Available online at www.sciencedirect.

com
Available online at www.sciencedirect.com
ScienceDirect
ScienceDirect
Procedia
Available Manufacturing00
online (2018) 540–545
atatwww.sciencedirect.com
Available
Procediaonline www.sciencedirect.com
Manufacturing00 (2018) 540–545 www.elsevier.com/locate/procedia
www.elsevier.com/locate/procedia
ScienceDirect
ScienceDirect 
Procedia Manufacturing 22 (2018) 533–538
Procedia Manufacturing 00 (2017) 000–000
www.elsevier.com/locate/procedia
11th International Conference Interdisciplinarity in Engineering, INTER-ENG 2017, 5-6 October
11th International Conference Interdisciplinarity in Engineering,
2017, Tirgu-Mures, Romania INTER-ENG 2017, 5-6 October
2017, Tirgu-Mures, Romania
Studying noise measurement and analysis
StudyingSociety
Manufacturing Engineering noiseInternational
measurement and
Conference analysis
2017, MESIC 2017, 28-30 June
2017, Ágoston
Vigo (Pontevedra),
a, Spain
Katalin *
Ágoston Katalina, *
P P0F

P P0F

Costing models for capacity optimization in Industry 4.0: Trade-off

a
PPetru Maior University of Tg.Mures, Tg.Mures, Romania
P

a
PPetru Maior University of Tg.Mures, Tg.Mures, Romania
P

between used capacity and operational efficiency

Abstract
Abstract
A. Santanaa, P. Afonsoa,*, A. Zaninb, R. Wernkeb
This paper presents some aspects regarding to sound characteristics, how the sound pressure level can be expressed, and how the
This
soundpaper presents
analysis can be someusedaspects regarding
for fault detection.
a to sound
Sound
University of characteristics,
level meter
Minho, how
is used
4800-058 the
for sound
Guimarães,sound pressure level can
measurement,
Portugal whichbe provides
expressed,measurements
and how the
sound
of soundanalysis canlevels
pressure be usedandfor fault detection.
displays b Sound
of dB.level
Unochapecó,
it in units meter
soundisChapecó,
89809-000
The usedmeter
level forSC,sound
Brazil
and measurement,
different which
types of usedprovides
transducersmeasurements
are briefly
of sound pressure
presented. levelsenvironment
In industrial and displaysnoise
it in units of dB. The
measurements aresound
madelevel meter
because and different
useful typesbe
sound cannot of separated
used transducers
from other are noises.
briefly
presented. In industrial
Motors’ working environment
is followed noise measurements
by vibrations are made
and oscillations whichbecause
are in useful
audible sound cannot domain
frequency be separated
so anyfrom other noises.
changes of the
Motors’ working
environmental noiseis can
followed
lead tobyuseful
vibrations and oscillations
information. which the
If a fault occurs, are produced
in audible frequency
vibration domain
sound overlapsso any
with changes
the base of the
sound
Abstract
environmental
signal noise can
and modulation lead to In
appears. useful
case information. If a fault
of different faults occurs,and
frequency the amplitude
produced vibration sound
modulations overlaps
appear. Thewith
latterthe
canbase
be sound
easily
signal
detectedanddue
modulation appears.frequencies
to the sideband In case of in different faults frequency
the frequency spectrum.and amplitude
Using Matlabmodulations
functions, theappear.
PowerThe latter can
Spectral be easily
Density of a
Under
detected the
signal can beconcept
due to ofKnowing
the sideband
obtained. "Industry 4.0",in production
frequencies
the specific the frequencyprocesses
frequencies spectrum.
due will
UsingbeMatlab
to the rotating pushed
element andtoanalyzing
be the
functions, increasingly
Power
the PSD ofinterconnected,
Spectral Densitynoise
measured of a
information
signal
faults can be based
obtained.on a real
Knowing timethebasis and,
specific necessarily,
frequencies due much
to the more
rotating efficient.
element In
and this context,
analyzing
localized. If sideband frequency around the fundamental frequencies is shown, then it has the meaning of an the capacity
PSD of optimization
measured noise
goes
faultsbeyond the traditional
can bepresence.
abnormality localized. aim of frequency
If sideband capacity maximization, contributing
around the fundamental also for isorganization’s
frequencies shown, then itprofitability
has the meaning and value.
of an
abnormality
Indeed,
© 2018 The leanpresence.
management
Authors. Publishedand continuous
by Elsevier B.V. improvement approaches suggest capacity optimization instead of
© 2018 The under
Peer-review
maximization. Authors.
The Published
responsibility
study by Elsevier
ofthe
of capacity B.V.committee
scientific
optimization andof costing
the 11th models
International
is anConference
importantInterdisciplinarity
research topic that in deserves
© 2018 The Authors.
Peer-review under Published
responsibility by Elsevier
ofthe B.V.committee
scientific of the 11th International Conference Interdisciplinarity in
Engineering.
contributions from both the practical and theoretical perspectives. This paper presents and discusses
Peer-review under responsibility of the scientific committee of the 11th International Conference Interdisciplinarity in Engineering. a mathematical
Engineering.
model for capacity management based on different costing models (ABC and TDABC). A generic model has been
Keywords: noise; sound; measurement; frequency spectrum; fault detection.
developed and it was used to analyze idle capacity and to design strategies towards the maximization of organization’s
Keywords: noise; sound; measurement; frequency spectrum; fault detection.
value. The trade-off capacity maximization vs operational efficiency is highlighted and it is shown that capacity
optimization might hide operational inefficiency.
1. Introduction
©
1. 2017 The Authors. Published by Elsevier B.V.
Introduction
Peer-review under responsibility of the scientific committee of the Manufacturing Engineering Society International Conference
Sound is a common part of everyday life. The sounds can be also unpleasant or unwanted, so called noise. The
2017.
Sound
level is a common
of annoyance part ofoneveryday
depends life.ofThe
the quality thesounds
sound can and be
ouralso unpleasant
attitude or it.
towards unwanted, so called
The sound noise. The
can damage and
level of annoyance depends on the quality of the sound and our attitude towards
Keywords: Cost Models; ABC; TDABC; Capacity Management; Idle Capacity; Operational Efficiency it. The sound can damage and

1. Introduction
* Corresponding author. Tel.: +40-743-794-319.
* E-mail
Corresponding author. Tel.: +40-743-794-319.
address:katalin.agoston@ing.upm.ro
The cost
E-mail of idle capacity is a fundamental information for companies and their management of extreme importance
address:katalin.agoston@ing.upm.ro
in modern production
2351-9789© systems.
2018 The Authors. In general,
Published it isB.V.
by Elsevier defined as unused capacity or production potential and can be measured
in several under
2351-9789©
Peer-review ways:
2018 Thetons of production,
Authors.
responsibility Published available
by Elsevier
ofthe scientific B.V.hours
committee of International
of the 11th manufacturing, etc. Interdisciplinarity
Conference The management of the idle capacity
in Engineering.
Peer-review underTel.:
* Paulo Afonso. responsibility
+351 253 ofthe scientific
510 761; committee
fax: +351 253 604of 741
the 11th International Conference Interdisciplinarity in Engineering.
E-mail address: psafonso@dps.uminho.pt

2351-9789 © 2017 The Authors. Published by Elsevier B.V.

Peer-review under responsibility of the scientific committee of the Manufacturing Engineering Society International Conference 2017.
2351-9789 © 2018 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of the scientific committee of the 11th International Conference Interdisciplinarity in Engineering.
10.1016/j.promfg.2018.03.078
534 Ágoston Katalin / Procedia Manufacturing 22 (2018) 533–538
Ágoston Katalin / Procedia Manufacturing 00 (2018) 540–545 541

destroy too. Sound measurements permit to determinate the level of the sound which may cause damage, or indicate
a fault, and is a useful tool in noise reduction.
The sound is the result of oscillations, variations in atmospheric pressure. These pressure variations must occur at
least 20 times per second to be heard. So the pressure variation per second is the frequency of the sound. The
hearing domain of sound for a young person is between approximately 20 Hz up to 20 kHz, while the range from the
lowest to highest note of a piano is 27.5 Hz to 4186 Hz.
The pressure variations travel through the medium. The speed of sound at room temperature is 344m/s. The
sound is characterized through the wavelength also. The wavelength is the distance from one peak to the next in case
of a sinusoidal sound pressure variation. The wavelength λ can be calculated from the speed ν and frequency f of the
sound: λ = 𝜈𝜈/𝑓𝑓 This equation shows that large wavelength corresponds for low frequencies and for small
wavelength corresponds to high frequencies (at 10 kHz the wavelength is 3.44cm).
A sound with one frequency is a pure tone. In practice there are no pure tones; sounds are made up of different
frequencies. The industrial noises consist of a mixture of frequencies which is called broad band noise. In
mathematics or measurements the white noise is often used, which is a sound with frequencies evenly distributed in
the hearing domain. [2]
The sound can be characterized through the amplitude of the pressure variations. The human ear can detect an
amplitude of 20 millionths of a Pascal. Express the sound amplitude in Pa results very large numbers. For this
reason for sound measure it is used the decibel or dB scale. Using the logarithms scale comparing the sound pressure
of one sound with another becomes easier.
The following definition gives the level of sound pressure p in decibels:

 p2 
SPL = 10 log10  2  dB (1)
 pref 
 

Where SPL is the Sound Pressure Level in dB; p is the sound pressure fluctuation (above or below atmospheric
pressure) and p ref is 20 micropascals (2 x 10-5Pa), which is approximately the threshold of hearing.
In this scale the normal conversation has 60-70dB, noise caused by traffic has 85-90dB, lawn mower 90-100dB,
chain saw 110dB.
The threshold of human hearing varies with frequency; it is most sensitive to middle frequencies and less
sensitive to lower and higher frequencies. To analyze noise, both frequency and amplitude need to be taken into
account. The sound or noise level measurements are made using A-weighted filter, and the measurement unit is A-
weighted decibels, dBA. The A-weighting filter has a frequency response corresponding to that of human hearing
(in level range about 40 dBA). The sound level expressed in dBA expresses the loudness of the sound. A detectable
change in level is circa 3-5dBA.
The ambient noise varies in level over time and to describe the noise variation with time there are used three
statistical characteristics. If a noise level having XdBA exceeded 20% of the time during the measuring period, this
result would be expressed with LA20 of XdBA. The LA90 characteristic is used to measure the average minimum
or background A-weighted noise level. The third characteristic is the LAeq (A-weighted equivalent continuous noise
level), and express the constant sound pressure level which is equivalent to the varying sound level over the
measurement time. These have the same acoustic energy. [1],[2],[3]

2. The sound level meter

The sound level meter provides measurements of sound pressure level and displays it in units of dB. The bloc
diagram of a sound level meter is shown on Fig. 1. [2],[4]
The input of the instrument is the microphone which converts the sound signal to an electrical signal. The
sensitivity of the microphone determinates the measurement precision. There are four types of microphones which
can be used in sound level meters: piezoelectric type, condenser type, electret and dynamic type. [6]
 Ágoston Katalin / Procedia Manufacturing 22 (2018) 533–538 535
542 Ágoston Katalin / Procedia Manufacturing00 (2018) 540–545

Fig. 1. Bloc diagram of the sound level meter.

In the piezoelectric microphone, a piezoelectric crystal generates an electric current (charge) when the air
pressure resulting from the sound waves act on it. These types of microphones are stable, robust and they are not
influenced by temperature.
The most used type of microphone in sound level meters is the condenser microphone. The condenser
microphone is constructed similar to a capacitor. A thin membrane or thin metal foil is close to a solid metal plate.
The membrane or diaphragm moves back and forth relative to the solid metal plate when sound waves hit the
diaphragm. So the capacitance changes to the rhythm of the sound and convert it to an electrical signal. But this
signal has low power and needs to be amplified through a preamplifier. These microphones are more accurate than
the other types, but they are influenced by dirt and humidity. The condenser microphones are used in precision
measurements.
The electret type differs very little on the condenser microphone. The potential difference is obtained by a
permanent electrostatic charge on the condenser plates and does not need polarization voltage.
By the dynamic microphone type the membrane is connected to a coil, which is placed in a magnetic field. The
movements of the coil generate a voltage proportional to the sound pressure variation. This microphone type is more
resistant but has poor frequency response.
The sound signal processing system begins with a weighting network. This is a relatively simple electronic circuit
whose sensitivity varies with frequency. The weighting can be "A", "B" and "C" weightings types which are
internationally standardized characteristics. The most used weighting network is “A” type. This circuit approximates
the signal to an inverted equal loudness low SPLs. [2]
To determinate the distribution of the sound (noise) level over a range of frequencies frequency analysis is used.
The most used analysis for noise studies is octave band analysis which gives more information about the sound.
For this the frequency range from 20 Hz to 20 kHz is divided into bands. This is done with electronic band-pass
filters which reject all sounds with frequencies outside the selected band. The bandwidth of these filters is one
octave or one third octave. An octave is a frequency band where the highest frequency is twice the lowest frequency.
For example, an octave filter with a centre frequency of 1 kHz admits frequencies between 707 and 1414 Hz. A third
octave covers a range where the highest frequency is 1.26 times the lowest frequency. The band-pass filters are
simple passive filters and some instruments automatically scan the whole range of frequency bands. Using these
sequential instruments the analyzed noise must remain constant both in amplitude and in frequency during the
analysis, 5 to 10 minutes. Other sound level meters have parallel filters, or use the fast Fourier transform. The results
of this frequency analysis are presented on a spectrogram which shows the level of the sound in each band. So the
pressure variation in time is converted to level variation (dB) in frequency (Hz). [10]
The resulting signal is amplified, and the Root Mean Square (RMS) value is determined. The RMS value is
important because it shows the energy quantity in the sound. The result is displayed in dB, but it can be obtained in
ac or dc signal form too.
The IEC 60651 standard gives the specifications for sound level meters and these characteristics refer to
directional characteristics, frequency weighting characteristics, time weighting, detector and indicator characteristics
and sensitivity to various environments. [4],[5]
536 Ágoston Katalin / Procedia Manufacturing 22 (2018) 533–538
Ágoston Katalin / Procedia Manufacturing 00 (2018) 540–545 543

If an instrument is connected into a circuit, the measured value will differ from the value without the instrument.
The instrument or sensor distorts the measured quantity. This is happening in the same way with the sound level
meter. The microphone orientation and type also distorts the sound field and influences the measurements. The used
microphone should have a uniform frequency response, and should have the same sensitivity in the whole frequency
range. Not only the sensor, but the environment also influences the measurement. Humidity, temperature, wind,
vibrations, pressure can reduce the measurement accuracy. To eliminate or minimize these influences the
measurement conditions must be kept constant. [5]

3. Noise measurement in fault detection

Sound measurements are made always together with noise measurements. All types of transducers and data
acquisition circuits introduce different types of noise making the measurement less accurate. By sound measurement
the environmental noise also disturbs the correctness of analysis, but sometimes the change of the environmental
noise leads to useful information. The equipments’ or motors’ working is followed by vibrations, oscillations which
are in hearing domain, so the monitoring of the right functioning can be made through noise measurement and
analysis. [9]
In practical measurements the detection of a sound caused by a fault will be difficult because of the background
noise. The noise floor is the level of background noise in a signal. Just signals above the noise floor can be detected.
Vibrations produced by a motor are complex; they contain continuous noise but random noise also. Because noises
cannot be foreseen, the frequency spectrum can change and alter the base spectrum. The frequency spectrum of a
motor or equipment is dependent of functioning parameters. Minor faults for example those caused by a bearing,
influence the vibration, the noise of the motor and the frequency spectrum.
The starting data acquisition is made at normal functioning parameters to obtain the reference spectrum. The
further measurement should be compared with the reference. The measurement condition must be the same,
temperature, load, background noise.
If a fault occurs, the vibration, the noise produced overlap with the base sound signal, and a modulation appears.
In case of different faults frequency and amplitude modulations appear. The gearbox faults cause frequency
modulation and bearings cause amplitude modulation. The last can be easy detected due to the sideband frequencies
in the frequency spectrum. [7]
In vibration analysis there are specific frequencies due to the rotating element. These specific frequencies are
proportional with the fundamental frequency (1xRPM). The faults produce a sound with a different frequency than
the fundamental frequency. At the frequencies due to the fault a peak appears in the spectrum. Analyzing the
frequency spectrum, if a peak appears at other frequency than specific fundamental frequencies, it denotes a problem
and the amplitude of this peak denotes the seriousness of the problem. In practice it is difficult to observe the
sideband frequencies because of noise and because the specific frequencies are also high. On figure 2 the bloc
scheme for efficient amplitude demodulation is presented.

Fig. 2. Bloc scheme for efficient amplitude demodulation.

The high-pass filter eliminates the low frequency components and isolates the modulation frequency. Then the
signal is demodulated through rectifier. We obtain positive impulses whose frequency is equal to the rotating
elements frequency. To eliminate the fluctuations a low-pass filter is used. Through this process the amplitude of the
useful signal is higher. This method is used successfully in detection of small fault in bearing.
On a typical demodulated spectrum the basic level (noise) is uniform and peaks appear whose amplitude relative
to the base is important. As the environment noise grows, the demodulated spectrum appearance is the same. Bigger
 Ágoston Katalin / Procedia Manufacturing 22 (2018) 533–538 537
544 Ágoston Katalin / Procedia Manufacturing00 (2018) 540–545

the fault is, higher are the peaks’ amplitude above the basic level. These peaks should be detected and analysed in
the noise spectrum. In case of amplitude modulations (bearings faults) sideband frequencies appear in the frequency
spectrum. [8], [11]

4. Noise analysis simulation

By measuring sound produced by vibrations we cannot observe or distinguish the different frequency signals. It is
hard to tell if there is or not a vibration frequency which is caused by a fault. Matlab has several functions to
determinate the frequency spectrum of a signal. After loading the acquired data using pwelch or spectrum.welch the
Power Spectral Density (PSD) of the signal can be obtained. These functions estimate the PSD with Welch’s method
and use Hamming window. [12]
Figure 3 presents the noise and its frequency spectrum which model the environmental noise.
Motor functioning produces periodical vibrations, sound whose frequencies are proportional with the rotating
elements’ frequencies. Figure 4 presents the acquired sound at motor functioning and its spectrum. Peaks can be
observed at two different frequencies, corresponding to the fundamental frequency due to the motor and bearing.

Welch Power Spectral Density Estimate

4 Power/frequency (dB/Hz) -26

2 -27
-28
0
-29
-2 -30
-4 -31
0 5 10 0 200 400
Frequency (Hz)

Fig. 3. Noise and frequency spectrum of noise.

Welch Power Spectral Density Estimate

Power/frequency (dB/Hz)

10 -15

5
-20
0
-25
-5

-10 -30
0 5 10 0 200 400
Frequency (Hz)

Fig. 4. Vibrations and PSD for motor functioning.

The bearing usage causes amplitude modulation because it produces a periodical vibration with a smaller
frequency than the fundamental frequency which superpose. On the measured vibration sound this variation cannot
be observed, but the frequency spectrum is different. In accordance with the fault magnitude smaller or larger
sideband frequencies appear. These sideband frequencies should be detected. Figure 5 presents the vibration sound
signal in presence of faults and its PSD. The sideband frequencies around the 400Hz can be observed.
538 Ágoston Katalin / Procedia Manufacturing 22 (2018) 533–538
Ágoston Katalin / Procedia Manufacturing 00 (2018) 540–545 545

Welch Power Spectral Density Estimate

10 -10

Power/frequency (dB/Hz)
5 -15

0 -20

-5 -25

-10 -30
0 5 10 0 200 400
Frequency (Hz)

Fig. 5. Vibration sound signal and PSD in presence of faults.

Knowing the fundamental frequency of different rotating elements and analyzing the frequency spectrum of the
signal the place and the size of the fault can be detected.

5. Conclusions

Measuring and analyzing sound produced by vibrations allows detecting faults in different parts of equipments.
Sound measurements are made always together with noise measurements. The equipments’ or motors’ working is
followed by vibrations, oscillations which are in hearing domain, so monitoring the right functioning of these can be
made through sound and noise measurement and analysis.
To analyze sound, both frequency and amplitude need to be taken into account. To determinate the distribution of
the sound level frequency analysis is used.
In noise measurement and analysis there are specific frequencies due to the rotating element. These specific
frequencies are proportional with the fundamental frequency. If a fault produces the vibration, the produced noise
overlaps with the fundamental sound signal and modulation appears.
In Matlab there are functions to determinate the Power Spectral Density of a signal. Using this type of analysis
faults can be detected and localized. Through simulation it was shown how and where the peaks appear in the
frequency spectrum caused by faults. In practice these analysis are more complex.

References

[1] Jane D. Mac, Noise & Noise Measurement, http://www.southwoodresources.com.au/southwood/pdf/planning/APPMSEMF.PDF .

[2] Brüel & Kjaer, Measuring Sound, http://www.bk.dk .
[3] Brüel & Kjaer, Environmental Noise measurement, https://www.bksv.com/media/doc/br0139.pdf .
[4] Malchaire J.,Sound Measuring Instruments, 1994, http://www.who.int/occupational_health/publications/noise6.pdf .
[5] Department of Environment and Heritage Protection, Queensland Government, Noise Measurement Manual, ver.4, 2013.
[6] *** Prosig, Noise & Vibration Measurement Handbook, Fourth Edition, http://www.prosig.com.
[7] A.L.Lance, W.D.Seal, F.Labaar, Phaise Noise and AM Noise Measurement in the Frequency Domain, chapter 7, Academic Press, 1984.
[8] ***Mobil Industrial AG, Analiza vibratiilor, Modularea demodularea in amplitudine, Analizorul FFT, Spectru de frecventa,
http://www.mobilindustrial.ro/current_version/online_docs
[9] C.Mason, How to analyze noise & vibration in rotating machines, http://blog.prosig.com/2011/08/09/how-to-measure-noise-vibration-in-
rotating-machines
[10] J.Wren, Understanding Filter Characteristic, http://blog.prosig.com/2011/10/04/understanding-filter-characteristics
[11] D.Davies, Bearing & Gearbox Vibration Analysis Using Demodulation Techniques, 2014, http://blog.prosig.com
[12] *** Matlab User Guide.