Noise generation by vacuum cleaner suction units. Part III. Contribution of structure-borne noise to total sound pressure level

Applied Acoustics 68 (2007) 521–537 www.elsevier.com/locate/apacoust

Noise generation by vacuum cleaner suction units. Part III. Contribution of structure-borne noise to total sound pressure level ˇ udina *, Jurij Prezelj Mirko C University of Ljubljana, Faculty of Mechanical Engineering, Asˇkercˇeva 6, 1000 Ljubljana, Slovenia Received 17 January 2006; received in revised form 19 July 2006; accepted 5 October 2006 Available online 29 November 2006

Abstract Noise emitted by a vacuum cleaner suction unit consists of airborne and structure-borne noise. The airborne noise is generated mainly by the turbo blower and the structure-borne noise is generated mainly by the driving electric motor. The structure-borne noise depends on the suction unit design and on operating conditions, and is especially distinct at partial flow rates when rotating stall and surge appear. Among geometrical parameters, the stator of the blower and the electric motor, or metal shield if any, have the greatest effect on the structure-borne noise. Therefore, in this part of the paper, the effects of vibrations of the electric motor structure on the noise characteristics have been measured and analysed at the design and off-design operation. The contribution of structure-borne sound to the total sound pressure level becomes relatively less important at higher flow rates and with a vaned diffuser built-in the blower. Ó 2006 Elsevier Ltd. All rights reserved. PACS: 43.50.Ed; 43.50.Pn; 47.27.Sd; 43.50.Gf Keywords: Suction unit; Structure-borne noise; Vibration; Measurements; Surge

1. Introduction Noise generated by a vacuum cleaner suction unit is a result of noise generated partially by the blower and partially by the electric motor and thus consists of airborne and *

Corresponding author. Tel.: +386 1 4771 443; fax: +386 1 251 8567. ˇ udina). E-mail address: [email protected] (M. C

0003-682X/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.apacoust.2006.10.001

522

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

structure-borne origins. The airborne noise is generated mainly by the turbo blower and the structure-borne noise is generated mainly by the driving electric motor. The contribution of the structure-borne noise to the total sound pressure level (SPL) depends on the geometry of the electric motor and on operating conditions (speed and load). Among the geometrical parameters, the stator of the electric motor (or metal shield if any), and among operating conditions the load, or appearance of rotating stalls and surge at partial flow rates, are the most influential parameters for generation of the structure-borne noise and to its contribution to the total SPL. The contribution of the structure-borne noise to the total SPL also depends, although indirectly, on the stator of the blower, i.e., the builtin vaned diffuser. Structure-borne noise is an audible sound that is structural in origin, i.e. it begins as vibration. Basically, each substructure of a suction unit (stator, rotor, yokes A and B, shield, blower cowl, etc., see Fig. 1 in Part I of the paper) has one of the resonance M5

Mic M1 M3

M2 M4

M1 M3

1m

M5

Mic

Fig. 1. Vibration measurement points: M1–M5 measurement positions, Mic-microphone position.

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

523

frequencies and can affect the overall vibration behaviour. The normal mode of vibration of each part of the suction unit depends on its mechanical properties. For the electric motor stator’s packet, the mechanical properties depend on the magnetostriction and attraction forces, on the number of teeth and slots, on material and windings, on fastening of the stator packet in the housing, etc. This is similar with modes of vibration of the metal shield around the stator, if any, and the blower cowl. However, the spectra indicate not only the resonance natural frequencies of the individual parts but also the frequencies introduced by the assembly process. The coupling process causes modification of the subsystem modes. There is also a superposition of discrete frequency tones due to the excited forces and due to the normal modes of vibration [3,4]. In this case, it is very hard to associate the obtained real resonance frequencies with the individual vibration mode, especially with such a small complex noise source as the suction unit. Additional difficulties appear when more discrete frequency tones are closely spaced in the frequency spectrum and when they have response curve as an overlap in form of broadband noise, which is also a result of chosen frequency resolution [5]. The structural resonances or natural frequencies (non-rotational noise) can easily be distinguished from the harmonic-order-related frequencies (rotational noise) within the spectra by continuously changing the voltage supply or rotational speed and observing the local vibration maxims. When a discrete frequency (or broadband group of frequencies) is not changing with changing the rotational speed, then this is a constant natural or resonant frequency (mode of vibration) as part of structure-borne noise, which theoretically does not depend on the excited forces. The inverse is also true: if a discrete frequency is changing with rotating speed, then this is a different harmonic of rotation and is not a result of structural resonances [1,2]. On this way, it is possible to determine the resonant phenomena or mechanical resonance frequencies and internal damping conditions of the assembled suction unit. Since the mechanically and electromagnetically generated noise is mostly in the form of discrete frequencies tones (rotational and non-rotational), the broadband turbulent noise, which is changing in frequency and magnitude when changing the rotational speed or flow velocity, is a form of airborne noise. In this paper, the contribution of the structure-borne noise to the total SPL is analysed at different geometries of the suction units observed, i.e., with and without a vaned diffuser, and at different operating conditions. 2. Test procedure Two different types of suction units were used in this part of the experiments: those having blowers without vaned diffusers and the others having blowers with vaned diffusers. Vibration and noise characteristics of the suction units were measured simultaneously on a special test stand placed in an anechoic chamber, which was described in Part II of the paper. Vibrations were measured at five different positions; among them three were in axial and two in radial directions (see measurement points M1–M5 in Fig. 1). Vibrations were measured by an accelerometer, B&K, Type 4371, mounted on different parts of the suction unit, such as blower cowl and stator of the electric motor or its shield, if any (Fig. 1 above). The SPL was measured by a half-inch microphone, B&K, Type 4155, placed at a distance of 1 m, in the vicinity of the suction unit and perpendicular to the brushes axis (Fig. 1 below). Both measurement results have been analysed by a two-channel B&K FFT analyser, Type 2032.

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

524

3. Experimental results and discussion 3.1. Vibration characteristics of suction units Vibration characteristics vary from unit to unit and depend on their structural design and modes of vibration, as well as on the operating conditions. In Fig. 2a, measurement results at a constant voltage supply (230 V) for suction unit No. 1 (without vaned diffuser) and in Fig. 2b the same for suction unit No. 8 (with a vaned diffuser and metal shield) are presented. For characteristics of the suction units observed, see Table 1 in Part II of the paper. From Fig. 2, we can see that vibration velocities vary by changing operation conditions and position of the measurement point. The minimum values are at the design point of operation and maximum at a zero flow rate and surge point. From Fig. 2, we can also see that vibrations in the axial direction (measurement points M3, M4 and M5) are higher than those in radial direction (measurement points M1 and M2). The maximum values are a result of the effect of the surge phenomena at a partial flow rate and fluctuating of the air flow within the blower causing intensive vibration of the suction unit as whole; for more about surge see Parts I and II of the paper. Higher vibration velocities in the axial direction are thus in contrast with the findings of Wang and Lai [3,4], who measured vibrations just on the stator of the electric motor, not being coupled with another machine, and who

a

120

Vibration mm/s

100 80 M3

60

M4

40

M5

20

M2 M1

0 0

10

20

30 Orifice mm

40

50

60

b 120 Vibration mm/s

100 80 60

M4

40

M3

20

M5

M2 M1

0 0

10

20

30 Orifice mm

40

50

60

Fig. 2. Vibration velocity of suction unit at constant voltage supply (230 V) in measurement points M1–M5 at different operating conditions: (a) for suction unit without vaned diffuser (No. 1) and (b) for suction unit with a vaned diffuser (No. 8).

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

525

Vibration velocity [mm/s]

Orifice φ 50 mm 15 10

BPF

5 RF

SB5008

20

SB2000

Vibration velocity [mm/s]

founded that the main vibrations are excited on the electric motor stator in a radial direction. A comparison between Fig. 2a and b shows that vibration velocities measured in the axial direction (measurement points M3 and M4 on yoke A) are much higher (up to two times) on suction unit No. 8 than those on suction unit No. 1. A reason for could be found in the presence of the vaned diffuser built-in the blower and the metal shield around the stator of the electric motor, made in one piece with the yoke A, with the suction unit No. 8. The presence of the vaned diffuser causes higher vibration of the suction unit, especially at partial flow rates when the surge phenomenon appears. However, vibrations measured in the axial direction are not reflected in the emitted noise, because the noise is measured by a microphone placed in the radial direction, at an angle of 90°, i.e., perpendicular to the axial direction, where axial vibrations have almost no effect on the emitted noise. The metal shield around the stator of the electric motor including the yoke A thus has low rigidity and obviously needs serious improvement. The spectra of velocity vibrations measured at the constant voltage supply (230 V) and in the axial direction, on measurement point M3, and at three different orifices (/50 mm – at the free delivery, /19 or /16 mm – at the BEP and /0 mm – at a zero flow rate) are presented in Fig. 3 (left) for suction unit No. 1 and in Fig. 3 (right) for suction unit

0 0

2000

4000

6000

8000

Orifice φ 50 mm 40 30 20

BPF

10

2BPF

RF

0 0

10000 12000 14000

2000

4000

50

Vibration velocity [mm/s]

Orifice φ 19 mm

20 15 BPF

10 RF

5

SB

SB5152

Vibration velocity [mm/s]

25

22nd

0 0

2000

4000

6000

8000

Orifice φ 16 mm BPF

20

2BPF RF

10 0 0

Vibration velocity [mm/s]

Vibration velocity [mm/s]

15 2BPF

SB

SB

5 0 0

2000

4000

6000

8000

Frequency [Hz]

12000 14000

2000

4000

10000

6000

8000

10000

12000

14000

Frequency [Hz]

20 RF

10

10000

30

10000 12000 14000

Orifice φ 0 mm

BPF

8000

40

Frequency [Hz] 25

6000

Frequency [Hz]

Frequency [Hz]

12000

14000

50

Orifice φ 0 mm

RF 40

BPF 30 20 10

2BPF

0 0

2000

4000

6000

8000

10000

12000

14000

Frequency [Hz]

Fig. 3. Spectra of vibration velocities at constant voltage supply (230 V) in measurement point M3 at orifices / 50 mm, /19 or /16 mm and /0 mm: for suction unit without vaned diffuser (No. 1) (left) and for suction unit with a vaned diffuser (No. 8) (right); RF – peaks at the rotational frequency, BPF – peaks at the blade passage frequency, SB – structure-borne peaks at the corresponding frequency.

526

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

No. 8. From the spectra in Fig. 3, we can see that main velocity vibrations appear at the rotational frequency (RF) and blade passage frequency (BPF), and their higher harmonics (with exception of the suction unit No. 1 at higher flow rates), and that the magnitude of vibrations are much higher at the suction unit No. 8 than those at the suction unit No. 1. The differences are from multiples of 2–6 and follow the same trends as in Fig. 2. The vibration velocity peaks at the RF are a result of unbalanced rotating masses and mechanical friction, and at the BPF are the results of interaction of the rotor blades and diffuser vanes within the blower, if any, or backflow vanes. In Fig. 3 (left) we can see two groups of less pronounced discrete frequency peaks: one is between the RF and BPF and the other is above BPF. They are a result of structural vibrations (modes), i.e., part of the structural borne (SB) noise (see peaks SB2000 and SB5008) and are non-rotational in nature. Similar peaks are also partially seen in Fig. 3 (right) although less noticeable due to the two times smaller scale. Both groups of vibration peaks, rotational as well as non-rotational, change with operating conditions, but those rotational, at the RF and BPF and their higher harmonics, change in frequency and magnitude, whereas those (two smaller groups of peaks), which are non-rotational and structure-borne in origin, at 2 and 5 kHz, change in magnitude only. From Fig. 3, we can also see that discrete vibration velocities, rotational as well as non-rotational, increase with the load of the blower, i.e., towards a zero flow rate when the rotating stalls and surge appear. 3.2. Contribution of noise of the electric motor to the total SPL The driving electric motor causes mainly structure-borne noise and only a small part of aerodynamically generated noise. The contribution of noise of the electric motor to the total SPL, depending otherwise on its design and operating conditions, was determined by a separate noise measurement of the complete suction unit and the same suction unit with a dismantled rotor of the blower and its cowl. The noise of the suction unit without a rotor of the blower and its cowl practically represents the noise generated by the electric motor alone. This includes all sources of noise generated by the electric motor, airborne (due to rotation of the electric motor rotor), electromagnetically generated and mechanically generated noise. Noise characteristics of the electric motor were measured on a special test stand, which otherwise is used for measuring the electric motor efficiency. The test stand enables reestablishing the same operating conditions (speed and load) the electric motor is expected to encounter in service (with mounted rotor of the blower). To exclude noise generated by the test stand, it was wrapped and covered with an acoustical shield. The mounted rotor of the blower also has an effect on excited forces caused by the electric motor alone, especially at partial flow rates when the rotating stalls and surge appear. These additional forces are not possible to simulate on the electric motor alone but this is also not necessary, because the noise generated by these forces can mainly be assigned to the blower. In Fig. 4, noise spectra of the suction unit No. 1 (without vaned diffuser) for the complete suction unit (upper spectra) and without the rotor of the blower are depicted , i.e. with the electric motor alone (lower spectra) at rotational speeds corresponding to the constant voltage supply (230 V) and three characteristic orifice diameters: /0 mm (zero flow rate), /19 mm (BEP) and /50 mm (free delivery). We can see that the noise generated by the electric motor depends on the operating point, i.e., on the load and rotational

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

(a) φ0 mm

80 SPL [dB(A)]

527

BPF

SPLtotSU = 88 dB(A)

70

22nd

60 50 40 SPLtotEM=75 dB(A) 30 0

2

4

6 8 Frequency [kHz]

10

12

(b) φ19 mm SPL [dB(A)]

80

BPF

SPLtotSU = 82 dB(A)

70

22nd

60 50 40 SPLtotEM = 71 dB(A)

30 0

2

6 8 Frequency [kHz]

10

12

(c) φ50 mm

80 SPL [dB(A)]

4

BPF

SPLtotSU = 87 dB(A)

70

22nd

2 3

60 1

4

50 40 SPLtotEM = 70 dB(A)

30 0

2

4

6 8 Frequency [kHz]

10

12

Fig. 4. Noise spectra of complete suction unit without vaned diffuser (No. 1) (thick curve) and with dismantled rotor of the blower (thin curve) at constant voltage supply (230 V) and different flow regimes: (a) at zero flow rate (orifices /0 mm), (b) at BEP (orifice /19 mm) and (c) at free delivery (orifice /50 mm); SPLtotSU – total sound pressure level of the suction unit, SPLtotEM – total sound pressure level of the electric motor.

speed. The noise spectra of the electric motor alone are characteristically narrow-band (lined spectra) with many discrete frequency tones, which are in the form of a harmonic series with the RF and are thus rotational in origin. Among them, the first four harmonics of the RF and commutator brush frequency (CBF) are especially pronounced. These tones are excited by a harmonically-acting vibration caused by unbalanced rotating masses, friction in commutator brushes and bearings. In addition, the noise spectra of the electric motor also contain discrete frequency tones of non-rotational origin, which are caused by modes of vibration (mostly of the stator packet of the electric motor, and metal shield if any). The electric motor also generates a broadband noise but usually at a much lower level than the discrete frequency tones . Differences between the discrete frequency tones and the broadband noise are up to 15 dB(A) and more (Fig. 4). When the rotor of the blower is mounted on the electric motor shaft, additional discrete frequency tones at the BPF and its higher harmonics are generated. In addition, a high

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

528

level of broadband noise is generated. Fig. 4 shows that the total SPL of the complete suction unit with a running blower is much higher than those generated by the electric motor alone; differences are up to 13 dB(A) at a orifice of /0 mm, up to 11 dB(A) at an orifice of /19 mm, and up to 17 dB(A) at an orifice of /50 mm. Similar results were obtained by Brungart et al. [6]. These differences are so great that we can say that the effect of the electric motor to the total SPL is rather small. However, detailed analyses of noise and vibration spectra measured on the complete suction unit have shown that their effect to the total SPL cannot be entirely neglected. For this reason, simultaneous measurement of noise and vibrations were performed on the same suction unit. The vibrations were measured by an accelerometer fastened to the stator of the electric motor (in a radial direction, measurement point M1 in Fig. 1), while the noise was measured by a microphone placed at a distance of 1 m from the suction unit. Fig. 5 shows

(a) φ0 mm

140

n= 600 rps

SPLtot = 87.2 dB(A)

80 70

120

60

110

50

100 2

4

10

12

(b) φ19 mm

90

SPL [dB(A)]

6 8 Frequency [kHz]

n=475 rps, SPLtot=81.5 dB(A) Latot = 143.6 dB 22nd

80

140 130

BPF 70

120

60

110

50

La [dB]

0

100 0

2

4

6 8 Frequency [kHz]

10

12

(c) φ50 mm

90

SPL [dB(A)]

130

Latot = 147.4 dB

La [dB]

BPF

140

n = 450 rps, SPLtot = 87.9 dB(A) Latot = 141.6 dB

80

130

22nd

BPF 70

120

60

110

50 0

2

4

6

8

10

12

La [dB]

SPL [dB(A)]

90

100

Frequency [kHz]

Fig. 5. Comparison between noise spectra (thick curve) and vibration spectra (thin curve) of suction unit No. 1 at constant voltage supply (230 V) and different flow regimes: (a) at zero flow rate (orifices /0 mm), (b) at BEP (orifice /19 mm) and (c) at free delivery (orifice /50 mm); n – rotational speed, BPF – blade passage frequency, SPLtot – total sound pressure level, Latot – total level of vibration velocity.

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

529

simultaneously measured noise spectra (thick curves) and vibration spectra (thin curves) for suction unit No. 1 at constant voltage supply (230 V) and three different orifices (/0 mm, /19 mm and /50 mm) for comparison. We can see typically narrow-band vibration spectra with many pronounced discrete frequency tones within entire frequency range. Among them, the peaks corresponding to the RF, BPF and CBF, and their higher harmonics are the most pronounced. Additionally, a broadband peak between 4 and 6 kHz is especially pronounced. All peaks previously listed are a result of the common action of the mechanically, electromagnetically and aerodynamically excited forces [4,7,8]. Among these of mechanical origin the most important are those excited by the unbalanced rotating masses of the rotors of the electric motor and the blower, as well as bearings and commutator-brush friction. Among these of aerodynamic origin the most important are those excited by the BPF, rotating stall and surge phenomenon. The level of these vibrations depends, amongst other factors, on the modes of vibration. The modes of vibration depend on structural design of the suction unit, or form, rigidity, thickness, mass and material used. The highest values of vibrations appear again at the partial flow rate, when the rotating stalls and surge phenomena appear, see Figs. 5 and 3. The pronounced peak in vibration spectra, between 4 and 6 kHz (Fig. 5), and pronounced discrete frequency tones within this peak are not in correlation with the RF, therefore they are non-rotational structure-borne in origin due to resonances. These peaks are not seen in noise spectra of the electric motor alone (suction unit without the rotor of the blower) in Fig. 4 (thin curves). This means that they are a result of excited forces caused by the processes within the blower, which are manifesting through the suction unit (electric motor) structure in form of vibrations of the structure. A cursory glance at the spectra in Fig. 5a–c shows that most of the peaks in the vibration spectra do not coincide with the peaks of the noise spectra. However, a detailed analysis of the peaks in vibration spectra at the low frequency below BPF has shown that these peaks fully coincide with the corresponding peaks in noise spectra for the electric motor alone (Fig. 4), although there are some new peaks caused by the operation of the blower. These peaks are not seen in the noise spectra of the complete suction unit (with rotor of the blower) because they are mostly masked by broadband noise. The broadband turbulent noise, which is mostly of aerodynamic origin caused within the blower and electric motor, and the jets flow due to the exiting air flow on the electric motor side, more or less overlaps the discrete frequency tones of the rotational as well as non-rotational noise until it fully dominates, especially at the free delivery (Fig. 4c). As a result of this, the discrete peaks in noise spectra (thick curves in Figs. 4 and 5) are mostly masked by the broadband noise, but they are evidently reflected in pronounced broadband noise peaks in corresponding frequency region. The measuring instrument makes an envelope from the more closely spaced peaks; it is also determined by the selected frequency resolution. 3.3. Structure-borne noise at different operating conditions Fig. 6 shows the noise spectra of suction unit No. 1 (without vaned diffuser) at three different rotational speeds (196, 296 and 460 rps), and 10 different loads (orifices /0, /6.5, /10, and through the series to /50 mm) as a parameter, and Fig. 7 shows noise spectra of the same suction unit No. 1 at three different loads (orifices /0, /19 and /50 mm) and four different rotational speeds (200, 300, 375 and 450 rps) as parameter for comparison.

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

530 80

n=196 rps

70 3820 Hz

2085 Hz

10

60

914 Hz

SPL [dB]

BPF

50

2BPF 3BPF

50 30

40 515 Hz

40 30

0

20 0

2000

55th

44th

22nd

10

4000

6000

8000

10000

12000

Frequency Hz

914 Hz

S P L [dB]

70

10 BPF

60

3609 Hz

80

n=296 rps

22nd

37th 50

40

30

50 40

0

30

10 19

20 0

2000

4000

6000

8000

10000

12000

80

BPF

2BPF

22nd

60 50 4400 Hz

SPL [dB]

70

2085 Hz

914 Hz 1335 Hz

Frequency [Hz]

40

0

10

30

n=460 rps

20 0

2000

4000

6000

8000

10000

12000

Frequency [Hz]

Fig. 6. Noise spectra of suction unit without vaned diffuser (No. 1) at three rotational speeds (n = 196, 296 and 420 rps) and 10 different flow regimes (orifice diameter /0, /6.5, /10, . . . to /50 mm).

Similarly, Fig. 8 shows noise spectra of suction unit No. 5 (with a vaned diffuser) at three different rotational speeds (205, 285 and 450 rps), and ten different loads (orifices /0, /6.5, /10, and through the series to /50 mm) as parameter, and Fig. 9 shows noise spectra of the same suction unit No. 5 at three different loads (orifices /0, /16 and / 50 mm) and four different rotational speeds (200, 285, 350 and 420 rps) as parameter

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

375 rps

SPL [dB]

60

531

φ 0mm

450 rps 22nd

22nd

50 40

BPF

30

22nd

200 rps 300 rps

20 0

2000

4000

6000

8000

10000

12000

Frequency [Hz] 70

BPF

φ19 mm

450 rps

SPL [dB]

60

22nd

375 rps

50 40

300 rps

30

200 rps 20

0

2000

4000

6000

8000

10000

12000

Frequency [Hz] 70

BPF

375 rps 450 rps

22nd

SPL [dB]

60 50 BPF

40 200 rps

22nd 300 rps

30

44th

φ50 mm

20 0

2000

4000

6000

8000

10000

12000

Frequency [Hz]

Fig. 7. Noise spectra of suction unit without vaned diffuser (No. 1) at four rotational speeds (n = 200, 300, 375 and 450 rps) and three different flow regimes: (a) at zero flow rate (orifice /0 mm), (b) at BEP (orifice /19 mm) and (c) at free delivery (orifice /50 mm).

for comparison. Similar spectra are of the all others suction units observed. Figs. 7a and 9a show noise spectra at an orifice of /0 mm (similar noise spectra are obtained at orifices of /6.5 mm and /10 mm), Figs. 7b and 9b show spectra for orifices of /19 mm and /16 mm, respectively (similar noise spectra are at orifices of /16 or /19 mm and /13 mm), and Figs. 7c and 9c show the same for an orifice of /50 mm (similar spectra are at orifices of /40 and /30 mm). From Figs. 6–9 we can see that all spectra consist of broadband noise with pronounced discrete frequency tones. The discrete frequency tones can be structure-borne or airborne,

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

532 80

2BPF

4BPF

70

SPL [dB]

3BPF

944 Hz BPF

60

5BPF

6BPF

50 40 30

n=205 rps 20 0

2000

4000

6000

8000

10000

12000

Frequency [Hz] 80

SPL [dB]

3BPF

2BPF

70

4BPF

944 Hz BPF

60 50 40 30

n=285 rps 20 0

2000

4000

6000

8000

10000

12000

Frequency [Hz] 80

864 Hz

SPL [dB]

70

BPF

2BPF

3BPF

60 50 40 30

n=420 rps 20 0

2000

4000

6000

8000

10000

12000

Frequency [Hz]

Fig. 8. Noise spectra of suction unit with a vaned diffuser (No. 5) at three rotational speeds (n = 205, 285 and 420 rps) and ten different flow regimes (orifice diameter /0, /6.5, /10, . . . to /50 mm); BPF – blade passage frequency.

rotational and non-rotational in origin. Figs. 6–9 show pronounced discrete frequencies tones at the BPF and its higher harmonics, and two less-pronounced broadband peaks at the low frequency band, one between 500 and 2500 Hz, and the other between 3000 and 4500 Hz. These two pronounced broadband peaks, having more pronounced discrete

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

533

90 BPF420

80 2BPF

960 Hz

350 rps

SPL [dB]

70

420 rps

60 50 40 200 rps

30

285 rps

φ0 mm

20 0

2000

4000

6000

8000

10000

12000

Frequency [Hz] 90 2BPF

80

3BPF

SPL [dB]

70 420 rps

BPF

60 50 40 30 350 rps

20 0

2000

200 rps

φ16 mm

285 rps

4000

6000

8000

10000

12000

Frequency [Hz] 90 80

2BPF 350 rps

SPL [d B]

70

420 rps

60 50 40 200 rps

285 rps

30

φ50 mm

20 0

2000

4000

6000

8000

10000

12000

Frequency [Hz]

Fig. 9. Noise spectra of suction unit with a vaned diffuser (No. 5) at four rotational speeds (n = 200, 285, 350 and 420 rps) and three different flow regimes: (a) at zero flow rate (orifice /0 mm), (b) at BEP (orifice /16 mm) and (c) at free delivery (orifice /50 mm); BPF – blade passage frequency.

frequency peaks, are structure-borne in origin due to structural resonances. The measurement results have shown that, at the constant voltage supply (230 V) and suction units without vaned diffuser, the pronounced peaks of structural resonances appear at discrete frequencies 1232, 2800 and 4528 Hz for suction unit No. 1, at 640, 1550 and 2816 Hz for suction unit No. 2, at 704, 1248, 1664, 1920 and 4300 Hz for suction unit No. 3, and at 960, 1408, 2240 and 4624 Hz for suction unit No. 4. For suction units with a vaned diffuser, the pronounced peaks appear at discrete frequencies 515, 600, 867, 960, 1171,

534

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

1875, 2367 and 2460 Hz within the first broadband peak, and at 4289, 5484 and 6890 Hz within the second broadband peak. Similar results have been obtained by Braungart et al. [6] who performed measurements on a vacuum cleaner. Although, in principle, the frequency of the structural resonances does not depend on operating conditions (speed and load), detailed analyses have shown that frequencies of the pronounced peaks listed above are changed by modifying the operating conditions by approximately 1–3%, [5,10]. This is partially a result of the chosen frequency resolution and partially the effect of coupled frequency, or combined action of different exciting forces (mechanical, electromagnetic and aerodynamic in origin), which are changing by modifying the operating conditions. The levels of these peaks are changing much less than the levels of the high frequency part of spectra, being mostly of aerodynamic in origin. We must distinguish between contribution of the structure-borne noise to the total SPL at the suction unit without and with a vaned diffuser. For the suction unit without a vaned diffuser, the structure-borne noise at the low frequency range is prevailing at all rotational speeds when the flow rate is small enough, whereas at higher flow rates and sufficiently high rotational speeds, the broadband turbulent noise of aerodynamic origin prevails within the entire frequency range. This is a result of the fact that the magnitude of the aerodynamically generated non-rotational broadband (turbulent) noise increases with rotational speed faster than the structure-borne noise, as was stated in Part II of the paper. The increase in flow rate causes an increase in aerodynamic noise due to higher flow velocity and greater effect of the laminar and turbulent boundary layer vortex shedding, as well as due to the stronger effect of the jet flows from the electric motor openings. Turbulent noise of aerodynamic origin generated in this way increases steadily with flow rate and gradually overlaps the structure-borne non-rotational noise (pronounced at the low frequencies) as well as the rotational noise at the RF and BPF, and their higher harmonics. At higher rotational speeds (above 350 rps) and for higher flow rates (above an orifice diameter of /40 mm), the broadband turbulent noise of aerodynamic origin prevails within the entire frequency range and overlaps the structure-borne noise at the low frequency range and almost overlaps even all rotational noise at the BPF and its higher harmonics (Figs. 6c and 7c). This means that the total SPL is determined mostly by the structure-borne noise at lower flow rates and lower rotational speeds, as well as by the aerodynamically generated noise at higher flow rates and higher rotational speeds. For the suction units with a blower having a vaned diffuser, the structure-borne noise is of the same order of magnitude as those for the suction units without a vaned diffuser (compare Fig. 8 for suction unit No. 5 and Fig. 6 for suction unit No. 1). However, due to the much higher level of rotational noise at the BPF and its higher harmonics (differences are up to 15 and 20 dB), its effect to the total SPL is much smaller. For the BEP contribution of the structure-borne noise to the total SPL is almost negligibly small, as was stated in Part II of the paper. In Fig. 6, one more pronounced peak at a constant frequency is seen, at approximately 10.8 kHz. At the rotational speed of 196 rps, it corresponds to the 55th harmonic of the RF (Fig. 6a), at 296 rps it corresponds to the 37th harmonic of the RF (Fig. 6b), and at 460 rps it corresponds to the 23th harmonic of the RF (in Fig. 6c it is masked by broadband turbulent noise). According to Suh et al. [9], this peak could be assigned to the vibration modes of the stator caused by electromagnetic origin. For actual rotational speeds of the suction unit in vacuum cleaner applications, with rotational speeds above 450 rps, it is masked by broadband turbulent noise and thus not important.

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

535

3.4. Effect of rotating stalls and surge on structure-borne noise In the majority of the suction units observed, a surge phenomenon appears at a partial flow rate. The surge phenomenon is connected with a backflow in the suction unit that causes a fluctuation of the airflow and unstable performance characteristics. The surge is repeated periodically in a frequency between 5 and 10 times per second, as was stated in Part II of the paper. In addition, the surge causes a steep increase in vibrations of the suction unit as a whole, and so also of the electric motor stator (or metal shield, if any) and blower cowl (see Figs. 2 and 3), as well as the emission of additional structure-borne noise. Thus, the generated structure-borne noise appears in the form of pronounced discrete or broadband peaks, which are in connection with the modes of vibration. The modes of vibration depend, first of all, on the stator of the electric motor (or metal shield if any) and on the stator of the blower (with or without a vaned diffuser) and its cowl, and their fastening to the yokes A and B. The peaks caused by rotating stalls increase the levels of the structure-borne noise at partial flow rates and reach maximum values at the surge point, then start falling again towards a zero flow rate. For the suction units without a vaned diffuser, the maximum values of these peaks can even exceed the magnitude of the BPF (Fig. 10a). The structure-

a

0.12

Sound pressure [Pa]

n = 440 rps 0.10

867Hz 0.08

1828 Hz

BPF

0.06 0.04

2BPF 0.02

22nd

0.00 0

2000

4000

6000

8000

10000

12000

Frequency [Hz]

b

0.45

2BPF

Sound pressure [Pa]

0.40 0.35

n=420 rps

BPF

0.30 0.25

3BPF

0.20 0.15 0.10 0.05 0.00 0

2000

4000

6000

8000

10000

12000

Frequency [Hz]

Fig. 10. Effect of surge phenomenon on sound pressure at different flow regimes (orifices diameter */0, /6.5, /10, . . . to /50 mm): (a) for suction unit without vaned diffuser (No. 3) and rotational speed 440 rps, (b) for suction unit with a vaned diffuser (No. 5) and rotational speed 420 rps; BPF – blade passage frequency.

536

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

borne noise at the surge point prevails within the spectra at all flow rates when the rotational speeds are low enough. For the suction unit with a vaned diffuser, the structure-borne noise due to the surge is still the same order of magnitude as those at the suction unit without vaned diffuser (in both cases the sound pressures are approximately 0.05 Pa, compare Fig. 10a and b), but its effect on the total SPL is much smaller due to the much higher level of the rotational noise at the BPF and its higher harmonics (Fig. 10b). To reduce the structure-borne noise, vibrations of the suction unit structure have to be reduced. To reduce vibrations of the suction unit, the stiffening of its structure (especially outer surfaces) should be increased, e.g. by bending or ribbing and tuning with regard to the most pronounced discrete frequency or group of pronounced frequencies. A metal shield around stator of the electric motor is also recommended, but it must be properly fastened and manufactured in a proper form with proper position, form and dimensions of air openings. It can also be lined by a layer of an elastomer with good acoustical properties with regard to the excitation frequency, which we want to attenuate. An increase of the outlet blade angle of the blower leads to the surge phenomenon and to an increase of the structure-borne noise. Therefore, to reduce the structure-borne noise due to the surge phenomena, a smaller rotor outlet blade angle of the blower must be made. Finally, to reduce structure-borne noise, the vaned diffuser must be omitted (see also Part II of the paper). 4. Conclusions The emitted noise of a suction unit is generated partially by the turbo blower and partially by the driving electric motor. The blower is the main source of aerodynamically generated noise, whereas the electric motor is the source of mechanical and electromagnetic noise origin causing most of the structure-borne noise. The structure-borne noise depends on the suction unit geometry; first of all, on the stator of the electric motor, or the metal shield if any, and the stator of the blower, or built-in vaned diffuser, and on operating conditions (speed and load), primarily on appearing of the rotating stalls and surge. For the suction unit having a blower without vaned diffuser, the structure-borne noise is pronounced at the low frequency range and usually determines the total SPL at partial flow rates, whereas at higher flow rates the aerodynamically generated broadband (turbulent) noise increases so much that it dominates the structure-borne noise. The aerodynamically generated broadband (turbulent) noise at the free delivery also dominates the rotational noise of aerodynamic origin at the BPF and its higher harmonics, and prevails within the entire spectrum. For the suction unit having a blower with built-in a vaned diffuser, the structure-borne noise, although being of the same order of magnitude as those at the suction unit without a vaned diffuser, is of lesser importance due to the much higher rotational noise at the BPF and its higher harmonics. To reduce the structure-borne noise, the stiffening of the suction unit structure (especially outer surfaces) should be increased. A metal shield around the stator of the electric motor is recommended, but it must be manufactured in proper form and stiffness (by bending or ribbing), and also in proper position, form and dimensions of the air openings. To reduce the structure-borne noise due to the surge phenomena, a smaller rotor outlet blade angle of the blower must be made.

M. Cˇudina, J. Prezelj / Applied Acoustics 68 (2007) 521–537

537

Acknowledgement The authors gratefully acknowledge the support of the Slovenian Ministry of Education, Science and Sport. References [1] Gargano E, Bartolini A. Acoustic evaluation of hydraulic power unit by spectrograms. In: Proceedings of the sixth international congress of sound and vibration, 5–8 July 1999, Copenhagen, Denmark. p. 2915–20. [2] Dimarogonas A. Vibration for engineers. second ed. NJ: Simon& Schuster, Prentice-Hall; 1996. [3] Lai JCS, Wang C. Prediction of noise radiation from induction motors. In: Proceedings of the sixth international congress on sound and vibration, Copenhagen, Denmark; 5–8 July 1999. p. 2449–56. [4] Wang C, Lai JCS. Vibration analysis of an induction motor. J Sound Vib 1999;224(4):733–56. ˇ udina M, Prezelj J, Rejec J. Noise generation by vacuum cleaner suction units. In: C ˇ udina M, Portorozˇ, [5] C editors. Proceedings of the first congress of Alps Adria acoustics association and third congress of Slovenian Acoustical Society, Slovenia; September 1–2, 2003. p. 155–64. [6] Brungart TA, Lauchle GC, Ramanujam RK. Installation effects on fan acoustic and aerodynamic performance. NCEJ 1999;47(1):3–7. [7] Timar PL. Noise and vibration of electrical machines. Budapest: Akademiai Kiado; 1989. [8] Paljan D. Some problems in the calculation and measurement of magnetic noise on cage-type induction motors. Elektrotehnika 1975;8(1):23–8. [9] Suh SJ, Chung J, Lim BD, Hwang CH. Case history: noise source identification of an automobile alternator by RPM dependent noise and vibration spectrum analysis. NCEJ 1991;37(1):31–6. ˇ udina M, Prezelj J. Noise generation by rotating stall and surge in a vacuum cleaner suction unit. Fan [10] C Noise 2003. In: Proceedings of the second international symposium, an international INCE symposium, 23– 25 September 2003. Senlis (France): CETIM, Technical centre for mechanical industries; CETIAT; 2003, p. 1–8.