Article
Silencer Design for the Control of Low Frequency Noise in
Ventilation Ducts
Edoardo Alessio Piana 1, * , Ulf Erik Carlsson 2 , Adriano Maria Lezzi 3 , Diego Paderno 3 and Susann Boij 2






Citation: Piana, E.A.; Carlsson, U.E.;
Lezzi, A.M.; Paderno, D.; Boij, S.
Silencer Design for the Control of
Low Frequency Noise in Ventilation
Ducts. Designs 2022, 6, 37. https://
doi.org/10.3390/designs6020037
Academic Editors: Tiago
Pinto Ribeiro and Julian D. Booker
Received: 17 February 2022
Accepted: 4 April 2022
Published: 6 April 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Designs 2022, 6, 37. https://doi.org/10.3390/designs6020037
1 Applied Acoustics Laboratory, University of Brescia, Via Branze 38, 25123 Brescia, Italy
2 MWL, Department of Engineering Mechanics, KTH—Royal Institute of Technology, Teknikringen 8,
SE-100 44 Stockholm, Sweden; ulfc@kth.se (U.E.C.); sboij@kth.se (S.B.)
3 Department of Mechanical and Industrial Engineering, University of Brescia, Via Branze 38,
25123 Brescia, Italy; adriano.lezzi@unibs.it (A.M.L.); diego.paderno@unibs.it (D.P.)
* Correspondence: edoardo.piana@unibs.it
Abstract: The control of noise propagating along ventilation system ducts has always been an
important issue in the building and vehicle sectors. This problem is generally tackled by selecting
noise-reducing components with a suitable transmission loss, possibly verifying their effectiveness at
a later time. The aim of this article is to characterize the nature of the problem and propose a design
approach focusing directly on the perceived effect, that is, on the sound pressure level downstream
of the outlet. Because the nature of the noise emission depends on various generation mechanisms,
different methods can be applied. Usually, it is more difficult to realize good attenuations at low
frequencies because of the limits of sound absorbing materials in such frequency range. For this
reason, the ability of reactive components to attenuate the noise below the cut-on frequency will
be investigated. This goal is reached by applying the transfer matrix approach to a duct system,
with the implementation of the transfer matrices of each single element, and then assembling a
system capable of acoustically describing the source and the duct structure. The coupling between
the duct system with source and receiver impedances allows one to predict the sound pressure level
at a given distance from the outlet. The proposed methodology is implemented in a user-friendly
calculation tool with possible academic and professional application. Predictive capability, usability,
and intuitiveness of the proposed design procedure are validated against experimental results by real
potential users, who express positive feedback.
Keywords: low frequency noise; transfer matrix method; duct acoustics; reactive silencers; ventilation
noise
1. Introduction
The control of noise in duct systems is a topic of interest in many different sectors. All
ducts through which gas flows generate acoustic noise. Sometimes the generated noise
causes too high noise exposure or discomfort for operators, drivers, passengers, etc. In such
cases, adequate noise reduction measures must be taken. Some examples are vehicles [1–5],
buildings [6,7], ships [8–10], industries [11,12], and energy plants [13–16].
Choosing suitable noise reduction measures requires expert knowledge on how the
noise is generated and transmitted from the sources to the observation point. Normally, the
duct flow noise contains both tonal and random components. The random components,
caused by turbulent flow, are distributed over a wide frequency range, whereas the tonal
components appear at distinct frequencies. Tonal components are caused by periodic
processes or phenomena such as ones involving fans [17–20] or compressors [7,21,22]
running at constant speed or periodic eddy generation [23,24] causing whistling noises.
From a noise comfort point of view the tonal components cause more discomfort than the
wide-band turbulence flow noise. Therefore, the noise reduction measures often focus
on the tonal components. After a comparison of the problematic frequencies with the
https://www.mdpi.com/journal/designs
Designs 2022, 6, 37 2 of 22

characteristics of the possible generation mechanisms, the source causing problems can be
identified. Once the sources have been identified, it is possible to choose how to reduce the
noise, working on the source generation mechanism or on the transmission via the duct.
This type of problem is assuming an ever-growing interest with the expansion of
heat recovery ventilation systems. Indeed, the comfort inside buildings is a particularly
challenging issue given the extremely low background noise that can be found in some
very quiet environments [25,26]. For the same reason, it is important to properly design the
building elements against tapping noise, especially for lightweight structures [27]. because
many solutions can also be used for façade sound insulation, the thermal properties of
classic and sustainable materials can also be of interest [28,29].
As concerns the design of silencers suitable for reducing tonal noise transmitted
along the duct system, there are two fundamentally different types of elements: resistive,
based on acoustic energy dissipation, and reactive, based on wave interference. Typical
resistive elements are porous materials, orifices, and perforated plates. Recently, micro-
perforated plates and 3D printed porous materials have becoming more widely used in
the aerospace industry because their acoustic properties can be easily tuned to obtain high
performance silencers working at specific frequencies or over a broad band frequency
range. Another technique used to reduce noise in ducts is based on active control. Even
if it is very interesting and a subject of a number of scientific publications, this type of
sound reduction strategy will not be considered in this paper because it requires additional
electronic components (DSP microphones and actuators), which make it expensive to apply
in real cases. In contrast to most of the resistive elements, reactive silencer elements can be
tuned to have very high noise reduction capabilities at specific frequencies. Hence, reactive
elements are the natural choice for the reduction of low frequency tonal noise.
An interesting way to describe a system through an electro-acoustic analogy was
introduced by Schönfeld [30]. The main advantage is the ability of the method to describe
the elements through a lumped parameters approach, thus allowing an easy implementa-
tion and fast calculations. Later, this theory was mainly applied to exhaust systems. The
approach requires the characterization of the source and of the outlet impedances. A contri-
bution to the modelling of fluid machines as sources of sound in duct and pipe systems is
given in [31]. Later, Glav and Åbom developed a general formalism for analyzing acoustic
two-port networks [32]. Such work led to the realization of lumped parameters software
for simulating the sound propagation in complex duct systems based on the transfer matrix
method [33]. This paper shows how the transfer matrix method can be used to design
reactive silencers suitable for solving the specific noise problem at hand. It is interesting
to note that the transfer matrix approach can also be adopted for the acoustic design of
building elements, as shown by Caniato [34].
One positive property of the transfer matrix method, also known as the T-matrix
method or TMM, is that it describes each silencer element with its T-matrix, and the
complete silencer properties can be easily obtained by a combination of the elements’
T-matrices [35].
In particular, the overall sound transmission loss (TL) can be obtained as a direct
out-put of the model. The general approach is, therefore, to estimate the TL required to
fulfill the design specifications and to select and combine components that can provide
it while respecting some practical constraints, such as cost and size limits. However, this
approach does not return the impact of the solution on the perceived noise, leaving this
verification to a later time. In recent years, comfort evaluations are gradually moving
towards an occupant-centric paradigm, which involves the adaptation of traditional design
methods to more detailed, easy-to-interpret approaches.
This paper fills the gap by proposing an expanded TMM method (xTMM), integrating
the classic TMM procedure to build a reactive silencer with the possibility to evaluate its
effectiveness directly in terms of perceived noise—namely, sound pressure level at a certain
distance from the ductwork outlet. A self-developed tool in MATLAB is also described
Designs 2022, 6, 37 3 of 22

that can be used in academic and professional courses to promote noise-oriented design
approaches.
The paper presents the formulation of noise in ducts in Section 2. Section 3 describes
the xTMM method, based upon coupling the TMM system with source and receiver
impedances, and introduces the self-developed design tool and its characteristic compo-
nents. Section 4 presents the procedure adopted in the design stage. Section 5 discusses the
case study, and finally Section 6 draws the conclusions.

2. Noise Generation and Propagation in Ducts

There are different generation mechanisms that build up noise in ducts. Some of
them are related to rotating machinery or vibrating membranes, some others have a
fluid-dynamic origin such as vortex shedding and turbulence. Usually, once the noise is
generated by a source, it propagates inside the duct until it is radiated by an outlet. In the
case of ventilation systems, noise is caused by two main sources: a fan or a compressor
and the turbulence generated by the air flow in the ducts [36]. At constant speed, the
ventilation system fan drives an air volume flow through the ventilation ducts. The volume
flow has a non-zero mean value and a superimposed pulsating part caused by the fan
blades. At constant speed, the pulsating part causes a tonal acoustic noise. The mean flow
itself generates a wide-band random character noise caused by turbulent vortex shedding
in the boundary layer between the air flow and the duct walls, and at the fan blades. At
specific flow conditions, some components can generate a whistling noise. All these sound
generating mechanisms cause volume flow disturbances that radiate acoustic noise from
the duct outlet to the receiving environment.
The cut-on frequency is the frequency below which only plane waves propagate in the
duct. For example, in a circular duct such frequency is defined as:
c
f cut−on = 1.841 (1)
2πD
where c is the speed of sound in the duct and D is the duct diameter. Above the cut-on
frequency, the shape of the wave front is no longer plane, and a good attenuation can
be easily achieved by placing absorbing material on the duct walls [37]. For this reason,
an advanced design procedure is not necessary to reduce noise components above the
cut-on frequency. The cut-on frequency discriminates the low-frequency range from the
high-frequency range and gives a first numerical indication on the frequency interval
where an acoustic design based on the plane wave assumption is feasible. Below the cut-on
frequency, alternative methods must be applied to reduce the noise. Such methods refer to
the use of reactive elements, which must be designed to reflect a part of the sound wave
with an opposite phase with respect to the sound propagating in the duct. To have such an
effect, an abrupt change in the impedance characteristics of the duct is necessary. There
are different reactive elements able to introduce an impedance change that can be used to
achieve a reduction of the noise at the outlet, and they are: area discontinuities, expansion
chambers, and side branches. Each one of these elements has a different behavior that can
be exploited to build a system w a broad band sound attenuation. An overview of the
devices that can be used for the acoustic design is given in the following sections.
Reactive elements are effectively used to attenuate noise in specific frequency ranges,
generally related to the motor rpm and number of blades of a fan [37]. The “passing by
frequency” can be computed as:
rpm
f pb = × bn (2)
60
where rpm is the number of revolutions per minute of the motor and bn is the number of
blades of the fan. This is the lowest frequency at which an attenuation may be required.
Higher-order harmonics can be computed as:
rpm
f pb = × nb × n (3)
60
Designs 2022, 6, 37 4 of 22

where n is a natural number. Usually, the first two harmonics (n = 2, 3) correspond to the
loudest components, which are the target when designing reactive elements.
The amount of attenuation given by a duct element is usually expressed using two
different descriptors: transmission loss (TL) and insertion loss (IL). The TL is defined as the
acoustical power difference between the forward travelling incident pressure wave at the
inlet of the element and the forward travelling transmitted pressure wave at the outlet of
the element. Expressed in decibels, the equation for determining the TL reads:

Win
TL = 10 log = 20 log pin /pout + 10 log Sin /Sout (4)
Wout

where W is the sound power, p is the sound pressure, S is the cross section, and the “in” and
“out” subscripts indicate the inlet and the outlet of the considered element, respectively.
The IL is defined as twenty times the logarithm of the pressure measured at a certain
point in two different system configurations a and b:

IL = 20 log p a /pb (5)

As an example, for regulatory applications concerning vehicle noise, the IL is used to

compare the sound pressure level (SPL) at 1 m distance, 45◦ from the outlet of an exhaust
system without a silencer to the SPL in the same position when a silencer is fitted on the
exhaust system. From a practical point of view, the IL is usually the easiest and most
common way for evaluating the sound reduction properties of a silencer system. However,
this may not be enough to ensure acceptable conditions in some practical situations, such
as the assessment of acoustic comfort in indoor environments with mechanical ventilation
systems. In such cases, the SPL in a given position must be estimated, which requires the
knowledge of quantities such as source impedance, the nature of the ground (reflection
coefficient), and the radiation model of the outlet [38,39].

3. Silencer Design with xTMM

3.1. Overview
The proposed design method is based on a scientific approach. More specifically, the
lumped parameters acoustic modeling is positioned halfway between a purely empirical
approach and the application of computationally burdensome FEM or BEM simulations.
Nevertheless, this method holds some of the attractive features of both approaches, such
as quick implementation, short calculation time [40], and consistency with experimental
results [41]. A method based on the same element model has been used recently for
designing the exhaust systems of vehicles [42] as well as many other systems [43–46]. The
method can also be adapted to be used outside the automotive field, for example to design
industrial mufflers as well as HVAC ducts or thermal recovery ventilation systems, with
the fire frequency replaced by the blade passing frequency of fans generating noise.
In this approach, the duct is split into fundamental blocks, each one having an inlet
and an outlet. Such blocks can be characterized by different transfer matrices relating two
variables: the sound pressure, p, and the volumetric flow rate, Q. Applying an electro-
acoustic analogy where the pressure is equivalent to voltage and the volumetric flow
to current, the transfer matrix approach can be considered as the equivalent of the two
Kirchhoff’s circuit laws. The interconnection of different blocks can be expressed by a
product of the different transfer matrices characterizing the single blocks describing the
system (Figure 1).
Under linear conditions, the T-matrix method can be used to describe the acoustic
properties of a silencer system. All models described hereafter assume 1-dimensional plane
wave propagation along the duct axis. This means that the models are valid up to a higher
frequency limit where the cross section is small compared to half a wavelength. Each
block corresponds to a physical component and has a specific effect on the overall sound
attenuation of the system. Detailed references to the mathematical models of the single
 Designs 2022, 6, 37 5 of 22

Designs 2022, 6, x FOR PEER REVIEW

elements described hereafter can be found in [36]. The selection of these blocks and the
ability to simulate their combined effect is therefore the key to a good acoustic design of
the system.
Figure 1. Equivalent acoustic model of a single element—transfer matrix representation.

Under linear conditions, the T-matrix method can be used to describe the acoust
properties of a silencer system. All models described hereafter assume 1-dimensiona
plane wave propagation along the duct axis. This means that the models are valid up to
higher frequency limit where the cross section is small compared to half a wavelength
Each block corresponds to a physical component and has a specific effect on the overa
sound attenuation of the system. Detailed references to the mathematical models of th
single elements described hereafter can be found in [36]. The selection of these blocks an
the ability to simulate their combined effect is therefore the key to a good acoustic desig
of the system.
Figure In this work,
1. Equivalent a software
acoustic developed
model of in MATLAB matrix
a single element—transfer has been used to perform the simu
representation.
Figure 1. Equivalent acoustic model of a single element—transfer
lations. Below the cut-on frequency, the software can model the system matrix representation.
as a network o
In this
blocks work,the
because a software developed
continuous in MATLAB
field variables has beenpressure
describing used to perform the sim-flow va
or volumetric
ulations.
Under
iations Below
(p(x,t) the conditions,
linear
and cut-on
Q(x,t))frequency,realthe
in thethe software
T-matrix
system can
method
can modelcanthebesystem
be discretized used
into as
toa describe
elementsnetwork the on
in which ac
ofofblocks because
the quantities the continuous
can be assumed field variables describing pressure or volumetric flow
properties of a silencer system. toAll be models
a locally described
spatially invariant
hereafter (p(t)assume
and Q(t))1-dimen
[36].
variations (p(x,t) and Q(x,t)) in the real system can be discretized into elements in which
plane The elements can be selected among several
This pre-determined elements. These eleu
one ofwave propagation
the quantities along
can be assumed the duct
to be axis.
a locally means
spatially that(p(t)
invariant theandmodels are
Q(t)) [36]. valid
ments
higher The are the
frequency most
elements can common
limit reactive
whereamong
be selected the cross devices
severalsection used to attenuate
is small
pre-determined low
compared
elements. frequency
Theseto noise
half a wavel
elements an
will
are be
the introduced
most common in the next
reactive section.
devices Figure
used to 2 gives
attenuate
Each block corresponds to a physical component and has a specific effect an
low example
frequency of a complete
noise and ventilatio
willon the o
system
be and
introduced its
in acoustic
the next T-matrix
section. model.
Figure 2 The
gives T-matrix
an example method
of a is
completetailored for analysis o
ventilation
sound attenuation of the system. Detailed references to the mathematical models
cascade-coupled
system and its acousticsystems. The
T-matrix overall
model. Thebehavior
T-matrix can be described
method is tailored forby analysis
a “total”of T-matri
single elementssystems.
cascade-coupled
described hereafter can be found in [36]. The selection of these bloc
equal to the product ofThe
theoverall
singlebehavior
T-matrices can be
of described by a “total” T-matrix equal
its constituents:
the ability
to the to simulate
product their
of the single combined
T-matrices effect is therefore the key to a good acoustic
of its constituents:
of the system.
N = (6
In this work, a software developed T = ∏ in Tn MATLAB has been used to perform(6) the
lations. Below the cut-on frequency, the n=1 software can model the system as a netw

blocks because the continuous field variables describing pressure or volumetric flo
iations (p(x,t) and Q(x,t)) in the real system can be discretized into elements in whi
of the quantities can be assumed to be a locally spatially invariant (p(t) and Q(t)) [3
The elements can be selected among several pre-determined elements. The
ments are the most common reactive devices used to attenuate low frequency noi
will be introduced in the next section. Figure 2 gives an example of a complete vent
system and its acoustic T-matrix model. The T-matrix method is tailored for anal
cascade-coupled systems. The overall behavior can be described by a “total” T-
equal to the product of the single T-matrices of its constituents:

Figure2.2.AAventilation
Figure ventilation system
system with
with silencer
silencer (top)
(top) =
andand its linear
its linear acoustic
acoustic system
system modelmodel (bottom). Th
(bottom).
fan is represented by its acoustic volume flow, Q , and its acoustic impedance, Z
The fan is represented by its acoustic volume flow, QK , and its acoustic impedance, ZK . The silencer
K K . The silencer sy
tem by
system byits
itsT-matrix, T,and
T-matrix, T, andthe
thereceiving
receiving room
room (load)
(load) withwith its acoustic
its acoustic impedance,
impedance, ZL . ZL.

The
Thetransmission
transmission lossloss
of a of
silencer is a measure
a silencer of the of
is a measure isolated silencersilencer
the isolated noise reduction
noise reductio
capacity. It is a useful tool when a silencer component is tuned to a certain problematic
capacity. It is a useful tool when a silencer component is tuned to a certain problemati
frequency or frequency range. The insertion loss, on the other hand, also accounts for the
frequency or frequency range. The insertion loss, on the other hand, also accounts for th
system source and receiver impedances. Hence, it is ideal as a measure of the capability of
a silencer to reduce the noise when it is installed in the system.
Designs 2022, 6, 37 6 of 22

The system transmission loss, expressed using the duct system T-matrix components
and the impedances at the duct inlet and outlet, is:
!
Zout t12 t22 Zin 2
DTL = 10 log ·t + + t21 Zin + (7)
4Zin 11 Zout Zout

where tij are elements of the system T-matrix. If the impedances at the inlet (Zin ) and
outlet (Zout ) are equal to the corresponding duct plane wave specific impedance, ρ0 c/Sin
and ρ0 c/Sout , respectively, reflection free in- and outlets have been assumed. Using this
assumption, it is possible to analyze the system sound reduction properties without any
influence from the source and the receiver impedances.
The insertion loss, expressed in terms of the system acoustic properties, is:
2
t11 ZL + t12 + t21 ZK ZL + t22 ZK
D IL = 10 log (8)
tr,11 ZL + tr,12 + tr,21 ZK ZL + tr,22 ZK

where tij and tr,ij are T-matrix elements for the case ”with” silencer and the reference case
“without” silencer, respectively.
In the construction and validation phases, a silencer system was built and its insertion
loss measured. The measured insertion loss could be used to compare and validate the
predicted insertion loss. A relevant comparison requires a good estimation of the source
and receiver impedances, ZK and ZL . In the MATLAB scripts, the receiver properties
are described with the “free space” impedance model, while the source impedance is
described with the impedance measured on the loudspeaker used during the validation
measurements.
It is worth noting that the xTMM method retains the same limitations of the classic
TMM, in that:
• it is only valid under the duct cut-on frequency;
• because it is applicable in the plane wave range, the results are not valid in the vicinity
of acoustic treatments or close to duct shape or size variations;
• it neglects evanescent coupling between the silencer elements [47];
• In the following, the reactive elements that can be combined to design the silencer are
briefly recalled, based on the formulae in [36]. The time dependence adopted in this
work is exp(iωt). As shown in the following Section 3.2.3, such an assumption helps
to remove discontinuities (infinity points) to the TL computed for quarter-wavelength
resonators. In the case of particularly narrow ducts, Stinson’s model [48] can also
be applied. The transmission loss plots in Section 3.2 refer to the actual elements
used for the experiments in Section 5. Components are assumed to be characterized
by an internal friction factor, Xi , which can be used to estimate pressure drops (see
Section 3.3).

3.2. T-Matrices for Acoustical-Physical Elements

3.2.1. Straight Pipe
Straight pipe elements are found in all duct systems, and they usually add only a very
small dissipation per meter length to the noise propagating in the duct. The T-matrix for a
straight pipe, length L (Figure 3), is:
   
t11 t12 cos(KL) i ( Z/S) sin(KL)
T= = (9)
t21 t22 iS sin(KL)/Z cos(KL)

where K = ω/c is the wave number, c is the speed of sound in air, ω is the angular frequency,
ρ is the air density, S is the pipe cross section area, and Z = ρc is the acoustic impedance of
the air inside the duct.
 Designs 2022, 6, 37 7 of 22
esigns 2022,6,
signs 2022, 6,xx FOR
FORPEER
PEERREVIEW
REVIEW 77 of
of 22
22

Figure 3.
Figure 3. Straight
Straight pipe
Figurepipe sketchpipe
sketch
3. Straight (left)sketch
(left) and example
and example of TL
of
(left) and TL obtained
obtained
example with
with
of TL TMM (right).
TMM
obtained (right).
with TMM (right).

3.2.2. Expansion
3.2.2. Expansion Chamber Chamber
3.2.2. Expansion
Chamber
This element
This elementThisfeatures
element
features anfeatures
an area expansion
area expansion followed
an areafollowed
expansion byfollowed
by an area
an area contraction,
contraction, resulting in
by an area resulting in resulting
contraction,
aa double in a double
double reflection.
reflection. The
The reflection.
attenuation
attenuation The attenuation
depends
depends on the
on thedepends on the
inlet/outlet
inlet/outlet cross
crossinlet/outlet cross
section areas,
section areas, section areas,
main
main
chamber cross
chamber mainsection
cross chamber
section area,cross
area, andsection
and area,
the length
the length ofand
of the length chamber.
the expansion
the expansion of the expansion
chamber. chamber.
Anexpansion
An expansion An expansion
cham-
cham-
ber can be chamber
considered canasbe considered
being made upas being
of three made
straight up of three
pipes, straight
where
ber can be considered as being made up of three straight pipes, where the middle pipe has the pipes,
middle where
pipe hasthe middle
aa larger pipe
larger cross has
cross section a larger
section area cross
area (Figure section
(Figure 4).
4). Its area
Its T-matrix(Figure
T-matrix is: is: 4). Its T-matrix is:

=
= ∙∙ T ∙∙= T1 ·T2 ·T3 (10)
(10) (10)
where TT22 is
where is the
the intermediate
intermediate pipe
pipe T-matrix.
T-matrix.
where T is the intermediate
2 pipe T-matrix.

Figure 4.
Figure 4. Expansion
Expansion chamber
Figure 4.chamber sketch
sketch
Expansion (left)sketch
(left)
chamber and example
and example of TL
of
(left) and TL obtained
obtained
example with
with
of TL TMM (right).
TMM
obtained (right).
with TMM (right).

When using
When using an expansion
When
an expansion chamber, the
using an expansion
chamber, the maximum
chamber, TL is
is found
the maximum
maximum TL found by:
TLby:
is found by:

= =nc
= f n,,= = 1,3,
1, 3, 5,=
5,
, n … 1, 3, 5, . . .
… (11)
(11) (11)
44 4LC
where cc is
where is the
the speed
speed of of sound andand LL is
is the
the length
length of of the
the expansion
expansion chamber.
chamber. This This device
device
where c is thesound
speed of sound and L is the length of the expansion chamber. This device
hasaa broadband
has broadband attenuation,
attenuation, which
whichis ispositive
positive in inthe the case
case of
of wideband
wideband noise. On Onthe
the other
other
has a broadband attenuation, which is positive in the casenoise.
of wideband noise. On the
hand, the
hand, the attenuation
attenuation is is not
not extremely
extremely high,
high, soso ifif aa large
large attenuation
attenuation isis required
required atat specific
specific
other hand, the attenuation is not extremely high, so if a large attenuation is required
frequencies, itit is
frequencies, is suggested
suggested to to use
use other
other types
types of
of reactive
reactive devices
devices such
such asas aa quarter-wave-
quarter-wave-
at specific frequencies, it is suggested to use other types of reactive devices such as a
length resonator
length resonator or a Helmholtz
or a Helmholtz
quarter-wavelength resonator.
resonator.
resonator or a Helmholtz resonator.
3.2.3. Quarter-Wavelength
3.2.3. Quarter-Wavelength ResonatorResonator
Resonator
3.2.3. Quarter-Wavelength
Quarter-wavelength
Quarter-wavelength resonators
resonators
Quarter-wavelength behave like
behave
resonators like expansion
expansion
behave chambers
chambers
like expansion chambers in that
in that they
inthey are are based
are
that they
based on
based on multiple
multiple reflections
reflections
on multiple but are
but
reflections are characterized
butcharacterized by aa more
by
are characterized more selective
selective
by a more behavior
behavior
selective given by
given
behavior by
given by the
the finite
the finite size
size ofsize
of
finite the side
the side
of cavity.
thecavity. AA quarter-wavelength
quarter-wavelength
side cavity. A quarter-wavelength resonator
resonator is aa type
is
resonator type of side-branch.
of
is a type side-branch.
of side-branch. In the
In the
In the plane
plane wave
plane
wave wavefrequency
frequency
frequency region, aa side-branch
region,
region, side-branch T-matrix
a side-branch cancan
T-matrix
T-matrix can be written
be written
be written as: as:
as:
1 00
= = t121 .. 1
 
= t11= 0 (12)
(12)
T= /
/ =1 1 . (12)
t21 t22 Ss /Zs 1
Designs 2022, 6, x FOR PEER REVIEW 8 of 22
Designs 2022, 6, 37 8 of 22

where SSs isisthe

where theside-branch
side-branchinletinletcross
crosssection
sectionarea
areaand Zs Z
and is its specific acoustic input im-
s s is its specific acoustic input
pedance seen from its inlet. The specific acoustic input
impedance seen from its inlet. The specific acoustic input impedance impedance for thefor
quarter-wave-
the quarter-
length resonator is:
wavelength resonator is:
Zs ==−− iρ0 c· cot (KLs ). ).
∙ cot( (13)
(13)
where LLss is
where is the
the resonator
resonator length.
length. Losses
Losses in
in the
the resonator
resonator areare accounted
accounted forfor by
by the
the medium
medium
complex compression modulus, β = β ·
complex adiabatic compression modulus,  = ∙(1 + i), where  is the medium internal
adiabatic (1 + iη), where η is the medium internal
loss factor. For
loss factor. Forthis
thisreason,
reason,thethespeed
speedofofsound, = c·(1
sound,c =c c∙(1 +/2),
+ i iη/2),
andand the wave
the wave number,
number, K=
K = ω/c,
ω/c, turn turn
out tooutbeto be complex-valued.
complex-valued.
Equation (12) is
Equation (12) is the
the T-matrix
T-matrix for
for the
the side-branch
side-branch seen seen from
from its
its coupling
coupling surface
surface with
with
the main duct. For the real physical side-branch, there are also inlet and outlet
the main duct. For the real physical side-branch, there are also inlet and outlet pipes, both pipes, both
with length, L; see Figure 5. Consequently, the physical quarter-wavelength
with length, L; see Figure 5. Consequently, the physical quarter-wavelength resonator T- resonator
T-matrix corresponds
matrix corresponds toto the
the matrix
matrix product:
product:
T ==T1 ·T∙ 2 ·T∙3 (14)
where T
where T11 and
and T T33 are
are the
the T-matrices
T-matrices of
of the
the straight
straight pipes
pipes with
with length,
length, L,
L, and cross section
and cross section
area S.
area S. TT22 is
is the
theside-branch
side-branchT-matrix,
T-matrix,Equation
Equation (12),
(12), with
with side-branch
side-branch input
input impedance,
impedance, Zs,
Zs, expressed according to Equation
expressed according to Equation (13). (13).

Figure 5. Quarter-wavelength resonator sketch (left) and examples of TL obtained with TMM (cen-
Figure 5. Quarter-wavelength resonator sketch (left) and examples of TL obtained with TMM (center
ter and right).
and right).

This component
This component provides
providesaahigh
highTL
TLatatsome
somespecific
specificfrequencies.
frequencies.The
The maximum
maximum TLTL
is
is obtained for:
obtained for:
nπ
KLs = (15)
=2 (15)
2
with
with 2π f
K= (16)
c2
= (16)
thus
nc
fn = , n = 1, 2, 3, . . . (17)
thus 4Ls
Equation (15) shows that the quarter-wavelength resonator size can be significant. For
= , = 1, 2, 3, … (17)
4
example, for tuning the element at a frequency of 100 Hz, the quarter-wavelength resonator
side branch length must have a characteristic length of 0.85 m. It is worth noting that a
Equation (15) shows that the quarter-wavelength resonator size can be significant.
quarter-wavelength resonator with such a length sometimes cannot be easily fitted in a
For example, for tuning the element at a frequency of 100 Hz, the quarter-wavelength
real system. A common solution for this issue is to bend the resonator or fold it along the
resonator side branch length must have a characteristic length of 0.85 m. It is worth noting
main duct axis so that it occupies a space compatible with the overall duct dimension while
that a quarter-wavelength resonator with such a length sometimes cannot be easily fitted
maintaining its characteristic length.
in a real system. A common solution for this issue is to bend the resonator or fold it along
the main
3.2.4. duct axis
Helmholtz so that it occupies a space compatible with the overall duct dimension
Resonator
while maintaining its characteristic length.
The Helmholtz resonator (Figure 6) is a side-branch with a specific input impedance:

Zs = R + iωρ0 Ln + ρ0 c2 Sn /iωVc . (18)

 3.2.4. Helmholtz Resonator
The Helmholtz resonator (Figure 6) is a side-branch with a specific input impedance:
= + +ρ /iω . (18)
Designs 2022, 6, 37 9 of 22
where Vc is the resonator volume given by Lc and Sc (see Figure 6). Resonator losses are
often generated by a damping material placed in the resonator neck. Such a term is ac-
counted for by introducing the resistance, R. The resistance is determined by the resistiv-
where Vc is the resonator volume given by Lc and Sc (see Figure 6). Resonator losses
ity r [Pa∙s/m
are] and
2
oftenthe dampingby
generated material length,
a damping Lr:
material placed in the resonator neck. Such a term
is accounted for by introducing the= resistance,
. R. The resistance is determined
(19) by the
resistivity r [Pa·s/m2 ] and the damping material length, Lr :
The T-matrix is calculated with (12), where the side-branch acoustic inlet impedance,
Zs, is obtained from Equations (18) and (19). R = rLr . (19)

Figure 6. Helmholtz
Figure 6. resonator
Helmholtzsketch (left)sketch
resonator and example of example
(left) and TL obtained with
of TL TMM with
obtained (right).
TMM (right).

The Helmholtz resonator

The T-matrix is can be considered
calculated with (12), as where
a mass-spring system where
the side-branch aninlet
acoustic addi-impedance,
tional inductance (mass of
Zs , is obtained fromair Equations
in the neck (18)of and
the (19).
resonator) is placed between the main
duct and the cavityThe (compliance of the air can
Helmholtz resonator in theberesonator
considered volume acting as a spring).
as a mass-spring systemThewhere an ad-
Helmholtzditional
resonator is highly(mass
inductance selective and
of air provides
in the neck of anthe
important
resonator)attenuation for one the main
is placed between
duct andthat
specific frequency the cavity
depends (compliance
on the sizeofofthe airmain
the in the resonator
duct and onvolume acting as
the geometry of athe
spring). The
resonator. Helmholtz
The Helmholtz resonator is highly
resonance selectivenot
frequency, and provides an
considering theimportant
damping attenuation
material, for one
specificas:
can be computed frequency that depends on the size of the main duct and on the geometry of the
resonator. The Helmholtz resonance frequency, not considering the damping material, can
be computed as:
= s (20)
2 ( c ) Sn
fH = (20)
2π Ln ( Lc Sc )
It is worth noting that, due to the very low value reached by the side branch imped-
It is worth
ance at the resonance noting that,
frequency, the due to the
particle very low
velocity at value
the neckreached byresonator
of the the side branch
can beimpedance
extremely at the For
high. resonance frequency,
this reason, a sortthe particle velocity
of “blockage” appearsat the neck
at the of the
inlet resonator
of this can be extremely
component,
requiring ahigh. For this
“length reason, a In
correction”. sort of “blockage”
this appears pipe
case, for a circular at thewith
inletaofdiameter,
this component,
D, the requiring
neck length,a “length
Ln, can be correction”. In this case,
modified according to for a circular pipe with a diameter, D, the neck length,
[36]:
 L
Circular , can be modified according
n pipe in baffle: ΔL = 0.82 D/2; to [36]:
 • circular
Free end Circularpipe
pipeΔLin=baffle: ∆L = 0.82 D/2;
0.61 D/2.
• Free end circular pipe ∆L
A last comment concerns the effect of 0.61 D/2. the inner losses of a Helmholtz Res-
= increasing
onator (as well A aslast
of acomment concerns the
quarter-wavelength effect of When
resonator). increasing the inner
a material losses aofcer-
expressing a Helmholtz
tain flow resistivity is placed in a side branch, the effect is to slightly reduce the maximumexpressing
Resonator (as well as of a quarter-wavelength resonator). When a material
a certainat
of the attenuation flow
the resistivity
resonance isfrequency/ies,
placed in a side branch,atthe
increasing theeffect
sameistime
to slightly
the fre-reduce the
quency widthmaximum of the This
of the peak. attenuation
behavior at can
the be
resonance
used to frequency/ies, increasing
make the attenuation at the same time
capability
the frequency width of the peak. This behavior can be used
of the element less sharp, and then to make it more flexible in the reduction of noise in to make thethe
attenuation
capability of the element less sharp, and then to make it more
frequency range around the resonance peak. To take into account such behavior, the soft- flexible in the reduction of
noise in the frequency range around the resonance peak. To take into
ware can compute the effect of an absorbent layer with a flow resistivity, r, and a thick- account such behavior,
the software
ness, Lr, placed over thecan compute
neck the effect of an absorbent layer with a flow resistivity, r, and a
cross section.
thickness, Lr , placed over the neck cross section.
Designs 2022,
Designs 6, x6,FOR
2022, 37 PEER REVIEW 10 of
1022
of 22

3.3.3.3.
Outlet Impedance
Outlet ImpedanceModel
Model
ToTo
account
account for theoutlet
for the outletimpedance,
impedance,aaformulation
formulation for
for the radiation impedance,
impedance,ZZrr,, by
bySilva
Silvaet
etal.
al.for
forthe
theun-flanged
un-flangedcase
caseisisused
used[39]:
[39]:
( − ) − i − (−i ) 2
= (d − n ) − ika − d2 (−ika) (21)
Zr =2 + (1 + 1 )i + (−i ) 2 (21)
2 + (d1 + n1 )ika + d2 (−ika)
where a is the duct radius, k is the wavenumber, and the coefficients d1 = 1.393, d2 = 0.457,
where
and a is theare
n1 = 0.167 duct
theradius, k isathe
result of wavenumber,
numerical fitting.and
Thethe coefficients
radiation d1 = 1.393,
impedance, Zr, dis
2= 0.457,
then
and n
multiplied = 0.167 are the result of a numerical fitting. The radiation impedance,
1 by ρ0c/Sout to obtain the outlet impedance, ZL. Figure 7 shows the real and Z , is
r im- then
multiplied by
aginary parts of the ρ c/S
0 outlet to obtain the outlet impedance,
out impedance as a function of frequency Z L . Figure 7 shows the real
for a duct of diameter and
imaginary
45.2 mm. parts of the outlet impedance as a function of frequency for a duct of diameter
45.2 mm.

Figure 7. Real
Figure andand
7. Real imaginary parts
imaginary of outlet
parts impedance
of outlet of aof45.2
impedance mm
a 45.2 diameter
mm tube.
diameter tube.

3.4.3.4. Pressure
Pressure DropDrop Estimation
Estimation
Aerodynamic
Aerodynamic losses losses cause
cause a pressure
a pressure drop
drop when
when thethe
airair flows
flows through
through a duct.
a duct. This
This
energy
energy loss
loss reduces
reduces thethe
flowflow speed
speed andand
thethe system
system efficiency.
efficiency. Therefore,
Therefore, it should
it should be be made
made
as small as possible. A good duct and silencer design has a high noise
as small as possible. A good duct and silencer design has a high noise reduction and a low reduction and a low
pressure drop. Because the pressure drop is caused by the mean flow, we assume that thethe
pressure drop. Because the pressure drop is caused by the mean flow, we assume that
pressure
pressure drop
drop caused
caused byby side-branches
side-branches with
with zero
zero mean
mean flow
flow is negligible.
is negligible. In In
thethe expansion
expansion
chamber we have a mean flow and, hence, a possibly significant
chamber we have a mean flow and, hence, a possibly significant pressure drop. Below pressure drop. Belowa a
simple pressure drop estimation method, Darcy’s friction factor
simple pressure drop estimation method, Darcy’s friction factor method is described.method is described.
TheThe method
method divides
divides thethe pressure
pressure drop
drop in in
twotwo generation
generation mechanisms.
mechanisms. TheThe first
first is the
is the
contribution
contribution from
from thethe friction
friction force
force between
between thethe flowing
flowing medium
medium and and
thethe duct
duct wall.
wall. TheThe
second is the contribution from the flow separation appearing at discontinuities such as
second is the contribution from the flow separation appearing at discontinuities such as a
a sudden cross section area increase. The pressure drop caused by friction between the
sudden cross section area increase. The pressure drop caused by friction between the flow-
flowing medium and the duct wall is estimated with:
ing medium and the duct wall is estimated with:
L1 1 2
∆∆Pfriction== X D 2 ρ0 U (22)
(22)
2
where
whereX is
X Darcy’s friction
is Darcy’s factor,
friction UU
factor, is the mean
is the meanflow
flowspeed, L isL the
speed, pipe
is the pipelength, and
length, D is
and D is
thethe
pipe hydraulic
pipe (wet)
hydraulic diameter.
(wet) diameter.Normally,
Normally,thethe
hydraulic
hydraulicdiameter
diameter is equal to to
is equal thethe
pipe
pipe
inner diameter.
inner diameter.AtAt
a sudden
a sudden cross section
cross sectionarea increase,
area wewe
increase, estimate
estimate thethe
flow separation
flow separation
pressure drop
pressure dropwith:
with:
1
∆Pseparation = Xs1 ρ0 U12 (23)
Δ = 2 (23)
2
where the “friction” factor Xs is
Designs 2022, 6, 37 11 of 22

where the “friction” factor Xs is

 2
S
Xs = 1− 1 (24)
S2
where S1 and S2 are the cross-sectional area at the inlet and at the outlet of the sudden
cross-section increase or decrease, respectively. The pressure drop caused by a sudden
cross section area decrease is assumed to be negligible. U1 in Equation (23) is the mean
flow speed at the inlet where the cross-section area is S1 . The total pressure drop over the
n-th element is the sum of its friction and flow separation losses:

∆Pn = ∑ ∆Pn,friction + ∑ ∆Pn,separation (25)

Finally, the total pressure drop over the duct system is the sum of the contributions
from all N elements:
N
∆Ptot = ∑ ∆Pn (26)
i =1

This quantity can be useful when sizing the ventilation system. Because this work
focuses on the acoustic aspects of the design procedure, pressure drop results will not be
presented in the following.

4. Materials and Methods

4.1. Outline of the Complete Design Procedure
The TMM theory reported in Section 3 is the core of the design procedure outlined in
Figure 8, which can be roughly summarized in the following steps:
1. Definition of objectives and design constraints;
2. Identification of the frequency components to attenuate;
3. Selection of reactive elements and of their sequence;
4. Application of TMM method;
5. Verification of objective achievement in terms of SPL;
6. Verification of design constraint respect;
7. Definition of the final solution.
Steps 1–2 form the inputs to be passed to the xTMM method. Step 2 consists of
looking at the one-third octave spectrum and identifying the one-third octave bands where
noise reduction is required. Then, a second check on the narrow band sound pressure
level spectrum allows one to identify the narrow band frequencies where noise reduction
measures are needed. The narrow band frequencies are the ones to be used for dimensioning
the silencer elements.
Steps 3–6 are carried out through a manual iterative process: in particular, if design
criteria, such as size limits, or SPL objective are not fulfilled with the current trial solution,
a new element combination must be devised, and steps 3–6 must be repeated. This manual
iterative block can be replaced by an optimization algorithm to automatically provide the
best element sequence, which is beyond the scope of this work, without modifying the
procedure framework.
Step 3 is carried out by investigating the transmission loss for the available silencer
elements (quarter-wavelength resonator, Helmholtz resonator, and expansion chamber). In
this way, it is possible to make an estimate on which elements to apply, especially when it
is necessary to have a high transmission loss at more than a single frequency. In this case,
it is interesting to note that while a quarter-wavelength resonator expresses a high TL at
odd “resonance frequencies” with a high selectivity (sharp peaks), an expansion chamber
has a less sharp TL characteristic (see Figures 4 and 5, respectively). This behavior allows
one to have a good TL over a wider frequency range. The position of the frequencies with
the higher TL is usually determined by the length of the element, while the amount of
reduction for an expansion chamber is controlled by the diameter of the expansion chamber.
The real silencer system is built up using silencer elements and straight coupling pipes. The
 source and receiver impedances, thus the loudspeaker impedance must be determined.
When the IL has been calculated, the system contribution to the sound in the observation
point can be computed and compared with the requirement. The fulfillment of the SPL
objective is checked at this stage (step 6) and, if necessary, adjustments to the solution are
Designs 2022, 6, 37 made, from a simple rearrangement of the elements to the complete re-design of the 12 ofsys-
22
tem, until all the requirements are satisfied. Depending on the specific situation, pressure
drops can also be part of the acceptability criteria.
The solution
minimum distance provided
between twoin step 7 is aelements
silencer silencer that
is setsatisfies the noise attenuation
by the dimensions objec-
of the coupling
tive while respecting all the system design
elements (T-couplings, area changes, and muffs). constraints.

Figure8.8.Flowchart
Figure Flowchartshowing
showingthe
thesteps
stepstotobe
befollowed
followedduring
duringthe
thedesign
designprocess.
process.QW,
QW,HR,
HR,and
andEC
EC
indicate quarter-wavelength resonator, Helmholtz resonator, and expansion chamber, respectively.
indicate quarter-wavelength resonator, Helmholtz resonator, and expansion chamber, respectively.

When all silencer elements are created, they can be combined to build the silencer
system, and it is possible to calculate the overall properties: the length, the TL, and the
IL (step 4 in the procedure outlined above). The elements must follow a sequence for the
correct implementation of the T-matrix product, with the first element representing the
element closest to the source, and so on. The total TL is calculated, and the first check on the
dimensions is performed (step 5). As discussed in Section 3, the IL depends on the source
and receiver impedances, thus the loudspeaker impedance must be determined. When the
IL has been calculated, the system contribution to the sound in the observation point can
be computed and compared with the requirement. The fulfillment of the SPL objective is
checked at this stage (step 6) and, if necessary, adjustments to the solution are made, from a
simple rearrangement of the elements to the complete re-design of the system, until all the
requirements are satisfied. Depending on the specific situation, pressure drops can also be
part of the acceptability criteria.
The solution provided in step 7 is a silencer that satisfies the noise attenuation objective
while respecting all the system design constraints.
 Designs 2022, 6, x FOR PEER REVIEW 13 of 22

Designs 2022, 6, 37 13 of 22

4.2. Validation of the Silencing System

Once a system
4.2. Validation satisfying
of the Silencing all the requirements has been modeled, a prototype can be
System
builtOnce
for experimental
a system satisfying all theArequirements
validation. simple sketchhas
showing all the relevant
been modeled, information
a prototype can be
can be prepared at this stage, taking care of specifying all the necessary internal
built for experimental validation. A simple sketch showing all the relevant information dimen-
can
sions. The construction part must follow a certain path:
be prepared at this stage, taking care of specifying all the necessary internal dimensions.
The
1. construction
Cut the pipepart must follow
elements a certainthe
and assemble path:
silencer system according to the sketch;
2.
1. Measure the background noise;
Cut the pipe elements and assemble the silencer system according to the sketch;
2.3. Measure
Mount the thesilencer system
background in a test rig and measure the SPL frequency spectrum in
noise;
3. the observation
Mount the silencerpoint at a desired
system in a testdistance from the the
rig and measure outlet;
SPL frequency spectrum in
4. the
Measure the reference
observation point atsystem SPLdistance
a desired in the observation point;
from the outlet;
4.5. Measure
Calculatethethereference
silencer system SPLinsertion
in theloss from the measured
observation point; data.
5. Calculate the silencer
It is important to buildsystem insertion
the silencer loss such
system fromthat
the measured data.
it agrees with the silencer design
stage. An
It is accurately
important tobuilt
buildsilencer system
the silencer has asuch
system goodthat
chance to show
it agrees withgood agreement
the silencer with
design
the theoretical
stage. resultsbuilt
An accurately in the experimental
silencer systemvalidation.
has a goodFor this reason,
chance to show thegood
following speci-
agreement
fications
with must be carefully
the theoretical checked
results in for compliance
the experimental with the
validation. design
For sketch: the following
this reason,
specifications
 must be carefully
Center-to-center distanceschecked
between forside
compliance
brancheswith the design sketch: resonators,
(quarter-wavelength
• Center-to-center
and Helmholtz distances between side branches (quarter-wavelength resonators, and
resonators);
 Helmholtz resonators);
Internal expansion chamber lengths;
• Internal
Distancesexpansion chamber lengths;
from components, side branch centers, expansion chamber inner walls,
• Distances from components,
etc., to duct inlet/outlet. side branch centers, expansion chamber inner walls, etc.,
toAccording
duct inlet/outlet.
to the definition, the silencer IL can be estimated as the difference of the
SPLAccording to system
between the the definition, the silencer
with a reference ductIL can be estimated
(Lp,without as the difference
) and the system of the
with the silencer
SPL between
(Lp,with): the system with a reference duct (L p,without ) and the system with the silencer
(Lp,with ):
D IL = = ,
L p,without −− ,
L p,with (27)
(27)
InInthe
thevalidation
validationmeasurement,
measurement,thetheobservation
observationpoint
pointisisplaced
placedon
onthetheduct
ductaxis.
axis.For
For
example,ininFigure
example, Figure99the
theobservation
observationpoint
pointisisplaced
placed0.5
0.5mmfrom
fromthetheoutlet.
outlet.The
Thesource
sourceused
used
in the measurement must be the loudspeaker used in the simulations.
in the measurement must be the loudspeaker used in the simulations.

Figure 9. IL measurement outline.

Figure 9. IL measurement outline.

Theprocedure
The proceduretotofollow
followtotocalculate
calculatemeasurement-based
measurement-basedILILis:is:
1.1. Attach
Attachthe
thereference
referencepipe
pipeto tothe
theloudspeaker;
loudspeaker;
2.2. Place the microphone in the observation
Place the microphone in the observation pointpointon
onthe
theduct
ductaxis
axisatatthe
thedesired
desireddistance
distance
from the outlet;
from the outlet;
3. Measure the sound pressure, pwithout , using pink or white noise excitation signal;
 Designs 2022, 6, x FOR PEER REVIEW 14 of 22

Designs 2022, 6, 37 14 of 22

3. Measure the sound pressure, pwithout, using pink or white noise excitation signal;
4. Replace the reference pipe with the silencer duct;
4.5. Replace
Measurethe thereference
sound pressure,pipe with the
pwith silencer
, using theduct;
same signal;
5.6. Measure
Switch off thethesound
source; pressure, pwith , using the same signal;
6.7. Switch
Measure offthethebackground
source; sound pressure, pbackground;
7.8. Measure
Remove the contributionsound
the background of thepressure,
background pbackground
noise; from the measured RMS sound
8. Remove
pressuresthe forcontribution of the background noise from the measured RMS sound
both the measurements;
9. pressures ,
for
= both the
− measurements;;
9.10. pe 2 = p 2 − p 2 ;
Calculate the reference
meas,corr e meas e duct and silencer duct sound pressure levels;
background
10.
11. Calculate
Calculate the the reference
IL. duct and silencer duct sound pressure levels;
11. Calculate the IL.
5. Case Study
5.5.1.
Case Study and Application of xTMM Design Procedure
Objectives
5.1. Objectives and Application of xTMM Design Procedure
The validation of the procedure has been carried out at the University of Brescia as
partThe of anvalidation
advancedof the procedure
course addressedhas been carried
to students out at the University
and practitioners in the fieldofofBrescia as
acoustics
part of an advanced course addressed to students and practitioners
and vibration. To verify the capability of the method and the usability of the calculation in the field of acoustics
and
tool,vibration.
four groups To of verify the capability
participants were askedof thetomethod
carry outand the
the usability
complete of the calculation
procedure to design
tool, four groups
a silencer, includingof participants
selection of were asked to carry
components out the complete
and experimental procedure
validation to design
of the system. a
silencer, including selection of components and experimental validation of the system. The
The steps to follow involve designing, building, and experimentally validating a silencer
steps to follow involve designing, building, and experimentally validating a silencer setup
setup mimicking a small ventilation system located in a building. The objective was to
mimicking a small ventilation system located in a building. The objective was to reduce
reduce the sound pressure level at 0.5 m from the outlet of a pipe arrangement represent-
the sound pressure level at 0.5 m from the outlet of a pipe arrangement representing a
ing a ventilation system. In detail, the A-weighted Sound Pressure Level (SPL) at an ob-
ventilation system. In detail, the A-weighted Sound Pressure Level (SPL) at an observation
servation point placed at 0.5 m from the outlet on the duct axis must not exceed 55 dB(A).
point placed at 0.5 m from the outlet on the duct axis must not exceed 55 dB(A). The
The reason for imposing such a limit is that, during a normal conversation, the A-
reason for imposing such a limit is that, during a normal conversation, the A-weighted
weighted SPL measured at 1 m distance from the speaker is approximately 65 dB(A). By
SPL measured at 1 m distance from the speaker is approximately 65 dB(A). By requiring
requiring the ventilation system contribution to be lower than 55 dB(A), a reasonably good
the ventilation system contribution to be lower than 55 dB(A), a reasonably good sound
sound environment for a conversation is achieved [49].
environment for a conversation is achieved [49].
As the starting point, a real or simulated ventilation system sound and a calibration
As the starting point, a real or simulated ventilation system sound and a calibration
file for scalingthe
file for scaling thesound
sounddata datawereweremade
madeavailable.
available.These
Thesefiles
filescan
canbe beused
usedtotocharacterize
characterize
the basic ventilation system, consisting a straight pipe, 3
the basic ventilation system, consisting a straight pipe, 3 m long, with an inner diameterm long, with an inner diameterof
of 45.2 mm. The ventilation system has only one outlet. For
45.2 mm. The ventilation system has only one outlet. For practical implementation reasons, practical implementation rea-
itsons, it is assumed
is assumed that the that the silencer
silencer components
components can be can be placed
placed anywhere
anywhere fromfrom
0.5 m0.5 m from
from the
the inlet to 0.3 m from the outlet. Any cascade combination
inlet to 0.3 m from the outlet. Any cascade combination of silencer element types is allowed. of silencer element types is
allowed. The silencer elements are connected through
The silencer elements are connected through straight pipes, 45.2 mm diameter, meaning straight pipes, 45.2 mm diameter,
meaning
that that the
the cut-on cut-on frequency
frequency of the main of duct
the main
is 2200duct
Hz.is The
2200chosen
Hz. The chosenelements
silencer silencer are
ele-
ments are mounted in series. In the practical implementation,
mounted in series. In the practical implementation, for comparison reasons, the total length for comparison reasons, the
total
of the length of the duct,
duct, including theincluding
silencing the silencing
elements, mustelements,
be exactlymust3 mbe(from
exactly 3m
inlet to (from
outlet,inlet
see
Lto outlet,
system in see
Figure L 10).
system in Figure 10).

Figure 10. Example of cascade-coupled silencer system with three silencer elements.
Figure 10. Example of cascade-coupled silencer system with three silencer elements.

Beforeselecting
Before selectingthethecomponents
componentsableabletotoproperly
properlyattenuate
attenuatethetheSPL
SPLatatthe
theoutlet
outletand
and
designaasuitable
design suitablesilencer
silencer system,
system, thethe SPL
SPL frequency
frequency spectrum
spectrum at observation
at the the observation
pointpoint
was
was determined
determined as a signal
as a signal representing
representing a measurement
a measurement of the of the generated
noise noise generated by a ven-
by a ventilator,
tilator, together
together with aquote
with a certain certain quote band
of broad of broad band
noise. The noise.
SPL was Thederived
SPL wasby derived by post
post processing
the waveform Audio File Format, simulating the signal recorded by a microphone atby
processing the waveform Audio File Format, simulating the signal recorded thea
reference position at 0.5 m from the duct outlet. The post processing was carried out
Designs 2022, 6, x FOR PEER REVIEW 15 of 22

Designs 2022, 6, 37 15 of 22
microphone at the reference position at 0.5 m from the duct outlet. The post processing
was carried out by performing an FFT analysis of the signal and computing the one-third
octave band spectrum. The latter information is extremely useful because it allows one to
by performing an FFT analysis of the signal and computing the one-third octave band
determine which frequencies give the highest contribution. Based on the narrow band
spectrum. The latter information is extremely useful because it allows one to determine
spectrum, it is possible to make a first estimation of the dimensions of the reactive com-
which frequencies give the highest contribution. Based on the narrow band spectrum, it is
ponents by using Equations (11), (17), and (20). The signal needs to be properly averaged
possible to make a first estimation of the dimensions of the reactive components by using
to obtain a clear idea about the amplitude of the reductions. For this reason, a synthesis of
Equations (11), (17), and (20). The signal needs to be properly averaged to obtain a clear
the FFT spectrum to one-third octave band analysis allows an estimation of the broad
idea about the amplitude of the reductions. For this reason, a synthesis of the FFT spectrum
band noise and of the A-weighted sound pressure level at the observation point. In the
to one-third octave band analysis allows an estimation of the broad band noise and of the
case at hand, the
A-weighted noise
sound at the receiver
pressure position
level at the for a 3point.
observation m long straight
In the pipe
case at is shown
hand, in at
the noise
Figure 11. It is position
the receiver worth observing
for a 3 m that
long the noisepipe
straight presents some
is shown in well-defined
Figure 11. It iscomponents at
worth observing
192.5 Hz, 385 Hz, 576 Hz, 768 Hz, and 961 Hz. These are the frequencies to be
that the noise presents some well-defined components at 192.5 Hz, 385 Hz, 576 Hz, 768 Hz,targeted by
theand
implementation
961 Hz. These of are
specific reactive elements.
the frequencies to be targeted by the implementation of specific
reactive elements.

Figure 11. 11.

Figure Simulation of the
Simulation sound
of the pressure
sound level
pressure measured
level at 0.5
measured m from
at 0.5 thethe
m from outlet of aof3 am3 long
outlet m long
straight pipe.
straight pipe.

TheThe one-third
one-third octave
octavebands
bandswhere
where the A-weighted SPL
the A-weighted SPLexceeds
exceedsthe themaximum
maximum al-
allowed
lowed level of 55 dB(A) were then decremented by a “first trial” insertion
level of 55 dB(A) were then decremented by a “first trial” insertion loss of 20 dB (see loss of 20 dB
(see Table1).1).The
Table The20 20 dB
dB value
value is merely
merely an an indicator
indicatorthatthatpoints
pointsoutoutforfor
which
which bands
bands it isit is
necessary
necessary to reduce
to reducethethe
noise, andand
noise, it does notnot
it does specify thethe
specify amount
amountof reduction
of reductionrequired
required by by
thethe
individual
individual silencer
silencercomponents.
components.Once Oncethethe20-dB
20-dBIL ILisis applied
applied to thethe selected
selectedbands,
bands,the
theresulting
resultingoverall
overallsound
soundpressure
pressurelevel was
level successfully
was successfully checked
checkedto verify thatthat
to verify it was below
it was
the objective
below valuevalue
the objective of 55ofdB(A). Had not
55 dB(A). Hadthis
notbeen
thisthe
beencase,
theadditional IL should
case, additional also have
IL should
been
also have applied for the for
been applied highest remaining
the highest one-third
remaining octave octave
one-third bands. bands.
The required A-weighted
The required A-
SPL at the receiver,
weighted SPL at the receiver,L , was then calculated as:
A,req LA,req, was then calculated as:

L,A,req( ( )f )== L, A,without

( ()f − ( ()f )
) − D, IL,necessary (28)(28)
The results of the calculations are reported in Table 1. It can be observed that in this
The results of the calculations are reported in Table 1. It can be observed that in this
example it is necessary to apply a reduction at the 200 Hz, 400 Hz, 630 Hz, 800 Hz, and
example it is necessary to apply a reduction at the 200 Hz, 400 Hz, 630 Hz, 800 Hz, and
1000 Hz 1/3 octave bands.
1000 Hz 1/3 octave bands.
The estimation of the dimensions of the elements required to properly attenuate the
targeted frequency components was therefore carried out by recalling the shape and the
frequency values of the different elements, and by proceeding with heuristic reasoning.
For example, the first group of participants started from a 192.5 Hz component, where a
good attenuation can be obtained with a quarter-wavelength resonator. The length of a
resonator tuned at this frequency, assuming a speed of sound equal to 343 m/s, is 0.445 m.
Such elements can also be used to attenuate other frequencies and can be obtained by
multiplying the fundamental frequency by odd natural numbers (for n = 3, f3 = 577.5 Hz,
for n = 5, f5 = 962.5 Hz). This means that, with a single quarter-wavelength resonator, it
 needed in that band.

fn [Hz] LA, Without [dB(A)] DIL, Necessary [dB] LA at the Receiver [dB(A)]
125 37.5 0 24.9
160 38.1 0 39.5
Designs 2022, 6, 37 200 78.2 20 18.1 16 of 22
250 40.2 0 40.0
315 41.3 0 38.9
is possible
400 to cover three-out-of-five
71.3 noise components.
20 The other component 49.7 at 385 Hz
can 500
be removed by using 43.0a 0.222 m expansion chamber,0 whereas the peak 35.7 at 768 Hz is
attenuated
630 with a second, smaller
66.6 quarter-wavelength20 resonator. To summarize,
21.5 the final
implementation
800 of the system
67.5 devised by the first
20group is made by using a
38.3 0.222 m long
expansion
1000 chamber for eliminating
64.6 the 385 Hz peak,
20 a 0.112 m long quarter-wavelength
28.9
resonator to attenuate the 768 Hz component, and a 0.445 m quarter-wavelength resonator
Total level 79.7 51.2
to attenuate the other three frequencies of interest. For the sake of completeness, the TLs of
the individual elements used in this solution coincide with the graphs in Figures 4 and 5,
The estimation of the dimensions of the elements required to properly attenuate the
whereas the TL of the complete system is shown in Figure 12.
targeted frequency components was therefore carried out by recalling the shape and the
frequency values of the different elements, and by proceeding with heuristic reasoning.
Table 1. Example illustrating how the synthesized third octave band spectrum derived from Figure 11
For example, the first group of participants started from a 192.5 Hz component, where a
must be modified to achieve an overall SPL lower than 55 dB(A). The numerical value 20 dB is not a
good attenuation can be obtained with a quarter-wavelength resonator. The length of a
requirement on minimum reduction. It is merely an indicator that reduction measures are needed in
resonator
that band.
tuned at this frequency, assuming a speed of sound equal to 343 m/s, is 0.445 m.
Such elements can also be used to attenuate other frequencies and can be obtained by
multiplying the fundamental
fn [Hz] frequency
LA, Without [dB(A)] by D odd natural [dB]
IL, Necessary numbers (forLA atnthe= Receiver
3, f3 = 577.5 Hz,
[dB(A)]
for n = 5, f125
5 = 962.5 Hz). This means that, with a single quarter-wavelength resonator, it is
37.5 0 24.9
possible to160 cover three-out-of-five38.1 noise components. 0 The other component39.5 at 385 Hz can
be removed 200by using a 0.222 78.2 m expansion chamber, 20whereas the peak at 768 18.1Hz is atten-
uated with 250a second, smaller 40.2 0
quarter-wavelength resonator. To summarize, 40.0
the final im-
plementation of the system devised by the first group is made by using a38.9
315 41.3 0 0.222 m long
400 71.3 20 49.7
expansion500 chamber for eliminating 43.0
the 385 Hz peak, 0
a 0.112 m long quarter-wavelength
35.7
resonator 630to attenuate the 76866.6 Hz component, and a20 0.445 m quarter-wavelength 21.5 resonator
to attenuate800 the other three frequencies
67.5 of interest. 20For the sake of completeness,
38.3 the TLs
of the individual
1000 elements used 64.6in this solution coincide
20 with the graphs in28.9 Figures 4 and
5, whereas
Total the
levelTL of the complete79.7 system is shown in Figure 12. 51.2

Figure 12.12.
Figure TLTL
ofof
the complete
the system
complete computed
system using
computed the
using TMM.
the TMM.

5.2. Results and Discussion

In this section, the outcome of the design procedure applied to the case study described
in the previous Section is presented, together with an experimental validation. In particular,
the results refer to the solution devised by the first group, including two PhD students in
acoustics and vibration subjects, and a third-year Mechanical Engineering student. The
drawing showing the practical application of the proposed solution scheme is represented
in Figure 13.
 5.2. Results and Discussion
In this section, the outcome of the design procedure applied to the case study de-
scribed in the
In this section, the previous
outcome Section
of the is presented,
design proceduretogether
applied with
to theancase
experimental
study de- validation. In
particular, the results refer to the solution devised by the first
scribed in the previous Section is presented, together with an experimental validation.group, including
In two PhD
students in acoustics and vibration subjects, and a third-year Mechanical
particular, the results refer to the solution devised by the first group, including two PhD Engineering stu-
Designs 2022, 6, 37 dent. The schemeofis
17 22
students in acoustics anddrawing showing
vibration subjects,the
andpractical application
a third-year of theEngineering
Mechanical proposed solution
stu-
represented
dent. The drawing showing in Figure 13. application of the proposed solution scheme is
the practical
represented in Figure 13.

Figure 13. Sketch of the implementation of the silencer system proposed by the first group. The
source
Figure 13. Sketch is placed
of the at the left side
implementation and
of the the outlet
silencer at the
system right side.
proposed by the first group. The
Figure 13. Sketch of the implementation of the silencer system proposed by the first group. The
source is placed at the left side and the outlet at the right side.
sourceAsisconcerns
placed at the
the left side and thedimension
cross-section outlet at theofright
the side.
expansion chamber, it was chosen to
be equal
As concerns to 134 mm for
the cross-section practical of
dimension reasons related tochamber,
the expansion the availability from the
it was chosen tolocal shops of
pipes As
forconcerns
the the cross-section
implementation of the dimension
test rig. of the
The expansion
following stepchamber,
was to it wasthe
build chosen to
actual
be equal to 134be mm for practical
equal to 134 mm reasons related reasons
for practical to the availability
related to from the local shops
the availability fromofthe local shops
pipes for the system by cuttingofpipes
implementation
of pipes for the implementation
with
the test rig.theThe
same
of the
length of thewas
following
test rig. step
components
The followingto build used
stepthewas
during the design
actual
to build the actual
stage.pipes
system by cutting Afterwithpreparing
the the length
same 3 m long of straight
the pipe (reference
components used pipe), the
during all the parts represented
design
system by cutting pipes with the same length of the components used during the design
in Figure
stage. After preparing 13 were
the 3preparing cut. The
m long straight real silencer system was developed in collaboration with the
stage. After the 3 mpipe long(reference
straight pipepipe), all the parts
(reference pipe),represented
all the parts represented
Technical
in Figure 13 were cut. Theand Industrial
real silencer Design Laboratory of the University of Brescia, which develops
in Figure 13 were cut. Thesystem was developed
real silencer system was in collaboration
developed in with the
collaboration with the
Technical anddevices for research
Industrial Design and teaching
Laboratory of purpose
the based on
University of the “study—model”
Brescia,
Technical and Industrial Design Laboratory of the University of Brescia, which developswhich developsapproach typical
of industrial
devices for research and design product based development [50].
devices forteaching
researchpurpose
and teaching purpose on the “study—model” approach typical
based on the “study—model” approach typical
of industrial design Being
product cheap, effective, quick
development [50]. to build, and based on commercial parts were key re-
of industrial design product development [50].
quirements
Being cheap, effective, forquick
this application.
to build, and The arrangement was therefore conceived as a modular
Being cheap, effective, quick tobased
build,on andcommercial parts were
based on commercial key re-
parts were key require-
system
quirements forments based
this application. on push-fit PVC
The arrangement wastewater pipes.
was therefore The straight
conceived parts were
as a modular the only custom-
for this application. The arrangement was therefore conceived as a modular system
system based onized pieces
push-fit and were
PVC wastewatercut to well-defined
pipes. The sizes.
straight For the construction of the expansion cham-
based on push-fit PVC wastewater pipes. Theparts
straightwereparts
the only
werecustom-
the only customized
ized pieces andbers,
were
pieces a and
range
cut to of pipes
cut towith
well-defined
were different
sizes.
well-defined diameters
Forsizes.
the and
construction
For the lengths was
of the expansion
construction provided.
of the cham-Thechambers,
expansion end of thea
cylinder
bers, a range of pipes and
with the piston
different body
diameterswere made
and of
lengthsplywood,
was and
provided.
range of pipes with different diameters and lengths was provided. The end of the an O-ring
The end was
of placed
the in acylinder
recess
of
cylinder and theandthe piston
piston to
body were
the piston ensure
bodymade airtightness.
were of madeplywood, The inlet and
and anand
of plywood, the
O-ring outlet pipes
was placed
an O-ring were
was in madein a recess ofmm
a recess
placed using 50 the
external
of the piston topiston diameter
ensuretoairtightness. pipe
ensure airtightness. terminations,
The inletThe andinlet as illustrated
the outlet
and the pipes
outletin
were Figure
made
pipes 14.
were Quarter-wavelength
using
made 50 using
mm 50 mm exter- res-
onators
external diameter pipewere
nal diameter made
terminations, using asaillustrated
pipe terminations, T-branch and a vertical
in Figure
as illustrated in14. pipe.14.Plywood
Quarter-wavelength
Figure pistons
Quarter-wavelengthres-with resonators
an O-ring
seal
wereand
onators were made madea rod
using usingfitted
a T-branch with
andaahandle
a T-branch vertical
and for adjusting
pipe.
a vertical Plywood thepistons
pipe. Plywood position of the
with
pistons an sliding
O-ring
with piston
an O-ring sealwere
and
integrated
seal and a rodafitted
rod fitted in
withwith the design
a handle
a handle of
for forthe components.
adjusting
adjusting thetheposition
positionof ofthethe
sliding
slidingpiston
pistonwere
were integrated in
integrated in the
thedesign
designof
ofthe
thecomponents.
components.

Figure 14. Cross section showing an example of the expansion chamber structure.
Figure 14. Cross section showing an example of the expansion chamber structure.
Figure
Figure 14. Cross section 15 shows
showing the system
an example assembled
of the expansion beforestructure.
chamber the tests. To avoid flanking transmis-
sions,Figure 15 shows
the source was the system
enclosed inassembled
a wooden boxbefore thewith
filled tests.absorbing
To avoid material.
flanking transmis-
The pipes
sions,
Figure 15andshows the source
thethe
other was enclosed
components
system assembledwerein a wooden
suspended
before box filled
using
the tests. with
wooden
To avoid absorbing material.
legs. transmis-
flanking On the top parts The pipes
of the
and
pipes
sions, the source the
was other
used
enclosedcomponents
to build
in athe were
boxsuspended
quarter-wavelength
wooden using wooden
filled withresonators,
absorbing the legs. On
handles
material. Theforthe top parts
moving
pipes of the
the pistons
and the otherpipes
are used to
clearly
components build
were the
visible. quarter-wavelength
suspended using wooden resonators,
legs. Onthe thehandles
top partsforofmoving
the the pistons
are clearly
To visible.
pipes used to build the quarter-wavelength resonators, the handles for moving the pistons analyzer was
measure the SPL at 0.5 m from the outlet, an OROS Type OR36
connected to a Bruel & Kjaer Type 4189 microphone. The measurement chain was calibrated
are clearly visible.
using a Bruel & Kjaer pistonphone Type 4228. The loudspeaker was fed with a broadband
white noise ranging from 50 Hz to 5000 Hz.
A series of measurements were made in the reference situation (3 m long straight
pipe) and on the complete system to evaluate IL. In detail, the background noise, the noise
coming from the outlet using the straight pipe, and the noise coming from the outlet when
the silencer system is installed were evaluated at the receiver position. The results of these
measurements are shown in Figure 16. It is worth noting that, because the pink noise
Designs 2022, 6, 37 18 of 22

Designs 2022, 6, x FOR PEER REVIEW 18 of 22

is generated starting from 50 Hz, below such frequency there is only a slight difference
between the three measurements.

Figure 15. Practical implementation of the sketch reported in Figure 14. The source is placed at the
right side and the outlet at the left side.

To measure the SPL at 0.5 m from the outlet, an OROS Type OR36 analyzer was con-
nected to a Bruel & Kjaer Type 4189 microphone. The measurement chain was calibrated
using a Bruel & Kjaer pistonphone Type 4228. The loudspeaker was fed with a broadband
white noise ranging from 50 Hz to 5000 Hz.
A series of measurements were made in the reference situation (3 m long straight
pipe) and on the complete system to evaluate IL. In detail, the background noise, the noise
coming from the outlet using the straight pipe, and the noise coming from the outlet when
the silencer system is installed were evaluated at the receiver position. The results of these
measurements are shown in Figure 16. It is worth noting that, because the pink noise is
generated starting from 50 Hz, below such frequency there is only a slight difference be-
Figure
tween
Figurethe15. Practical
three
15. implementationofofthe
measurements.
Practical implementation thesketch
sketchreported
reportedininFigure
Figure14.
14.The
Thesource
sourceisisplaced
placedatatthe
the
right side and the outlet at the left side.
right side and the outlet at the left side.

To measure the SPL at 0.5 m from the outlet, an OROS Type OR36 analyzer was con-
nected to a Bruel & Kjaer Type 4189 microphone. The measurement chain was calibrated
using a Bruel & Kjaer pistonphone Type 4228. The loudspeaker was fed with a broadband
white noise ranging from 50 Hz to 5000 Hz.
A series of measurements were made in the reference situation (3 m long straight
pipe) and on the complete system to evaluate IL. In detail, the background noise, the noise
coming from the outlet using the straight pipe, and the noise coming from the outlet when
the silencer system is installed were evaluated at the receiver position. The results of these
measurements are shown in Figure 16. It is worth noting that, because the pink noise is
generated starting from 50 Hz, below such frequency there is only a slight difference be-
tween the three measurements.

Figure
Figure16.16.
SPL spectra
SPL measured
spectra measuredat at
0.50.5
mm from
fromthethe
outlet forfor
outlet thethe
background noise
background (blue),
noise 3m
(blue), long
3m long
straight pipe (yellow), and the silencer system (orange).
straight pipe (yellow), and the silencer system (orange).

The
Themeasured
measuredSPLs
SPLswithout
without and
andwith thethe
with silencer system
silencer system were
werecorrected
corrected to to
consider
consider
thethe
background
backgroundnoise,
noise,thus
thusititwas
waspossible
possible to
to compute
compute the IL.IL. Figure
Figure1717shows
showsaacomparison
compar-
between
ison measured
between measuredandand
predicted curves.
predicted Below
curves. 50 Hz,
Below 50 the
Hz,effect of theofsilencer
the effect system
the silencer
can hardly be spotted, but above this frequency the prediction follows quite well the
experimental data. The discrepancy around 700 Hz and 750 Hz between the predicted and
the measured IL may be due to the noise radiated by the low-density commercial pipes used
for the experiments (0.27 kg/m). For the same reason, as can be seen in Figure 15, there are
some frequency components in the measured SPL for the complete system (which is longer
than the basic one) that are higher than the ones measured for the 3 m long straight pipe.

Figure 16. SPL spectra measured at 0.5 m from the outlet for the background noise (blue), 3 m long
straight pipe (yellow), and the silencer system (orange).

The measured SPLs without and with the silencer system were corrected to consider
the background noise, thus it was possible to compute the IL. Figure 17 shows a compar-
ison between measured and predicted curves. Below 50 Hz, the effect of the silencer
 and the can
system measured
hardly ILbemay be due
spotted, but to the noise
above radiated the
this frequency by the low-density
prediction commercial
follows quite well
pipes
the experimental data. The discrepancy around 700 Hz and 750 Hz between the in
used for the experiments (0.27 kg/m). For the same reason, as can be seen Figure
predicted
15, there are some frequency components in the measured SPL for the complete
and the measured IL may be due to the noise radiated by the low-density commercial system
(which is longer
pipes used than
for the the basic one)
experiments (0.27that are For
kg/m). higher than the
the same onesas
reason, measured for in
can be seen theFigure
3m
long straight
15, there arepipe.
some frequency components in the measured SPL for the complete system
Designs 2022, 6, 37 19 of 22
(which is longer than the basic one) that are higher than the ones measured for the 3 m
long straight pipe.

Figure 17. Comparison between the predicted IL (red) and the measured IL (blue).

Applying
Figure the obtained IL,thethe resulting (red)
A-weighted SPL at the (blue).
receiver is equal to
Figure17.
17.Comparison
Comparisonbetween
between thepredicted
predicted IL
IL (red) and
and the
the measured
measured IL
IL (blue).
51.2 dB(A), and thus the requirement of SPL < 55 dB(A) is fulfilled. In Figure 18, the pre-
dictedApplying
SPL spectrum
Applying the at the receiver
the obtained
obtained IL,
IL, the is plottedA-weighted
the resulting
resulting in narrow band
A-weighted SPL
SPLat and
at theinreceiver
the one-third
receiver octaveto
isis equal
equal to
bands.
51.2
51.2 dB(A),
dB(A), and
and thus
thus the
the requirement
requirementof ofSPL
SPL<<5555dB(A)
dB(A)isisfulfilled.
fulfilled.InInFigure
Figure18,
18,thethepre-
pre-
dicted
dicted SPL spectrum
spectrumatatthe
thereceiver
receiver is plotted
is plotted in narrow
in narrow bandband and
and in in one-third
one-third octaveoctave
bands.
bands.

Figure 18. Resulting SPL narrow band and one-third octave band spectra at the receiver. Blue lines:
Figure 18. Resulting SPL narrow band and one-third octave band spectra at the receiver. Blue lines:
without silencingsystem.
without silencing system.Red
Redlines:
lines:with
withsilencing
silencingsystem.
system.

FigureAs
18.aResulting SPL narrow
final remark, it is band
worth and one-third
noting thatoctave bandgroups
all four spectra involved
at the receiver. Blue
in the lines:
project
As asilencing
without final remark, itRed
system. is worth
lines: noting
with that allsystem.
silencing four groups involved in the project man-
managed to successfully design their silencing system by proposing different solutions. All
aged to successfully design their silencing system by proposing different solutions. All
participants provided positive feedback on the intuitiveness of the procedure and on the
participants
As a of provided
final positive feedback on theall
intuitiveness of the procedure and onman-
the
usability theremark, it is worth
tool, showing thatnoting that
the method canfour
be groups involved
efficiently appliedinin
the project
real contexts.
usability
aged to of the tool, showing
successfully that the
design their methodsystem
silencing can bebyefficiently
proposingapplied in real
different contexts.All
solutions.
participants
6. Conclusionsprovided positive feedback on the intuitiveness of the procedure and on the
usability
The of the coming
noise tool, showing that the method
from ventilation can
systems bebe
can efficiently applied
an annoying infor
issue realapplications
contexts.
requiring high quality standards. While it is usually relatively simple to find a solution for
the noise propagating above the cut-on frequency, below this limit it is necessary to use
reactive components that can be represented through an electro-acoustic analogy. However,
the evaluation is generally carried out in terms of sound transmission loss of the system,
which is not immediately relatable to human perception.
This article has shown how the classic transfer matrix method theory for the prediction
of a reactive silencer’s transmission loss can be expanded to allow the direct evaluation of
sound pressure level at a given distance from the outlet. A self-developed tool with possible
academic and professional application has been used to demonstrate the potentiality of
the method and validate it against experimental results by four groups of students and
practitioners attending an advanced course in acoustics and vibration. The outcomes of this
study are two-fold: on the one hand, the method, though relying upon relatively simple
Designs 2022, 6, 37 20 of 22

models of components and acoustic loss factors, provided noise reduction results close
to the experimental evidence; on the other hand, the implementation of this expanded
framework into a user-friendly tool proved to be an effective way to foster noise-oriented
solutions and contribute to creating comfortable environments. Possible future develop-
ments include the implementation of optimization strategies to automatize the choice of
the components, their characteristic dimensions, and their sequence in order to obtain the
desired noise reduction performances with given design constraints.

Author Contributions: Conceptualization, U.E.C. and S.B.; methodology, U.E.C. and S.B.; software,
U.E.C., E.A.P. and S.B.; validation, E.A.P. and U.E.C.; formal analysis, S.B.; investigation, E.A.P.
and D.P.; resources, E.A.P., A.M.L. and D.P.; data curation, E.A.P. and U.E.C.; writing—original
draft preparation, E.A.P.; writing—review and editing, U.E.C. and A.M.L.; visualization, D.P. and
E.A.P.; supervision, A.M.L. and S.B. All authors have read and agreed to the published version of
the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Song, H.; Cho, H. Flow Characteristics and Noise Reduction Effects of Air Cleaners of Automobile Intake Systems with Built-in
Resonators with Space Efficiency. J. Eng. Res.-Kuwait 2021, 9, 294–302. [CrossRef]
2. Bae, M.-W.; Ku, Y.J.; Park, H.-S. A study on effects of tuning intake and exhaust systems upon exhaust noises in a driving car of
gasoline engine. T. Korean Soc. Mech. Eng. A 2021, 45, 31–40. [CrossRef]
3. Dahlan, A.A.; Muhamad Said, M.F.; Latiff, Z.; Mohd Perang, M.R.; Abu Bakar, S.A.; Abdul Jalal, R.I. Acoustic Study of an Air
Intake System of SI Engine Using 1-Dimensional Approach. Int. J. Automot. Mech. Eng. 2019, 16, 6281–6300. [CrossRef]
4. Yakar, Ö.; Erol, H. Acoustical System Modelling of the Heavy Truck Air Intake System. Int. J. Heavy Veh. Syst. 2019, 26, 158–174.
[CrossRef]
5. Nadareishvili, G.G.; Galevko, V.V.; Rakhmatov, R.I. Improvement of Acoustic Characteristics of Motor Vehicle Intake System
Based on Calculation and Experimental Research. In International Conference on Industrial Engineering; Lecture Notes in Mechanical
Engineering Book Series; Springer: Cham, Swizerland, 2019; pp. 2311–2322. [CrossRef]
6. Vidal, V.; Mann, A.; Verriere, J.; Kim, M.; Ailloud, F.; Henner, M.; Cheriaux, O. Towards a Quiet Vehicle Cabin Through Digitalization of
HVAC Systems and Subsystems Aeroacoustics Testing and Design; SAE Technical Paper; SAE International: Pittsburgh, PA, USA, 2019.
[CrossRef]
7. Caniato, M.; Bettarello, F.; Schmid, C.; Fausti, P. The Use of Numerical Models on Service Equipment Noise Prediction in
Heavyweight and Lightweight Timber Buildings. Build. Acous. 2019, 26, 35–55. [CrossRef]
8. Borelli, D.; Gaggero, T.; Rizzuto, E.; Schenone, C. Onboard Ship Noise: Acoustic Comfort in Cabins. Appl. Acoust. 2021,
177, 107912. [CrossRef]
9. Han, H.; Lee, C.; Lee, K.; Jeon, S.H.; Park, S. Estimation of the Sound in Ship Cabins Considering Low Frequency and Flowing
Noise Characteristics of HVAC Duct. Appl. Acoust. 2018, 141, 261–270. [CrossRef]
10. Borelli, D.; Schenone, C.; Di Paolo, M. Application of a Simplified “Source-Path-Receiver” Model for HVAC Noise to the
Preliminary Design of a Ship: A Case Study. In Proceedings of the 21st International Congress on Sound and Vibration
2014, ICSV 2014, Beijing, China, 13–17 July 2014; International Institute of Acoustics and Vibrations: Beijing, China, 2014;
Volume 2, pp. 1257–1264.
11. Altabey, W.A. An Exact Solution for Acoustic Simulation Based Transmission Loss Optimization of Double-Chamber Silencer.
Sound. Vib. 2020, 54, 215–224. [CrossRef]
12. Ray, E.F. Aerodynamics of Ducted Systems and Flow Affects upon Silencers, Combustion Turbines, and Machine Performance. In
Proceedings of the 41st International Congress and Exposition on Noise Control Engineering 2012, New York, NY, USA, 19–22
August 2012; INTER-NOISE 2012. Volume 4, pp. 3080–3091.
13. Yadav, S.; Sharma, G.S. Noise Reduction Techniques for FD & ID Fan in Thermal Power Plant Using Chamber Based Silencer. Int.
J. Mech. Prod. Eng. Res. Dev. 2019, 9, 409–420. [CrossRef]
14. Mimani, A.; Kirby, R. Design of Large Reactive Silencers for Industrial Applications. In Proceedings of the INTER-NOISE
2018—47th International Congress and Exposition on Noise Control Engineering: Impact of Noise Control Engineering, Chicago,
IL, USA, 26–29 August 2018; Institute of Noise Control Engineering: Chicago, IL, USA, 2018.
15. Papini, G.S.; Pinto, R.L.U.F.; Medeiros, E.B.; Coelho, F.B.G. Hybrid Approach to Noise Control of Industrial Exhaust Systems.
Appl. Acoust. 2017, 125, 102–112. [CrossRef]
Designs 2022, 6, 37 21 of 22

16. Limacher, P.; Spinder, C.; Banica, M.C.; Feld, H.-J. A Robust Industrial Procedure for Measuring Modal Sound Fields in the
Development of Radial Compressor Stages. J. Eng. Gas Turb. Power 2017, 139, 062604. [CrossRef]
17. Khelladi, S.; Kouidri, S.; Bakir, F.; Rey, R. Predicting Tonal Noise from a High Rotational Speed Centrifugal Fan. J. Sound Vib. 2008,
313, 113–133. [CrossRef]
18. Velarde-Suárez, S.; Ballesteros-Tajadura, R.; Pablo Hurtado-Cruz, J.; Santolaria-Morros, C. Experimental Determination of the
Tonal Noise Sources in a Centrifugal Fan. J. Sound Vib. 2006, 295, 781–796. [CrossRef]
19. De Laborderie, J.; Moreau, S. Prediction of Tonal Ducted Fan Noise. J. Sound Vib. 2016, 372, 105–132. [CrossRef]
20. Holewa, A.; Weckmüller, C.; Guérin, S. Impact of Bypass Duct Bifurcations on Fan Noise. J. Propul. Power 2014, 30, 143–152.
[CrossRef]
21. Kabral, R.; Du, L.; Åbom, M.; Knutsson, M. A Compact Silencer for the Control of Compressor Noise. SAE Int. J. Engines 2014, 7,
1572–1578. [CrossRef]
22. Figurella, N.; Dehner, R.; Selamet, A.; Tallio, K.; Miazgowicz, K.; Wade, R.; Karim, A.; Keller, P.; Shutty, J. Effect of Inlet Guide
Vanes on Centrifugal Compressor Acoustics and Performance. Noise Control Eng. J. 2014, 62, 232–237. [CrossRef]
23. Nagarhalli, P.V.; Maurya, A.; Kalsule Ceng, S.; Titave, U. Optimizing an Automotive HVAC System for Enhancement of Acoustic
Comfort. In Proceedings of the SAE Technical Papers; SAE International: Jodhpur, India, 22 September 2021.
24. Xue, F.; Sun, B. Experimental Study on the Comprehensive Performance of the Application of U-Shaped Corrugated Pipes into
Reactive Mufflers. Appl. Acoust. 2018, 141, 362–370. [CrossRef]
25. Harvie-Clark, J.; Chilton, A.; Conlan, N.; Trew, D. Assessing Noise with Provisions for Ventilation and Overheating in Dwellings.
Build. Serv. Eng. Res. Technol. 2019, 40, 263–273. [CrossRef]
26. Caniato, M.; Schmid, C.; Gasparella, A. A Comprehensive Analysis of Time Influence on Floating Floors: Effects on Acoustic
Performance and Occupants’ Comfort. Appl. Acoust. 2020, 166, 107339. [CrossRef]
27. Bettarello, F.; Fausti, P.; Baccan, V.; Caniato, M. Impact Sound Pressure Level Performances of Basic Beam Floor Structures. Build.
Acous. 2010, 17, 305–316. [CrossRef]
28. Caniato, M.; Marzi, A.; Monteiro da Silva, S.; Gasparella, A. A Review of the Thermal and Acoustic Properties of Materials for
Timber Building Construction. J. Build. Eng. 2021, 43, 103066. [CrossRef]
29. Caniato, M.; Cozzarini, L.; Schmid, C.; Gasparella, A. Acoustic and Thermal Characterization of a Novel Sustainable Material
Incorporating Recycled Microplastic Waste. Sustain. Mat. Technol. 2021, 28, e00274. [CrossRef]
30. Schönfeld, J.C. Analogy of Hydraulic, Mechanical, Acoustic and Electric Systems. Appl. Sci. Res. 1954, 3, 417–450. [CrossRef]
31. Bodén, H.; Åbom, M. Modelling of Fluid Machines as Sources of Sound in Duct and Pipe Systems. Acta Acust. 1995, 3, 549–560.
32. Glav, R.; Åbom, M. A General Formalism for Analyzing Acoustic 2-Port Networks. J. Sound Vib. 1997, 202, 739–747. [CrossRef]
33. Elnady, T.; Åbom, M. SIDLAB: New 1D Sound Propagation Simulation Software for Complex Duct Networks. In Proceedings of
the Thirteenth International Congress on Sound and Vibration, Vienna, Austria, 2–6 July 2006; Volume 5, pp. 4262–4269.
34. Caniato, M. Sound Insulation of Complex Façades: A Complete Study Combining Different Numerical Approaches. Appl. Acoust.
2020, 169, 107484. [CrossRef]
35. Bilawchuk, S.; Fyfe, K.R. Comparison and Implementation of the Various Numerical Methods Used for Calculating Transmission
Loss in Silencer Systems. Appl. Acoust. 2003, 64, 903–916. [CrossRef]
36. Åbom, M. An Introduction to Flow Acoustics; KTH: Stockholm, Sweden, 2010.
37. Wallin, H.P.; Carlsson, U.; Åbom, M.; Bodén, H.; Glav, R.; Hildebrand, R. Sound and Vibration; KTH: Stockholm, Sweden, 2014.
38. Boij, S.; Abom, M.; Pieters, R. Reflection Properties of a Flow Pipe with a Small Angle Diffuser Outlet. In Proceedings of the 17th
International Congress on Sound and Vibration 2010, ICSV 2010, Cairo, Egypt, 18–22 July 2010; Volume 1, pp. 706–712.
39. Silva, F.; Guillemain, P.; Kergomard, J.; Mallaroni, B.; Norris, A.N. Approximation Formulae for the Acoustic Radiation Impedance
of a Cylindrical Pipe. J. Sound Vib. 2009, 322, 255–263. [CrossRef]
40. Elnady, T.; Elsaadany, S.; Herrin, D.W. Investigation of the Acoustic Performance of After Treatment Devices. SAE Int. J. Passeng.
Cars Mech. Syst. 2011, 4, 1068–1075. [CrossRef]
41. Elsahar, W.; Elnady, T. Measurement and Simulation of Two-Inlet Single-Outlet Mufflers. SAE Int. J. Passeng. Cars Mech. Syst.
2015, 8, 1026–1033. [CrossRef]
42. Piana, E.A.; Uberti, S.; Copeta, A.; Motyl, B.; Baronio, G. An Integrated Acoustic–Mechanical Development Method for off-Road
Motorcycle Silencers: From Design to Sound Quality Test. Int. J. Interact. Des. Manuf. 2018, 12, 1139–1153. [CrossRef]
43. Jiménez, N.; Romero-García, V.; Pagneux, V.; Groby, J.-P. Rainbow-Trapping Absorbers: Broadband, Perfect and Asymmetric
Sound Absorption by Subwavelength Panels for Transmission Problems. Sci. Rep. 2017, 7, 13595. [CrossRef] [PubMed]
44. Long, H.; Cheng, Y.; Liu, X. Asymmetric Absorber with Multiband and Broadband for Low-Frequency Sound. Appl. Phys. Lett.
2017, 111, 143502. [CrossRef]
45. Boulvert, J.; Humbert, T.; Romero-García, V.; Gabard, G.; Fotsing, E.R.; Ross, A.; Mardjono, J.; Groby, J.-P. Perfect, Broadband, and
Sub-Wavelength Absorption with Asymmetric Absorbers: Realization for Duct Acoustics with 3D Printed Porous Resonators. J.
Sound Vib. 2022, 523, 116687. [CrossRef]
46. Long, H.; Shao, C.; Cheng, Y.; Tao, J.; Liu, X. High Absorption Asymmetry Enabled by a Deep-Subwavelength Ventilated Sound
Absorber. Appl. Phys. Lett. 2021, 118, 263502. [CrossRef]
47. Červenka, M.; Bednařík, M.; Groby, J.-P. Optimized Reactive Silencers Composed of Closely-Spaced Elongated Side-Branch
Resonators. J. Acoust. Soc. Am. 2019, 145, 2210–2220. [CrossRef]
Designs 2022, 6, 37 22 of 22

48. Stinson, M.R. The Propagation of Plane Sound Waves in Narrow and Wide Circular Tubes, and Generalization to Uniform Tubes
of Arbitrary Cross-sectional Shape. J. Acoust. Soc. Am. 1991, 89, 550–558. [CrossRef]
49. Rusnock, C.F.; Bush, P.M. Case Study: An Evaluation of Restaurant Noise Levels and Contributing Factors. J. Occup. Environ.
Hyg. 2012, 9, D108–D113. [CrossRef]
50. Paderno, D.; Bodini, I.; Villa, V. Proof of Concept as a Multidisciplinary Design-Based Approach. In International Conference of the
Italian Association of Design Methods and Tools for Industrial Engineering; Lecture Notes in Mechanical Engineering; Springer: Cham,
Swizerland, 2020; pp. 625–636. [CrossRef]