Poles and (d) 12 Poles.

4.24

0.5

Specific Loudness (Sone)

3.24

Specific Loudness (Sone)

5

5

6

(c)

5.5

Specific Loudness (Sone)

6

(a) Bands of frequency (bark)

Consequently, the lowest value of Loudness appears, in overall terms, when the machine operates at 12 poles, where it presents in the case of fc = 28, a Loudness value at which for almost all frequencies the sound quality is worse. Thus, this is the value to avoid in this case, see Fig. 4d. Therefore, the higher number of poles, the lower Loudness, whereas THD decreased, decreasing the emitted acoustic noise level for the same values of fc, especially at low frequencies. In this sense, as previous researches has shown [2,34], the influence of M = 15 can be seen since the greater the value of fc the more the harmonics fade, yielding a sound spectrum where only the space harmonics are meaningful. It happens because this configuration of the parameters (12 poles and M = 15) displaces the first significant harmonic toward a higher frequency to avoid the mechanical resonance produced by the electromagnetic noise and thereby reduces the emitted noise by the motor.

Bands of frequency (bark)

510

511

R. Lara et al. / Measurement 73 (2015) 503–514

4

10

3.5 3 15

2.5 2

20

1.5

Bands of frequency (bark)

4.5

Specific Loudness (Sone)

5.5 5

5

4.5 4

10

3.5 3 15

2.5 2

20

1.5

1 17

18

20

25

30

35

1

0.5

17

18

20

a 6

4

10

3.5 3 15

2.5 2

20

1.5 1 25

30

35

0.5

a

Bands of frequency (bark)

4.5

Specific Loudness (Sone)

Bands of frequency (bark)

5

5

20

35

0.5

6

(d)

5.5

18

30

a

(b)

17

25

5.5 5

5

4.5 4

10

3.5 3 15

2.5 2

20

1.5

Specific Loudness (Sone)

Bands of frequency (bark)

5

5

6

(c)

5.5

Specific Loudness (Sone)

6

(a)

1 17

18

20

25

30

35

0.5

a

Fig. 6. Specific Loudness values for technique HIPWM-FMTC2. (a) 2 Poles, (b) 4 Poles, (c) 6 Poles and (d) 12 Poles.

the lowest value of the control parameter k = 3.24. It manifests itself in the lowest frequency bands, coinciding and overlapping with those originating from the aerodynamic noise component. Thus, in this instance it is difficult to establish an optimum value for the control parameter as it depends on those frequency bands for which the improvement of sound quality is desired. When the number of poles increases, it was observed how a high level of Specific Loudness is maintained in low frequency bands even in the case of the lowest speed (Fig. 5d). As already discussed in Section 2.3.2, although the SLPWM technique is able to redistribute and displace greater level harmonics to high frequencies, it nevertheless shows subharmonic characteristics at the frequencies 5f and 7f with f being the fundamental frequency, in this instance 50 Hz. Therefore, this is the reason why high levels of Specific Loudness are maintained at low frequencies. In the case of a machine running at the lowest of the analyzed speeds, corresponding to a 12-pole configuration (Fig. 5d), the lowest Loudness levels appear at the highest control parameter values found at k = 3.84 and k = 4.24. Thus, these are those that yield a priori better acoustic quality. The best values for Specific Loudness are achieved for those frequencies where the interaction between both space and time harmonic is low. Increasing k, the first significant harmonic tends to displace toward the higher frequency avoiding the mechanical resonance radiated by the electromagnetic noise and thus reducing the overall noise emitted by the motor and also improving the electrical THD. 3.3. Analysis of the HIPWM-FMTC2 technique Fig. 6(a–d) depicts the Loudness level for each one of the control parameters used in the study of the

HIPWM-FMTC2 technique and shown in Table 1. As in previous cases, predominance of the aerodynamic component with regard to low pole number (high speeds) is clearly visible. Especially in the case of 2 poles (Fig. 6a), the influence of the control parameter a is practically null. In contrast, when the speed decreases, dependency the control parameter increases, to the point that it determines the level of Specific Loudness. It is clearly seen how at 6 poles (Fig. 6c) a value of a = 25° produces a very high Specific Loudness level at high frequencies, one even higher than is produced in any of the other configurations, including those in which the aerodynamics component is dominant. It is due to the electric spectrum of the current input supply shows a great number of harmonics with a significant magnitude. In the case of the electrical machines with an important number of poles, where both space and time harmonics is high, these are very difficult to avoid, so this increases the acoustic noise and Loudness. In this sense, changing the a parameters it is possible to scatter the harmonics toward high frequencies to avoid the mechanical resonance induced by electromagnetic noise and therefore to reduce the acoustic noise radiated by the motor. For this reason, in those circumstances that value of parameter a should be avoided at all costs as there is clearly better sound quality for values a = 18°. In the case of lower speed, where the electromagnetic component is the most important and consequently the influence of the strategy’s control parameter is greater, it is observed that although in overall terms the performance of acoustic quality is better, it is especially so for the value f a = 18°, for which it is considerably superior. Therefore, this is the value with the best acoustic performance in terms of acoustic quality.

512

R. Lara et al. / Measurement 73 (2015) 503–514

0.14 0.12 0.14 0.12 Where SPL is sound pressure level (dBA); L is Loudness (Sone); R is Roughness (Asper).

L

57.14 56.69 46.29 47.64 90.08 85.98 82.34 85.89 0.13 0.14 0.09 0.14 60.84 59.10 47.46 57.52 90.09 86.91 82.88 83.47 0.21 0.14 0.09 0.11 59.82 57.28 29.95 37.28 89.28 86.44 80.04 86.70 0.18 0.11 0.08 0.11 58.52 48.60 35.14 40.46 89.20 83.32 78.86 80.23 0.16 0.16 0.11 0.13 60.77 50.91 51.31 37.70 89.96 86.01 85.79 79.95 0.09 0.17 0.18 0.09 62.21 58.95 67.22 40.97 90.21 87.26 89.84 81.70 0.10 0.15 0.18 0.11

SPL L

60.26 60.87 74.28 42.80 90.50 87.95 91.10 85.59 2 4 6 12

29

R L SPL

28

R L SPL

27

R L SPL

24

R L SPL

22

R SPL

L 18

SPL

R 16.75

fc Poles

Table 2 Values of the sound pressure level, Total Loudness and Total Roughness according to the control parameters and the number of poles in the strategy HIPWM-FMTC for a modulation order M = 15.

In terms of Specific Loudness, it is difficult to determine which of the techniques provided better overall results given that a said variable achieves very different values for distinct combinations of a machine’s control parameters and pole numbers. This is due to the fact that the frequencies in which the machine shows worse results depend on one hand on the presence of harmonics of a spatial origin that result from constructive characteristics and the machine’s pole number, and on the other on the presence of harmonics of temporal origin that are distinct for every control parameter’s value at each technique [1,4,15,40,41]. As mentioned before, the results found in this research should be similar when another kind of induction motor with the same constructive parameters is used under the same operating conditions. Previously carried out studies confirm that the HIPWM-FMTC2 strategy achieves slightly inferior results than those obtained by the other two techniques [2]. However, in the present work it was observed that in terms of acoustic quality that conclusion is not easily reached by addressing exclusively the distribution of Specific Loudness. Simply with the presence of low-order harmonics, characteristic of the SLPWM technique, the results worsen when compared with the value of Loudness obtained by the other two techniques analyzed at the same frequencies, confirming it to be the one which gives worse results at low frequencies. Nevertheless, the same cannot be asserted for mid and high frequencies. Thus, the values of sound pressure level, Total Loudness and Roughness for each of the analyzed strategies are presented, HIPWM-FMTC (Table 2), SLPWM (Table 3) and HIPWM-FMTC2 (Table 4), with the objective of analyzing in which of the configurations and strategies is found the best value of sound quality and to compare with the classical study based on noise level. For the HIPWM-FMTC strategy (see Table 2), the lowest value of Total Loudness is present in the 6-pole and fc = 27 configuration, reaching a value of 29.95 Sones. In turn, SLPWM (see Table 3) reaches a minimum at the 12-pole and k = 3.84 configuration, yielding a value of 34.95 Sones. Finally, the HIPWM-FMTC2 strategy (see Table 4) presents a value of 35.63 Sones for the configuration 12-pole and a = 18. Based on these values, the first combination (strategy HIPWM-FMTC, 6 poles and fc = 27) could be considered of better acoustic quality. Although due to the fact that the differential margin between the obtained values is narrow, above all taking into account that it deals with the Total Loudness value for all frequencies, it was considered convenient to employ an additional criterion, utilizing another psychoacoustic study parameter that permitted deciding the parameter combinations that achieved a better result in terms of sound quality. Given this, the result of the Roughness parameter defined in Section 2.1.2 will be analyzed for those combinations that showed better Total Loudness. Those that presented a lower value of Roughness shall be understood as those that, in general terms, will provide better sound quality results.

R

3.4. General analysis

513

R. Lara et al. / Measurement 73 (2015) 503–514

Table 3 Values of the sound pressure level, Total Loudness and Total Roughness according to the control parameters and number of poles in the SLPWM strategy for a modulation order M = 15. Poles

k 3.24

2 4 6 12

3.54

3.84

4.24

SPL

L

R

SPL

L

R

SPL

L

R

SPL

L

R

89.79 89.17 87.02 84.07

58.15 60.94 50.77 41.77

0.12 0.14 0.13 0.08

88.70 89.13 81.70 84.03

58.65 64.51 44.45 46.43

0.14 0.12 0.09 0.16

90.20 89.72 87.70 78.69

58.47 65.03 61.15 34.95

0.13 0.15 0.16 0.12

89.68 88.54 85.46 81.87

63.59 59.51 54.17 38.19

0.14 0.18 0.12 0.12

Where SPL is sound pressure level (dBA); L is Loudness (Sone); R is Roughness (Asper).

Table 4 Values of the sound pressure level, Total Loudness and Total Roughness according to the values of the control parameters and number of poles in the HIPWMFMTC2 strategy for a modulation order M = 15. Poles

a 17

2 4 6 12

18

20

25

35

SPL

L

R

SPL

L

R

SPL

L

R

SPL

L

R

SPL

L

R

89.98 87.22 83.77 85.00

58.78 58.11 49.64 52.48

0.13 0.15 0.10 0.13

89.88 87.85 81.37 78.10

55.95 55.64 44.91 35.63

0.13 0.17 0.11 0.12

89.53 88.09 87.65 85.20

56.46 60.66 62.26 50.94

0.14 0.15 0.14 0.14

89.85 87.51 91.20 83.47

58.88 57.10 74.47 48.91

0.19 0.19 0.11 0.17

90.17 87.90 82.96 84.32

60.83 59.53 47.40 51.00

0.14 0.13 0.08 0.1

Where SPL is sound pressure level (dBA); L is Loudness (Sone); R is Roughness (Asper).

In the case of the HIPWM-FMTC strategy, the combination that exhibits the least Total Loudness was the one with 6 poles and fc = 27. In this instance, the Roughness value was 0.09 Asper (Table 2). In the SLPWM technique, the best combination was for 12 poles and k = 3.84, having a Roughness of 0.12 Asper (Table 3). Finally, the combinations with the best Total Loudness for the strategy HIPWM-FMTC2 were, on the one hand, that of 12 poles and a = 18, and on the other 6 poles and a = 45. These presented Roughness values of 0.16 Asper and 0.12 Asper, respectively (Table 4). As specified in Section 2.1.2, the just noticeable difference for the Roughness parameter is present for measures superior to 17% of the same. In the range of the obtained values, the just noticeable difference for 0.09 Asper will be 0.0153 Asper. For 0.12 Asper, the just noticeable difference stands at 0.02 Asper. This suggests that of the studied combinations, the one that shows the best performance in terms of acoustic quality thus is the first of these, strategy HIPWM-FMTC, given that the rest of combinations present values greater than the just noticeable difference. The sound pressure levels in these cases are respectively 80.04 dB for HIPWM-FMTC, 78.69 dB for SLPWM and 78.10 dB for HIPWM-FMTC2. Therefore, it can be confirmed that a given machine operation with a greater, overall noise emission level does not necessarily need to be accompanied by worse sound quality results [2].

of acoustic emission level. The psychoacoustic parameters used were Specific Loudness, Total Loudness and Total Roughness. It was shown that the sound-pressure level and acoustic quality of the noise emitted by the motor do not need to coincide for a given combination of parameters. Thus, depending on the result desired, different criteria and analyses can be utilized. For the strategies analyzed, the one that obtains a better result corresponds to that in whose conception prevailed less harmonic distortion (HIPWM-FMTC), over that which sought less heating of associated electronics (SLPWM), and above even that which sought a lower acoustic emission level (HIPWM-FMTC2). Moreover, the best configurations of control parameters, which reduce the acoustic noise, radiated and improve the sound quality for each technique has been given. In this sense, due to the relationship between the constructive parameters of the induction motors and acoustic noise emission, the results should be similar to another kind of induction motor operating under the same conditions and feeding on the same input current waveform. Finally, the use of sound quality in electrical machines could also help from a maintenance standpoint, as it could boost the detection of misalignments, shabby bearing and overloads among others comparing the results for machines in different operating conditions.

References 4. Conclusions Sound quality criteria were used to study the psychoacoustic performance of noise emitted by a four-speed, Dahlander-type motor when fed with several PWM modulation techniques which were previously analyzed in terms

[1] N. Hashemi, R. Lisner, D.G. Holmes, iEEE, Acoustic noise reduction for an inverter-fed three-phase induction motor, in: Conference Record of the 2004 Ieee Industry Applications Conference, Covering Theory to Practice, vol. 1–4, 2004, pp. 2030–2035. [2] A. Ruiz-González, M.J. Meco-Gutiérrez, F. Vargas-Merino, F. PérezHidalgo, J.R.H. Larrubia, Shaping the HIPWM-FMTC strategy to reduce acoustic noise radiated by inverter-fed induction motors,

514

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14] [15]

[16]

[17]

[18]

[19]

[20]

[21]

R. Lara et al. / Measurement 73 (2015) 503–514 in: 19th International Conference on Electrical Machines, ICEM 2010, Rome, 2010. A. Ruiz-Gonzalez, M.J. Meco-Gutierrez, F. Vargas-Merino, J.R. Heredia Larrubia, F. Perez-Hidalgo, New modulation strategy to reduce the acoustic noise and vibrations radiated by PWM-Controlled Induction Machine Drives, in: ICEM: 2008 International Conference on Electrical Machines, Vilamoura, Portugal, 2009, pp. 1318–1320. A. Ruiz-Gonzalez, F. Vargas-Merino, F. Perez-Hidalgo, M.J. MecoGutierrez, J.R. Heredia-Larrubia, Low switching PWM strategy to reduce acoustic noise radiated by inverter-fed induction motors, Bari (2010) 1353–1358. M. Meco-Gutierrez, J.R. Heredia-Larrubia, F. Perez-Hidalgo, A. RuizGonzalez, F. Vargas-Merino, Pulse width modulation technique parameter selection with harmonic injection and frequency modulated triangular carrier, IET Power Electron. 6 (2013) 954–962. A.C. Binojkumar, J.S.S. Prasad, G. Narayanan, Experimental investigation on the effect of advanced bus-clamping pulse width modulation on motor acoustic noise, IEEE Trans. Ind. Electron. 60 (2013) 433–439. J.F. Brudny, J.P. Lecointe, Rotor design for reducing the switching magnetic noise of AC electrical machine variable-speed drives, IEEE Trans. Ind. Electron. 58 (2011) 5112–5120. L. Mathe, F. Lungeanu, D. Sera, P.O. Rasmussen, J.K. Pedersen, Spread spectrum modulation by using asymmetric-carrier random PWM, IEEE Trans. Ind. Electron. 59 (2012) 3710–3718. B. Weilharter, O. Biro, H. Lang, G. Ofner, S. Rainer, Validation of a comprehensive analytic noise computation method for induction machines, IEEE Trans. Ind. Electron. 59 (2012) 2248–2257. D. Torregrossa, A. Khoobroo, B. Fahimi, Prediction of acoustic noise and torque pulsation in PM synchronous machines with static eccentricity and partial demagnetization using field reconstruction method, IEEE Trans. Ind. Electron. 59 (2012) 934–944. J. Le Besnerais, V. Lanfranchi, M. Hecquet, P. Brochet, Characterization and reduction of audible magnetic noise due to PWM supply in induction machines, IEEE Trans. Ind. Electron. 57 (2010) 1288–1295. J. Hur, J.W. Reu, B.W. Kim, G.H. Kang, Vibration reduction of IPMtype BLDC motor using negative third harmonic elimination method of air-gap flux density, IEEE Trans. Ind. Appl. 47 (2011) 1300–1309. Y.C. Chen, W.C. Ko, H.L. Chen, W.J. Wu, P.Z. Chang, C.K. Lee, A thin light flexible electromechanically actuated electret-based loudspeaker for automotive applications, IEEE Trans. Ind. Electron. 59 (2012) 4140–4147. W. Gieras, L. Cho, Noise of Polypahse Electric Motors, Taylor & Francis, Boca Raton, USA, 2006. A. Hubert, G. Friedrich, Influence of power converter on induction motor acoustic noise: interaction between control strategy and mechanical structure, IEE Proc. –Electr. Power Appl. 149 (2002) 93– 100. S. Chaithongsuk, B. Nahid-Mobarakeh, J.P. Caron, N. Takorabet, F. Meibody-Tabar, Optimal design of permanent magnet motors to improve field-weakening performances in variable speed drives, IEEE Trans. Ind. Electron. 59 (2012) 2484–2494. T. Kobayashi, F. Tajima, M. Ito, S. Shibukawa, Effects of slot combination on acoustic noise from induction motors, IEEE Trans. Mag. 33 (1997) 2101–2104. J. Le Besnerais, V. Lanfranchi, M. Hecquet, P. Brochet, Optimal slot numbers for magnetic noise reduction in variable-speed induction motors, IEEE Trans. Mag. 45 (2009) 3131–3136. J. Muster, P.K. Budig, R. Belmans, W. Geysen, P.K. Sattler, Audible noise in speed-controlled inverter-fed medium-sized inductionmotors, European Trans. Electr. Power Eng. 5 (1995) 5–13. B. Cassoret, J.P. Lecointe, J.F. Brudny, Influence of the pole number on the magnetic noise of electrical AC machines, Prog. Electromag. Res. B (2011) 83–97. A. Camacho, G. Pinero, M. De Diego, A. González, Exploring roughness perception in car engine noises through complex cepstrum analysis, Acta Acustica United with Acustica 94 (2008) 120–140.

[22] G.B. Rossi, F. Crenna, M. Codda, Measurement of quantities depending upon perception by jury-test methods, Measurement 34 (2003) 57–66. [23] H. Fastl, The psychoacoustics of sound-quality evaluation, Acustica 83 (1997) 754–764. [24] F. Vargas-Merino, M.J. Meco-Gutiérrez, J.R. Heredia-Larrubia, A. Ruiz-Gonzalez, Low switching PWM strategy using a carrier wave regulated by the slope of a trapezoidal modulator wave, IEEE Trans. Ind. Electron. 56 (2009) 2270–2274. [25] Z.Q. Zhu, L. Xu, D. Howe, I.E.E. Iee; IEE, Influence of mounting and coupling on the natural frequencies and acoustic noise radiated by a PWM controlled induction machine, in: Ninth International Conference on Electrical Machines and Drives, 1999, pp. 164–168. [26] ISO 532:1975, Acoustics – Method for calculating loudness level, 1975. [27] M.D. Redel-Macías, S. Pinzi, D.E. Leiva-Candia, A.J. Cubero-Atienza, M.P. Dorado, Influence of fatty acid unsaturation degree over exhaust and noise emissions through biodiesel combustion, Fuel 109 (2013) 248–255. [28] M.D. Redel-Macias, S. Pinzi, M.F. Ruz, A.J. Cubero-Atienza, M.P. Dorado, Biodiesel from saturated and monounsaturated fatty acid methyl esters and their influence over noise and air pollution, Fuel 97 (2012) 751–756. [29] M.D. Redel-Macias, S. Pinzi, D. Leiva, A.J. Cubero-Atienza, M.P. Dorado, Air and noise pollution of a diesel engine fueled with olive pomace oil methyl ester and petrodiesel blends, Fuel 95 (2012) 615– 621. [30] L.P.R. de Oliveira, K. Janssens, P. Gajdatsy, H. Van der Auweraer, P.S. Varoto, P. Sas, W. Desmet, Active sound quality control of engine induced cavity noise, Mech. Syst. Signal Process. 23 (2009) 476–488. [31] P. Daniel, R. Weber, Psychoacoustic roughness: implementation of an optimized model, Acustica 83 (1997) 113–123. [32] F.R. Lara, Modular semi-anechoic chamber for acoustic testing in rotating electrical machines, in: International Conference of Electrical Machines, Vilamoura, Portugal, 2008. [33] J. Datta, S. Chowdhuri, J. Bera, G. Sarkar, Remote monitoring of different electrical parameters of multi-machine system using PC, Measurement 45 (2012) 118–125. [34] A. Ruiz-González, M.J. Meco-Gutiérrez, F. Pérez-Hidalgo, F. VargasMerino, J.R. Heredia-Larrubia, Reducing acoustic noise radiated by inverter-fed induction motors controlled by a new PWM strategy, IEEE Trans. Ind. Electron. 57 (2010) 228–236. [35] M.J. Meco-Gutiérrez, F. Pérez-Hidalgo, F. Vargas-Merino, J.R. Heredia-Larrubia, Pulse width modulation technique with harmonic injection and frequency modulated carrier: formulation and application to an induction motor, IET Electric Power Appl. 1 (2007) 714–726. [36] F. Vargas-Merino, M.J. Meco-Gutierrez, J.R. Heredia-Larrubia, A. Ruiz-Gonzalez, IEEE, Slope modulation strategy for generated PWM, 2008. [37] P.K. Sadhu, G. Sarkar, A. Rakshit, A microcontroller-based variable voltage variable frequency sinusoidal power source with a novel PWM generation strategy, Measurement 45 (2012) 59–67. [38] H. Stemmler, T. Eilinger, Spectral analysis of the sinusoidal PWM with variable switching frequency for noise reduction in inverter-fed induction motors, Power Electron. Spec. Conf. PESC 94 (1994) 269– 277. [39] F.R. Lara, Establecimiento de un protocolo de ensayos normalizado para el estudio del ruido aéreo emitido por máquinas eléctricas rotativas, Congreso Hispano Luso de Ingeniería Eléctrica-CHLIE, Marbella, Spain, 2005. [40] W.R. Finley, Noise in induction-motors – causes and treatments, IEEE Trans. Ind. Appl. 27 (1991) 1204–1213. [41] W.R. Finley, IEEE, Noise in induction-motors – causes and treatments industry applications society, in: 37th Annual Petroleum and Chemical Industry Conference: Record of Conference Papers, 1990, pp. 55–63.