Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

Journal of Sound and Vibration 494 (2021) 115842

Contents lists available at ScienceDirect

Journal of Sound and Vibration


journal homepage: www.elsevier.com/locate/jsv

Analysis of the vibroacoustic characteristics of cross laminated


timber panels using a wave and finite element method
Yi Yang∗, Chiaki Fenemore, Michael J. Kingan, Brian R. Mace
Acoustics Research Centre, Department of Mechanical Engineering, The University of Auckland, Auckland 1142, New Zealand

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

Article history: This paper presents an analysis of the vibroacoustic characteristics of Cross-Laminated
Received 20 September 2020 Timber (CLT) panels using a Wave and Finite Element (WFE) method. Two WFE modelling
Revised 2 November 2020
approaches are investigated: the first models each layer of the panel as a homogeneous
Accepted 10 November 2020
orthotropic material, whereas the second approximates the entire panel as a single equiv-
Available online 13 November 2020
alent homogeneous orthotropic material. Results are also compared to those of a thin, or-
Keywords: thotropic plate model which is a model commonly used in the literature. The three meth-
Wave and finite element ods are used to calculate the plane-wave and diffuse-field sound transmission loss of a
Sound transmission CLT panel. It is seen that all models are in reasonable agreement at low frequencies, but
Cross laminated timber that the models predict different sound transmission loss at higher frequencies. The dis-
persion curves and mode shapes of the free waves propagating within the in-vacuo panel
are predicted for both WFE models. These results are used to interpret the vibroacoustic
behaviour of the panel. It is observed that, at high frequencies, the vibroacoustic behaviour
of the panel is complicated, with many wave modes cutting on. This behaviour is not
modelled by the equivalent orthotropic plate model, which is incapable of capturing those
higher-order waves. The contributions to the sound transmitted by the various wave modes
are estimated and the most significant modes identified. Finally, predictions of the sound
transmission loss of a number of CLT panels are calculated using the layer-wise WFE mod-
elling approach. These predictions are compared with experimental measurements, show-
ing reasonable agreement. The predictions clearly capture the important coincidence dips
present in the measured spectra, although there are some minor discrepancies between
the measured and predicted levels and coincidence frequencies. These discrepancies may
be partially due to the uncertainty and variability of the material properties of wood. A
series of simulations are performed to investigate the effect of the uncertainty of different
structural material properties on the sound transmission loss.
© 2020 Elsevier Ltd. All rights reserved.

1. Introduction

Cross-laminated timber (CLT) panels are constructed from a number of layers of wooden beams laid at right-angles in
adjacent layers and bonded with adhesive [2]. Layering the timber in this way gives the panel relatively high stiffness in
all directions. Note that CLT panels have relatively low mass compared with other traditional building materials such as
concrete or brick [3]. Because of these advantages, and the fact that CLT structures can be assembled quickly and easily


Corresponding author.
E-mail address: yi.yang@auckland.ac.nz (Y. Yang).

https://doi.org/10.1016/j.jsv.2020.115842
0022-460X/© 2020 Elsevier Ltd. All rights reserved.
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

on-site, CLT is gaining in popularity as a building material in many countries [4]. However, due to their low mass and high
stiffness, CLT structures are prone to poor acoustic performance.
There have been a number of recent experimental studies to determine the acoustic performance of CLT. For example,
Schoenwald et al. [5] made a series of measurements quantifying the impact noise of CLT floors, air-borne sound transmis-
sion through CLT walls and flanking transmission through junctions between CLT panels, which were used as input data
for evaluating the acoustic performance of a large building structure using ISO 15712 (EN12354). Anders et al. [6] collated
results from a large number of well-controlled laboratory measurements of the impact insulation of various CLT-based floor
constructions. The size of the data set allowed general trends for different types of floor construction to be identified. Other
studies on the impact sound insulation of different CLT floor systems are also presented by Zeitler et al. [7], Di Bella et
al. [8] and Caniato et al. [9]. Hoeller et al. [1] published the results of a comprehensive experimental study on the sound
transmission loss of CLT wall and floor systems and also presented a method for calculating the flanking sound transmission
based on measurements of the vibration reduction index of different types of panel junctions. Barbaresi et al. [10] presented
the results of an experimental campaign investigating flanking sound transmission. Di Bella et al. [11] presented a study
investigating the accuracy of the ISO 12354 noise prediction method to calculate the sound transmission through a CLT
structure.
In addition to these primarily experimental investigations, a number of studies have presented models for predicting the
vibroacoustic behaviour of CLT panels. However, because they have a complex structure and are made from wood, which has
uncertain and variable mechanical properties, predicting the vibroacoustic behaviour of CLT structures is difficult [12,13]. CLT
panels are made from various woods (e.g. spruce or pinus radiata) and the properties of the wood are dependent on many
factors including the tree’s growing conditions and what part of the tree the wood is taken from. The beams are cut from
tree trunks that have different mechanical properties in directions parallel and perpendicular to the grain pattern. Knots can
be found randomly distributed throughout the beams and the grain slope can also vary [3,13].
For the purposes of modelling the structural and vibroacoustic behaviour of CLT, the mechanical properties of the panel
need to be determined. There have been a number of studies to determine the material properties of wood. For example,
Bucur and Archer [14] assume that wood can be modelled as a homogeneous, orthotropic material and they describe an
experimental method for determining the corresponding elastic moduli and Poisson ratios for this model. It is also assumed
that these properties are frequency independent. Their method involves exciting a small wooden sample with ultrasonic
waves and measuring the wave speed in different directions. The material properties of the assumed model are then de-
duced using the measured wave speeds. According to Steiger et al. [15], the stiffness properties of CLT panels parallel and
perpendicular to the grain direction of the face layers can be derived on the basis of the mechanical properties of the raw
material (layers) using compound theory [13] or other methods. Bending tests using strip-shaped specimens cut from panels
is another method used to evaluate the stiffness properties of CLT. However, such tests are apparently difficult and are not in
all cases a reliable indicator of the CLT panel’s real mechanical performance [15,16]. Steiger et al. [15] measured the elastic
properties of CLT panels using modal analysis of an entire panel and static bending tests on CLT panels and strips cut from
the panel. They presented values for the Young’s and shear moduli for an equivalent homogenised orthotropic panel. Gsell et
al. [17] described an experimental method for determining the elastic properties of a CLT panel. The panel was modelled as
an equivalent homogenised orthotropic plate (using Reddy’s higher order plate theory) and the corresponding elastic prop-
erties of this plate were estimated by matching the measured modal frequencies with those calculated theoretically for the
homogenised panel. The purpose of the studies of both Gsell et al. [17] and Steiger et al. [15] appears to be to determine
mechanical properties for structural analysis. A number of other studies have also measured the material properties for the
purpose of predicting vibroacoustic behaviour. These typically model the panel as an equivalent homogenised orthotropic
panel with mechanical properties which vary with frequency. For example, Van Damme et al. [18] performed an experimen-
tal investigation to determine the mechanical properties of a 3-layer CLT panel. Material properties in a particular direction
were measured by cutting two, three-layer ‘beams’ from a CLT panel (with the cuts made in different directions) and the
natural frequencies of the bending modes of each beam were compared with those of “equivalent” Euler-Bernoulli and Tim-
oshenko beams with material properties (Young’s and shear moduli along the axis of the beam) determined to best match
the experimental measurements. This approach is based on that described in [19]. It was assumed that the elastic moduli
for both beams are equal to those of the panel along the axis of the beam when it forms part of the panel. Van Damme
et al. [18] observed that in order to obtain a good fit with experimental data, the material properties needed to vary with
frequency and only frequencies below 1500 Hz were considered. Santoni et al. [20] presented a method for determining the
frequency dependent elastic properties of an equivalent homogeneous orthotropic panel using measurements of the bending
wave speed. Other methods have also been used to measure these properties based on experimental measurements of the
wave propagation characteristics within the panel. Van Damme and Zemp [21] investigated two methods for measuring the
dispersion curves in a 3-layer CLT beam and used these methods to determine the elastic properties. They modelled the
CLT beam as either a Timoshenko beam or a three-layer sandwich beam with a soft core and the elastic properties were
determined such that the theoretical dispersion curves best matched the measured curves. These material properties did not
vary with frequency. Van Damme et al. [22] described a similar method for calculating the elastic properties of a three-ply
CLT panel.
There have been several attempts to develop simple analytical models for predicting sound transmission through, and
radiation from, CLT panels that utilise experimental methods for measuring the mechanical properties of the CLT panel.
For example, Santoni et al. [23,24], presented a model to predict the modal-average radiation efficiency of a CLT panel.

2
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

Fig. 1. Sound transmission through an infinite CLT panel: the incidence direction of plane waves to the normal to the lower surface is defined by angle θ .

The method assumes that the panel can be modelled as being equivalent to a thin, homogenous, orthotropic plate with
mechanical properties that are dependent on frequency. The mechanical properties of the equivalent plate were determined
so that the free-wavenumbers along the axes of orthotropy were equal to those measured. The wavenumber measurement
method was based on that described by Nightingale et al. [25]. Santoni et al. [26] presented a method for determining the
radiation efficiency of a point excited CLT panel and validated the method against experimental measurements. The method
calculates the stiffness characteristics using the technique described in Santoni et al. [27]. A similar method was applied to
calculate sound transmission loss in [27]. Lubos et al. [28] presented a method for calculating the radiation efficiency and
sound transmission loss of a CLT panel in which the panel was modelled as either an equivalent isotropic or orthotropic
plate with the equivalent plate having a bending stiffness calculated so that the longitudinal wave speed matched that
measured using an ultrasonic method. The calculated sound transmission loss was in modest agreement with measurements,
with relatively poor agreement at high frequencies. Winter et al. [29] analysed the vibroacoustic behaviour of isolated and
connected CLT panels using an energy distribution approach. They noted that “through-thickness effects” became important
at high frequencies and that thin plate models were not capable of modelling this behaviour. They therefore modelled the
CLT panels as thick, homogeneous, orthotropic panels with material properties calculated using a law-of-mixtures approach.
The current state-of-the-art, as described above, is to model the vibroacoustic behaviour as being identical to that of an
equivalent orthotropic plate or solid panel, whose elastic properties are determined by measurement or by a law-of-mixtures
approach. These properties might be frequency dependent. However, such simple models are an approximation that might
not accurately model the complicated wave motion in laminated panels. Alternatively, the vibroacoustic behaviour of the
panel can be modelled using a more accurate approach – such as the method described here.
In this paper, the vibroacoustic behaviour of a CLT panel is modelled using the wave and finite element (WFE) method
[30-33]. Two WFE models are presented. In the first, the CLT panel is modelled as a laminate with each layer modelled as a
thick orthotropic solid with constant material properties; in the second, the CLT panel is approximated as a single-layer or-
thotropic solid material with constant material properties. The parameters of the equivalent orthotropic panel are estimated
using a ‘law-of-mixtures’ or ‘smeared’ approach. The WFE method is computationally efficient and straightforward to imple-
ment as only a small segment of the structure is modelled using conventional finite element (FE) methods. The approach
is used to calculate the sound transmission through a CLT panel and also to understand the complex wave motion of the
panel. The vibroacoustic behaviour of the layered and the single layer homogenised models are compared with each other,
and also with results predicted for a thin orthotropic plate model. The two WFE models, and the equivalent orthotropic
plate model, predict similar behaviour at low frequencies, but the predicted wave modes and sound transmission loss differ
significantly at high frequencies. The contributions to the sound transmission from individual in-vacuo wave modes are es-
timated, allowing them to be ranked in importance. Results for the layered model are then compared with experimentally
measured sound transmission loss (TL), showing good agreement. Finally, the paper concludes with a study exploring the
effect of variations in material properties on the sound transmission loss of the panel.

2. Review of the WFE method

In this section, the WFE method presented in [34] for predicting sound transmission through panels is briefly described.
Consider the system shown in Fig. 1. The panel is assumed to be infinitely large with an acoustic fluid (air, here) on both
sides. A time-harmonic plane wave at frequency ω is assumed to be incident on the lower surface of the panel, and produces
a reflected and a transmitted wave. (Note that time harmonic dependence of the form exp(iωt ) is assumed throughout). The
relationships between the amplitudes of the pressures of the incident, reflected and transmitted waves pi , pr and pt and
transverse displacements (in the z-direction) w1 and w2 of the outer surfaces of the structure are given by

pi − pr = D f w1 , pt = D f w2 , (1)

3
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

Fig. 2. Three-layer CLT panel and FE model: (a) CLT panel with a total thickness of h and thicknesses (h1 , h2 ) of individual layers, length a, width b and
ply angles of 90◦ /0◦ /90◦ ; (b) FE model of a small segment of dimensions Lx × Ly × h; DOFs of the hyper-nodes at the corners are collected in vectors
q1 , q2 , q3 , q4 .

where
iρω2
Df = (2)
kz

is the dynamic stiffness of the air, ρ is the density of air, ω is the angular frequency, kz = k2 − k2x − k2y is the wavenumber
in the z direction, k = ω/c is the acoustic wavenumber, c is the sound speed in air and kx and ky are the trace wavenumber
components in the x- and y-directions. For a plane wave with an incidence angle θ to the z-axis and an azimuthal angle φ ,
kx and ky are
kx = k sin θ cos φ , ky = k sin θ sin φ . (3)
The through-thickness construction of the panel can be arbitrarily complicated (e.g., panels consisting of wood and damp-
ing layers) and thus the WFE method is particularly well-suited for modelling layered structures. Only a small segment of
the structure is modelled using conventional finite elements. Typically, this segment is discretised by using one element in
the x- and y- directions and a number of elements through the thickness. For example, Fig. 2(a) shows a three-layer, sym-
metric CLT panel with ply angles of 90◦ /0◦ /90◦ while Fig. 2(b) shows an example FE model of a small, rectangular segment
of the panel discretised using 2 solid elements per layer. If there are internal nodes in the FE model, dynamic condensation
can be used to eliminate the internal degrees of freedom (DOFs) [35]. The element size must not be too large compared
to the shortest structural/acoustic wavelength, otherwise the structural motion cannot be accurately captured by the WFE
model. For further information, the reader is referred to [36] which presents a discussion of the numerical issues which
can be encountered when using the WFE method and techniques to avoid or remove them. The equation of motion of the
segment can be written as
 
K − ω 2 M q = f + e, (4)
where K and M are, respectively, the stiffness and mass matrices of the segment, q is the vector of nodal DOFs, f is the
vector of internal nodal forces and e is the vector of external nodal forces induced by the acoustic pressures on the exterior
surfaces of the panel. Damping could be included by way of a viscous damping matrix C or, as here, by the stiffness matrix
being complex, e.g. K(1 + iη ), where η is the loss factor. The vectors can be partitioned as
 T  T  T
q= qT1 qT2 qT3 qT4 , f= fT1 fT2 fT3 fT4 , e= eT1 eT2 eT3 eT4 (5)
with each partition corresponding to a corner of the rectangular segment. The external nodal forces corresponding to the
out-of-plane displacements of the surfaces can either be lumped at the nodes or calculated as consistent nodal forces using
 
ej = ± p(x, y )N j (x, y )dxdy (6)
S

where e j is an element of the external nodal force vector for the j-th node, the ± signs apply to the lower and upper
surfaces respectively, p(x, y ) is the total acoustic pressure on S and N j (x, y ) is the relevant shape function associated with
the transverse displacement of the surface S at the j-th node.

4
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

Periodic structure theory and equilibrium conditions (Eqs. (15-16) in [34]) are used to postprocess the mass and stiffness
matrices and the nodal force vectors. The governing equation for the structural response to acoustic excitation by an incident
plane wave can then be written as

D(kx , ky , ω )q1 = 2 pi ε u1 , (7)


where D(kx , ky , ω ) = Ds + +ε D f u1 uT1 ε D f u2 uT2
is the total spectral dynamic stiffness matrix (DSM), where Ds = K − ω2 M
is
the structural DSM, q1 is the vector of the nodal DOFs of hyper-node 1, ε is a spectral coefficient (cf. Eq. (21) in [34])
and u1 and u2 are fluid/structure coupling vectors whose elements are all zero apart from those which correspond to the
transverse displacements w1 and w2 , which are equal to one.. Note that the fluid dynamic stiffnesses only change the diag-
onal elements of the structural DSM. The effects of fluid loading are typically negligible for applications involving CLT and
air, and hence those DSMs can be neglected, except around coincidence frequencies where they can add significant damp-
ing through radiation. Once Eq. (7) is solved to yield the DOF vector q1 , the amplitudes of the reflected and transmitted
pressures can be calculated by inserting w1,2 = uT1,2 q1 into Eq. (1). The plane-wave sound transmission loss is defined as
T L = −10log10 [τ (θ , φ )],where
 
 p2 
τ (θ , φ ) =  2t  (8)
p i

is the plane-wave power transmission coefficient. The diffuse-field sound transmission loss is defined as T Ld = −10log10 [τd ],
where
 2 π  θ0
τ (θ , φ ) sin θ cos θ dθ dφ
τd (ω ) = 0  02π  θ0 (9)
0 0 sin θ cos θ dθ dφ

is the diffuse-field power-transmission coefficient. For a perfectly diffuse sound field, the upper limit of integration θ0 should
equal 90◦ . Some authors suggest using a limit of approximately 78◦ , which has been observed to provide better agreement
between predictions and measurements [37].
For the case of in-vacuo free waves propagating in the structure, D f = 0 and there are no external nodal forces acting on
the nodes of the segment, i.e. e = 0 in Eq. (4). The wavenumbers can be determined by forming an eigenproblem involving
the frequency, ω, and the free wavenumber components, kx and ky . The form of the eigenproblem depends on the nature
of the solution sought and further discussion can be found in [38]. In the numerical examples below, the frequency and
one wavenumber component, say ky , are specified and the eigenproblem is solved to determine the wavenumbers, kx , of
propagating waves at this frequency: to determine the freely propagating waves, typically damping is neglected and real-
valued solutions for kx and ky are found.

2.1. In vacuo wave modes and contribution to the transmitted power

The response to an acoustic wave with trace wavenumbers kx and ky can be found by inverting the DSM, D in Eq. (7),
and the transmitted pressure then follows straightforwardly. In many cases, however, the fluid loading is light so that the
response can be approximated accurately by neglecting the fluid dynamic stiffnesses D f . The matrix Ds can then be inverted
by eigendecomposition [39] and explicitly written as the sum of contributions in each of the in-vacuo wave modes. In
essence this is equivalent to projecting D−1 onto the in-vacuo wave modes. Each wave mode then contributes an amount to
the transmitted power that depends on how well the wave mode couples to the surrounding fluid and whether the wave
is acoustically fast (in which case there is coincidence). The contribution of each of these modes to the transmitted power
can then be estimated and the wave modes ranked in importance. This is useful both for interpreting the vibroacoustic
behaviour of a structure and for identifying the dominant modes (see [40]).

3. Vibroacoustic modelling of CLT

3.1. Modelling approach

In this section, two WFE models (model 1 and model 2) and a thin, orthotropic plate model (model 3) of a CLT panel are
developed. Numerical predictions generated using these models are compared. The models are intended to represent a 0.2 m
thick CLT panel with a mean density of 450 kg m−3 and comprising six layers in a stacking sequence 0◦ /90◦ /0◦ /0◦ /90◦ /0◦ .
The first WFE model represents the CLT panel as a laminate comprising six orthotropic solid layers, so that, for example, the
shear in each layer can be different. The second WFE model represents the CLT panel as a single, homogenized, orthotropic
thick panel with properties derived based on the proportional thicknesses of the individual layers using a rule of mixtures
[13]: this form of model is sometimes used in the literature [29]. These WFE models have enough elements through the
thickness so that through-thickness effects and shear deformation are modelled accurately. The thin orthotropic plate model
is taken from [41] and is commonly used in the literature (e.g. [26,28].): the relevant properties of this plate are identical to
those of model 2. The orthotropic plate model uses the bending stiffnesses in the principle directions and an out-of-plane
shear stiffness for describing the out-of-plane motion while the shear deformation through the thickness of the panel is

5
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

Table 1
Material properties and stacking sequence. Material properties used in model 1 (laminated panel), model 2 (equivalent homogeneous or-
thotropic solid) and model 3 (orthotropic thin plate).

Material property Model 1 Model 2 and Model 3


−2
Ex [Nm ] 1.1 × 10 10
8.243 × 109
Ey [Nm−2 ] 3.67 × 108 3.123 × 109
Ez [Nm−2 ] 3.67 × 108 3.67 × 108
Gxy [Nm−2 ] 6.9 × 108 6.9 × 108
Gyz [Nm−2 ] 6.9 × 107 9.857 × 107
Gxz [Nm−2 ] 6.9 × 108 1.725 × 108
νyx 0.014 0.339
νzy 0.030 0.122
νzx 0.014 0.063
η 0.020 0.020
ρ [kgm−3 ] 450 450
Damping loss factor 0.02 0.02
Stacking sequence∗ 0◦ /90◦ /0◦ /0◦ /90◦ /0◦ 0◦

N.B. The local x-axis of each layer is along the grain and aligned at the angle relative to the global x-axis specified in the stacking
sequence row.

Fig. 3. Dispersion curves of CLT panel for waves propagating in the x-direction
(ky = 0): model 1, model 2 and model 3. Bending wave (1), in-plane shear wave (2) and extensional wave (3).

modelled using linear theory with a correction factor of 5/6 being adopted to account for the non-uniform distribution of
the through-thickness shear stress. The material properties used in all three models are given in Table 1 and are based
on a panel described in [29]. It is assumed that these properties do not vary with frequency. The purpose of this section
is to assess the differences between the vibroacoustic behaviour predicted by all three models. For this purpose, both the
dispersion curves and diffuse-field sound transmission loss are predicted using all three models and compared.
Each of the two WFE models uses an FE model of a small segment of the CLT panel with dimensions Lx = Ly = 5mm
and which comprises eighteen ANSYS SOLID185 elements through the thickness with 3 elements per layer. Each element
has eight nodes with three DOFs per node, corresponding to translations in the x-, y- and z-directions. In order to avoid
potential numerical issues [36], Zhong’s method [42] was adopted in the WFE models for calculating the dispersion curves

6
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

Fig. 4. Dispersion curves of CLT panel for waves propagating in the x-direction
(ky = 0): model 1 and model 2. Bending wave (1), in-plane shear wave (2) and extensional wave (3).

Fig. 5. Diffuse-field transmission loss of CLT panel (dB ref. 1): model 1 (laminate model), model 2 (homogenized solid model) and model 3 (thin plate
model).

7
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

Fig. 6. Oblique sound transmission loss of CLT panel, model 1 (dB ref. 1): θ = 10◦ , φ = 0◦ and θ = 15◦ , φ = 0◦ and θ = 20◦ , φ = 0◦ . Dips at 2427Hz, 2591Hz,
3237Hz and 4680Hz induced by coincidence.

Fig. 7. Dispersion curves of CLT panel and acoustic trace waves: bending wave (1) and extensional wave (3) and higher-order wave (5); Trace waves:
θ = 10◦ , φ = 0◦ , and θ = 15◦ , φ = 0◦ , and θ = 20◦ , φ = 0◦ .

8
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

z (a)
x
w
u

z (b)
x
w
u

z (c)
x
w
u

Fig. 8. Wave mode shapes of the CLT panel, WFE model 1, displacements u and w in the x- and z- directions: (a) branch 1, at 1 kHz; (b) branch 3, at
2427 Hz; (c) branch 5, at 4680 Hz.

of structures. For the orthotropic plate model, the analytical equations presented in chapter II of [43] are used to estimate
the wavenumbers.
Fig. 3 shows the in-vacuo dispersion curves predicted using the three models for waves propagating in the x-direction
(ky = 0). Damping is neglected for clarity, so that the propagating wavenumbers are pure real. At low frequencies (be-
low 10 0 0 Hz, Fig. 3), all three models predict behaviour similar to that expected of a thick plate with three typical wave
branches: a bending wave (branch 1) which also involves significant transverse shear, an in-plane shear wave (branch 2)
and an extensional wave (branch 3). There is good agreement between the wavenumbers predicted by all three models for
the extensional and shear waves. The dispersion curves predicted by both WFE models for the bending wave are in good
agreement below 400 Hz, but diverge as frequency increases above this value. The bending wave dispersion curve predicted
by the orthotropic plate model is in reasonable agreement with the two WFE models below 100 Hz, but the dispersion
curves diverge significantly as frequency increases above this value, partly because transverse shear effects become impor-
tant. Fig. 4 shows the dispersion curves for waves propagating in the x-direction predicted using the WFE models up to
50 0 0 Hz. This figure demonstrates how, at higher frequencies, the behaviour becomes much more complicated with many
new wave branches cutting on: this behaviour cannot of course be captured by the orthotropic thin plate model. At lower
frequencies, branch 3 is an extensional wave mode that also involves Poisson contraction. Branch 4 (first antisymmetric
shear mode) cuts on at 1160 Hz while branch 5 (first antisymmetric extensional mode) cuts on at 1500 Hz. Branch 6, which
cuts on at about 2350 Hz, involves a symmetric, through-thickness extensional motion, similar to the mass-spring-mass
resonance associated with double-skinned panels. Branches 3 and 6 couple strongly, veer apart and above 2500 Hz or so
are broadly a through-thickness resonant mode (branch 3) and an extensional wave (branch 6), the former being a strong
radiator. Above 2700 Hz, there are a number of higher-order wave branches cutting on at high frequencies, such as branch
7 at about 4829 Hz.
Fig. 5 shows the one-third octave band diffuse-field transmission loss calculated using all three models. The results are
in close agreement at low frequencies (below approximately 250 Hz) but the agreement between the WFE models and the
thin plate model deteriorates somewhat as frequency increases. In particular, the coincidence dip around 2500 Hz is not
captured by the orthotropic thin plate model because the associated wave mode involves through-thickness deformations

9
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

Fig. 9. Ratios of transmitted and incident powers τ , six-layer CLT panel for ky = 0, integral over kx : (a) total (dots) and contributions of individual wave
modes; (b) proportion of total transmitted power; propagating wave modes are labelled with numbers 1, 3, 5, 6, 7.

(branch 3 of the dispersion curves) which are not modelled. The good agreement at low frequencies is due to the fact
that the orthotropic plate model accurately captures the dominant wave modes responsible for sound transmission at these
frequencies. However, at high frequencies there are various higher-order wave branches which may contribute to the sound
transmission and which cannot be modelled using the orthotropic plate model. For this reason, the layered WFE model will
henceforth be used exclusively in this paper.

10
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

Fig. 10. CLT panel: (a) plane-wave power transmission coefficient; (b) dispersion curves for in-vacuo propagating waves in the (kx , ky ) plane (kx ≥ 0, ky ≥ 0)
at 500 Hz, neglecting damping. Branches 1, 2, 3 represent bending/transverse shear, in-plane shear and extensional waves respectively. The sonic
circle is indicated by the dashed red curve.

3.2. Mechanism of sound transmission through CLT panels

Fig. 6 shows the transmission loss for acoustic plane wave excitation with incidence angles (θ , φ ) = (10◦ , 0◦ ), (15◦ , 0◦ )
and (20◦ , 0◦ ), predicted using model 1. Dips occur at 2427 Hz, 2591 Hz, 3237 Hz and 4680 Hz, and are induced by coinci-
dence between the structural wavenumbers and the trace wavenumbers of the incident acoustic wave. This is demonstrated
in Fig. 7, which shows the dispersion curves predicted by the WFE model and the trace wavenumbers of the incident waves.
The wave mode shapes of branch 1 (bending/transverse shear wave, the first anti-symmetric transverse wave mode) at
1 kHz, branch 3 (first symmetric extensional wave mode) at 2427 Hz and branch 5 (higher-order, first antisymmetric ex-
tensional wave mode) at 4680 Hz are shown in Fig. 8. (Note that the shape of each wave mode changes with frequency).
The wave mode shape associated with branch 1 is antisymmetric: this mode is an out-of-plane bending/transverse shear
dominated wave mode. The wave mode shape associated with branch 3 is symmetric: at low frequencies this is an exten-

11
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

Fig. 11. CLT panel: (a) plane-wave sound power transmission loss coefficient; (b) dispersion curves for in-vacuo propagating waves in the (kx , ky ) plane
(kx ≥ 0, ky ≥ 0) at 3500 Hz, neglecting damping. Important wave branches are labelled with 1, 3 and 6. The sonic circle is indicated by the dashed
red curve.

sional wave with the out-of-plane motion associated with Poisson contraction, but at higher frequencies, above the through-
thickness resonance at 2300 Hz or so, significant out-of-plane motion occurs. Branch 5 involves antisymmetric extensional
motion of the panel, with higher-order through-thickness behaviour.
Fig. 9 shows the ratio of transmitted to incident powers for the purpose of illustrating the contributions of individual
wave modes to the total transmitted power. These calculations were performed using the method described in Ref. [40] with
the effects of cross-modal coupling neglected. The contribution of each individual wave mode to the total transmitted power
is calculated from Eq. (11) and Eq. (15) in [40]. Fig. 9(a) shows the total transmitted power and the power associated with
the individual wave modes, while Fig. 9(b) shows the proportion of the total transmitted power from the individual modes.
The curves are numbered such that they correspond to the wave branches shown in Fig. 4. As can be seen, only wave modes
1, 3, 5 and 6 make a significant contribution to the transmitted sound power at frequencies below 5 kHz. At low frequencies
(below 2200 Hz or so), the total sound power transmission ratio is dominated by that of the bending wave (branch 1). At
2359 Hz, 4483 Hz and 4829 Hz the behaviour becomes complicated due to the cut-on of higher modes and the coupling

12
Y. Yang, C. Fenemore, M.J. Kingan et al.
13

Journal of Sound and Vibration 494 (2021) 115842


Fig. 12. Diffuse field transmission loss (one-third octave band) (dB ref. 1): (a) comparison between 3-layer panel (curve 1), 5-layer panel (curve 2) and 7-layer panel (curve 3); (b, c, d) comparison with
measurement: WFE, measurement [1].
Y. Yang, C. Fenemore, M.J. Kingan et al.
14

Journal of Sound and Vibration 494 (2021) 115842


Fig. 13. Power transmission coefficient for the 3-layer CLT panel as a function of kx and ky at (a) 30 0 Hz, (b) 50 0 Hz and (c) 4.25 kHz; —— wavenumbers of propagating waves in vacuo; —— sonic circle (in (c)
the sonic circle lies outside the range of wavenumbers shown).
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

Fig. 14. Diffuse-field sound transmission loss of three-layer CLT panel (dB ref. 1): measurement [28] , WFE model, isotropic plate
model and orthotropic plate model.

between through-thickness modes and the in-plane waves. There are 4 main wave modes that radiate sound efficiently at
these higher frequencies. Between 2400 Hz and 4320 Hz, two wave modes (branches 1 and 3) are mainly responsible for
sound transmission.
Wave modes 1, 3, 5 and 6 dominate the sound transmission over different frequency ranges below 5 kHz. Other wave
modes do not make a significant contribution at any frequency. The primary reason is that the out-of-plane displacements
of these wave modes couple well with the air and, if excited, contribute significantly to sound transmission through the
panel. Conversely, other wave modes (e.g. 2), whose primary displacements occur in-plane, do not couple well with the
acoustic fluids and thus do not significantly influence the sound transmission loss behavior of the panel.
Fig. 10(a) shows the plane-wave power transmission coefficient (defined in Eq. (8)) as a function of the trace wavenum-
ber components kx and ky at 500 Hz while Fig. 10(b) shows the purely-real in-vacuo dispersion curves neglecting damping.
Overlaid on these curves is the “sonic circle” in the (kx , ky ) plane, i.e. the wavenumbers of sound in air: acoustic waves can
only be excited by structural wavenumber combinations inside the red sonic circle. Only one quarter of the wavenumber
plane is shown due to symmetry. At 500 Hz there are only three propagating wave modes which correspond to the three
wave branches (bending (1), shear (2) and extensional waves (3)) shown in Fig. 4. The dispersion curve of the bending wave
is near-elliptical, which indicates that the panel behaves like an orthotropic plate, while the other two branches interact
because of the anisotropy. The wavenumbers of the three wave branches are smaller than those of the sonic circle, indi-
cating that they are coincident waves. Fig. 10(a) shows that the sound transmission loss at this frequency is dominated by
transmission at wavenumbers close to those which coincide with the propagating bending wave, which couples well with
the acoustic field. There is little contribution from wavenumbers close to the dispersion curves of the other two free waves.
(cf. the proportion of power transmitted by this wave mode in Fig. 9(b) for ky = 0 at 500 Hz.)
In contrast, Figs. 11(a) and (b) show the plane-wave transmission loss coefficient, the in-vacuo, undamped dispersion
curves of propagating waves at 3500 Hz and sonic circles. Again, sound transmission is dominated by those wavenumber
components that correspond to well-coupled propagating waves. The peaks are associated with the bending/transverse shear
mode (branch 1) and the first through-thickness mode (branch 3), illustrated in Fig. 8(a) and (b) respectively, which both
transmit sound power efficiently. At this frequency, branch 3 is the stronger radiator (cf. Fig. 9(b) at 3500 Hz for ky = 0).

15
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

Fig. 15. Mean and range of diffuse-field transmission loss (three-layer CLT panel) (dB ref. 1): (a) Ex ; (b) Ey ; (c) Ez and (d) Gxy ; range and mean.

Table 2
Assumed material properties of the CLT panels and linings.

Property 3-layer panel 5-layer panel 7-layer panel Gypsum board


−2
Ex [Nm ] 11 × 10 9
11 × 10 9
11 × 10 9
3.37 × 109
Ey [Nm−2 ] 3.67 × 108 3.67 × 108 3.67 × 108 3.37 × 109
Ez [Nm−2 ] 3.67 × 108 3.67 × 108 3.67 × 108 3.37 × 109
Gxy [Nm−2 ] - -
Gxz [Nm−2 ] - -
Gyz [Nm−2 ] - -
vxy 0.42 0.42 0.42 0.30
vxz 0.42 0.42 0.42 0.30
vyz 0.30 0.30 0.30 0.30
ρ [kgm−3 ] 543.6 522.3 530 777.5
h [mm] 78 175 245 12.7
Lx =Ly [mm] 6.5 10 10 -

4. Comparison with measurements

4.1. Comparison with the measurements of Hoeller et al. [44]

The WFE method described in section 2 was used to model the diffuse-field sound transmission loss of the 3, 5 and
7 layer CLT panels which were used in the experiments reported by Hoeller et al. [44]. These experiments measured the
diffuse-field sound transmission loss of CLT panels with 2 layers of 12.7 mm gypsum board lining attached to one side of
the CLT panel. The properties of the CLT panels and the gypsum board linings used in the WFE models are presented in
Table 2. The material properties of the timber are the same as those used in Model 1 in Table 1, and the densities are given
in Hoeller et al. [44]. The material properties of gypsum board were averaged from several sources [45-47]. The CLT panels
are assumed to be constructed from layers of equal thickness. The FE models were constructed using a single stack of ANSYS

16
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

Fig. 16. Mean and range of diffuse-field transmission loss (three-layer CLT panel) (dB ref. 1): (a)Gxz ; (b) Gyz ; (c) ρ and (d) η; range and mean.

SOLID185 elements. The 3-layer model used 4 finite elements per CLT layer and 2 elements per gypsum board layer, and the
5- and 7-layer models used 3 elements per CLT layer and 1 element per gypsum board layer. This was done to ensure the
shapes of the elements were approximately cubic. The diffuse-field sound transmission loss was calculated by truncating
the integral over θ at 78°. This is a common method used to account for differences between testing a finite panel in a
reverberation chamber and a prediction made assuming an infinite panel. The truncation particularly affects the calculated
values of the diffuse-field transmission loss at low frequencies, as significant sound transmission occurs at or near θ = 90°
at those frequencies.
The predicted transmission losses are shown in Fig. 12(a) and compared with the experimental measurements in
Fig. 12(b, c, d). The shapes of the transmission loss curves are similar, although there are discrepancies in levels at some
frequencies – particularly close to the coincidence dips which occur at slightly different frequencies. Discrepancies can be
partly attributed to differences between the assumed and actual material properties for these panels which were estimated,
rather than measured. In particular, the predictions all exhibit two coincidence dips, one at low frequencies (between 100 Hz
and 500 Hz) and one at high frequencies (between 1500 Hz and 4500 Hz) with the transmission loss increasing exponen-
tially (linearly in dB) with frequency in the region between the two dips. The frequencies of the coincidence dips decrease
as the panel thickness increases. This behavior is also observed in the experimental results, although the exact location of
the dips is somewhat different between the predictions and measurements.
Fig. 13 shows the calculated value of τ (kx , ky ) for the 3-layer CLT panel in the (kx , ky ) plane at 300 Hz, 500 Hz and
4.25 kHz. The supplementary material contains animations of these curves (plotted in the (θ , φ ) plane) for the frequency
range 200 - 5000 Hz. The dispersion curves of the in-vacuo, undamped propagating waves are overlaid in black whilst the
red curve indicates the structural waves with sonic speed. At 300 Hz and 500 Hz, three dispersion curves are predicted.
These correspond to bending, shear and axial waves, as observed previously. However, only the bending wave corresponds
to large transmission due to its strong coupling with the air. At higher frequencies, higher order wave modes cut on. In
particular, there is the wave mode that involves symmetric motion of the panel, with higher-order through-thickness be-
havior. Coincidence with this wave also contributes significantly to the diffuse sound transmission loss as clearly shown in
Fig. 13 (c). This behavior is similar to that of the panel considered in section 3.

17
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

Table 3
Material properties and stacking sequence of three-layer CLT panel.

3-layer CLT panel

Ex = 1.07 × 1010 Nm−2 ,Ey = 6.31 × 108 Nm−2 ,Ez = 1.37 × 109 Nm−2 Gxz =
1.33 × 109 Nm−2 ,Gxy = 1.284 × 109 Nm−2 ,Gyz = 1.07 × 108 Nm−2 ,νxy =
0.462,νyz = 0.53,νxz = 0.422, stacking sequence: 90◦ /0◦ /90◦ .

4.2. Comparison with the measurements of Krajci et al. [28]

In this section, the diffuse-field sound transmission loss of a three-layer CLT panel is estimated using the WFE model
and the results are compared with experimental data and predictions from an orthotropic plate model and an equivalent
isotropic model presented in [28]. The effects of variability in the material properties on the sound transmission loss are
then presented.
The CLT panel considered in this section has overall dimensions of 4.18 m × 2.89 m × 0.08 m and consists of three layers
of thick softwood boards with an average density ρ = 522 kg m−3 . The first and third layers have the same thickness of
15 mm and the thickness of the middle layer is 50 mm. The material properties and stacking sequences of the layers are
taken from Table 3 of [12] and are listed in Table 3. The loss factor η is taken to be 0.05.
The panel is modelled using the full, layered WFE method described in section 2. A small segment of the panel with
dimensions of 2 mm × 2 mm × 80 mm was modelled using 16 ANSYS SOLID 185 elements each of identical size, with the
three layers being modelled by 3, 10 and 3 elements respectively. Each element had eight nodes with three DOFs at each
node: translations in the x-, y- and z-directions. The WFE model was used to calculate the diffuse-field sound transmission
loss. Unlike in the previous section, the results presented here account for the finite size of the panel using the spatial
windowing technique described in [48]. The predicted and measured one-third octave band sound transmission loss values
(digitised from [28]) are shown in Fig. 14. The WFE predictions are in good agreement with the measurements except
at low frequencies (below 200 Hz) and around the measured coincidence frequency (400 Hz). The discrepancies below
200 Hz could be due to the effect of panel resonances which are not captured in the models while the discrepancies around
the coincidence frequency could be caused by differences in the assumed and actual material properties. Also shown are
predictions (digitised from [28]) for an infinite orthotropic plate model and a finite equivalent isotropic plate model, whose
material parameters were calculated using the measured wave speeds in [28]: the plate models have poorer accuracy.
The differences observed in Fig. 14 between the measured and predicted sound transmission loss - particularly around
the coincidence frequency - might be partially due to errors in the assumed values of the material properties used in the
WFE model. In order to investigate the effects of variability in the properties of the CLT panel on its sound transmission
loss, a population of panels with elastic properties, densities and damping loss factors that have mean values listed in
Table 3 were considered. These material parameters are assumed to have a uniform distribution within the panel population
over the interval [m-r/2, m+r/2], where m and r are the mean and range of the parameter. The ranges, expressed as a
percentage of the mean value, are 20 % for density, 30 % for elastic properties and 100 % for the damping loss factor. A
number of simulations were performed to illustrate the effects of variability in the properties by varying one property at a
time. Figs. 15-16 show the mean and range of the diffuse-field sound transmission loss for variations in the Young’s moduli
Ex , Ey , Ez , the shear moduli Gxy , Gyz , Gxz , the density and the damping loss factor. It can be seen that the diffuse-field sound
transmission loss is sensitive to the along-grain Young’s modulus Ex between 183-536 Hz, i.e. below and around coincidence,
with a variance up to 7 dB and the shear modulus Gyz at higher frequencies, between 354-1800 Hz, with a variance up to
5 dB. Variability in density ρ has an important influence on the diffuse-field TL at and below coincidence, while variability
in loss factor η is important at and above coincidence. The TL is not significantly dependent on variability in the cross-grain
Young’s moduli Ey , Ez and the shear moduli Gxy and Gxz .

5. Conclusions

In this paper, the vibroacoustic characteristics of CLT panels were investigated using a wave and finite element (WFE)
method. A WFE model which represented the panel as a laminate of thick, orthotropic, solid layers was presented and
compared with a WFE model which represented the panel as a homogenous, orthotropic solid panel and an analytical
model which represented the panel as a ‘homogenised’ orthotropic plate. The dispersion curves and sound transmission
loss were calculated. It was seen that the CLT panel behaves like an orthotropic plate at low frequencies with three typical
wave branches: bending (with significant transverse shear), in-plane shear and in-plane extensional waves. At higher fre-
quencies, higher-order wave modes can propagate. These involve through-thickness deformations that the orthotropic plate
model cannot capture. Some of these wave modes, and in particular the first through-thickness resonance mode, contribute
strongly to acoustic transmission. The wave mode behavior predicted using the homogenized solid panel model also exhib-
ited quite different behavior to the layered model. This suggests that a model should take the laminated structure of the
CLT panel into account, together with transverse shear, in order to correctly predict the vibroacoustic behavior, especially
at higher frequencies. It was also shown how the WFE method can be used to identify acoustically important wave modes,
which typically have mode shapes with significant out-of-plane displacements or are acoustically fast. This was achieved by

18
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

projecting the DSM onto the in-vacuo wave modes and estimating the power radiated by each individual wave mode. Inspec-
tion of the wave mode shapes was also used to identify modes which have significant out-of-plane displacements, which
couple strongly with the air adjacent to the panel. The full, solid, layerwise WFE model was used to make predictions of
diffuse field sound transmission loss for a number of different CLT panels and the results were compared with experimental
measurements. In the simulations the material properties were assumed to be independent of frequency. The predictions
were observed to be in reasonable agreement with experimental data. Some differences between the predicted and mea-
sured levels were observed. These could be partly due to individual resonances of the finite panel, which are important at
very low frequencies, and partly due to errors in the assumed values of the material properties used in the WFE model,
which might also vary with frequency. Simulations were performed to investigate the effects of variation in the physical
properties of the panel on the sound transmission loss. It was seen that realistic variations of the elastic modulus Ex , shear
modulus Gyz , density and damping loss factor can produce significant changes in the calculated sound transmission loss. The
transmission loss was seen to be relatively insensitive to the other properties considered. The full WFE model can accurately
predict the vibroacoustic characteristics of CLT panels and can also be used straightforwardly for analysing the sensitivity of
the sound transmission loss to material properties without using approximations made by homogenised solid and thin-plate
models. While airborne sound transmission was seen to be dominated by a few wave modes (those that typically involve
significant transverse displacements of the surfaces), the other wave modes (e.g. in-plane shear or extension) can contribute
significantly to structure-borne transmission. Future work concerns the calculation of this structure-borne sound.

Declaration of Competing Interest

The authors declare that there is no conflict of interest.

CRediT authorship contribution statement

Yi Yang: Methodology, Validation, Writing - original draft. Chiaki Fenemore: Validation, Writing - original draft.
Michael J. Kingan: Methodology, Validation, Writing - original draft. Brian R. Mace: Methodology, Validation, Writing -
original draft.

Acknowledgments

The authors acknowledge the financial support provided by the New Zealand Ministry of Business, Innovation and Em-
ployment (MBIE) through a Smart Ideas Grant entitled “Predicting Sound Transmission in Lightweight Buildings”. Chiaki
Fenemore also acknowledges the support of the University of Auckland Acoustics Research Centre Doctoral Scholarship.

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jsv.2020.115842.

References

[1] C. Hoeller, J. Mahn, D. Quirt, S. Schoenwald, B. Zeitler, Apparent sound insulation in cross-laminated timber buildings, The Journal of the Acoustical
Society of America 141 (2017) 3479.
[2] Z. Wang, J. Zhou, W. Dong, Y. Yao, M. Gong, Influence of technical characteristics on the rolling shear properties of cross laminated timber by modified
planar shear tests, Maderas, Ciencia y tecnología 20 (2018) 469–478.
[3] F. Morandi, S. De Cesaris, M. Garai, L. Barbaresi, Measurement of flanking transmission for the characterisation and classification of cross laminated
timber junctions, Applied Acoustics 141 (2018) 213–222.
[4] R. Brandner, G. Flatscher, A. Ringhofer, G. Schickhofer, A. Thiel, Cross laminated timber (CLT): overview and development, European Journal of Wood
and Wood Products 74 (2016) 331–351.
[5] S. Schoenwald, B. Zeitler, I. Sabourin, F. King, Sound insulation performance of cross laminated timber building systems, in: INTER-NOISE and NOISE–
CON congress and conference proceedings, Institute of Noise Control Engineering, 2013, pp. 4337–4346.
[6] A. Homb, C. Guigou-Carter, A. Rabold, Impact sound insulation of cross-laminated timber/massive wood floor constructions: Collection of laboratory
measurements and result evaluation, Building Acoustics 24 (2017) 35–52.
[7] B. Zeitler, S. Schoenwald, I. Sabourin, Direct impact sound insulation of cross laminate timber floors with and without toppings, in: INTER-NOISE and
NOISE-CON Congress and Conference Proceedings, Institute of Noise Control Engineering, 2014, pp. 5742–5747.
[8] A. Di Bella, N. Granzotto, L. Barbaresi, Analysis of acoustic behavior of bare CLT floors for the evaluation of impact sound insulation improvement, in:
Proceedings of Meetings on Acoustics 22ICA, ASA, 2016.
[9] M. Caniato, F. Bettarello, P. Fausti, A. Ferluga, L. Marsich, C. Schmid, Impact sound of timber floors in sustainable buildings, Building and Environment
120 (2017) 110–122.
[10] L. Barbaresi, F. Morandi, M. Garai, A. Speranza, Experimental measurements of flanking transmission in CLT structures, in: Proceedings of Meetings on
Acoustics 22ICA, ASA, 2016.
[11] A. Di Bella, C.C. Mastino, L. Barbaresi, N. Granzotto, R. Baccoli, F. Morandi, Comparative study of prediction methods and field measurements of the
acoustic performances of buildings made with CLT elements, in: INTER-NOISE and NOISE-CON Congress and Conference Proceedings, Institute of Noise
Control Engineering, 2017, pp. 3555–3564.
[12] D.W. Green, J.E. Winandy, D.E. Kretschmann, Mechanical properties of wood, in: Wood handbook: wood as an engineering material, 113, WI: USDA
Forest Service, Forest Products Laboratory, Madison, 1999, pp. 4.1–4.45. 1999, General technical report FPL; GTR-113.
[13] J. Bodig, B.A. Jayne, Mechanics of wood and wood composites, Van Nostrand Reinhold, New York, 1982.
[14] V. Bucur, R. Archer, Elastic constants for wood by an ultrasonic method, Wood Science and Technology 18 (1984) 255–265.

19
Y. Yang, C. Fenemore, M.J. Kingan et al. Journal of Sound and Vibration 494 (2021) 115842

[15] R. Steiger, A. Gülzow, C. Czaderski, M.T. Howald, P. Niemz, Comparison of bending stiffness of cross-laminated solid timber derived by modal analysis
of full panels and by bending tests of strip-shaped specimens, European Journal of Wood and Wood Products 70 (2012) 141–153.
[16] R.-A. Jöbstl, T. Bogensperger, T. Moosbrugger, G. Schickhofer, A Contribution to the Design ans System Effect of Cross Laminated Timber, in: CIB W18,
39th Meeting, 2006.
[17] D. Gsell, G. Feltrin, S. Schubert, R. Steiger, M. Motavalli, Cross-laminated timber plates: Evaluation and verification of homogenized elastic properties,
Journal of structural engineering 133 (2007) 132–138.
[18] B. Van Damme, S. Schoenwald, M.A. Blanco, A. Zemp, Limitations to the use of homogenized material parameters of cross laminated timber plates
for vibration and sound transmission modelling, in: Proceedings of the 22nd International Congress on Sound and Vibration, International Institute of
Acoustics and Vibration Florence, Italy, 2015, pp. 12–16.
[19] P.-O. Larsson, Determination of Young’s and shear moduli from flexural vibrations of beams, Journal of Sound and Vibration 146 (1991) 111–123.
[20] A. Santoni, S. Schoenwald, B. Van Damme, P. Fausti, Determination of the elastic and stiffness characteristics of cross-laminated timber plates from
flexural wave velocity measurements, Journal of Sound and Vibration 400 (2017) 387–401.
[21] B. Van Damme, A. Zemp, Measuring dispersion curves for bending waves in beams: a comparison of spatial Fourier transform and inhomogeneous
wave correlation, Acta Acustica united with Acustica 104 (2018) 228–234.
[22] B. Van Damme, S. Schoenwald, A. Zemp, Modeling the bending vibration of cross-laminated timber beams, European journal of wood and wood
products 75 (2017) 985–994.
[23] A. Santoni, S. Schoenwald, B. Van Damme, H.-M. Tröbs, P. Fausti, Average sound radiation model for orthotropic cross laminated timber plates, Proc
EuroRegio (2016) 13–15.
[24] A. Santoni, P. Fausti, S. Schoenwald, H.-M. Tröbs, Sound radiation efficiency measurements on cross-laminated timber plates, in: INTER-NOISE and
NOISE-CON Congress and Conference Proceedings, Institute of Noise Control Engineering, 2016, pp. 5979–5989.
[25] T. Nightingale, R. Halliwell, G. Pernica, Estimating in-situ Material Properties of a Wood Joist Floor: Part 1—Measurements of the Real Part of Bending
Wavenumber, Building Acoustics 11 (2004) 175–196.
[26] A. Santoni, S. Schoenwald, P. Fausti, H.-M. Tröbs, Modelling the radiation efficiency of orthotropic cross-laminated timber plates with simply-supported
boundaries, Applied Acoustics 143 (2019) 112–124.
[27] A. Santoni, P. Bonfiglio, P. Fausti, S. Schoenwald, Predicting sound radiation efficiency and sound transmission loss of orthotropic cross-laminated
timber panels, in: Proceedings of Meetings on Acoustics 173EAA, The Journal of the Acoustical Society of America (2017) 015013.
[28] L. Krajči, C. Hopkins, J.L. Davy, H.-M. Tröbs, Airborne sound transmission of a cross laminated timber plate with orthotropic stiffness, Euronoise 2012
(2012).
[29] C. Winter, M. Buchschmid, G. Müller, Modeling of orthotropic plates out of cross laminated timber in the mid and high frequency range, Procedia
engineering 199 (2017) 1392–1397.
[30] B.R. Mace, D. Duhamel, M.J. Brennan, L. Hinke, Finite element prediction of wave motion in structural waveguides, The Journal of the Acoustical Society
of America 117 (2005) 2835–2843.
[31] D. Chronopoulos, B. Troclet, M. Ichchou, J. Lainé, A unified approach for the broadband vibroacoustic response of composite shells, Composites Part B:
Engineering 43 (2012) 1837–1846.
[32] C. Droz, Z. Zergoune, R. Boukadia, O. Bareille, M. Ichchou, Vibro-acoustic optimisation of sandwich panels using the wave/finite element method,
Composite Structures 156 (2016) 108–114.
[33] F. Errico, M. Ichchou, S. De Rosa, F. Franco, O. Bareille, Investigations about periodic design for broadband increased sound transmission loss of sand-
wich panels using 3D-printed models, Mechanical Systems and Signal Processing 136 (2020) 106432.
[34] Y. Yang, B.R. Mace, M.J. Kingan, Prediction of sound transmission through, and radiation from, panels using a wave and finite element method, The
Journal of the Acoustical Society of America 141 (2017) 2452–2460.
[35] Y. Yang, B.R. Mace, M.J. Kingan, Wave and finite element method for predicting sound transmission through finite multi-layered structures with fluid
layers, Computers & Structures 204 (2018) 20–30.
[36] Y. Waki, B. Mace, M. Brennan, Numerical issues concerning the wave and finite element method for free and forced vibrations of waveguides, Journal
of Sound and Vibration 327 (2009) 92–108.
[37] F.J. Fahy, P. Gardonio, Sound and structural vibration: radiation, transmission and response, Elsevier, 2007.
[38] B.R. Mace, E. Manconi, Modelling wave propagation in two-dimensional structures using finite element analysis, Journal of Sound and Vibration 318
(2008) 884–902.
[39] J.M. Renno, B.R. Mace, Calculating the forced response of two-dimensional homogeneous media using the wave and finite element method, Journal of
sound and vibration 330 (2011) 5913–5927.
[40] Y. Yang, B.R. Mace, M.J. Kingan, Ranking of sound transmission paths by wave and finite element analysis, Journal of Sound and Vibration, Rapid
Communications (2020).
[41] R. Szilard, Theories and applications of plate analysis: classical, numerical and engineering methods (2004).
[42] W. Zhong, F. Williams, On the direct solution of wave propagation for repetitive structures, Journal of Sound and Vibration 181 (1995) 485–501.
[43] L. Cremer, M. Heckl, Structure-Borne Sound, second edition, Springer-Verlag, Berlin, 1988.
[44] C. Hoeller, J. Mahn, D. Quirt, S. Schoenwald, B. Zeitler, NRC Research Report RR-335: Apparent Sound Insulation in Cross-Laminated Timber Buildings,”
National Reseach Council Canada, Canada, (2017).
[45] N. Jiang, S. Ma, Simplified calculation model and experimental study of latticed concrete-gypsum composite panels, Materials 8 (2015) 7199–7216.
[46] C. Petrone, G. Magliulo, G. Manfredi, Mechanical properties of plasterboards: experimental tests and statistical analysis, Journal of Materials in Civil
Engineering 28 (2016) 04016129.
[47] I. Rahmanian, Thermal and mechanical properties of gypsum boards and their influences on fire resistance of gypsum board based systems, The
University of Manchester (United Kingdom), 2011.
[48] D. Rhazi, N. Atalla, A simple method to account for size effects in the transfer matrix method, The Journal of the Acoustical Society of America 127
(2010) 30–36.

20

You might also like