Applied Acoustics 166 (2020) 107356

Contents lists available at ScienceDirect

Applied Acoustics
journal homepage: www.elsevier.com/locate/apacoust

Extensive investigation of multiphysics approaches in simulation of

complex periodic structures
Marco Caniato a,⇑, Federica Bettarello b, Paolo Bonfiglio c, Andrea Gasparella a
a
Free University of Bozen, Faculty of Science and Technology, Bozen, Italy
b
University of Trieste, Engineering and Architecture Department, Trieste, Italy
c
Materiacustica srl, Ferrara, Italy

a r t i c l e i n f o a b s t r a c t

Article history: The acoustic performance of complex periodic structures is of paramount importance in sustainable con-
Received 25 February 2020 structions. Nevertheless, traditional numerical methodologies like progressive impedance models,
Received in revised form 30 March 2020 double-leaf theory or standard methodologies cannot offer reliable results, comparable to measurements.
Accepted 31 March 2020
Even complete modelling using Finite Element (FE) simulations may not provide good results, due to the
calculation of many variables and the consequent error propagation. For these reasons, in this paper an
extensive investigation of numerical approaches accuracy is proposed and applied to complex periodic
Keywords:
timber structures.
Transfer matrix method (TMM)
Simplified FEM
Transfer Matrix Modelling (TMM) method and simplified Finite Element simulations were used and
Timber buildings compared to traditional methodologies. In order to obtain reliable results, a complete material and build-
Porous materials ing element characterization is carried out.
Thick insulated walls Both transmission loss and radiation efficiency were computed in order to understand advantages and
limitations of all these methods. Results clearly show that Transfer Matrix Method combined to simpli-
fied Finite Element analysis is the best method as regards to accuracy of results and that is able to cor-
rectly compute these two parameters as a function of frequency.
Ó 2020 Elsevier Ltd. All rights reserved.

1. Introduction However, such construction cannot be approached in the same

way as traditional heavyweight ones because (i) the realization
Recently, the industrial production of prefabricated elements technologies are profoundly different, (ii) the design follows differ-
has become a major part of civil construction. The quality of the ent lines of thought and (iii) the attainable performances can be
manufactured elements rose in time for the use of these kind of different, especially in the acoustic domain [4]. For example, the
assemblies. In addition, their cost can be reasonable thanks to results of the COST action FP90702 [5] led to the conclusion that
the standardization of built elements as well as the efficiency in complex periodic structures provide better acoustic insulation in
the use of their components [1]. Many structures can be found in mid- and high-frequency, but worse in low-frequency than tradi-
buildings such as prefabricated walls and floors, roofs, ventilated tional heavy ones. In this perspective, also the design and perfor-
façades and metal or wooden construction [2]. In the latter case, mance tools must evolve and adapt.
commonly entire buildings can be made up of complex periodic Transmission loss and radiation efficiency are nowadays two of
structures, such as timber frame partitions [3]. the most used parameters [6] to characterize the acoustic proper-
Furthermore, in recent years the number of wooden buildings ties of a partition (wall or floor). The former is related to the capa-
has been growing, thanks to a change in the awareness of design- bility to reduce the sound power of an incident wave. To this end,
ers, builders and end users. Timber buildings can be cost effective the sound transmission loss (TL), as a function of the frequency and
and store CO2. Additionally, they are based on a sustainable, as a single number rating, is a parameter universally utilized to
renewable and non-polluting material, neither for production, compare the performances of walls and floors.
nor for final disposal. Radiation efficiency is an indicator able to explain the link
between the structural velocities of a certain partition. Radiation
efficiency is getting more interest mainly within the context of
analytical models for the prediction of sound transmission loss.
⇑ Corresponding author.

https://doi.org/10.1016/j.apacoust.2020.107356
0003-682X/Ó 2020 Elsevier Ltd. All rights reserved.
2 M. Caniato et al. / Applied Acoustics 166 (2020) 107356

Available methodologies for acoustic simulation of transmission The last method deals with an analytical matrix approach,
loss and radiation efficiency can be summarized as follows: where many dedicated models could be introduced (when needed)
and solved in an all-inclusive model.

Empirical or Semi-empirical Standardized Models (ESSM) More specifically, TMM generally solves a two-dimensional

Statistical Energy Analysis (SEA) problem related to the impact of a flat acoustic wave on the surface

Finite Element Method (FEM) of a structure composed of two or more layers. In general, the

Progressive Impedance (PI) method method can be described analytically as follows in Eq. (3):

Transfer Matrix Method (TMM).
VðS1 Þ ¼ ½T V ðS2 Þ ð3Þ
The first ones (ESSM) are related to single partition composi- The vector V(S1) represents all the variables necessary to define
tions and could not include geometrical or layering variations the acoustic indicators (pressure, stresses, velocity, etc.) present on
[6]. The unique general model is provided by ISO 12354-1 standard the surface S1, while the vector V(S2) contains the same descriptors
[7,8]. This model is based on the transmission coefficient s, which for the surface S2. Elements of matrix T depend on physical and
depends on the dimensions of the partition, on the forced wave mechanical parameters, related to each specific layer.
radiance factor, on the free wave radiance factor, on the critical fre- In other words, the transfer matrix [T] describes the full trans-
quency, on the vibration velocity and on the radiated sound power. mission of sound waves through the layered structure. The size of
All these parameters are a function of complex laboratory mea- this matrix depends on the nature of each layer, such as solid, fluid,
surements and thus they are not generally available. poroelastic or viscoelastic. The materials within the various cou-
Regarding the second and third ones, recent works managed to pled layers are assumed in the first instance to be infinite in the lat-
solve the problem using SEA [9–12] or FEM analysis [13–14] trying eral parts, but, in this way, there could be significant differences
to cover other possibilities, but with very limited results and between the measured values and those simulated, especially at
applications. low frequency. This has been avoided by using the window with
The finite element method (FEM) is the most commonly used finite limitations [20].
numerical prediction technique for solving engineering problems. Assuming that the layered structure is enveloped in a semi-
It solves the distribution of one (or several) field variable(s) in a infinite fluid on both sides, it is possible to describe the relation-
continuous domain, governed by an appropriate (set of) partial dif- ship between the complex transmission T and the complex reflec-
ferential equation(s) and boundary conditions. tion coefficient R as described in Eq. (4):
Quite recently [15], new simplified FEM approaches have been
proposed to reduce significantly computation time for periodic det½ðDn þ 1Þ
T ¼ ð1 þ RÞ ð4Þ
structures. In practice, if the geometry (for example a wall) can det½D1
be considered as a combination and repetition of the same unity
(usually called ‘‘unit cell”), numerical simulations are carried out being [Dn+1] and D1 matrices obtained from a complete matrix D
only on the unit cell using specific boundary conditions (Bloch- (combination of transfer matrix of each layer, coupling matrices
Floquet conditions for instance), applied to the lateral boundaries and proper boundary conditions) once columns n + 1 and 1 have
of the unit cell itself. been deleted.
The fourth one (PI theory [16]) generalizes the problem, starting The complex reflection coefficient R in Eq. (4) is defined as fol-
from the materials properties and including the 2D wave propaga- lows in Eq. (5):
tion. This method defines the surface acoustic impedance parame-
ter defined as follows in Eq. (1): Z s cos h  Z 0
R¼ ð5Þ
Z s cos h þ Z 0
iz2 cot ðkc lÞ þ zc
z1 ¼ z c ð1Þ where Z0 = q0 c0 represents the characteristic impedance of the
z2  izc cot ðkc lÞ
fluid. This is a function of density q0 and speed of sound c0.
where z1 and z2 are the surface impedances at respectively x = 0 and Zs represents the surface impedance at the interface between S1
x = l of a layer, Zc is the complex impedance for dissipating materi- domain and Layer 1 and can be calculated as Eq. (6):
als and kc is the complex wave number.
det½D1 
If this model is applied to multi-layered systems, Eq. (1) is used Zs ¼  ð6Þ
det½D2
for each individual interface and generalized in the following Eq.
(2): Finally, values of transmission coefficient are averaged over all

 possible incidence angles and diffuse field sound transmission loss
i
izi1 cot kc li þ zic is calculated.
zi ¼ z c
 ð2Þ For these reasons, more effective methods, such as TMM, sim-
i
zi1  iz cot kc li plified FEM or a combination of them, need to be studied and val-
idated [21]. These methods have the advantage of not having to
where li is the thickness of every element constituting the completely redesign the geometry of the sample and of being able
composition. to include dedicated analytical methods focused on specific struc-
This method is widely utilized for systems involving porous or tures [22]. TMM approach has already been applied to metal
fibrous materials and in general for composite systems [17], used frame-structures and also to plasterboard double leaves walls
in the field of sound absorption. However, when the method is and massive partitions [23,24].
applied to transmission loss, the results are not always reliable TMM was successfully applied to (i) multi-layered infinite plate
because they tend to under or overestimate the impedance for [22], (ii) finite stiffened and orthotropic structures lined by porous
elastic materials such as wood, metals, etc. Furthermore, it fails materials in a multi-layered configuration [25], (iii) double fiber-
to take into account any resonance of the system or rigid or vis- board partitions and sandwich structures [26], (iv) sound absorp-
coelastic connections present in the composition [18,19]. Finally, tion prediction [27], (v) External Thermal Insulation Composite
no formulations, taking into account for structural connections System (ETICS) influence on masonry wall [28], but no application
(point or linear studs), are available to authors’ knowledge. to periodic structures is currently available. Some recent studies
 M. Caniato et al. / Applied Acoustics 166 (2020) 107356 3

[29,30] dealt with the TMM simulation of cross-laminated timber

walls.
Simplified FEM models were successfully applied to automotive
field [31], using new FEM elements [32] or applied to very simple
heavyweight structures made of concrete [33], bricks [34,35] or
simple composite plates [36–38]. For lightweight timber partitions
some attempt were made for solid walls [39–40], but no attempt
was made on sound reduction prediction of complex periodic
structures.
Therefore, there is need to understand whether TMM and sim-
plified FEM models are able to provide reliable results for periodic
partitions with respect to the apparent sound insulation index and
radiation efficiency.
This research aims to perform an extensive analysis among the
described methods, suitable for complex periodic structures. The
obtained results are compared to laboratory ones, considering the
apparent sound insulation index R (Transmission Loss TL) and radi-
ation efficiency in frequency. Accuracy is taken into account in
order to classify the best procedure. Finally, a combination of sim- Fig. 2. Investigated structure– bare configuration (up) and bare + ETICS (down).
plified FEM and TMM is studied and considered, as it is the best
both from the point of view of accuracy and reliability of the results.
1. OSB panel, thickness 2.2 cm
2. Wooden beams 16 cm thick and 8 cm wide
2. Materials and methods 3. Filling in mineral wool 16 cm thick and density 40 kg/m3
4. OSB panel, thickness 1.4 cm
In this work, numerical models are applied and studied in order 5. ETICS constituted by glass wool, 8 cm thickness and density
to compare them and determine their accuracy in relation to the 60 kg/m3
prediction of sound reduction and radiation efficiency of periodic 6. Plaster layer 7 mm thick and density 1515 kg/m3
structures. These latter ones are usually constituted of multilayer
lightweight timber wall. For this reason, this construction is used In order to proceed also with a parametric study, useful to prop-
to investigate this topic. Thus, the analysis of the materials influ- erly identify every single layer or junctions’ influence, a two-steps
ence on the propagation of the acoustic wave is analyzed. In analysis was carried out as follows:
Fig. 1, the studied structure is shown, highlighting the periodic
structure provided by timber studs, fixed together with the exter- (i) simulation of the bare timber frame partition (bare
nal Oriented Strand Boards (OSB). Between the studs, it is possible configuration)
to notice the rock wool layer. The results are then compared to (ii) implementation of the timber wall with the ETICS layering
ESSMs [7] and to laboratory measurements. (Bare + ETICS).

2.1. Investigated structures

2.2. Experimental: Materials characterization
A stratified wooden partition including a frame solution is iden-
tified and tested [41] and consists of the following elements All materials were mechanically and acoustically characterized
(Fig. 2): as follows. Complex elastic modulus of OSB and plaster was mea-
sured using the method described in Ref. [42]. A beam of length
equal to 30 cm was mounted on a shaker on one end and freely
suspended on the other end, as depicted in Fig. 3.
Acceleration response over equally spaced (1 cm) 20 points was
measured once an exponential sweep between 200 Hz and 4000 Hz
was generated by the shaker.
Using the inhomogeneous wave correlation method and disper-
sion relation for Kirchhoff beams, firstly the complex wave number
was determined as a function of the frequency and successively the
real and imaginary part of the complex modulus were calculated.
An example of complex wavenumber as a function of the fre-
quency is depicted in Fig. 4 for OSB beam.
This approach permits to model OSB and plaster layers as
homogeneous and isotropic linear viscoelastic solids.
Mechanical properties of rock and glass wool were measured
using a well-established quasi-static method [43] (Fig. 5). Material
(indicated in Fig. 5 as 1) is mounted between a moving plate, con-
nected to a shaker. Transmitted force F(w) is measured using a
force transducer (indicated in Fig. 5 as 3). Longitudinal displace-
ment d1(w) is measured using an accelerometer (indicated in
Fig. 5 as 2). From the ratio between transmitted force and longitu-
Fig. 1. Picture of the laboratory measurement (bare partition side), highlighting the dinal displacement (assuming a nearly zero Poisson’s ratio), it is
inner layering. possible to calculate the complex stiffness and from geometrical
4 M. Caniato et al. / Applied Acoustics 166 (2020) 107356

Fig. 3. Example of material characterization for OSB (left) and plaster (right). 1: Shaker, 2: Accelerometer, 3: tested beam.

ing the surface impedance values measured by means of transfer

function method and impedance tube [44].
This approach is used to model rock wool and glass wool as
equivalent porolelastic media [18]. According to Biot’s model, the
poroelastic material can be replaced on a macroscopic scale by
an equivalent fluid of effective density q and effective bulk modu-
lus K, coupled with the elastic frame (i.e. the skeleton of the mate-
rial). This latter parameter can be described as a function of
Young’s modulus, Poisson’s ratio and damping loss factor.

2.3. Experimental: Laboratory complete wall measurements

In order to obtain an experimental characterization, this parti-

tion was tested in laboratory according to the series of standards
ISO 10140:2010 [41]. This procedure was useful to study the fre-
quency behavior of partition transmission loss, in order to under-
stand the robustness and reliability of the model and in order to
validate the numerical simulation and the assumptions related to
it. The real timber frame partition was mounted in laboratory (cou-
Fig. 4. Example of obtained complex wavenumber as a function of the frequency pled chambers) and an airborne omnidirectional noise source was
for OSB sample. used to excite the sample.
In addition, vibration measurements were performed in the two
different cases, using a Dytran 500 mV/g accelerometer posed on
the receiver room wall side in several different positions (Fig. 6),
both in correspondence to wooden beams (studs) and to OSB pan-
els (between studs) [45]. Distance between positions is 0.45 m.
These measurements were performed in order to understand
the vibrational phenomena related to airborne sound excitation
and propagation on timber frame partition and in order to calcu-
late the radiation efficiency.

2.4. Numerical: TMM simulations

The basic assumption of this approach is to consider the trans-

mission through different media and through acoustic bridges
(beams) as two independent and additive paths. The total trans-
Fig. 5. Test-rig for the determination of complex elastic modulus for glass wool and
rock wool.

thickness and area of the sample it is possible to determine the

complex modulus.
Fluid phase parameters (i.e. airflow resistivity r, open porosity
/, tortuosity a and thermal characteristic lengths K and K’) were
determined by means of a minimization based approach, consider- Fig. 6. Description of the accelerometer positions during laboratory measurements.
 M. Caniato et al. / Applied Acoustics 166 (2020) 107356 5

mission loss TL is determined by adding a correction term to the At a final stage, additional layers are added in a new homoge-
transmission coefficient of the main structure without beams. nized TMM model (hTMM), where the bare wall has been replaced
According to this, timber frame beams can be considered as sound with the equivalent homogeneous plate, generally having a fre-
bridges, as described by Eq. (7): quency dependent Young’s modulus and loss factor, while the Pois-
  son’s ratio is fixed in advance to a standard value.
Wi Wi
TL ¼ 10 log ¼ 10 log ð7Þ
W2 W 2;P þ W 2;B 2.5. Numerical: FEM simulations

where Wi is the incident power on the primary wall and W2 is the In the present section an overview of the simplified finite ele-
radiated power from the layering. It could be divided into two parts: ment model is given, as depicted in Fig. 9 for the bare panel.
the power W2,B radiated from the bending wave near field due to A 2D dimensional model was created for the simulation of the
the beam and the power W2,P due to the transmission through sound transmission loss. In particular:
the main structure (Fig. 7).
In order to investigate which procedure is more suitable to be
The upper domain A represents a fluid domain (defined by den-
adapted to different partition components, various simulation sity and sound speed), where an incident sound pressure level
steps were taken: of 94 dB at each frequency of interest was applied. The lower
(B) represents the receiving semi-infinite fluid domain. Over
a. Once defined the elastic properties of the external shells the red line, the sound intensity in normal direction with
(OSB), the partition is modelled, as if the rigid connections respect the bare panel was averaged.
between the two external panels were not present.
The domains related to OSB layers (horizontal yellow) and studs
b. The structural connections are inserted. (vertical orange) were modelled as elastic solids defined by
c. The glass wool and plaster are introduced (as poroelastic and standard mechanical properties (Young’s modulus, Poisson’s
elastic solids, respectively) in order to simulate the ETICS ratio and damping loss factor) and density.
construction.
The domain related to internal mineral wool was modelled as a
poroelastic medium using Biot’s theory for poroelasticity [18].
In all the previous phases, the OSB panels are considered linear This model requires five physical parameters (airflow resistiv-
elastic solids. ity, open porosity, tortuosity and viscous and thermal charac-
Available formulations considering the structural links have teristic lengths) and mechanical properties.
been proposed mainly for lightweight partitions made of two lay-
ETICS layering was modelled using an elastic solid for the plas-
ers of elastic solid (wood, plasterboard, metal, etc.) separated by ter and a poroelastic medium for the glass wool.
sound absorbing materials. When supplementary layers are added
(such for example ETICS systems), some limitation can arise due to The incident sound pressure (and thus the emitted sound inten-
not well-defined coupled conditions between main partition and sity) was investigated at each frequency of interest (in this case 9
added layer. To overcome this limitation, an alternative procedure frequency for each 1/3 octave band between 100 Hz and
can be used. 5000 Hz) and for equally-spaced 40 angle of incidence between
In particular, as a first stage the bare partition is modeled using 0° and 90°.
the standard TMM method described before (steps (i) to (iii)). Once For each frequency and angle of incidence h, the sound trans-
the transmission loss of the bare partition has been calculated, mission loss was calculated as Eq. (8) [46]:
such a partition is simulated as a unique layer (having the same
equivalent density) using a homogenization procedure, as TLðf ; hÞ ¼ Lp ðf ; hÞ  LI ðf ; hÞ  6 ð8Þ
described in Ref. [28]. In practice, a frequency dependent Young’s
modulus of an equivalent linear viscoelastic solid has been deter- where Lp is the average sound pressure level in the source room, LI
mined, guaranteeing that its transmission loss was identical to is the sound intensity level of the receiving room, 6 is the adapta-
the one obtained with traditional TMM scheme of the complete tion term provided by standard.
wall (Fig. 8). Successively, an integration over all possible angles of incidence
was applied to numerical results and a spatial windowing function
[20] was used in order to calculate the diffuse field sound trans-
mission loss.
All the domains were meshed using quadratic tetrahedron ele-
ments (six elements per wavelength, referring to a maximum fre-
quency of 6000 Hz). Boundary condition of Floquet-Bloch theory
are applied to lateral dimension (left and right) [47–49].

2.6. Numerical: Empirical, semiempirical and standard models

In order to analytically calculate the transmission loss, the

parameter s has to be computed as described in Ref. [7], using
the following Eq. (9):
TL ¼ 10 log s ð9Þ
In order to compute the frequency transmission loss, the s
parameter need to be calculated as equations (10):

 2  0:5 2 !
2q0 c0 ðl1þ l2 Þfc
2
r
s¼ 2r f þ 2 forf < fc
2pfm0 2
l1 þ l2 f g
Fig. 7. Decoupled approach to taking into account structural link in TMM method.
6 M. Caniato et al. / Applied Acoustics 166 (2020) 107356

Fig. 8. Equivalent Young’s modulus for bare wall obtained from the homogenization process.

Fig. 9. Simplified finite element model of the bare structure.

 2    2 
2q0 c0 p fc r The radiation efficiency r is of paramount importance for the
s¼ for f ffi fc ð10Þ
2pfm0 2 f g prediction models related to building acoustics. This parameter
defines the relation between the radiated sound power Wrad of a
 2   moving partitions and the one hypothetically related to a rigid pis-
2q0 c0 p r2 ton with the same surface area S, moving with same mean square
s¼ for f > fc
2pfm0 2 g velocity [7,46] as Eq. (12):

W rad
where f is the frequency [Hz], fc is the critical frequency [Hz], g is r¼ ð12Þ
the global damping [-], m’ is the mass per unit area [m2] and l1 qo c0 Shv 2 i
and l2 are the linear dimension of the partition [m]. The radiation where q0 is the air density [kg/m3], c0 the speed of sound within the
factor for forced transmission rf is described by the following equa- fluid [m/s], S is the radiating surface area [m2] and hvi is the aver-
tions (11): aged values of measured velocity in the receiving side. In this equa-

 tion, the Waterhouse correction has been applied [50], assuming
rf ¼ 0:5 lnðk0 ðl1 l2 Þ0:5  K the diffuse field approach and taking into account the boundary
  effects as depicted in Eq. (13):
l2 l2 5l2 1  
K ¼ 0; 964  0:5 þ ln þ  ð11Þ A St k 0
pl1 l1 2pl1 4pl1 l2 k20 W rad ¼< p2 > 1þ ð13Þ
4qo c0 8V
1 where hpi is the averaged values of the sound pressure level in the
k0 ¼ 2pf
c0 receiving room [Pa], A is the absorption area [m2], V is volume of the
 M. Caniato et al. / Applied Acoustics 166 (2020) 107356 7

Table 1 3.3. TMM simulations

Results of materials characterization.

By realizing numerical simulations using the method of transfer

matrices, the frequency behavior of the various layers reported in
the previous paragraph was investigated. In Fig. 13, the frequency
trends of the numerical simulations carried out exclusively with a
single OSB panel are reported.
As described is Section 2.2, OSB layer is characterized under the
hypothesis of being homogeneous and isotropic material [51]. Due
to the complexity of the structure, such hypothesis needs to be
investigated. To this aim, starting from data related to a static
orthotropic characterization [51], TMM simulations have been car-
ried out and the comparison in terms of sound transmission loss is
depicted in Fig. 13.
From Fig. 13, it is possible to observe a good reliability of the
hypothesis of isotropic material for OSB layers.
As a further step, the simulation of the bare wall is compared
receiving room [m3], k0 is the wavelength [m] and St is global with experimental test. Thanks to the decoupled approach, TMM
absorbing surface of the receiving room [m2]. allows the computation of airborne and structural-borne contribu-
tion to the overall sound transmission loss. The comparison is
shown in Fig. 14. From the graph, it is interesting to observe that
2.7. Accuracy investigation the sound transmission loss is fully governed by the structural
path, as the sound insulation of the main wall without studs would
In order to estimate the quality of the fit, a standard deviation is be hugely higher at almost all frequencies of interest.
calculated by taking into consideration the measured values as the As described above, when adding the ETICS layering, the simu-
average data (l) for each of n frequency bands (1/3 octave) and the lation changes, because a new layer modifies the physical behavior
calculated values as experimental ones (Eq. (14)): of the partition. Some problems in the middle-high frequencies are
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi presents and could not be solved, due to TMM coupling limitation.
u
u1 X n
In order to avoid this, a homogenization process was realized and
rdev ¼ t ðxi  lÞ2 ð14Þ results (Fig. 15) highlight a better agreement between the latter
N i¼1
simulation and the laboratory measurements.
In addition, the mean difference is computed (Eq. (15)): Finally, TMM simulated radiation efficiencies rbare,TMM and
rbare,ETICS,TMM are reported in Fig. 16.
1 Xn
Dmean ¼ jxi  lj ð15Þ 3.4. FEM simulations
N i¼1

Furthermore, in order to graphically compare trends, two- By carrying out numerical simulations using simplified FEM, the
populations Q-Q plots are used to identify the quantile influences. frequency behavior of the various layers was investigated. In
The square regression index r2 is calculated and reported. Fig. 17 left, the frequency trends of the numerical simulations are
reported, while in Fig. 17 right the simulated radiation efficiency
frequency trends are depicted.
3. Results
3.5. Empirical, semiempirical and standard models
In this section, the results of the performed tests and simulation
are presented. The calculated transmission loss frequency trends are reported
in Fig. 18, while the rbare,standard and rbare,ETICS,standard overlap the
3.1. Materials characterization measured ones.

The results of materials characterization are summarized in 3.6. Combination of simulation methodologies
Table 1.
Taking into account all the positive results obtained from the
presented analysis, a combination of methodologies usable for
3.2. Laboratory measurements the determination of transmission loss and radiation efficiency of
complex periodic structures is described below. This procedure is
The results obtained by the transmission loss test conducted in to be followed in order to raise the accuracy at the maximum level
the laboratory are used as a basis for the validation of the designed and thus to obtain the best predicted results, compared to simu-
models. The frequency trend is shown in Fig. 10, highlighting the lated ones:
typical resonance and coincidence zones for this type of partition.
In addition, vibration measurements were carried out on the a. The combined procedure can only be used as long as an
receiving wall side, using a dodecahedral source as exciting source acoustical and mechanical characterization of any single
in the emitting room. component is carried out. Only in this way, reliable result
In Fig. 11, the mean acceleration spectrum for the two kind of can be achieved. If these data are not available, they have
configuration is reported, highlighting the frequency trend mea- to be measured or found in literature. Particular attention
sured on the rigid connection and on the OSB panel. should be paid to the frequency dependency of some param-
The measured radiation efficiencies rbare,meas and rbare,ETICS,meas eters, characterizing the behavior of elements, such as
are reported in Fig. 12. Young’s Modulus or damping.
8 M. Caniato et al. / Applied Acoustics 166 (2020) 107356

Fig. 10. Laboratory measurements of transmission loss as a function of frequency.

Fig. 11. Mean acceleration spectrum for the two kind of configuration for bare wall (left) and bare wall plus ETICS (right).

Fig. 12. Comparison between measured radiation efficiency of the bare wall and with the ETICS adding.
 M. Caniato et al. / Applied Acoustics 166 (2020) 107356 9

Fig. 13. Numerical simulations for different OSB models.

Fig. 14. Air-borne and structure-borne contribution to the overall sound transmission loss for the bare wall.

b. The exciting source must be identified for the determination Once the frequency trend of the transmission loss is obtained,
of the final parameter. an additional layer can be added. To this aim, two ways are possi-
ble. Again, if a parametric analysis is needed, it is preferable to use
For the bare structure, it could be chosen whether or not to the TMM. However, to avoid problems of intrinsic limits to this
perform the parametric analysis. If so, the use of the TMM method [30], it is necessary to homogenize the bare wall (hTMM).
method is preferable, since it can easily provide this type of As before, for the above mentioned reasons, the low frequency
study. However, since this procedure is analytical-energetic, it range can be covered more easily by the simplified FEM, which is
will not be able to provide a correct prediction at low and also used for the area inherent to the coincidence present on the
medium low frequency, due to the presence of structural reso- bare wall. In fact, since the additional layer modifies the structural
nances [18]. These effects are very present in the bare configu- propagation, intervening precisely on the frequency of coincidence
ration and are caused by the direct exposure of the supporting (Fig. 11), the simplified FEM is better suited to anticipate this phe-
structure to acoustic excitation. For this reason, at low and nomenon of structural nature. The frequency composition of the
medium-low frequencies (100 Hz160 Hz) the use of simplified results obtained by the combination of hTMM and simplified
FEM is to be preferred. The frequency composition of the trend FEM provides the final trend.
obtained by the simplified FEM and TMM provides final results The various steps to be followed for the determination of radi-
(Fig. 19 left). ation efficiency parameter are described and explained below.
10 M. Caniato et al. / Applied Acoustics 166 (2020) 107356

Fig. 15. Numerical TMM simulations for bare wall and ETICS system.

Fig. 16. Comparison of radiation efficiency results for TMM.

a. As for the transmission loss, the combination of numerical In Fig. 19, the whole procedure is graphically depicted and, in
simulations can only be used when the precise acoustic Fig. 20 the results for transmission loss are reported, highlighting
and mechanical characteristics of the involved elements the agreement between the measured results and the combination
are known. Only in this way, at some frequency reliable of numerical simulation approaches.
results can be achieved. If these data are not available, they
have to be measured or found in literature.
b. For the bare configuration, both methods have to be used to 3.7. Accuracy
obtain good frequency trend results. For low frequency
range, as for the transmission loss, the simplified FEM is As described in Section 2.7, the calculated standard deviation,
indicated, while for middle low and high frequency ranges the mean difference and the squared regression coefficient are
(2500 Hz–5000 Hz) the TMM has to be used. Unfortunately, reported in Table 2. Two-populations Q-Q plots considering two
for the layer addition, no numerical or analytic process could different groups are reported in Fig. 21 for transmission loss and
provide reliable results. Fig. 22 for radiation efficiency.
 M. Caniato et al. / Applied Acoustics 166 (2020) 107356 11

Fig. 17. Comparison of transmission loss results for simplified FEM (left) and for radiation efficiency (right).

Fig. 18. Comparison of transmission loss results for analytical model.

4. Discussion it from the standard frequency domain. This behavior is also high-
lighted in Fig. 12, where the radiation efficiency is depicted.
In this work, an extensive investigation of numerical As regards to the acceleration, it could be understood that the
approaches is presented including experimental, analytical and influence of the external layer on the measured values results in
numerical simulations, in order to understand if the available a clear change of the vibration behavior for the frequency range
methods may be valid tools to predict the radiation efficiency from 160 Hz to 250 Hz. In fact, for the bare configuration it is evi-
and the transmission loss of complex periodic timber structures. dent that the studs act as a principal path for the propagation of
the structure-borne sound, while the addition of the new external
layer involves a substantial change on the acceleration peak
4.1. Results towards the internal membrane.
As regards to the analytical standard model, results clearly show
Experimental measurements show a real increase in sound that it is not able to correctly predict the transmission loss param-
insulation given by the application of the ETICS layering especially eter, especially when adding the ETICS layering and that, for the
at mid-to-high frequencies. This is also demonstrated by the accel- determination of radiation efficiency, experimental measurements
eration measurements (Fig. 11) carried out on the wall receiving are needed in a real laboratory facility. Below the critical frequency,
side of the two configurations. In this case, it is clear that in the for both configuration this model tends to underestimate the sound
bare configuration, the partition moves even at mid-high frequen- insulation, even if for the bare wall the coincidence frequency is
cies (from 1250 Hz to 2500 Hz), clearly identified by the coinci- correctly highlighted. This fact shows that this analytical model
dence frequency. When an external layer is added (ETICS), this may not be used for this kind of partitions, mainly because the pres-
phenomenon is modified by the new configuration, thus excluding ence of studs is not considered in the equations.
12 M. Caniato et al. / Applied Acoustics 166 (2020) 107356

Fig. 19. Combination of numerical simulations scheme. Transmission loss (left) and radiation efficiency (right). L stands for ‘‘low frequency range” (100 Hz–160 Hz), ML for
‘‘middle-low frequency range” (200 Hz–800 Hz), MH ‘‘middle-high frequency range” (1000 Hz–2000 Hz) and H for ‘‘high frequency range” (2500 Hz–5000 Hz).

Fig. 20. Combination of numerical simulations results. Transmission loss for bare configuration (left) and bare + ETICS (right). Radiation efficiency of bare structure (bottom).

Another unsolved issue, related to this latter reason, is that By using the transfer matrix method, (Eqs. (3) to (6), Fig. 14), it
equations parameters include the measured sound pressure levels is also possible to explore the influence of the propagation paths as
and the measured sound velocity in the radiating side. This fact part of a parametric study.
implies that the application of this procedure is not easy, because In this view, the isotropic or orthotropic nature of the OSB panel
measurements on the real sample are always necessary. was investigated (Fig. 13) and it was concluded that the first one
 M. Caniato et al. / Applied Acoustics 166 (2020) 107356 13

Table 2
Accuracy for different numerical methods.

Methodology Bare Bare + ETICS

Numerical method
TL Radiation efficiency TL Radiation efficiency
dev. st. Dmean r2 dev. st. Dmean r2 dev. st. Dmean r2 dev. st. Dmean r2
Standard models 5.9 4.8 0.931 – – – 9.2 7.7 0.931 – – –
TMM 2.8 2.0 0.951 3.0 2.6 0.891 5.4 4.2 0.973 8.22 6.6 0.788
hTMM – – – – – – 2.6 2.3 0.987 8.22 6.6 0.788
Simplified FEM 3.4 3.0 0.959 3.3 3.3 0.902 2.7 2.3 0.990 8.12 6.9 0.826
(h)TMM + simplified FEM 2.4 1.9 0.965 3.0 2.6 0.920 2.3 1.9 0.992 8.15 6.6 0.879

Fig. 21. Q-Q plots for transmission loss.

could be considered a reliable approximation. The related litera- potentially reach extremely high transmission loss values in fre-
ture [52–55] agrees that the dissipative viscoelastic effect of the quency, being governed exclusively by the mass-spring-mass phe-
polymer, which acts as a strengthening, plays a very important role nomenon. However, when rigid structural connections are added
on the propagation of the wave within recycled materials. across the entire partition thickness, the performance drops up to
As shown in Fig. 14, the partition could be extremely perform- 60 dB due to solid propagation. In this case, neither the mass of the
ing, if it did not include the structural connections passing from connection nor its stiffness is very important; as a matter of fact, it
one side to the other [56]. In fact, airborne sound insulation could has been set with a real literature values [6], as the wooden beams
14 M. Caniato et al. / Applied Acoustics 166 (2020) 107356

Fig. 22. Q-Q plots for radiation efficiency.

are very thin compared to their length. Furthermore, they present this progression. Almost the same consideration could be done for
in reality a high flexural elastic modulus and a consequent excel- the bare plus ETICS structure, where the hTMM case is included.
lent transmission of the sound wave [57–59]. This latter simulation provides more accurate results, compared
In addition, the use of the TMM method has shown to be the to TMM and very similar to simplified FEM. The combination of
possibility to study the partition parametrically, in good agreement hTMM and simplified FEM, depicted in Fig. 19 (left), presents the
with past researches [60,61]; it may therefore provide a very pow- best agreement also in this case.
erful tool for the acoustic performance characterization of this type A dedicated discussion is reserved for the prediction of the radi-
of partitions [62–64]. The limits of this method lie exclusively in ation efficiency. For the bare wall configuration, reliable results are
the deep knowledge of the physical mechanisms governing the provided by TMM (Fig. 16) and simplified FEM (Fig. 18) in the mid-
propagation of the acoustic wave in different materials, as well dle low and high frequency ranges, while for the middle ones the
as the possibility of measuring them or finding reliable data in lit- trends are in good agreement; nevertheless, they fall within the
erature. In fact, it has been demonstrated that the model variation range of 5 dB. Unfortunately, for the bare wall plus ETICS layer-
of the OSB panel alone provides or does not provide good results in ing, numerical simulations provided results in agreement with
accordance with the experimental one. Furthermore, Fig. 15 shows the measured ones only for middle-low frequency (until 630 Hz).
how the TMM methods hardly predict the TL especially at high fre- In other frequencies, the predicted values do not agree with the
quencies; thus, the use of homogenization process is highly recom- measured. This fact could be related to the few simplifications
mended, as it can really improve provided results. assumed in the simulations process. Actually, there is no set of
The simplified FEM results show a very good agreement dedicated equation to the best authors’ knowledge to solve this
between the calculated transmission loss and the measured one. problem. It is then hard to predict correctly the velocity, once
In comparison with the TMM results, the FEM ones could fit better structural connections (studs) are used. In facts, many of the avail-
the laboratory values for both configuration in terms of general fre- able approaches are limited to a decoupled approach only for radi-
quency trend and for the coincidence zone. ated sound, as applied using Eq. (7). Such approach has been
The accuracy study (Table 2) shows that and improvement is mainly investigated for lightweight partitions with linear and point
conceivable. It is also possible to classify the models according to structural connections without the adding of any additional layer
the used statistical parameters. The two different structure provide (ETICS for instance).
very similar accuracy trends. As a matter of fact, for both cases In this research, this limitation was tried to be overcome by
standard models show to be the worst ones even if, for the bare using the homogenization procedure of the bare wall. This proce-
structure, results are slightly better than the bare plus ETICS one. dure has shown advantages in prediction of Sound Transmission
The combination of (h)TMM and simplified FEM in frequency Loss, but it could be that important pieces of information were lost
depicted in Fig. 19 (left) provides the best results compared to about the real velocity distribution (which is in reality a combina-
measured ones for both cases (Fig. 20). tion of velocity along main partition and in correspondence of
For the bare case, it is possible to highlight that standard devi- structural links). This is the reason that may lead to a not correct
ation and mean difference are comparable. The simplified FEM estimation of radiation efficiency with such complicated multi-
model present a higher agreement to the measured parameters layer system.
in comparison to TMM and standard models. This fact is high- On the other side, FEM analysis should better grasp the physics
lighted also graphically by the Q-Q plots (Fig. 21), where it can and coupling among field variables; it is worthy of mention that
easily be noticed how, starting from standard models which cannot what has been proposed is a simplified 2D model without real
fit the measured ones, moving to TMM, then to simplified FEM and boundary conditions (which is coherent for TMM methodology).
finally to the combination of TMM and simplified FEM, the quantile The fact that in principle the 2D system has an infinite extent
curve moves closer to the y = x one. Furthermore, regressive coef- (which is ‘‘resized” with finite size correction only for the radiated
ficients r2 resumed in Table 2 show a positive increment, following sound) may have limitations on velocity distribution of a multi-
 M. Caniato et al. / Applied Acoustics 166 (2020) 107356 15

Table 3
pros and cons comparison.

Methodology Advantages Limitations

Analytical Reduced computational time and a deep parametric Only related to homogenous structures (without rigid connection embedded). Need
models study could be performed. laboratory measurements
TMM
High flexibility for any multilayer system and for
Partitions are supposed to have infinite lateral extent, thus real boundary conditions are
different excitations difficult to apply and no modal behaviour can be detected. TMM deals with planar sur-

Structural links effects can be evaluated within the faces although it can be extended to curved panels and cylindrical pipes. Anyway, it is
framework of TMM using decoupled approaches not possible to simulate directly complex geometries systems
Simplified
High flexibility for any system having a complex
High frequency limit depends on the size of the mesh (at least 6 element per wavelength
FEM geometry and for different excitations should be used). This result implies prohibitive computational time when geometry is
complex. Suitable alternatives can be found for periodic structures

layer wall. In addition, this approach has been proved to be reliable have been described and their parametric influence has been dis-
for sound transmission loss, but presents an overall limitation due cussed and argued.
to missing behavior of the wall, which is orthotropic in geometry The results provided by this approach were then validated by
and this aspect may lead to uncertainties in the velocity distribu- laboratory measurements, showing an excellent agreement for
tion. Thus, this aspect surely requires an extensive experimental the transmission loss parameter, while good results are highlighted
set, as future development. for radiation efficiency.
The accuracy study (Table 2) shows that and very few improve- It has been shown that very good results can be achieved, also
ments are conceivable. For the bare case, both standard deviations compared to traditional methodologies or standard ones and that
and mean differences highlight that the predicted results using parametrical investigations on every step of the analysis are possi-
TMM and simplified FEM are similar and that they could have ble and useful to highlight paramount effects:
some agreement with the measured ones. On the other hand, they
also evidence how no improvement is possible even combining the
isotropic behavior of external panel and its consequence on
TMM with the simplified FEM depicted in Fig. 19 (right). In the bare coincidence,
plus ETICS case, statistical parameters show a bad agreement
viscoelastic influence of bare partition at middle-high
between test and simulations and that again no improvement is frequencies,
possible even with methods combinations (Fig. 20).
predominant sound wave path transmission through solid
Q-Q plots clearly show also graphically that the predicted junctions.
results in both structure cases do not produce a good agreement
with the tested ones. The bare case identify two different quantile Finally, it is demonstrated that a combined approach based on
distributions; although they are from the same family, they are (i) a deepened materials characterization, (ii) a TMM with homog-
very different (very different standard deviation). In the bare plus enization process and (iii) a simplified FEM numerical simulations
ETICS case the quantile plot delineate a big difference between may robustly provide reliable results, raising the accuracy results.
the two trends. Since the two curve are not aligned and they lay In the end, since the combined method is based on general
far from each other, it could be concluded that the content of the assumption, if materials characterization may be performed or is
included data is very different. already available, it could be generalized to other complex and
In the end, summarizing the pros and cons of the extensive composite structure with rigid connections (stud) and multiple
investigation (Table 3), it could be highlighted that not all the pre- layering.
sented models deals with the determination of the transmission
loss of complex partition containing rigid connections.
Author contributions
Different approaches are required for overall partition charac-
terization. With the TMM method, it is possible to obtain an opti-
M. C. and F. B. developed the research. M.C. and F.B. defined the
mal final value and, by means of the homogenization process, it is
methods and comparisons. M. C. and P. B. performed the numerical
possible to add other layers with known acoustic properties. In
simulations and the experimental measures. M. C. and F. B. per-
addition, it is possible to study the frequency partition behavior,
formed laboratory measurements. M.C. designed and tested all
thanks to parametric analysis, which is able to deeply focus to
the steps of the extensive investigation and comparisons with
the differentiation between airborne and structure-borne sound.
the measured data. A. G. overviewed and supervised the research.
The whole process takes few minutes of calculation.
M. C. wrote the paper.
The simplified FEM, however, is able to fully simulate the struc-
tures. While obtaining the same result as the TMM procedure, the
simplified FEM has the advantage of being able to choose one or Declaration of Competing Interest
more points within the geometry and, only in those areas, to inves-
tigate the behavior. However, this procedure involves much longer The authors declare that they have no known competing finan-
calculation times and requires periodic structures. cial interests or personal relationships that could have appeared
to influence the work reported in this paper.
5. Conclusions
Acknowledgement
An extensive investigation is developed to study the transmis-
sion loss and radiation efficiency of complex periodic partitions. This work was financed by EFRE 2014-2010 1095 E2I@NOI CUP
The adopted methodology comprises analytical, empirical and D56C18000180009 and by ‘‘Klimahouse and energy production”
semiempirical models, TMM and simplified FEM simulations. All in the framework of the programmatic-financial agreement with
input data were previously measured using dedicated procedures the Autonomous Province of Bozen-Bolzano of Research Capacity
and instrumentation. Necessary factors for numerical simulation Building, which are gratefully acknowledged.
16 M. Caniato et al. / Applied Acoustics 166 (2020) 107356

This work was also partially finance by Interreg BIGWOOD pro- [28] Santoni A, Bonfiglio P, Davy JL, Fausti P, Pompoli F, Pagnoncelli L. Sound
transmission loss of ETICS cladding systems considering the structure-borne
ject, ITAT 1081 CUP: I54I18000310006.
transmission via the mechanical fixings: numerical prediction model and
experimental evaluation. Appl Acoust 2017;122:88–97.
[29] Santoni A, Bonfiglio P, Fausti P, Schoenwald S. Predicting sound radiation
References efficiency and sound transmission loss of orthotropic cross-laminated timber
panels. In Proceedings of meetings on acoustics, 173EAA, vol. 30; 2017, p.
015013.
[1] Navaratnam S, Ngo T, Gunawardena T, Henderson D. Performance review of
[30] Santoni A, Bonfiglio P, Mollica F, Fausti P, Pompoli F, Mazzanti V. Vibro-
prefabricated building systems and future research in Australia. Buildings
acoustic optimisation of wood plastic composite systems. Constr Build Mater
2019.
2018;174:730–40.
[2] Baghchesaraei A, Baghchesaraei OR. Using prefabrication systems in building
[31] Yao L, Tian W, Li L, Yao L. Numerical investigations of a partition-of-unity
construction. Int J Appl Eng R 2015:44258–62.
based ‘‘FE-Meshfree” QUAD4 element with radial-polynomial basis functions
[3] Ramage MH, Burridge H, Busse-Wicher M, Fereday G, Reynolds T, Shah DU,
for acoustic problems. Appl Math Mod 2016;40(13–14):6199–217.
et al. The wood from the trees: the use of timber in construction. Renew Sust
[32] Xu JP, Rajendran S. A partition-of-unity based ‘FE-Meshfree’ QUAD4 element
Energ Rev 2017;68(Part 1):333–59.
with radial-polynomial basis functions for static analyses. Comput Methods
[4] Caniato M, Bettarello F, Ferluga A, Marsich L, Schmid C, Fausti P. Acoustic of
Appl Mech Eng 2011;200(47–48):3309–23.
lightweight timber buildings: a review. Renew Sust Energ Rev
[33] Jean P, Siwiak H, Joubert G. A decoupled vibro-acoustic development of FEM:
2017;80C:585–96. https://doi.org/10.1016/j.rser.2017.05.110.
application to laboratory modelling. J Build Acoust 2006;13(2):83–9883.
[5] COST action FP0702. Acoustic design of lightweight timber frame construction,
[34] del Coz Díaz JJ, Álvarez Rabanal FP, García Nieto PJ, Serrano López MA. Sound
final ebook. Available at http://extranet.cstb.fr/sites/cost/ebook/Forms/
transmission loss analysis through a multilayer lightweight concrete hollow brick
AllItems.aspx (accessed on 22nd September 2019); 2012.
wall by FEM and experimental validation. Build Environ 2010;45:2373–86.
[6] Mak CM, Wang Z. Recent advances in building acoustics: an overview of
[35] Jacqus G, Berger S, Gibiat V, Jean P, Villot M, Ciukaja S. A homogenised
prediction methods and their applications. Build Environ 2015;91:118–26.
vibratory model for predicting the acoustic properties of hollow brick walls. J
[7] International Standard Organization. ISO 12354-1:2017 Building acoustics –
Sound Vib 2011;330(14):3400–9.
estimation of acoustic performance of buildings from the performance of
[36] Wyngaert JCE, Schevenels M, Reynders EPB. Predicting the sound insulation of
elements – Part 1: Airborne sound insulation between rooms.
finite double-leaf walls with a flexible frame. Appl Acoust 2018;141:93–105.
[8] Granzotto N, Di Bella A, Piana EA. Prediction of the sound reduction index of
[37] Tsai YT, Pawar SJ, Huang JH. Optimizing material properties of composite
clay hollow brick walls. Build Acoust 2020. https://doi.org/10.1177/
plates for sound transmission problem. J Sound Vib 2015;335:174–86.
1351010X20903144.
[38] Winter C, Buchschmid M, Müller G. Modeling of orthotropic plates out of cross
[9] Kouyoumji JL, Guigou C. Predicting sound transmission loss of timber framed
laminated timber in the mid and high frequency range. Proced Eng
walls and floors using SEA. In ‘‘AcuBois” project of the French wood industry,
2017;199:1392–7.
Proceedings of Internoise 2015, San Francisco (US), 9–12 August 2015.
[39] Bard D, Negreira J, Kouyoumji JL, Borello G, Catherine G. Challenges for
[10] Kouyoumji JL. Prediction sound transmission loss on lightweight timber
acoustic calculation models in ‘‘silent timber build”, Part 1- FEM. In 43rd
framed construction using sea. In Proceedings of Internoise 2013, Innsbruck
international congress on noise control engineering: improving the world
(AUSTRIA), 15–18 September 2013.
through noise control, Proceedings of internoise; 2014. p. 4424–9.
[11] Wang T, Li S, Rajaram S, Nutt. Predicting the sound transmission loss of
[40] Kouyoumji JL, Bard D, Borello G, Catherine G. Challenges for acoustic
sandwich panels by statistical energy analysis approach. J Vib Acoust
calculation models in ‘‘silent timber build”, Part 2. In 43rd International
2010;132(1):1–7.
congress on noise control engineering: improving the world through noise
[12] Kouyoumji JL. Predicting sound transmission loss of timber framed walls and
control, Proceedings of internoise; 2014.
floors using SEA. In ‘‘AcouBois” project of the French wood industry,
[41] International Standard Organization. ISO 10140-1:2010 Acoustics – laboratory
Proceedings of Internoise 2015, San Francisco (USA), 9–12 August 2015.
measurement of sound insulation of building elements – Part 1: Application
[13] Mahn J, Hopkins C, Filippoupolitis M, Schanda U, Völt R, Krajči L. The
rules for specific product; 2019.
optimization of a wooden floor design based on validated finite element
[42] Santoni A, Bonfiglio P, Fausti P, Pompoli f. Alternative method to the Oberst
models. In Proceedings of Internoise 2014, Melbourne (Autralia), 16–19
technique to measure the complex elastic modulus of visco-elastic materials.
November 2014.
Noise Control Eng; 67(1).
[14] Henning P, Kirkegaard L, Vabbersgaard A. FEA of the variations in sound
[43] Bonfiglio P, Pompoli F, Horoshenkov KV, Rahim MIBSA, Rodenas LJ, Bécot F-X,
insulation in nominally identical prefabricated lightweight timber structures.
Gourdon E, Jaeger D, Kursch V, Tarello M, Roozen NB, Glorieux C, Ferrian F,
In Proceedings of Internoise 2013, Innsbruck (AUSTRIA), 15–18 September
Leroy P, Briatico Vangosa F, Dauchez N, Foucart F, Lei L, Carillo K, Doutres O,
2013.
Sgard F, Panneton R, Verdiere K, Bertolini C, Bär R, Groby J-P, Geslain A, Poulain
[15] Renno JM, Macea BR. Calculation of reflection and transmission coefficients of
N, Rouleau L, Guinault A, Ahmadi H. How reproducible are methods to
joints using a hybrid finite element/wave and finite element approach. J Sound
measure the dynamic viscoelastic properties of poroelastic media? J Sound Vib
Vib 2013;332(9, 29):2149–64.
2018;428:26–43.
[16] Fringuellino M, Guglielmone C. Progressive Impedance Method for the
[44] Bonfiglio P, Pompoli F. Inversion problems for determining physical
classical analysis of acoustic transmission loss in multilayered walls. Appl
parameters of porous materials: overview and comparison between different
Acoust 2000;59(3):275–85.
methods. Acta Acust U Acust 2013;99(3):341–51 (11).
[17] Fortini M, Granzotto N, Piana EA. Vibro-acoustic characterization of a
[45] Mahn J, Hoeller C, Zeitler B. Measurement of the airborne and resonant
composite structure featuring an innovative phenolic foam core. Appl Sci
radiation efficiencies. In Proceedings of the 22nd international congress on
2019;9(7). https://doi.org/10.3390/app9071276 (Article number 1276).
sound and vibration, ICSV 2015, Florence Italy; 12 July 2015–16 July 2015.
[18] Allard J, Atalla N. Propagation of sound in porous media: modelling sound
[46] International Standard Organization. ISO 15186-1:2000 Acoustics –
absorbing materials. 2nd ed. Chichester, UK: John Wiley & Sons Ltd.; 2009.
measurement of sound insulation in buildings and of building elements
[19] Piana E, Milani P, Granzotto N. Simple method to determine the transmission
using sound intensity – Part 1: Laboratory measurements.
loss of gypsum panels. In 21st international congress on sound and vibration
[47] Floquet G. Sur les ´equations diff ´ erentielles lin ´ eaires‘acoef-ficients p
2014, ICSV 2014, vol. 5, Beijing China, 13 July 2014–17 July 2014; 2014. pp.
´ eriodiques. Annales scientifiques de l’ ´Ecole NormaleSup ´ erieure
3700–6.
1883;12:47–88.
[20] Bonfiglio P, Pompoli F, Lionti R. A reduced-order integral formulation to
[48] Bloch F. Uber die Quantenmechanik der Elektronen in Krist-allgittern.
account for the finite size effect of isotropic square panels using the transfer
Zeitschrift f ̈ur Physik A 1929;52(7–8):555–600.
matrix method. J Acoust Soc Am 2016;139(4):1773. https://doi.org/10.1121/
[49] García PG, Fernández-Álvarez J-P, Floquet-Bloch. Theory and its application to
1.4945717.
the dispersion curves of nonperiodic layered systems. Math Prob Eng 2015.
[21] Yu X, Zhang Q, Kang J, Cui F. Predicting integrated thermal and acoustic
doi: 10.1155/2015/475364 (Article ID 475364).
performance in naturally ventilated high-rise buildings using CFD and FEM
[50] Uosukainen S. On the use of the Waterhouse correction. J Sound Vib 1995;186
simulation. Build Sim 2018;11(3):507–18.
(2, 21):223–30.
[22] Munjal M. Response of a multi-layered infinite plate to an oblique plane wave
[51] Morales EAM, Da Silva Bertolini M, Do Nascimento MF, Lahr FAR, Ballarin AW.
by means of transfer matrices. J Sound Vib 1993;162(2):333–43.
Study of Brazilian commercial Oriented Strand Board panels using stress wave.
[23] Sharp BH. Prediction methods for the sound transmission of building
Wood Res 2013;58(2):295–306.
elements. Noise Cont Eng 1978;11(2):53–63.
[52] Johnson DL, Koplik J, Dashen R. Theory of dynamic permeability and tortuosity
[24] Vigran TE. Sound transmission in multilayered structures–introducing finite
in fluid-saturated porous media. J Fluid Mech 1987;176:379. https://doi.org/
structural connections in the transfer matrix method. Appl Acoust 2010;71
10.1017/S0022112087000727.
(1):39–44.
[53] Champoux Y, Allard J. Dynamic tortuosity and bulk modulus in air-saturated
[25] Ghinet S, Atalla N. Sound Transmission Loss of Insulating Complex Structures.
porous media. J Appl Phys 1991;70:1975–9. https://doi.org/10.1063/1.349482.
Can Acoust/Acoustique canadienne 2001;29(3–26).
[54] Benkreira H, Khan A, Horoshenkov KV. Suistainable acoustic and thermal
[26] Dijckmansa A, Vermeir G. Sound transmission through finite lightweight
insulation materials from elastomeric waste residues,. CHEM Eng Sci
multilayered structures with thin air layers. J Acoust Soc Am 2010;3513.
2011;66:4157–71.
https://doi.org/10.1121/1.3500698.
[55] Sun Z, Shen Z, Ma S, Zhang X. Novel application of glass fibers recovered from
[27] Verdière K, Panneton R, Elkoun S. Transfer matrix method applied to the
waste printed circuit boards as sound and thermal insulation material. J Mater
parallel assembly of sound absorbing materials. J Acoust Soc Am 2013;4648.
Eng Perform 2013;22(10):3140–6.
https://doi.org/10.1121/1.4824839.
 M. Caniato et al. / Applied Acoustics 166 (2020) 107356 17

[56] Davy J. Predicting the sound insulation of walls. J Build Acoust 2009;16:1–20. [61] Brunskog J, Hammer P. Prediction model for the impact sound level of
https://doi.org/10.1260/135101009788066546. lightweight floors. Acta Acust U Acust 2003;89(2):309–22.
[57] Bucur V, Lanceleur P, Roge B. Acoustic properties of wood in [62] Brunskog J, Hammer P. Prediction models of impact sound insulation on
tridimensional representation of slowness surfaces. Ultrasonics 2002;40: timber floor structures; a literature survey. J Build Acoust 2000;7(2):89–112.
537–41. [63] Morandi F, De Cesaris S, Garai M, Barbaresi L. Measurement of flanking
[58] Bucur V. Acoustics of wood. Berlin, Heidelberg: Springer Verlag; 2006. transmission for the characterisation and classification of cross laminated
[59] Beall FC. Overview of the use of ultrasonic technologies in research on wood timber junctions. Appl Acoust 2018;141:213–22. https://doi.org/10.1016/j.
properties. Wood Sci Technol 2002;36:197–212. apacoust.2018.07.009.
[60] Brunskog J, Hammer P. Design possibilities for impact noise insulation in [64] Barbaresi L, Morandi F, Belcari J, Zucchelli A. Optimizing the mechanical
lightweight floors – a parameter study. In Proceedings of euronoise 2003, characterization of a resilient interlayer for the use in timber construction. In
Naples (Italy). Proceedings of ICSV24, London; July 2017.